PR 2
Transféré par
Kiran AdhikariPR 2
Transféré par
Kiran AdhikariUniversité Catholique de Louvain
Faculté des Sciences
Département de Physique
Bound State Description in Quantum
Electrodynamics and Chromodynamics
Binding Energy Effects on Annihilation Rates and Spectra
Christopher Smith
Dissertation présentée en vue
de l obtention du grade de
Docteur en Sciences
(Groupe des Sciences Physiques)
Louvain-la-Neuve
Mai 2002
Remerciements
Je voudrais tout d abord remercier mon promoteur, Jean Pestieau, pour toutes ces années de
soutiens et de collaborations ef caces, ainsi que pour les innombrables discussions sur la physique
des particules en général. C est bien sûr sous son impulsion que cette thèse a vu le jour, et, tout
au long du travail, son soutien fut très important pour faire face aux nombreuses dif cultés tant
techniques
que conceptuelles, et pour persévérer malgré les vicissitudes de la communication sci-
enti que.
Je remercie également les professeurs Jean-Marc Gérard et Jacques Weyers pour leur aide dans
ce long apprentissage de la théorie des champs. Les différents aspects de cette théorie qu ils m ont
fait découvrir durant mes premières années de thèse furent essentiels à l aboutissement de ce travail.
Merci à tous les lecteurs de cette thèse, qui ont accepté de faire partie du Jury, les professeurs
J.-P. Antoine, R. Gastmans, J. Govaerts et L. Sehgal.
Finalement, je remercie bien entendu toute ma famille et tous les amis, et en particulier Stéphanie
Trine pour son soutien patient, pour son aide aux premiers stades de ce travail, et sa lecture critique
du présent manuscrit.
i
Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
1 Overview of Positronium Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Theoretical Expectations (1934-1952) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 Discovery and First Measurements (1951) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Current Situation (2002) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.1 Experimental Achievements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.2 Theoretical Achievements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Theoretical Description of Positronium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.1 Pirenne-Wheeler Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.2 Convolution Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.3 Bethe-Salpeter Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2 A New Basis for QED Bound States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1 Critical Analysis of Standard Approach Decay Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1.1 Problematic Properties of the Convolution Formula . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.1.2 Dispersion Analysis of the BS Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2 Lowest Order Decay Amplitudes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.2.1 A Simple Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2.2 First Example: Parapositronium to Two Photons . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.3 Second Example: Paradimuonium and Low s Theorem . . . . . . . . . . . . . . . . . . . . . . . 26
ii Contents
2.2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.3 Orthopositronium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.3.1 Ore-Powell Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.3.2 Point-like Amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.3.3 Coulomb Form factor and Ore-Powell Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.3.4 A Simple Ansatz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.3.5 A Scalar and Pseudoscalar Resonance Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.4 Application to Other Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.4.1 Orthodimuonium and Photon Vacuum Polarization . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.4.2 Pion Decay to Orthopositronium and Radial Excitations . . . . . . . . . . . . . . . . . . . . . . 46
2.4.3 Hyper ne Splitting and Mass Renormalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.4.4 Possible Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.4.5 Higher Orders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3 Overview of Quarkonium Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.1 Quarkonium Physics Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.1.1 Heavy Quarkonium Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.1.2 Heavy Quarkonium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.1.3 Other Quarkonium Physics Observables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.2 Quarkonium Annihilation Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2.1 Decay Rate Formula and Selection Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2.2 Pseudoscalar Quarkonium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.2.3 Vector Quarkonium Decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.3 Analysis of Inclusive Charmonium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.3.1 Pseudoscalar Charmonium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Contents iii
3.3.2 Vector Charmonium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.3.3 Ratios of Wavefunctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.3.4 Conclusion: Fit to the Charmonium Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.3.5 Vector Bottomonium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4 Binding Energy Effects on Quarkonium Annihilation Rates . . . . . . . . . . . . . . . . . . 79
4.1 Binding Energy Effects on Quarkonium Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.1.1 QCD Quarkonium Wavefunctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.1.2 Strong Coupling Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.1.3 Rho-pi Puzzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.1.4 Photon Spectrum in Quarkonium Inclusive Radiative Decays . . . . . . . . . . . . . . . . . . 85
4.1.5 Transitions among Quarkonium States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Conclusions and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
A Resources for QED Bound states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
A.1 Parapositronium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
A.1.1 Details of the Dispersion Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
A.1.2 Sommerfeld Factor and Form Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
A.1.3 Radiative Corrections to Parapositronium Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
A.1.4 Dispersion Relations and Kaon Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
A.2 Orthopositronium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
A.2.1 Phase-space Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
A.2.2 Euler-Heisenberg Lagrangian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
A.2.3 The Photon Two-point Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
A.2.4 The Photon Four-point Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
iv Contents
A.2.5 The Ansatz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
B Resources for QCD Bound States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
B.1 Running of the Strong Coupling Constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
B.1.1 One-Loop Running . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
B.1.2 Running Beyond One-Loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
B.2 Standard Annihilation Rate Computations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
B.2.1 Pseudoscalar Quarkonium Decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
B.2.2 Vector Quarkonium Decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
C Useful formulae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
C.1 Wavefunction for Hydrogen-like Atoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
C.1.1 Momentum Space Wavefunctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
C.2 Loop Integrals, Gamma Functions and Related . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
C.2.1 Dirac Matrix Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
C.2.2 Feynman Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
C.2.3 Complex logarithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
C.2.4 D-Dimensional Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
C.2.5 Digamma, Polylogarithms and Riemann Functions . . . . . . . . . . . . . . . . . . . . . . . . . 141
C.2.6 Hypergeometric Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
C.3 Cross Section, Phase Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
C.3.1 Two-Body Phase Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
C.3.2 Three-Body Phase Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
C.3.3 Recursive Calculation of Phase-Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
v
Introduction
Positronium is the bound state made of an electron with a positron, in a sea of virtual particles (photons and
electron-positron pairs). The electromagnetic interaction is responsible for the binding hence positronium shares
many properties with the hydrogen atom. An important difference, however, is the possibility of annihilation
of the electron when it meets the positron, giving gamma rays. Positronium is a short-lived system. Another
speci city is the absence of strong interaction effects: the proton in the hydrogen atom is not an elementary
particle, but is a complicated aggregate of quarks, antiquarks and gluons. Positronium is purely electromagnetic
QED is suf cient, in principle, to predict all its properties to a very high accuracy. Both the weak and strong
interactions are present, but through highly suppressed corrections, and do not enter at the present level of
experimental or theoretical precision.
There are two motivations for studying the positronium system: because it is a standard test for QED, and
because it is a rst approach to the more complicated problem of the quark structure of hadrons, and con nement
in QCD.
QED Standard Tests
The Standard tests for QED, or alternatively the domains in which QED has been very successfully applied, are
of two types: elementary particle processes and low-energy bound state properties. For elementary particles, one
can distinguish:
1- High energy processes, like
,
with any lepton ( ). Such processes
typically test QED at several , at facilities like which proceed by colliding positron
with electron beams. At such an energy scale, QED must be supplemented by strong and weak interaction
effects. No disagreement has been reported.
2- The electron and muon anomalous moments and . By this is understood the deviation of the
magnetic moment from its value of as predicted by the Dirac theory. The current achievements, both
theoretically and experimentally are impressive. On the theoretical side, the computation of thousands of
multi-loop diagrams has been done. The agreement between theory and experiment is a great success of
QED. However, beyond the present level of precision, the anomalous moments can no longer be used as
pure tests for QED because strong interaction effects will dominate.
Our main concern will be bound state-type QED tests. Again, we distinguish two types:
3- Spectroscopic properties of atom-like systems. The various spacings between the energy levels of hy-
drogen have been both theoretically
!"#$
and experimentally very well studied. The most famous historically
is the Lamb shift: the spacing % &'()* of the hydrogen levels. The +ne and hyper+ne structure of
the hydrogen, helium and muonium (the hydrogen atom with a muon in place of the proton) also provide
very accurate tests.
4- Positronium spectroscopy and lifetime.
For a more detailed presentation of all those tests, we refer to [16]. The distinction made between elementary
particle and bound state tests is profound. One of the most natural frameworks of quantum +eld theory is
perturbation theory: the Feynman graphs with their associated rules. For example, to describe the process
vi Introduction
,-,. / 0-0.
, one starts with a simple lowest order approximation
to which corrections are then calculated, like for example
5 67
The theoretical computation is then ordered as a series in the 1ne-structure constant 2 3 4 4 : each coupling
;
between a photon and a charged particle line brings in a factor of 8 9 : < 2, with 8 the electric charge. Because
the constant 2 is much smaller than one, the whole computation method is under control. The only dif1 culty
is the various divergences occurring in some graphs. Renormalizability of QED ensures that a meaningful
answer can be obtained. Also, infrared divergences have been shown to exponentiate. At the end of the day, the
agreement between theory and experiment is a very stringent test of QED in its perturbative regime, and of the
associated theoretical tools.
On the other hand, a bound state is intrinsically non-perturbative. It is the permanent exchange of photons
between the constituents that binds them together
The ordering of processes in increasing orders of 2 collapses. Perturbative quantum 1eld theory is rather devoid
in front of such problems. Many standard theoretical methods explicitly assume the non-existence of bound
states.
Fortunately, there is a way out. The idea is to change the basis on which perturbation theory is built, i.e.
the lowest order, and to consider alternative expansion parameters. For all spectroscopic observables, a natural
small parameter is the ratio of the binding energy over the masses of the constituents. For the hydrogen atom, it
EFG HI JK DD LHI M E N DOPQ
is =>?@AB C D . In other words, the natural energy scale of any process inside the
S
hydrogen atom is at the atomic level, a few R . Equivalently, for such loosely bound systems, the constituent
velocities are small, T UV WW X, which is the expansion parameter used in practice. Since the system is essen-
tially non-relativistic, to Yrst order in the new expansion parameter, one recovers the non-relativistic Schrödinger
equation with a Coulomb potential. That simpler problem can be solved exactly (i.e., non-perturbatively), and
will serve as a new basis for a perturbation theory. Graphically, the lowest order is the so-called ladder approx-
vii
imation (technically, crossed ladders are neglected, and only Coulomb photon exchanges are considered in the
non-relativistic limit)
Both radiative and relativistic corrections to this Zrst approximation are then computed, in increasing order of
the small parameters ([ and \]^ say). These are the very successful techniques used for the spectroscopic tests.1
Positronium is also a bound state, and the same techniques have been used. The small binding energy and
^
\ ] are the expansion parameters: the system is treated non-relativistically to lowest order, and radiative and
relativistic corrections are then computed perturbatively. All this would be Zne, except for one point. The
positronium annihilates into photons. That process is intrinsically relativistic. The Schrödinger equation cannot
`
describe the disappearance of positronium into photons. Further, the atomic energy scale of a few _ is no
longer natural, because of processes like
where the photons carry energies of the order of the electron mass. The photons can serve as probes, giving
information on the internal structure of the bound state. When they are hard, they essentially see a slowly moving
electron-positron pair, while when they are soft, they are sensitive to the non-perturbative binding of the pair. In
this respect, positronium is a unique system. It requires a full understanding of bound states in the relativistic
quantum Zeld theory framework. Most theoretical approaches rely on the same type of approximations (mainly a
factorization of the bound state wavefunction and the annihilation mechanism). The subject of the present thesis
is to test the theoretical consistency of those models.
a
QCD and Con nement
Quarks are subject to the strong interaction. There are three types of strong charges, called colors. Quantum
chromodynamics is the theory describing the interactions between quarks, antiquark and gluons. This theory
is similar to QED, except that gluons do carry a strong chargef and therefore can interact among themselves.
e
Technically, QCD is a gauge theory with the non-abelian bc d as gauge group. The eight gluons belong to the
adjoint representation, while the three colored quarks belong to the fundamental representation.
g qrsss
For the hydrogen atom physics, yet another small expansion parameter has proven to be useful: hijklmn o p . This parameter
is important in the treatment of the proton recoil effects.
viii Introduction
In principle, the consequences of the gluon charge lead to a small increase of complexity: compared to QED,
one only has to consider new diagrams like
Unfortunately, the main characteristic of the strong interaction is, evidently, of being strong. In other words, the
coupling constant of QCD is much greater than one at ordinary energies, and perturbation theory breaks down:
the two diagrams shown above cannot be ordered in decreasing magnitude.
This picture of the strong interaction is supported by two main facts:
1- There is another consequence of the gluon charge. As the energy increases, the strong charge decreases.
t
At suf ciently high energy, above a few GeV, QCD is perturbative, and can be directly tested. Various
processes at high energy are consistently described within the context of QCD, and this is a great success.
The measured coupling constant, when extrapolated to low-energy, is indeed greater than unity.
2- Quarks and gluons have never been observed as free particles. Instead, one always sees aggregates of
t
quarks and gluons, the baryons (like protons, pions,...). This phenomenon is called con nement. This is
understood as follows. If the interaction is very strong, a colorful particle cannot exist freely, its stable
xy
state will always be inside a colorless aggregate. For a group like uv w , the most trivial aggregates are
t
the quark-antiquark and the three-quark bound states. This nicely ts the observed spectrum of particles.
It should be emphasized that no rigorous proof that the stable states of QCD are indeed the observed hadrons
exists. Nobody has ever been able to compute the mass of the pion{ or the proton in terms of the quark masses
}
and QCD coupling constant. Further, low-energy reactions (below z | ) among hadrons are not quantitatively
described in terms of the underlying quark processes. This is a serious limitation. By comparison, the most
stringent tests on QED do involve low-energy bound states. For QCD, quantitative tests are only possible at high
energy. The most promising tests of QCD at low energy seem to be lattice calculations, but they are still very far
from the precision achieved in QED.
t
The study of bound states in Quantum Field Theory is already a very dif cult non-perturbative problem.
t
Con nement adds to this complexity, and most methods used in QED are just no longer applicable. There is one
exception: heavy quarkonia.
{ }
Such states are made of a heavy quark ~ or with its corresponding antiquark. Being
t
very heavy (of a few | ), one can hope that a rst order non-relativistic approximation is reliable. Of course,
the quarkonium wavefunctions are not known, since non-relativistically, QCD cannot just be a Coulomb-like
t
interaction (such a potential does not lead to con nement). By studying quarkonium states, one gets information
ix
on the QCD potential. Quarkonium is therefore a crucial intermediate step to understand low-energy QCD.
The physics of quarkonium is very similar to that of positronium. The spectroscopy is analogous, while
quarkonium decays through the annihilation of its constituents, like positronium. The techniques used for
quarkonium are inspired from the positronium ones. In particular, the same factorization hypothesis is used
in both problems to disentangle the perturbative and non-perturbative parts of the dynamics. The applicability
of the positronium models to quarkonium will be analyzed in the present thesis.
Brief Outline of the Thesis
The goal is to address the relevance and consistency of the positronium models. We will show that a basic re-
quirement of Quantum Field Theory is violated in the current computations of positronium decay rates. The
probability
amplitudes for positronium decay are not analytical. This is rather technical, but far reaching a mod-
i cation of the model is unavoidable. In fact, we will see that the perturbation method designed for spectroscopic
studies cannot be used for decay processes. The binding energy expansion is in contradiction with quantum eld
theory analyticity requirements. A natural solution is then to construct a new basis for an alternative perturba-
tion theory. This has to be done non-perturbatively in the binding energy. There is an additional motivation for
this. In quarkonium physics, typical binding energies are much greater. Being able to treat them exactly will
unavoidably lead to interesting advances.
The thesis is organized in four chapters, with the structure
Chapter 1 Positronium General Introduction and Standard Approaches
Chapter 2 Original Contributions
Chapter 3 Quarkonium General Introduction and Standard Approaches
Chapter 4 Original Contributions
The original discussions in chapter 2 and 4 have been made as self-contained as possible. Chapter 1 and 3 serve
to set the stage, and introduce the standard material needed for our presentation.
Chapter 1 begins with a brief historical overview of positronium physics. Then, the present status of both
experiment and theory is shortly reviewed. The nal section describes the evolution of the theoretical models
designed to describe the annihilation of positronium. By models is meant a scheme of computation: the de nition
of a lowest order approximation, exactly calculable, and of the perturbation theory. All those models introduce
x Introduction
some kind of factorization of the bound state dynamics from the decay process
For instance, the simplest of all formulas is
¡¡
¢£¤¥¦§ (1)
with ¨ the Schrödinger wavefunction encoding the non-perturbative bound state dynamics, and © ª«¬« ® ¯°±
is the scattering cross section representing the perturbative decay process. More advanced techniques will be
described, and their factorization ²
hypothesis emphasized. ²
Chapter 2 begins with a clari cation of the physics of the approximation done in de ning the lowest order
amplitude in standard approaches. In doing so, it will be seen immediately that this basis is in contradiction with
analyticity. By factorizing the process as pictured above, the photons are always emitted from charged particles.
If the photon is very soft, however, it should behave as emitted from a neutral boson (i.e. the positronium)
instead. It cannot discern the charged content of the bound state if its wavelength is much larger than the typical
³´µ¶
separation. The binding of the electron and positron into a neutral boson becomes very important for a
soft photon, or in other words, when the energy of the photon is of the order of the binding energy.
The goal is therefore to construct amplitudes so that the photons are always aware of the binding of the
charged particles. We propose such an alternative construction for the lowest order amplitude, with no factoriza-
xi
tion. Graphically, it is
The decay amplitudes are constructed as closed loops. The bound state dynamics is introduced via a form factor
which is related to the Schrödinger wavefunction, hence contains the effects of Coulomb photon exchanges.
Technically, by closing the loop (i.e. without factorization), the resummation of the Coulomb photon is no
longer independent of the decay process. As a result, the ·nal state photons do ¸feel¸ the binding, and behave as
if emitted from a true neutral bound state.
For this construction to be a well-de·ned lowest order amplitude, it has to be calculable. To this end, we
found a very simple procedure, in which our new lowest order amplitudes are obtained from standard QED ones.
We then show that indeed, our approach correctly describes soft photons. Many other interesting, but more
technical, properties of our approach will also be discussed.
The chapter ends with a systematic study of various decay processes. The implication for spectroscopic
studies is also sketched. Finally, some possible extensions are reviewed.
Chapter 3 is a very short introduction to the physics of the ¹¹ (charmonium) and the ºº (bottomonium)
heavy quark bound states. It begins with the spectroscopy of charmonium and bottomonium as established
experimentally. Then, the description of decay processes is presented. Most studies rely on the extension of
the
¾ positronium decay formula (1) to the quarkonium case. The predictions for the decay rates of »¼½ and
¼ quarkonium states are systematically studied. Comparison with experimental branchings, along with the
»
extraction of ¿À , closes the chapter.
The last chapter presents the many problems and puzzles of quarkonium physics. It is shown that binding
energy effects can explain most of them. A de·nite quantitative study has not been performed though, because
of the lack of information on the QCD potential. However, the results obtained are qualitatively independent of
the detailed form of it.
Three appendices supplement the text with calculation details, proofs and numerical studies. The ·rst ap-
pendix contains additional information and detailed studies relevant for the chapter 2, while the second appendix
is related to chapter 3 and 4. The ·nal appendix contains useful formulas like special functions, dimensional
regularization formulas, phase-space analysis,...
Finally, the thesis ends with a bibliography. The references are ordered in sections (with no direct relation
to the sections of the text), and in each section, in chronological order. This bibliography is intended to be more
or less complete for positronium (but some omissions are unavoidable). For quarkonium, a complete reference
list would be too long, and only the papers directly related to our studies are quoted.
Chapter 1
Overview of Positronium Physics
This chapter presents the basic observables of positronium physics. This chapter sets the stage for more
elaborated studies. For this presentation, we have chosen a more or less historical Áow.
The Ârst section begins with the exposition of the theoretical status at the end of the forties, just before
the experimental evidence. Then, the Ârst confrontation of QED theory in positronium physics is presented.
In the second section, the most recent theoretical and experimental results are given and compared for both
spectroscopic and lifetime observables.
In the Ânal section, the evolution of the theoretical description of positronium decay is reviewed. The
hypotheses done are described. Their relevance will be discussed in the next chapter.
1.1 History
ÄÅÆ
Historically, it was Ârst postulated in à by Mohorovicic [59] that an electron and its associated antimatter
particle could form a quasi-stable atom, one year after the experimental discovery of the positron by Anderson.
ÄÆÇ
That state was Ârst called positronium by Ruark in à [33], who was unsuccessfully trying to observe it.
É
The properties of positronium, both spectroscopic and lifetime, were studied theoretically for about È years.
ÄÇ
The Ârst experimental observation Ânally came in à Ã, in agreement with the expectations.
1.1.1 Theoretical Expectations (1934-1952)
The Ârst approach to the positronium bound state is non-relativistic. The quantum mechanical treatment of it is
exactly the same as that of the hydrogen atom, except for a change of reduced mass. For instance, the energy
levels
2 Chapter 1 Overview of Positronium Physics
Ê
and corresponding wavefunctions can be speci ed by the same set of quantum numbers ËÌ ÍÌ ÎÏ Ì with Í Ð
Ñ Ø Ù ÚÚÚØ Ø
Ì ÒÒÒÌË Ó Ô and ÕÖ × Ó . The energy levels exhibit the well-known degeneracy in and ÕÖ :
ä
Û Ö
ÜÝ ÝÞß à áâã
åæä
å ñòó ô õ ò÷ ø
where ç è éê , é the electron mass. The ground state energy is therefore ëìíìíì è îïð î
ö
ùúû
î eV (half that of the hydrogen). ýþ
To the above quantum numbers, one has to add that of the spin. Two particles of spin ü can give a total
spin angular momentum of ÿ or ü. The spin four-fold degeneracy of each of the levels drawn on the gure is
split into a singlet spin ÿ state (parapositronium, - ) and a triplet spin ü state (orthopositronium, - )2
ü
Singlet þ
Triplet
The spin then combines with the orbital angular momentum to give the nal total angular momentum. For
example, using the spectroscopic notation with !! the twelve " states arrange as
&
'
' Parapositronium *+,- 3 states
( Orthopositronium ./,0 1 state
" #12 states $ %
'
' ./,- 3 states
) ./,1 5 states
Graphically,
2
Fine and Hyper ne Structure
At the lowest order, there is a high degeneracy. This does not survive at higher orders. Pirenne (1947) [61]
and Berestetski (1949) [62] were the 3 rst to compute the 3ne and hyper3ne structure of the positronium energy
levels. Their results were later corrected by Ferrell (1951) ([64], see also [26]), who found
? [\]^^
4567689: ; <=> DEFGHII O GFGHII _` \a^ h
@A? B C JKL KGMI with N JKL KGMI P R SSW X YZbc e
QTUV d fg
i j
For the hydrogen atom, this classi cation is not relevant because of the very small proton magnetic moment kl m pq
nor ss tu v
ywx
z{ . For positronium, the magnetic moment is the same (in magnitude) for both constituents.
1.1 History 3
where
¦ ª«¬
§ ®«¬¯°®±«¬ª° ²³ ´ µ ¶ ·
¨
|} ~
¡ ¢£¤ ¥ ¸ º»º¼½¾
¹ ¿À Á Â (1.1)
§
© Ç
à ÄÅÆ
É ÇÊ ËÌ Í Î Ï Ð
ÅÈ ÅÆ
Hence, we can modify our spectral representation
with the transition frequencies (in GHz),
ÒÓÔÕÖ× Ø
a Ñ ßàá â ãä Hyper ne Splitting
b ÙÚ c Û d Û e Û fÜ Þ
Ýçèé Ý Lyman å
g æéèç
h æ
çèêë
i Lamb Shift
We have also indicated the commonly used names for the most remarkable transitions. Physically, one has the
following selection rule for the dominant radiative transitions:
ïð ñ òó
í ïô ñ òó
M1 ÿ
E1 ì ï õö
î ÷ø ù úûüý þþ
where E1 are the electric dipole (or allowed) transitions (plain lines on the picture), by far the strongest transi-
tions. The forbidden transitions are the M1 magnetic dipole transitions (dashed lines).
The most interesting splitting is the hyper ne splitting of positronium, because it is the most
accessible both theoretically (1S states are relatively simple to treat) and experimentally (the higher excited states
are predicted to cascade decay into lower states quite rapidly). In (1.1), the state,
, appears separately
the spin-spin and annihilation potentials between and
because of the so-called annihilation graph. The hyper ne splitting is computed from the expectation values of
states
4 Chapter 1 Overview of Positronium Physics
(the other hyper ne operators contribute equally to both states, and are included in the exchange graph above).
The corresponding contributions to the splitting are [26]
!"#$ %&' ()()(/*+,-
Ortho '
Para
,-
Fine Struct. (Spin-Spin)
Annihilation . '
(1+ ,-'
(1.2)
0
Hyper ne Splitting
0
0
Karplus and Klein (1952) [67], through their now standard computation, obtained the rst radiative correc-
tion to the hyper ne splitting:
2345 6 789 :<=; > ? :AB C DE== FF G HIJKJLM GHz
@
Many additional graphs contribute to that order, and we shall not detail their techniques here.
Annihilation Rates
selection rules (Yang,
MNOI
[202] and Wolfenstein and Ravenhall,
MNOH
The positronium is a short-lived system. It decays into photons. The number of which can be deduced from
[66], see also [9]).
P Q
A positronium state is an eigenstate of the parity , charge conjugation and of the combination 3
. The QP
eigenvalues of those operators can be obtained in terms of those of as RST
U V WXYZ[\]
^ _ `abcdef
gh i jklmnop
These expressions are obtained from the intrinsic parities of the constituents and the antisymmetry requirement
klqrlm
on the exchange of the constituents. Parapositronium (orthopositronium) is a CP eigenstate with eigenvalue
s
. Since a system of photons has a charge conjugation eigenvalue of
t usvw x uyzw{
we obtain the selection rule
uyzw|}~
We therefore write, up to
(the principal quantum number can take any value)
Number of Dominant Annihilation mode
¡
¢££ even §¨©ª
¤¥ ¦ « ¬¬¬
® °±
¯ odd odd
²³ µ
´¿ÀÁ²µ¶ · ¸¸¸
¹º» ¼½½ º¾ ÂÃ Ä ÅÅ
ÀÁÆ ÇÈÈ even ÉÊ Ì
ÑÒÓ ÔÕÕ even ÒÖ Ë×ØÙÚÛ
ÉÍÎ Ï
Ü ÐÐÐÐ
ÝÝ
ÞßÚ àáâ eveneven
ãä åñòóô õ ëë
æçèé ê
æçè ìíí odd îï ðýþ ööö
ñò÷ øùù odd úû ü ÿ
This property, due to the special nature of positronium as a bound state made of an electron with a positron, has raised some interest in
positronium as a test of the discrete symmetries.
1.1 History 5
The only mode that needs comment is
. The two-photon mode is forbidden by the (Landau-
Pomeranchuk-)Yang theorem (1950) [202]: a vector particle cannot decay into two photons, because of
gauge invariance and special relativity. Most positronium state annihilation modes will be suppressed compared
to their electric dipole transitions to lower states. Only the ground states and are dominantly decaying
into photons, because the magnetic dipole transition is highly suppressed.
In 1946, Pirenne [61] and Wheeler [60] separately developed a theoretical model to compute the decay rates
of positronium. Their model is
!"# $ %&'()*+ %,-./ 0 1233
456789
? CD
with :;<= the relative velocity of the > >@ pair in their center-of-mass frame, and A B the positronium Schrödinger
E
wavefunction, in con guration space, at zero separation. One can understand this model as a replacement of the
scattering cross section initial Fux factor by the probability of contact in the bound state. Obviously, this model
CD C
applies to G-state annihilations only, since H B is zero for I K J . MN
Pirenne and Wheeler were able to apply their formula to parapositronium L O P, while, in 1949, Ore and
QN
Powell [63] used the same model to compute the orthopositronium L MP decay rate. The results were
\]
R ST bc^d e fgh ijklm
- U V W XXY Z [^ _ `a
{|} ~
n op
-q r s tttu v wxy z
The orthopositronium lifetime is roughly1000 times longer than the parapositronium one. At rst sight, one
would have expected a relative factor of . Phase-space is responsible for an additional factor of .
1.1.2 Discovery and First Measurements (1951)
The discovery came in 1951.
Martin Deutsch found evidence of orthopositronium formation in gases [34]. His
technique is based on the very different lifetime between both positronium species, and the paramag-
netic properties of the gas. This molecule has an isolated electron, hence is a catalyst for - -
by electron exchange. When positrons are slowed in a gas, both orthopositronium and parapositronium
are produced. The lifetime of the parapositronium is too short for its decay photons to be observed. Adjunc-
tion of a small amount gas therefore changes the observed gamma ray spectrum. This is an evidence for
orthopositronium formation.
6 Chapter 1 Overview of Positronium Physics
By varying the gas pressures, the lifetime of orthopositronium at zero pressure can be extrapolated. The
result was ¡¢
£¤ ¥¦ ¥«¬ ® ¯°±² ³ ´µ¶·¸
- § ¨© ª
This agrees with the result of Ore-Powell4 .
¹ »¼½
The hyper ne splitting was measured in º , still by M. Deutsch and his colleagues [34]. By using mag-
¹
netic elds (Zeeman effects) and radio frequencies (rf quenching techniques), they found
¾¿ÀÁ ÇÈÉÊÇ Ë ÈÊÉÌ
ÃÄ Å Æ GHz
Í
In good agreement with the theoretical expectation. Note that the precision is suf cient to test the presence of
Í
the annihilation potential contribution to the hyper ne splitting.
Í ÏÐÑ
The rst measurement of the parapositronium lifetime was performed in Î by Hughes [36], with the
result ÒÓ
ÔÕ Ö× ÖÑÜÝÝ Þ ßÜààá â ßàã äåæçè
- Ø ÙÚ Û
in agreement with the Pirenne-Wheeler result.
Later, many transitions were observed, in good agreement with the theory. For example, in 1975 [40], the
Lamb shift was measured at é
êëîìí ôõö÷ô ø ùõùùöú
ïðñ ò ó GHz
ûüýþ
ÿ GHz
where the rst order radiative corrections account for MHz to
.
The agreement between experiment and theory is excellent. Even if the accuracy is not formidable, in view
of the theoretical complexity, and of the many different types of effects already included, this was a great success
for QED.
1.2 Current Situation (2002)
A peculiarity of positronium physics is the recurrent occurrence of puzzles: at different stages, discrepancies
occurred between theory and experiment. Ultimately, those disappeared through corrections to the computations,
or more careful measurements. The recollection of the historical developments can be found in many places and
we do not intend to present it here [85], [87]. Instead, we directly state the most recent achievements, but it
should be clear that many groups were involved, especially on the theoretical side, for the computation and
recomputation of the many loop diagrams necessary for high order radiative corrections.
1.2.1 Experimental Achievements
The latest experimental results for the decay rate of the parapositronium are
!"" #$$% & $'( )*+,-
Yale (Gas), 1967 [38]
./001 2334 5 678 9:;<=
- Ann Arbor (Gas), 1982 [42] (1.3)
>?@@A@ BCDE F CAG HIJKL
Ann Arbor (Gas), 1994 [56]
M OP
Q By that time, the Ore-Powell prediction was not the only one on the market. The N accuracy of the Deutsch measurement was
suf cient to discriminate among the various theoretical computations.
1.2 Current Situation (2002) 7
SSS
The orthopositronium lifetime is experimentally simpler to measure, being R times larger than the para-
positronium one. Results of various groups are
\ `abcdb efbg hijklm
]
] nopqrp snpt London (Gas), 1978 [41]
]
] uvwxlm
^ Mainz (Vacuum), 1987 [46]
TUVW nopqyz s{|} ~
X-Y Z [ Tokyo (
Powder), 1995 [57] (1.4)
]
]
]
] Ann Arbor (Gas), 1989 [49]
_ ¡¢£¤
Ann Arbor (Vacuum), 1990 [50]
The most recent Ann Arbor and Tokyo results are mutually exclusive. This is known as the orthopositronium
lifetime puzzle 5 . Note also that the medium used varies. It is instructive to represent the evolution of the
experimental measurements over the last three decades. References to the original experimental works can be
found in [101].
§¨ ¶·¸
Year ¥¦©-ª « Technique Year µ©- ª « Technique
® °¯®
1951 6.8 ¬ gas 1978 7.050 ¬ vacuum
¯® °´®
1968 7.29 ¬ gas 1978 7.122 ¬ vacuum
°±® ±®
1973 7.262 ¬ gas 1982 7.051 ¬ gas
°±® ®
1973 7.275 ¬ gas 1987 7.031 ¬ vacuum
²® °¯®
1976 7.104 ¬ SiO³ 1987 7.0516 ¬ gas
´® °¹®
1976 7.09 ¬ vacuum 1989 7.0514 ¬ gas
® °²®
1978 7.056 ¬ gas 1990 7.0482 ¬ vacuum
²® ´º®
1978 7.045 ¬ gas 1995 7.0398 ¬ SiO³
Graphically, the decrease in the measured rate is obvious (theoretical predictions are discussed in the next sec-
tion)
Gas experiments are performed at various pressures, and then the result is extrapolated to vacuum, assuming
a linear dependence. In principle, the closer to zero pressure one can get, the more precise is the result. In
»
practice, a compromise has to be found because the ef ciency of positronium formation quickly decreases with
the pressure. Typically, at ¼ atmosphere, as many as 25-50% of the incoming positrons bind into a positronium.
½
In fact, it should be called the second orthopositronium lifetime puzzle. In the mid- and late seventies, a similar situation occurred. It
was later resolved in favor of the smallest lifetime measurement.
8 Chapter 1 Overview of Positronium Physics
The Tokyo experiment uses ¾¿ÀÁ powder. For grained materials, the same strategy as for gases could in
principle be used  namely the measurement at various densities, and then extrapolation to vacuum. However,
there is concern that the extrapolation to zero density may not be linear for powder, because positronium is
formed inside the powder grains. Varying the density only accounts for the effects of the atomic collisions once
the positronium has escaped the grain. Therefore, different data analysis techniques have to be designed. The
Tokyo method does not rely on any sort of zero density extrapolation.
In conclusion, there seem to be arguments for the Tokyo result, while the consistency of both the gas and
vacuum measurements of Ann Arbor is in their favor. The situation is unclear and further measurements are
necessary.
Ã
Concerning hyper ne splitting measurements, the most precise date back to the late seventies- early eighties.
It was found, including the other measured transitions,
Name Splitting
ÎÏÐ Ñ ÒÓÔ ExperimentÒ ÄGHz
ß
Å Reference
ÆÇÈÉÊ Õ Ö×ØÙØÚÛÜ Ý Þ
ËÌ Í Brandeis, 1974 [39]
Ö×ØÙØÚàáâ ãäåæ Yale, 1984 [43]
çèéêë òóô õ öòóô ö÷øøùúûü÷ öýþ ÿ öúû
ìíîïð ñ Stanford, 1989 [48]
ö÷øøùúûü÷ öù
Stanford&Bell, 1993 [52]
Ann Arbor, 1987 [45]
!!"# $%&'( $)''( Mainz, 1993 [53]
*+,-. 56 9:; <=>?? <= @=AB @AB
/0123 4 3 78 Ann Arbor, 1987 [45]
<=>? < C @DEB @<FGH
8 8 Mainz, 1993 [53]
IJKLM STU V WXY W
NOPQ R Z [\]^_] ` ab `_b Ann Arbor, 1987 [45]
]W c `W b
[\ [ [ Brandeis, 1975 [40]
W g hifj g
wz [\] cd[ `ef Mainz, 1993 [53]
klmno tvw x y w {{ {}{ ~{
pqrst u | 1993 [54]
{{ {} ~ ~
| 1994 [55]
Other experimental results include the study ofwthe Zeeman and Stark effects, tests of discrete symmetries
C, P, T and CP, CPT (by setting upper bounds on
or , by the analysis of angular
correlations in ,...) [44], [51], search for new light particles (axions, SUSY,...), search for the mirror
universe, search for positronium molecules, etc. Those studies lie somewhat out of our main concerns, and will
not be discussed here.
1.2.2 Theoretical Achievements
The parapositronium and orthopositronium decay widths are
§¨ ®
Â
¬ ¶ · ¿ÀÁ
-
© ± ²³ ²¼
¡¢ £ ¤ ª « § ¯° ´ µ ´
ÆÇÆÈÉÊ ËÌÍÎ Ï ÌÊÐÑ Õ ¥ ¦ ² ¸ ¹º» ¼ ¸ ½ ¾
à ÄÅ ÒÓÔ
æ ï
Þß à áâã ä å ôø
Ö×Ø é äê ò ó
Ù- Ú Û åç è ë ì
í î ìð ñ þ
ô õ ö÷ ù ø ú ûüý
ÿ
Ü Ý
!" # $%&'( ü
)
with coef cients *
2 N E LFGHPQQFQ KLFM
+ , - . /01 O
3 9 :;<= R E JJGST UTVW
4 5 678 O
> @ A A: = XY Z S[S\] U\W
4 5 ? 7 8 8
BCD E FGHIJ KLM ^_` a bcde fdg
1.2 Current Situation (2002) 9
The hij and hkj are the contributions from the four photon and lve photon lnal states, respectively.
pq
The para-
positronium result agrees with the experiment, but the precision is not suflcient to test m no radiative correc-
tions. For orthopositronium, the theoretical
vw
value tends to favor the Tokyo group result, but again, experimental
r
precision is not suf cient to test the s tu radiative corrections (see picture in the previous section).
x
The hyper ne splitting is computed to the same order, but with an analytic result
yz {
| } ~
¡ £¤ ® ® µ
±² ³ ´ µ¸
¢
¥¦§¨
©ª « ¬ ®¦¦ ©ª « ¯ ° µ² ¶ · ¯
» È Î × ØÙ Ú
¹º ÛÜÝ Þ
» ½¾ À
¿ ÁÂ ÃÄÂ Å Æ Ç É¿ ÃÄ Ê Ë ÌÍ
ÏÐ Ñ ÒÓ ÔÕ Ö
¼
â ãäåæçèè é äåèêèè ð
ß àá ë é äåìäçèëí îï ë é äåèêìçëí é ååå
û
ñ òóôõôö÷øö ùú ÷
GHz
õø ôõü
This result exceeds the experimental values of [43] and [39] by ò and experimental standard deviations,
respectively. The discrepancy is not too serious, but new measurements would really be welcomed.
Concerning references, most papers cited in the ýPositronium Theoryý section of the bibliography concern
þ
the computation of a part of the decay rate or hyper ne result. The latest results are presented in [124], [130]
and [131].
Other theoretical predictions have been obtained. In summary (see [132] and references cited there)
Name Quantity Theory
ÿ
GHz
"#$ $%& '%(
!
)*+ 123 4 5678 59::;<=>9999GHz ?;@
,-./ 0 GHz
AB 9678 4 96CD EFGHIFJK LIM
GHz
NO JPQR S TUVW XYZ[ XT\ X ]^_
`ab ghi j klm k GHz
cde f{|} ~ }n} opq rst uvw GHz
xy z z
GHz
¡ ¢£¤¥¦
§¨- ¬® ¯ °
± ²³± ´µ¶°··¸³ ¹º¶» ¼ ½¾¿ÀÁ
-© ª «
Some Peculiarities
The positronium perturbative series have two peculiarities: corrections in ÂÃ Ä appear, and the overall conver-
Å
gence is rather slow. In this subsection, we intend to explain qualitatively why, leaving more re ned quantitative
arguments to the next section.
The slow convergence feature is traced to the contribution of so-called binding graphs
ÇÈ ÇÈ
which account for roughly Æ of ÉÊ [69] and roughly Ë of ÉÌ [75], respectively. These diagrams have many
interesting features, and we will discuss about them later. For now, one can understand that such diagrams are
important because the exchanged photon is a binding photon, i.e. it is responsible for the binding of the electron
10 Chapter 1 Overview of Positronium Physics
and positron. For such photons, the naive counting in powers of Í fails. More precisely, the dynamics of the
above process is such that, for ÎÏÎÐ near the mass shell, there is a huge enhancement factor, compensating the
additional coupling constant.
Logarithmic corrections arise typically from Ñwould-beÑ infrared singularities, i.e. singularities occurring if
the electron and positron are on-shell. For example, the following diagrams introduce logarithmic corrections
[74]
In the Òrst case, one can see that logarithmic IR divergences will be introduced if the constituents are on-shell
because of the intermediate electron and photon propagators.
ÛÜÝ At low momentum, however, such divergences are
cut-off by the binding energy ÓÔ Õ Ö × ØÙ Õ ×ÙÚ . Hence, symbolically, one can understand that Þß à
corrections can arise from integrations like
á âã è
è é ê ëì íî
ï ðë
ìñ
âäåæ ç
That logarithmic correction is a soft-scale effect, i.e. arising from very soft momentum integration. ò ó
Similarly, the two-loop photon vacuum polarization insertion in the second graph introduces a ñ ô õ cor-
rection [70]. To see it, one just has to know that the fourth-order vacuum polarization has threshold singularities
behaving like û
õò óô ö÷ ø ùúû ý
ö ü ý
ö ý
ö ý
where þÿ is the momentum transfer owing through the diagram. In the case depicted on the gure, that mo-
mentum is þÿ hence
It is important to note that such logarithmic corrections are really speci c to bound state observables for
all of them, logarithmic corrections like or occur. They are intimately linked to the binding energy
(mass defect). The current approach to these corrections is however quite different in that it uses the language of
anomalous dimensions and renormalization group ow [128].
1.3 Theoretical Description of Positronium Decay
Now we turn to the theoretical description of bound state decays. It is not immediately obvious how to extend
the quantum mechanical formalisms or the Bethe-Salpeter one those were suited to the computation of energy
levels or wavefunctions. To describe a decay process, some kind of model has to be constructed. Usually, such
model will involve the bound state wavefunction in an essential way, so that all the previously cited formalisms
will still be important.
A good model (or scheme) is a model allowing for a perturbative expansion. The lowest order approximation
should be well-de ned, while corrections should be obtained in a systematic way. These two points are not so
easy to meet. Indeed, as we will see, issues of double counting plague decay computations. Further, there are
1.3 Theoretical Description of Positronium Decay 11
!
many dif culties associated with gauge invariance and infrared divergences. Finally, the relevance of the lowest
order approximation is often problematic, as will be exposed at length in the next chapter.
1.3.1 Pirenne-Wheeler Formula
The simplest of all decay formula has already been encountered " it is the Pirenne-Wheeler formula ([60], [61])
# $% & ' ()* + , 2 345 6 789:;<= 7>?@A B CDEE
-. / 0 1 1 FGHIJK (1.5)
M NOP Q
with L L the Schrödinger wavefunction. This formula relates the decay rate to the scattering
\]^ _ of the con-
stituentsc at
\ d rest. The limit
f RSTU V W enforces the correct selection rule, i.e. X - Y Z V [ and `-Y Z V
[ ] a b , because an e eg pair at rest can decay into two (three) photons only if in a total spin h ( b) state.
The results for the parapositronium and orthopositronium decay rates have already been discussed, so we will
be very brief and concentrate on i -j k l mm. n o pq
The lowest order result is obtained from ( r with p the center-of-mass energy)
³
s tuvwx y zz{|}~}
¡¢ £¤¥ ¦ §¨ °®¬ ´
© ®¯ª«
« ±²
» ¼»½ ¾ ¿
With µ¶·¸ ¹ º , we Ànd
Î ÒÑÔ
Á ÂÃ Ë ÌÍ Î ÏÐÑ
-Ä Å Æ ÇÇÈ É Ê ÈÊ ÒÎ Ó Õ
Radiative Corrections
The computation of the radiative corrections to Ö - × Ø Ù ÚÚ was achieved by Harris and Brown [65], [69],
[99], using the above formalism. The diagrams contributing, up to one-loop, to the scattering cross section are
12 Chapter 1 Overview of Positronium Physics
with in addition the crossed processes. The Ûrst one (Ü ) is the tree-level. Details of the calculations are in the
appendix A.1.3, and we quote
ì
ç èéêëì íîï ò ó ô ÷ ó õø
Ý Þß
-à á â ããä å æ ð ì ñ õ ö ù ú û þÿ
üý (1.6)
The mass and coupling constant are renormalized by the and
diagram, respectively.
Note that some IR divergences proportional to ÿ have been cancelled by the limit ÿ
[99]. This
cancellation is a selection rule cancellation: IR divergences in
the ! ( rst order) cancel against
" ###$
the
%& IR bremsstrahlung divergences of . Since
%* +,-,. / 001 this last decay cannot contribute for an initial
3
' state, the corresponding IR divergences of () disappear when the initial 2 24 pair is in a
56
7 state.
The singularity for zero relative velocity is also an IR divergence, coming from the diagram 8 , the binding
graph. The treatment of this divergence is special. In fact, it has to be dropped because it is already included :
in
the wavefunction. Indeed, the wavefunction accounts for all the Coulomb photon exchanges between the 9 94.
<
The ; =>?@ comes from the Coulomb part of the photon in A.
In a somewhat different reasoning, following Harris and Brown [69] (see also Schwinger [7], [Link], pp99
and Merzbacher [3], pp249), one can C
remember from non-relativistic quantum mechanics that the long range
E
Coulomb interaction between the B BD modi es wavefunctions as
U
FG HIJ FK L MN OPQ MK RST VWXY h ijk l mn o pq
Z [ \]^_`abcde f g g rstu v
y
where w is the initial x xz wavefunction entering the scattering process (not to be confused with the bound state
wavefunction). This correction factor is called the Sommerfeld factor. As a result, the Pirenne formula should
be written {|
} ~
¡¢£¤
where, to order ¥, ¦§¨©ª «
¬¬® ¯ °±² ³´ µ ¶· ÀÁÀ à ÄÅÆ
¸¹º» ¼ ½ ¾ ¿
The Çnal form for the correction is now simply
Ó ÙÝ à
È ÉÊ Ö× Ø ÛÜ
- Ë Ì Í ÎÎÏ Ð ÑÒ
Ô Õ Ù Ú Þ ß (1.7)
1.3.2 Convolution Formula
To avoid the kind of pathological velocity singularity encountered above, a more sophisticated model has to be
built. To
ä
express the amplitude for the bound state decaying into a given ánal state, in terms of the amplitude for
its âãâ constituents to scatter into that ånal state, the årst step is to relate the positronium state vector to the
constituent state vectors. This is done through a convolution integral (see for example [17], [24]):
æç è éêëì í îïð ñ òóô õûø ü
õö÷øù ú ýþÿ ýþÿ
(1.8)
The state vector of the bound state is the sum of the combined state vectors of the constituents weighted by the
momentum space Schrödinger wavefunction for the bound state. For the positronium ground state,
! " # $%&
' ( & ' '
1.3 Theoretical Description of Positronium Decay 13
+, - 6
with ) * the con guration space wavefunction at zero separation and ./ 0 1/ 2 3 45 (7 the electrom mass
9:; =>
and 3 the positronium mass). The 8 and < are just normalization factors for free states
?@A B CDE H KL
FGI J
and M NM O stand for polarization states, which combine to give a total spin P to the bound state.
This formula exhibits two important features : Qrst, the constituents are always on-shell. We are just aver-
aging over residual velocity, with weights given by their respective probability to occur inside the positronium,
TU
i.e. by R S . On-shellness of the VW is also obvious from their state normalization. Second, non-perturbative
TU
effects are introduced non-relativistically, through the function X Y .Z
From the expression of the state vector of the bound state, one de nes the amplitude for its decay as
[ \] ^ \_` a bc` d efg h ijk l o r z{ |}~ ~
lmnop q k stuv stuv
w xy (1.9)
Now the amplitude inside the convolution integral is a scattering amplitude. It can be treated using quantum eld
perturbation theory. Finally, the width is calculated as
¨ ©ª « ©¬ ® ¯° ±
¡ ¢ £¤¥¦ § § (1.10)
³
with ² ´µ the symmetry factor for the ¶nal state of ´ identical photons.
In the static limit, i.e. when · ¸ ¹, the convolution formula is equivalent to the Pirenne formula (1.5).
Indeed, if the velocity-dependent part of the scattering amplitude is neglected, we can approximate
Å ÆÇÈ Ð ÑÒÓ
º »¼ ½ »¾¿ À Á¿ Ã Ä ÉÏÌ ÚÛ ÜÝ ÛÞß àáâã äå æ çèå
ÉÊËÌÍ Î ÒÔ Õ Ö×Ø Ù éêéëìí
ôõö
î ï ðñò ó õ úû üýþÿ þ
÷ øù
!
When integrated over phase-space, the factor in brackets is proportional to "#$%&' .
Radiative Corrections
If the static limit is not taken, one can expect that the smearing introduced by the convolution will kill the
velocity divergence encountered before. This is indeed the case, and it is interesting to see in some detail how it
happens (see [84], [87]). All follows from the recursive property of the Schrödinger wavefunction
./0 67
( )*+ , -
*1 2 31 4 89
5:;< 8> ? B C DEF
= @A
which is the Bethe-Salpeter equation with a Coulomb ladder photon as kernel. Graphically,
(1.11)
14 Chapter 1 Overview of Positronium Physics
Now, looking back at the binding graph, one can see that when introduced in the convolution integral, its
Coulomb piece will reproduce the lowest order result
Mathematically, the Coulomb part of the diagram G, when the constituents have momentum H, has a piece like
K L L
NOP H
L LM
G I J MQ K R SSS (1.12)
H
U VWXY
This is of order T since K . Now, when convoluted with the wavefunction, the integral is
Z [\] ^ _]` abc de d n o pq r s
ghi tu
de d f fj k l mmm
This explains how the lowest order is recovered. Therefore, by considering only one-loop graphs, one vnds (1.7).
There is also a straightforward relation with the Sommerfeld factor. If the term (1.12) is expanded around
w xy
, one gets z
w }| }
{| } ~ ~
}| }
This is a contribution to the amplitude. When squared, the result (1.6) is found. Obviously, this singularity
originates in the improper ordering of the limits , .
1.3.3 Bethe-Salpeter Loop
Using Bethe-Salpeter (BS) analyses, one starts with a four-dimensional integral [87]
¢ £¤¥ ¦§ ªª £«¬ ®¯° ± ²³ ´µ ®¯° ¶ ·¸ ¹ º»¸¼
¡
¨© (1.13)
with ½¾¿ÀÁÁ the off-shell scattering amplitude and  the positronium momentum. For example, in the case of
Ã
- Â Ä Å ÆÆ , it is
Õ é
ÇÈ ËË Ì ÍÎ Î ÕÖ
ÉÊ ÏÐÑÒ Ó Ô ×Ø ÙÚ Û Ü æ ç è éêë ç
ìí è î ñ ò ò ô õö
÷ø ù
Ý Þß à ÞáÜ â ã äå ï ð ðó î
ú
The BS amplitude (1.13) has a loop-like structure. This becomes apparent after identi cation of the BS vertex
[19] ûü
ýþ ÿ
Hence
'(
!" # $ % & 012 3 5 6 H I J
DD EFGF KL M
)*+ ,' - . / 444 8
7 9 : NO P Q T U V W (1.14)
; <= > ? @ABC R S
1.3 Theoretical Description of Positronium Decay 15
In this representation, the BS vertex is an effective form factor for the bound state. Graphically, for the case of
X
- Y Z [ \\
Barbieri-Remiddi Reduction
As is well-known, the BS equation cannot be solved exactly, and one has to rely on some approximations
for the vertex (or wavefunction). The most popular one for positronium is that of Barbieri-Remiddi [73], which
can be viewed as a four-dimensional generalization of the usual Schrödinger wavefunction. In general, the
_` a b _cd e fgd
wavefunction is approximated as ] ^ ^ in (1.13), so that the decay amplitude is given by a
three-dimensional integral representation (see for example [87], [113]).
h
Speci cally, the Barbieri-Remiddi wavefunction (BRW) is
i st uv o p uy
jklm no p q r wx z { |} ~
where ¡ denotes the ground state, spin ¡. The de¢nitions are
£ ²
¤ ¥ ¦§¨ © ª« ¬ ® ¯ °±
³ ´® ¬µ¶ ¯ · Â
« ¸ ¹º ¸ ¼½ ¾ ¿ÀÁ
µ® »
ÃÄ ÅÆÇ È É Î
ÊË Ì Ï ÐÑ Ò Ó Ô ÕÖ ×ØÙ
Ì Í
For the wavefunction spin part Ú Úá
ØØ Û Ü Ý Þ é
â ã ä ê
å æ çêè ë
Ý Ý ßà
where ìí is the polarization vector of the orthopositronium state. Finally, the Schrödinger wavefunction at the
heart of the BRW is as usual îï
ï ö÷
ðñ ò óô õ
ðø ù ÷ø ñø
ú
To recover the three-dimensional convolution integral, it suf ces to note that the energy-dependent part of
the BRW is a delta-function representation
!
34 -
"# $ %&' ( )* (+,-. / 01 2
ûüý
þÿ
which amounts to neglecting the binding energy in that part of the wavefunction. In that limit, up to normalization
factors, the decay rate is
>?
5 267 8 9: - ; < @= D E @?C AGHIJ
F KLM N OPQ R STU VW XYZ[\]^^ _`a bcd efg bhij k lmn opq rstu
ABC
(1.15)
16 Chapter 1 Overview of Positronium Physics
y
with v w x z{|}, v~
and for parapositronium, for orthopositronium with polariza-
tion vector . The
factor is the spin part of the wavefunction. Note well that compared to (1.13),
the scattering amplitude is now evaluated for an on-shell electron and positron.
The Starting Point
In general, there is great latitude in the treatment of the projectors. Usually, those are simplied to [121]
¡ ¢£¤ ¥ ¦§¨ © ª«¬ ³° ®µ¶·¸ Î
®¯°± ² ´ ¹º ¾ »¿¼½ ÀÁÂÃÄÅÅ ÆÇÈ ÉÊË ÌÍ ÉÏÐÑ Ò ÓÔÕÖ (1.16)
×
The difference between both formulas being treated as perturbations. Such a simpli cation of the projectors
is very delicate, because of gauge invariance issues ((1.15) is gauge invariant thanks to the ØÙ Ú Û Ü and Ý
Þ ßà
åå
projectors on both sides of áâãä ). However, the formula (1.16) is the starting point of most modern applications.
This is probably because (1.16) can be ë obtained simply from (1.9) by working out the constraint æçæè é ê.
îï
Note well that this formula is suf cient to obtain the leading order approximation for the decay rate. Hence,
it appears as a good perturbation basis. Further, it is rather simple since ì í is the Schrödinger wavefunction.
ë
In other words, summing only the effect of the Coulomb instantaneous interaction into the wavefunction is
suf cient. A similar situation occurs for energy levels, since the Balmer formula is a rather good approximation.
Wavefunction energy dependences, or departure from the simple Coulomb Schrödinger form, or corrections
the spin and projector parts are treated as (relativistic) perturbations. They are typically of the order of ð ,
ñòto
i.e. óô . To the relativistic corrections, one must add radiative corrections to the scattering amplitude õö÷øùù. For
instance, order ú corrections are simply computed using (1.16) as we have seen in the previous sections. At the
end of the day, a consistent perturbation basis is built.
This closes the review of basic formulas used to compute decay rates in the standard approach. We have
reached a very general form (1.13). So general, in fact, that approximation methods are compulsory. Practically,
simpler expressions are used, and a consistent perturbation theory is built. Further, the non-relativistic QED
(NRQED) effective theory gives a systematics to this perturbation theory (see for example [16], [93], [105],
[109], [110], [112], [123]). This is the tool used in all recent advances in positronium physics, and more generally
in QED bound state theory.
Chapter 2
A New Basis for QED Bound States
The present chapter is the core of the thesis. It consists of three main parts: (1) the contradiction of standard
approaches with analyticity, (2) the resolution of the problem and (3) application to various processes. The
results of this chapter were presented in our papers [29], [30], [31] and [32].
In the ûrst part, problems with the standard approach perturbation theory are exposed. It is shown that the
standard basis chosen is inappropriate, because it is in contradiction with a fundamental theorem of Quantum
Field Theory (QFT), namely Low üs theorem. The predictions of the standard basis for the energy spectra of
the photons produced in the annihilation of positronium are wrong. This ûrst result is very straightforward in
the context of dispersion relations. To be more precise, we will clarify the approximation done in reducing the
Bethe-Salpeter loop into the three-dimensional convolution formula used as a perturbation basis. In this way, the
effects expanded perturbatively will be characterized.
In the second part, an alternative scheme of computation is presented. It amounts basically to a change of
lowest order amplitude. Compared to the perturbative expansion in the standard approach, this new lowest order
amplitude is a non-perturbative resummation of an inûnite class of corrections. This resummation will be shown
to be essential in order to deal with soft photons. Indeed, in the standard non-relativistic QED (NRQED)
approach, there is an inconsistency in the treatment of ýnal state photons when their energies are of the
order of the binding energy, because the NRQED momentum scaling rules are violated.
To be able to use our new basis in practical applications, a simple computation method will be proposed.
We will show that our non-perturbative bound state decay amplitudes are simply related to basic QED processes,
easily dealt with using QFT perturbative techniques. At the end of the day, a consistent, well-deûned lowest
order basis for QED bound state decay computations is obtained.
The ûnal sections contain the application of the method to various processes. The analytical properties of
the amplitudes are analyzed and agree with Low üs theorem. Further, the behavior of the series expansion is better
þ
than in NRQED approaches. Finally, extensions to other bound state problems (positronium formation, radiative
transitions, hyperûne splitting,...) are brie y discussed.
2.1 Critical Analysis of Standard Approach Decay Formula
The purpose of this section is to analyze the reduction of the four-dimensional Bethe-Salpeter loop (1.13)
ÿ
!" # $% & !" ' ()*+
(2.1)
to the three-dimensional convolution formula (1.16)
:;
, -. / 0 123 4 567 8 9 <;? A NO [
TU VWX Y Z V\]^ _ `ab c
<=>?: @ HLIJ M PQRR S
=BC D E FG K (2.2)
fg ij
where d e is the Barbieri-Remiddi wavefunction and h e the Schrödinger one. Let us recall that the reduction
mj mj n o mp q r stq
usually proceeds by approximating the energy part of k l as k l l and neglecting some u
dependences in the projectors. Then, the standard approach is to treat the difference between both formulas
18 Chapter 2 A New Basis for QED Bound States
as higher order perturbations. In other words, (2.2) is the basis of standard perturbation theory for decay rate
computations.
Our goal is to use dispersion relation techniques to see whether the above choice for the lowest order is
appropriate. Also, concerning the many effects expanded perturbatively, we want to address the behavior of the
series expansion. The information gathered in the present section will motivate the form of the new basis we
propose in the next section.
2.1.1 Problematic Properties of the Convolution Formula
We would like to answer the following questions, which we divide into three classes.
Factorization
The basic factorization of the bound state dynamics from the annihilation process. As we have seen, the
four-dimensional loop (2.1) is essentially a loop model with a form factor for the bound state (see (1.14)). Under
that form, both the bound state and decay dynamics are intimately linked. On the other hand, the convolution
formula exhibits a higher degree of factorization of both processes, since the integration is only carried over
three dimensions.
Another way to look at this reduction is to remark that in the loop (2.1), the constituents are off-shell, while
in the convolution formula, the constituents are on-shell. It is crucial to analyze how the off-shellness of the
constituents can be reintroduced perturbatively yzstarting from (2.2). This is especially important in view of the
present theoretical consideration, at order v wx . Indeed, the non-perturbative yz
phenomena responsible for the
off-shellness of the electron and positron inside the positronium are of v wx (since positronium mass minus
twice the electron mass is of that order).
Violation of energy conservation ? Positronium being a bound state, { | }~ ( the positronium mass
and ~ the electron mass). In (2.2), the total energy of the on-shell electron-positron pair entering the scattering
process is } , clearly greater than
, and getting worse as increases in the integration. The same
observation can be made about all the convolution-type formulas. An interpretation for this fact would be
welcomed.
Projectors and Gauge Invariance
The enforcement of gauge invariance may be problematic. This is more technical. In general, one has to
project the electron-positron pair into a spin state compatible with the total spin of the bound state. This bound
state spin state is usually treated non-relativistically, introducing some potential threats to the gauge invariance
of the decay amplitude. More precisely, Ward identities are satis ed if there are projectors on both sides of the
scattering amplitudes, like for example as
¡ ¢ £¤¥ ¦§¨ ©ª«¬
±²³² ³² ³´
In the static limit ® ¯ ® ° , this collapses to
µ ¶ ·¸¹º »¼ ½ ¾¿ À ÁÂÃ Î Ô Õ Ö×ØÙ
ÄÅÆÆ ÇÈÉ ÊËÌ Í Ì ÈÏ ÐÑÒ Ó
and one recognize the projector of (2.2). As a result, it appears that (2.2) is gauge invariant only in the static
limit. If one is interested in the lowest order static result, (2.2) is just Úne. If one wants to go beyond it, care is
needed in the treatment of the projectors.
2.1 Critical Analysis of Standard Approach Decay Formula 19
Another source of gauge invariance problems arises from the form of the Bethe-Salpeter loop itself (2.1).
Û
This is in fact more problematic. To understand it, it suf ces to remember that to enforce the Ward identities,
one needs cancellation among the various amplitudes contributing to a given process (for example, there is six
diagrams contributing to Ü- Ý Þ ß ààà ). Because of the momentum dependences of the form factor (i.e. of the
wavefunction), this cancellation is spoiled, and the Ward identities are violated.
Analyticity
The last, and more important point, is the analytical behavior in the soft photon limit. The consequences of
gauge invariance are well-known: one gets very tough constraints on the structure of amplitudes in the form of
ä
Ward-Takahashi Identities. For instance, any amplitude involving an external photon, á â ãäå , must verify
ä çè
the Ward Identity æäå , with æä the photon momentum and éä its polarization
í
vector. In addition to gauge
äê
invariance, since any probability amplitude is an analytical function, å æë ììì admits a Laurent expansion in
each of its variables. It was F.E. Low who, in the î fties [204], îrst realized that the Ward identity restricts the
form of the îrst two terms of the Laurent expansion in the external photon energy.
Low ïs theorem is a model-independent result, valid to all orders: for a complete amplitude, the soft-photon
limit only depends on the quantum numbers of the external particles, and not on the details of the intermediate
subprocesses. At the level of observables, the low-energy end of the photon spectrum is obtained by combining
the amplitude behavior with that of the phase-space. The most characteristic spectra are
ð Charged particles and photons in external states. The well-known bremsstrahlung emissions lead to an
òó ó ô õ ó
amplitude in ñ for ( is the energy of one of the emitted photons). At the decay rate or cross
section level, the IR divergent amplitude generates characteristic IR divergent spectra.
ó ó ô
ð Only neutral self-conjugate bosons, including photons in external states. The amplitude is in for
õ ó
( is again the energy of one of the photon). This statement is much stronger than what is sometimes
ö
thought of for a non-bremsstrahlung process. On general ground, it shows that feeling con dent with ÷
an
IR safe computation is theoretically incorrect. IR safety at the cross section or decay rate level is de nitely
÷
not suf cient. When one constructs a model designed to describe some processes among neutral bosons
and photons, one must ensure that the amplitude vanishes in the soft-photon limit.
It will be shown in the following that the standard approach amplitudes are in contradiction with Low øs the-
orem, at each order. The origin of the problem can already be explained from the occurrence of on-shell charged
particles in the intermediate stage. Indeed, models derived from Bethe-Salpeter analyses, or from QED non-
relativistic effective theory (NRQED), always connect the process of annihilation of bound charged particles to
÷
that of scattering of real, asymptotic charged particles. The dif culty with such approaches is thereby appar-
ent: asymptotic and bound charged particles have drastically different radiation properties: the former exhibit
bremsstrahlung-type radiations, while the later do not radiate zero energy photons (for very low-energy photons,
a positronium state is just a neutral, self-conjugate boson, hence it does not radiate in that limit). In the literature,
it is assumed that the intermediate charged particle bremsstrahlung radiations are disposed off simply by project-
ing onto the required positronium spin state. This is indeed the case for the singular part of the bremsstrahlung
radiation, but not for the constant part. In other words, Bremsstrahlung radiations typically lead to a Laurent
expansion for the amplitude as ùú û û û û
üý þþþÿ
û
üÿ û ÿ üÿ
û û
üÿ and ÿ must disappear [204]. While the cancellation of
sütheorem
Low ÿ
states that both the terms of
terms is automatic from selection rules, that of
ÿ terms is much more delicate. At the end of the
day, the amplitudes fail to vanish in the soft-photon limit.
20 Chapter 2 A New Basis for QED Bound States
2.1.2 Dispersion Analysis of the BS Loop
Most of the previous questions can be answered simply in the context of dispersion relations. To begin with, let
us take a four-dimensional loop model with a general form factor.
A Generalized Loop Model
Basically, we assume a loop structure for the decay amplitudes, similar to the Bethe-Salpeter loop. Positro-
nium decays into a virtual electron-positron pair which subsequently annihilates into real or virtual photons (an
odd number for ortho-states, an even number for para-states). The coupling of the positronium to its constituents
is described by a form factor , times a Dirac matrix structure consistent with the bound state quantum num-
bers. Remark that a constant form factor would amount to consider positronium as a point-like bound state. As
we will see, is in fact related to the bound state wavefunction.
For parapositronium decay into two photons, our model is represented
The corresponding amplitude is
- !" # $ !& # '()*
+,-.. 01 2 43 56/ 78 9:;<=>?@
% ; (2.3)
with A the positronium four-momentum and BC D EF GHIJK L MN the form factor. The tensor OPQ RSTUU is the
scattering amplitude for off-shell VWVX, with incoming momenta ZY [ \ ] and ZY [ ^ _, into two photons :
`ab cdeff hij klmn o pqrs tuv w x yz { | }~
cdeff g `ab
¡ ¦ ª«
¢ £¤ ¥ £ ¡ ¥§¨© (2.4)
Remarks :
1) The model is extended to orthopositronium decays through the replacement of ª¬ by £ .
2 ) The electron and positron in the loop are never on-shell, because ® ¯°.
©
3 ) ±² contains all the information about the bound state. Let us postulate a form for this coupling as (in
the positronium center-of-mass frame)
±² ³ ´µ¶· ¸¹º»¼½¾ ¿ÀÁ  ÃÄÅ (2.5)
Í
with Æ a constant, and ÃÄ Ç ÈÄ É ÊËÌ related to the binding energy ÎÏ Ð Ñ Ò ÓÔ. In QED, ÕÏ
á á
and Ö× are related to the Øne structure constant as ÙÚ Û ÜÝÞßà and â ß Û ÝßÞßà .
2.1 Critical Analysis of Standard Approach Decay Formula 21
The Loop Model Reproduces Standard Decay Amplitudes
ã
We can state our result in three steps (details are found in the appendix A.1.1). First, we de ne
äåæ çèéê ë ìíî ïð-ñò ïñóô õ ööô (2.6)
with the absorptive part given by the two vertical cuts
Then, using an unsubtracted dispersion relation (see [6], [219]) with ÷ ø ùú :
û üýþÿ
(2.7)
(
!" #$%& since $ ' ()), and changing variables, one recovers a factorized form (i.e. a
convolution type amplitude):
* +,--. / 001 2 3( 4 8567< =>?FGH @ABCDE LM NOP RQ STUV YY Z[\ ^] _`
9:; IJ K WX (2.8)
with a given by abc
VWXYY Zde fgh ijk fglh m nnopqrstuvw and x y z{|}~ } , i.e. the same as
in (2.2). Finally, if one further identi es
¡¢£
¡ ¨ ©ª ¨ª
by neglecting the dependence in the projectors ¤¥¦ § and «¥¬ § , (2.8) reduces to (2.2).
´µ¶·¸ ¹ º »¼½¾.
As a corollary, the energy-dependence of the form factor is demonstrated to be irrelevant. Indeed, the
dispersion relation xes ®¯ ° ±. From now on, we will simplify the notation and identify ² ³
Discussion
First, factorization appears simply as a manifestation of the optical theorem: the appearance of on-shell in-
Å integral ¿shifts¿ them off-shell: the amplitude
termediate states is expected in the imaginary part. The dispersive
at the physical point À Á ÂÃ is purely real, i.e. only off-shell Ä ÄÆ circulate inside the loop. Also, the apparent
Ç
non-conservation of energy is explained, since the dispersive integral is done along the loop model imaginary
part cut, where the initial energy is indeed suf cient to get on-shell constituents. In other words, the standard
approach convolution amplitudes are constructed like in Èold-fashioned perturbation theoryÈ. In the context of
Ç
quantum eld theory, their natural framework is dispersion theory.
Second, the spin projections are treated covariantly, since we took the BS vertex ÉÊË with its spin structure
replaced by ÌÍ or Î Ï (ÐÑ ÒÓÔÕÖ× is a scalar), a replacement dictated by the bound state properties under parity
and charge conjugation. This allows us to identify the correct covariant spin projectors for the ØÙØÚ pair,
and to preserve manifest gauge invariance. This is more adequate since it is the moving electron-positron pair
annihilating into photons that must be constrained to the required spin state.
exact. Starting with a four-dimensional BS form factor or vertex function ÛÜÝ Þ ßà áâãäåæ, the dispersion
Using dispersion relations to reduce the BS loop has the great advantage over approximate methods of being
22 Chapter 2 A New Basis for QED Bound States
relations alone enforce çè é ê, i.e. the energy-dependence is set to zero, and no approximation for the energy
part of the wavefunction are needed. This allows us to work in a formally relativistic environment ë standard
Quantum Field Theory techniques are thereby applicable.
Finally, when there is a virtual photon, or for three or more photon decay channels, the standard approach
formula appears hopelessly wrong, since it neglects all the oblique cuts
Indeed, it is important to realize that only the vertical cuts need to be considered to reproduce (2.2). Since
Cutkoski rule implies that all the cuts must be taken into account in order to produce analytical decay amplitudes,
there is no way, using the standard formalism, to get analytical results. In other words, a perturbation theory
starting with (2.2) as the lowest order basis will predict incorrect photon spectra.
2.2 Lowest Order Decay Amplitudes
í just seen that positronium amplitudes are built as loop amplitudes: the positronium couples to a vir-
We have
tual ì ìî loop, to which a given number of photons are attached. This loop structure is essential, because
oblique cuts have to be included. The coupling of positronium to its constituents is essentially determined by the
ï
positronium quantum numbers and wavefunction. Since the Schrödinger wavefunction contains the effects of the
exchange of in nitely many Coulomb photons among the constituents (in the ladder approximation) [73], these
effects are included also in the form factor. This choice of lowest order basis solves all the problems mentioned
in the previous section.
A very simple method can be used to effectively compute such loop amplitudes: any lowest order loop am-
plitude with a Coulomb form factor is the derivative with respect to the positronium mass of the corresponding
point-like amplitude. By point-like amplitude is meant the loop amplitude obtained by replacing the compli-
cated Coulomb form factor for the bound state by a constant form factor. Symbolically, for parapositronium
(orthopositronium) decay to an even (odd) number of real or virtual photons, the amplitude is
ô ÷
õõ øø
ð ñ õõõ øø
øø
ñòó õõ
ö ùø
2.2 Lowest Order Decay Amplitudes 23
where ú represents the Schrödinger wavefunction form factor, i.e.
and dashed photon lines stand for Coulomb photons.
ü
The loop amplitude will constitute our new lowest order basis for perturbation theory. Its properties will be
discussed later on. Extension to higher orders will be brie y commented at the end of the chapter.
2.2.1 A Simple Formula
Let us express mathematically the previous qualitative assertion, using the language of dispersion relations. The
point-like loop amplitude (subscript ý ) can be computed from its imaginary part
þÿ
using an unsubtracted dispersion relation with :
)*
! "#$% & ' 9
( +,- 6 78 (2.9)
./
<=> ? @A B EFG / 0 12 34 5
(:6 ; CD since H I JK). The Schrödinger form factor which accounts for the non-trivial
coupling of the bound state to its constituent is of the form
LM NOP QR S TUVW XYZ[ XYZ \ ]^_ gh
with ` abcd e ijkfl hkmk (2.10)
where one can recognize n opqr s tuv wxyz as the fundamental {| } ~ -wave Schrödinger momentum space
wavefunction. When expressed in terms of the dispersion relation variable, this form factor is only a function of
the initial energy (see appendix A.1.1):
}
The core of the derivative approach emerges from the observation that inserting in (2.9) is equivalent to
taking the derivative with respect to
¤
«¬ ® ¯°
¡¢
¤ ¥£ ¦§ ¨© ª
Ã
± ²³´µ¶·¸¹ º
ª ½¾ Ã ÅÂÆÇÈÇ ÉÊ ËÌ
» ¼¿ÀÁ Ä
hence ÍÎÏÐ
Ñ ÒÓÔÕ Ö ×ØÙÚÛÜÝÞß à ä åæçè (2.11)
àáâ ã
which is the desired result. The case of other parapositronium decay channels, or orthopositronium decay modes
í
is similarly treated (simply replace éê by ìë with ì the orthopositronium polarization vector).
Remark that the form factor is inserted directly into the dispersion integral, and not in Feynman amplitudes
(like in (2.1) or (2.3)). These two approaches are equivalent only in the punctual case (consider the momentum
24 Chapter 2 A New Basis for QED Bound States
î
ow through the form factor in each case). Working at the level of dispersion relations is much more in the spirit
of the ðBethe-Salpeter equation. Indeed, the wavefunction is extracted from the four-point Green ïs function, i.e.
in con gurations with an off-shell bound state (above threshold), and on-shell constituents:
The picture shows that both the Bethe-Salpeter
ð vertex and the dispersion integral (2.9) make use of the form
factor ñò with the same kinematical con guration (diagrams with bremsstrahlung radiation off the electron
lines are treated similarly). Further, we have seen that because of the momentum dependence of the form factor,
neither (2.1) nor (2.3) is gauge invariant. Now, since the imaginary part of the point-like amplitude is gauge
invariant, the decay amplitudes are constructed in a manifestly gauge invariant way.
The properties of our approach will be discussed at the end of the present section. Before, we will give two
illustrative examples.
2.2.2 First Example: Parapositronium to Two Photons
The point-like amplitude for the decay ó - ô õ ö ÷÷ is obtained by replacing the positronium and its Coulomb
form factor by an elementary, point particle with a pseudoscalar ÷ø coupling to the electron current (with unit
coupling constant)
plus the crossed process. The point bound state amplitude is fully relativistic, and standard loop integration
techniques can be used. Explicitly, the point-like amplitude is (the subscript ù is a reminder for point-like):
úû üý
-þ ÿ
! " #$ &'(&)*%+,-. 01 2345
/ 012067/89
:;< >?: >@7=89 ;A CD EFGH
B CDECIJBKL
Carrying the trace, we readily obtain
MN OP-QR S TTU V WXYZ[\]^_`abc_ a[c` \] dabe \^ da[ e fg lhijk
j
where the dimensionless6 loop integral form factor is
mn opqr s vwtu z{ |}~
xy with (2.12)
This criterion comes in because the dispersion relations (2.9) is built on the whole amplitude, and not just the loop form factor.
2.2 Lowest Order Decay Amplitudes 25
for can be dened from the above by analytic continuation). From the
(the form factor function
amplitude, the decay width is
¡¢-£¤ ¥ ¦¦§ ¨ ©ª«¬ ®¯ °±² ³´µ¶··µ ¸ ¹º» µ ÀÁ ÂÃÄ Â ÉÊËÌÍÎ Ï Î
° ¼½¾ ¿ Å ÄÆ ¿Ç Ä È ÐÑÒ (2.13)
Ù
where Ñ Î Ó ÔÎ ÕÖ×Ø is related to the binding energy ÚÛ Ü Ý Þßà.
Coulomb Form Factor
áâ ãäåæ çèéê
To get the physical positronium decay amplitude and rate, simply replace in (2.13) the loop form factor
by its derivative
ë ìíîï ð ñòóôõö÷øù ú
úûüýþ ÿ (2.14)
where is the -wave fundamental state Schrödinger wavefunction at zero separation, and is
obtained by matching the static limit i.e.
with the well-known lowest order result (See
-
Appendix A.1.2). This gives
, *
!"# $ %&'( *) +- ./01.2 ,34
9 < = >?@?ABC, the decay rate into two
The factor in square brackets is equal to 5 in the limit 3 6 7. Using 8 :;
DE >MN b
F-GH I JJK L O PQR SVTUUW XZY [\]^[_ `Ya X
photons is
XX XX (2.15)
The decay rate is a decreasing function of the binding energy: X X
Enhanced Convergence
c. Indeed, b econtains
The result (2.15)
using d
fb g hbi j k fbcofbithej, the
some effects Coulomb interactions among the constituents, at all orders in
form factor can be expanded as (see the discussion in the
appendix A.1.2)
l mn-op q rrs t vw z{ |}~ z
u x y ~ y
¢
¡
¥¦
£ ¤ § ¨©ª«¬®¯°± ² ¨°³´´
26 Chapter 2 A New Basis for QED Bound States
In other words, the binding energy effects included in our lowest order computation already account for a great
deal of the relativistic and radiative corrections as presented in the literature (see chapter 1), namely,
µ¶ ·¸ ¹ º»¼
- ½ ¾¿ÀÁÂà ÄÅÆ - (2.16)
where
ÁÂà ÄÅ Ç Èà ÉÊ Ë Ì Í ÎÏ ÑÐ Ò ÓÔ ÑÕÍÍ Ö ×ØÚÛÙ ÜÝÞ ß â ãäå æç è åãí
-
É òóôõ Ø àá ã é êëì í
(2.17)
îâ÷ ïø ðñùúûü ýóþÿ
÷ ø
with ö
Numerically, we can write
#$ * + ,-./
- !" !% & '( ) - 012
3
with the same series (2.17), but with the reduced coef cients
456 7 89:;<=
>?@ A BCDBE
FGH I JKLMNM
This last form is very interesting because binding energy corrections (i.e. O dependent) are singled out, while
the resulting purely perturbative (radiative and relativistic) corrections are much reduced. This means that one
could, at least in principle, express the decay rate as a lowest order result, non-perturbative in the binding energy,
times a rapidly converging perturbation series of radiative corrections. This is exactly what we have achieved.
Note Pnally that the corrections obtained here do not originate from oblique cuts, since there are no such
cuts for the two real photon decay. Rather, they originate in the many relativistic effects we are summing by
considering a four-dimensional loop model, and from the inPnite sum of Coulomb photon contained inside
the form factor. In the appendix A.1.2, the relation between the above correction and the spreading of the
wavefunction (and to the Sommerfeld factor), are exposed.
2.2.3 Second Example: Paradimuonium and Low Qs Theorem
The paradimuonium decay R - ST U VWXYZ is the simplest QED bound state decay process where Low [s
theorem implications can be illustrated. The paradimuonium is the \]^ , _X_Z electromagnetic bound state
[95], [111]. In appendix A.1.4, the decay `a b cdcef is presented. This elementary particle decay process
shares many interesting similarities with the present QED bound state case, especially regarding the enforcement
of analyticity.
Following the same steps as for g -hi j ff , we krst consider the point-like amplitude
2.2 Lowest Order Decay Amplitudes 27
(plus the crossed diagram) with the result
lm no -pq r stsuvw x yz}{| ~
(2.18)
¥®
with
¡ ¤ ¡¢£¥ ¨ ©¤ ¡ ¥¦ª « ¢£¬
¦§ ¯° ± ²³´µ
¶· ¸¹º»¼½¾ ¿ ÀÁ Æ Ç Ë ÌÍÐÎÏ ÕÖ×
ÂÃÄÅ ÈÉ Ê Ï Ñ Ò Ó Ô Ø × Ù Ü ÝÞ
Ù ÚÛ ê íê
ã
where ß is the muon mass, à the dimuonium mass, áÜ the reduced photon energy â Üäà , åæ ç èéëì ,
ï õ
mass, and with the function î de ned in (2.12). In the limit ðñ ò ó, the spectrum ô ö÷ôøù
éë the electron øú
goes to zero as ù as predicted by Low ûs theorem (the amplitude behaves as üý , and an additional factor üý
comes from phase-space).
Taking the derivative of þÿ to get the corresponding Coulomb form factor, we nd
þ
!
,-
" #$ %&' 2 -
( *)+ . /01 /3 4 ?@ A
5 6 789:
;< => = (2.19)
9:B
HI STUVW
where CD E FJ I GKL M N
OP Q R
The decay rate is then obtained by replacing XY by XZ[\] in (2.18)
lm mq w uv
^ _` efghi j k z
- ab c d no pr q st { |~
x y | |
|}
}
Binding Energy Expansion and Analyticity
It is very instructive to analyze in some details this result. Consider the static limit for the form
factor: ¡
¢£ §¨© ® ¯°± ² ³ ´µ ¶·¸
¤¥¦ ª« ¬ ¹º» (2.20)
In that limit, the standard lowest order result for the decay rate is recovered
É ÓÔ ×
¼ ½¾ Ã Ø ÛÜÝÞ
-¿À Á  ÂÄÅÆ ÇÈ ÎÏ ÐÕ ÑÒ Ö×Ø Ù Ú ×ß
Ê ËÌÍ
Ø
àá
However, the differential rate Ö Ö×Ø has a wrong behavior when ×Ø â ã. The spectrum is linear (in äå ) in the
limit æ â ã, in contradiction with Low çs theorem. Therefore, it appears that, contrary to the two real photon
case, the limit è é ê is far from smooth. It is inconsistent to consider both the soft-photon limit and the on-shell
limit simultaneously. Explicitly, the incompatibility of the two limits is obvious if ëì í î is taken ïrst
ðñò ÷øùúû üýþ ÿ
óôõö
(2.21)
$& ( 2
!"#
$ %'( & ) +* , ./0
- ) 111
28 Chapter 2 A New Basis for QED Bound States
Mathematically, what these considerations show is that the limit 3 4 5 does not exist at 67 8 5. It should be
clear that one of the main virtues of our approach is its non-perturbative treatment of 3 , leading to correct photon
spectra.
The contradiction between analyticity and the static limit, i.e. the basis around which the NRQED perturba-
tion theory is built, can be traced back to two different sources.
?@ABC D
First, the second term of 9:;<= > (2.19) comes from the oblique cuts in the imaginary part (i.e.
processes like E -FG H IJIKL times IJIK MNO P QRQS TUV)
First term :
Second term :
Those are the processes neglected in standard approaches, because they are formally of order U relative to
vertical cuts processes (see (2.19)). This is an inconsistent approximation since it is the sum of the two terms of
WXYZ[ \]^ _`a b
that enforces Low cs theorem. This can be seen by plotting
r s vw s
d efg hijkl m n o pq
qvtu r q
w tx
tu tx pq
for y z { and various values of |
Therefore, two terms that would be treated as different orders in the NRQED perturbation series are
needed simultaneously to enforce Low }s theorem.
The second source of the NRQED problems is apparent on the picture and in (2.21). The form factor is not
analytical on the whole phase-space when ~ . As the picture shows, even if ~ for QED bound
states, the limit ~
is not to be taken because it is singular. A breakdown occurs when the energy of the
photon, , is of the order of . Furthermore, as a matter of principle, since the non-perturbative result is
2.2 Lowest Order Decay Amplitudes 29
not analytic, it cannot be expanded as a series in , and the NRQED perturbation series is not to be expected to
converge. Needless to say, this is a serious drawback.
Total and Differential Rates
The binding energy effects on the spectrum and differential rate amount to overall suppression. As function
of the binding energy, we have, for
The various curves in the rst graph correspond to (top to bottom). The distortions generated
by the form factor on the low-energy end of the spectrum are not apparent, because of the spectrum function
, which is highly peaked near . In particular, the singular transition from a ¡¢ to a linear £¢
behavior near £¢«¬¤ ¥« when ¦ § ¨ does not appear. Note that, as is the case for the two-photon decay mode, it
is the factor ©ª which is responsible for the non-vanishing of the rate as ® § ¯.
More importantly, it can be shown by a numerical integration that the contribution of the oblique cuts to the
« order ® and higher (their interference with vertical cut is of order ®, while alone, they
total rate is of relative
stand at the order ® ). In other words, while their presence is essential to get a correct spectrum, they can be
omitted if one is interested in the total rate to lowest accuracy. The singular behavior of the spectrum is smeared
out by the phase-space integration.
2.2.4 Conclusion
A new lowest order basis is proposed, along with a simple computation scheme. This new basis is manifestly
gauge invariant, and formally relativistic. We want now to spend some time to analyze the enhanced convergence
property, and then the analytical properties.
To understand why the binding energy correction we obtained can be interpreted as part of the relativistic
° °
and radiative corrections computed in the literature, it suf ces to notice that the rst order radiative corrections
do contain a Coulomb photon exchange. Further, some relativistic corrections are simply generated in our
approach because the propagators of the loop are treated exactly (no approximation for the spinors), and because
the spin part of the wavefunction is taken as covariant. What is interesting is that our approach offers a simple
computation scheme accounting for many non-trivial corrections. The remaining perturbation series converges
much more rapidly. The special feature of bound state perturbation series, i.e. their slow convergence, is clearly
°
identi ed as a binding energy effect, and we are now able to factorize it from the start. Further, it is especially
clean in separating the effects of the binding energy from the purely radiative corrections. This is of interest,
especially regarding QCD bound states, since there, the binding energy is not simply related to ±² .
The most interesting part of our result is of course its analytical properties. That is where it is really superior
to the NRQED approach. The reason is simple to understand: when a soft photon is emitted, the scaling rules of
NRQED are violated. The emitted photon energy is of the same order as the binding energy. One can therefore
expect that a resummation of NRQED amplitudes will be necessary. Let us see how this arises.
30 Chapter 2 A New Basis for QED Bound States
In the context of NRQED, one would consider the oblique cuts as arising from higher order corrections,
In the second graph, we have illustrated a possible higher order reconstruction of the oblique cut. However,
each of these graphs´is separately badly behaved analytically, because at all stages one is selecting an initial
state made of free ³ ³µ. In other words, for a given vertical cut, the corresponding oblique cuts are always
introduced at higher orders. Obviously, it is only by summing the whole series that the emitted photon will truly
feel the binding of the charged particles. Our method of integrating the loop with its Coulomb form factor sums
¶
a minimal class of contributions so that the amplitude is well-behaved analytically. In other words, the in nite
Coulomb photon summation contained in the form factor (or wavefunction) is not factorized from the decay
process. This is a radical
¶ difference between NRQED and our approach.
The fact that an in nite summation is necessary is especially obvious¼ from the behavior of the non-perturba-
tive result we found. The energy spectrum of the photon in · -¸¹ º » »½¾ cannot be expressed perturbatively
as a series in ¾. Its low-energy point requires an exact treatment of ¾ .
In conclusion, NRQED is unable to describe the end-points of energy spectra, because of a violation of the
momentum ¿ scaling rules at those phase-space points. Our method is a consistent resummation scheme, and,
being suf ciently non-perturbative in the binding energy, the photon spectrum becomes well-behaved over the
whole phase-space. We therefore obtain an analytical basis for perturbation theory. The perturbative treatment of
higher orders will be discussed later. For total rates, our method offers a possibility to factorize binding energy
corrections from purely radiative corrections.
2.3 Orthopositronium Decay
We now apply our method to the orthopositronium decay to three photons. This is a very interesting decay
process.
¿
In the rst subsection, we shortly recall what is the lowest order basis chosen in standard computations,
namely the Ore-Powell amplitude [63], and emphasize its analytical defects.
In the second subsection, the lowest order result
¿ we will obtain by integrating the loop with the Coulomb
form factor inserted will be seen to be well-de ned analytically. Further, our result will again show that a
¿ ¿
perturbation around À Á Â is ill-de ned. Also, we will again nd that the bulk of the radiative corrections is
accounted for already at our lowest order.
¿
In the appendix A.2, one can nd a detailed analysis of the three-photon phase-space, including tools to
integrate over that three-body phase-space numerically. Also, the derivation of the Euler-Heisenberg Lagrangian,
as well as the explicit expressions for the two-point photon function to one and two loops and for the four-point
photon function to one loop are given.
2.3.1 Ore-Powell Spectrum
¿
From the requirement of gauge invariance, and because of the quantum numbers of the initial and nal particles
involved (neutral self-conjugate bosons), Low Ãs theorem predicts that the decay amplitude must vanish linearly
2.3 Orthopositronium Decay 31
when the energy of one of the photons is going to zero
Ä ÅÆ Î
Ï Ð ÑÒÓ
-Ç È É ÊÊÊË ÌÍ
with Ò the energy of one of the photons. The squared modulus of the amplitude therefore behaves as Ô ÕÒÖ× for
small photon energy. Aside from that, the three-photon phase-space alone (i.e. with a constant decay amplitude)
gives a differential rate as Ø
Ù ÚÛ
-Ü Ý Þ ßßßà
áâ ãä åæçèéêë æìè í î (2.22)
ã
ã
ó ó ã
with ñò the reduced photon energy, the orthopositronium mass. Combining the amplitude with the
î ï ð
phase-space, one ônds that the low-energy end of the photon spectrum must behave as
õö ÷ø
-ù ú û üüüý
þÿ
The standard approach at lowest order states that the orthopositronium decay rate is to be computed with the
formula
!!!""
-
#$%&'(
. 23
with )*+, the relative velocity of the - -/ in their center-of-mass frame, and 0 1 the positronium Schrödinger
wavefunction at zero separation (4 is the electron mass).
As the picture shows (crossed processes are not drawn), this formula states that in 5rst approximation, the
positronium decay rate can be computed from the static limit of the scattering cross section 6768 9 ::: (initial
particles at rest). Equivalently, it is found from the squared modulus amplitude for an ;<;8 pair at rest into ::: .
Summed over photon polarizations, this is easily shown to be [63]
= H Z [ \] ^ _`ab f ^ _b ab f ^ _ c ab
HI JKLMLNO b b b̀ b
H PQRSTU V WWWX YY _ _ _ _ _b̀ _b
>?@ABCDAEC?FG b c de c d e b
Y
_
This modulus squared behaves as a constant when one of the g is vanishing (as can be seen by using energy-
momentum conservation hi j hk j hl m n, see appendix A.2 for details on the three-photon phase-space
k
kinematics), while it should vanish as hg from Low os theorem. In turn, the well-known differential rate inherits
an incorrect analytical behavior p
q rs p }~
- t u v wwwx z
yz { |
32 Chapter 2 A New Basis for QED Bound States
where the spectrum function is
¥
¡ ¢¢¢£ ¤¤¦§¨©ª«¬ª«
¤
º» ¼ ½¾¿ À ¾¿Á
® ¯ °¯ ± ²³´ ¯ °¶ ± ²³´ ·²³ Â É ÊË Ì ÍÎÏ
ÀÁ¾  ÃÄ
Å À ¾¿Æ Ç È (2.23)
²³ µ °¯ ± ²³´ µ¸ ¹ Â
Ã
Ð Ñ Î Ô Õ Ö×Ø ÙÚ ×Ù Û Ü
ÒÓ near
The (normalized) differential rate can be plotted
In the Ore-Powell model, the photon energy spectrum vanishes only linearly near zero, instead of the required
Ý Þ× Ùß Û à áâã
äå. In other words, this decay spectrum does violate the basic requirement of analyticity.
For completeness, recall that it is this differential rate that gives the total width
ñò ó ôõ øù
æ çè þ ö
-é ê ë ìììí î ï ð ôö ÷ since úü û ýþ
ûÿ û
2.3.2 Point-like Amplitude
The point-like amplitude is the standard light-by-light box diagram
plus ve other ordering of the photon insertions. The amplitude can be found in many places [113], [205], see
appendix A.2. The tensor describing the transition from an off-shell photon to three on-shell photons
of helicity states
is such that
#$%
! " ( )* , - ( ) * , - ( ) * ,.
& ' "+ " + + "
2.3 Orthopositronium Decay 33
1234 5 4
67 869 86: 8 ; 6< ABCD B G HI
where / 0 /0 ( is the reduced energy of the photon = and > ? @ ). The E F=
are given in terms of individual dimensionless helicity amplitudes as
]^ a
JKLM N O Q RST XYZ[ \ \_ `ab efgh i
EF L P UVV W \ \ ccc d i (2.24)
P \ \ i
P t wx }
j kl tu v |}~
}|~
}~|
t y zz { { {
a
k km no p qrs
} ¤
{ £
¡¢ £
£
£
§¨
The helicity amplitudes ¥©ªª
¦ are complicated functions of «¬ and given in appendix A.2. The decay rate of
a point-like vector positronium to three photons is then
® ± Ç ÈÉÊËÌ Í Î ÏÊÉËÌ Í Î ÏËÉÊÌÐ
½¾¿ À Á ÂÃÄÂÃÅ Æ
¯ ° - ² ³ ´ µµµ¶ · ¹̧ ¼º»
ÝÞß Ô
with the three-body phase space written in terms of reduced photon energies as Ñ ÒÓÔ Õ Ö× Ø ÙÚÛÙÚ×Ü
(see [113], [23], appendix C.3).
Of specialãäinterest is the behavior of the low-energy end of the differential rate. For various values of the
åã å é êëì
ratio à Õ áâ (i.e. of the binding energy æç è , the photon spectrum, normalized to the total
rate is
í ï ï
Suf ciently close to zero, the behavior is always like îð, as required by Low ñs theorem. As ò increases, the ó
behavior is getting more and more pronounced. In the limit ò ô õ, the normalized spectrum is
ö ÷
ø þ
÷
ÿ
ø ùúû üý
ù
which is, as expected, the spectrum obtained from the Euler-Heisenberg effective theory [201]
$% &')* + , 0 12345
!" # &'( - .(&' /
2.3.3 Coulomb Form factor and Ore-Powell Spectrum
6 6
To insert the Schrödinger wavefunction form factor, we de ne modi ed helicity amplitudes according to (2.11)
9:
78;<<=>? CD EF GHI J KLMNOPQRS YZ
@A B T [\\ ]^_ `abcd
TUV WX
34 Chapter 2 A New Basis for QED Bound States
The resulting decay amplitude, and decay rate are constructed as in the punctual case, and we reach
uvw w
e fg
-h i j kkkl m nop rqs t x w y t{z
|} ~
with
is given by (2.24) with in place of . Integrating over ¡¢ , we get the photon spectrum
for various values of £
(2.25)
¬® ¯ ¦°
As long as £ ¥¤ ¦, the spectrum behavior is in §¨© close to zero, i.e. roughly in the·range § © ª « .
®
The properties of this spectrum as varies are completely similar to that of ±-²³ ´ µ¶ ¶¸, and one can show
that the limits µ ´ ¹ and º» ¼ ¹ are again incompatible. Contrary to that case, however, it is interesting to
see that because of the highly correlated three-photon phase-space, the behavior of the spectrum at high energy
is strongly dependent of its behavior at low energy. Therefore, the small effect of the binding energy on the
emission of soft photons completely alters the shape of the spectrum.
Looking at the wavefunction, it also appears that the behavior of the spectrum can be understood as
½ ¾¿ÀÁ  ÄÅ ÒÓÔ Õ Ö×ØÙ ÚÛÜ
Ã Ë Î Ï Ð Ñ Ore-Powell
ÆÇÈ É ÅÈÊÈ Ë Ì Ý
Í Þ ß à á âãäå æ çèéêë ì Constant í Euler-Heisenberg
í
So that the whole range of possible wavefunctions, from the free constituents to the point-like bound state case
is spanned as ê varies from î to ï. Correspondingly, one gets the spectrum evolution from the Ore-Powell to
the Euler-Heisenberg one (i.e., the point-like spectrum in the limit ð ï).
í
In the context of dispersion relations, the contributions originating in the oblique cuts (processes like ñ-
òó ÷ times øùø ÷û ü ýý þýÿ in the imaginary part, gure ÿ below) are essential to maintain a
í ôõôö
physical spectrum (i.e. in ö ú
agreement with Low s theorem). Again, those oblique cuts are neglected in standard
2.3 Orthopositronium Decay 35
approaches. Here, they are automatically accounted for since they are included in the point-like result.
Note also that the spectrum for a point-like bound state at threshold corresponds to the spectrum for
in the Coulomb form factor case. The bound state decay spectra for are unattainable in the point-like
bound state case.
Concerning the total integrated rate, we recover the Ore-Powell result at threshold (oblique cuts are propor-
tional to
and
)
$% & '()*+ ./ 0 8
-
! "# MeV , - 12 345 6 7
?
As 9 increases, the total rate quickly decreases (remember 9 : ;<=> = @ A)
By a numerical analysis, we found
B ^ _
HI J KL MN O NPQRST U STT UX U\ SQ T S]^ U QQQ
CDE F G VW V V W V^
X Y ZZ[ \ [ Y
For orthopositronium, we can express ` in terms of a, and we bnd the binding energy corrections to the total
rate cd u
qr s t x y xz{x|
e-f g h iiij k lmn o p
tv w v } ~ } (2.26)
This series is to be compared to the one presented in the literature (chapter 1), which is
¥© ® ³ ´ ¹ Ã
£ ¤ ¨ © « ¬ ²
- ¡ ¢ ¥ ¦§ ª ¯° ± µ ¶· ¸º »¼ ½ ¸Â
º
Ä ¸ ¾ ¿ÀÁ Â
with coef cients ÅÆ Ç ÈÉÊËÌÍÍÉÍ ÎÈÉÏ Ç È
Ñ ÓÊÓ Õ Î ÏÈ
ÐÑ Ç ÒÒÊÓË ÎËÍÏ Ô Ø Ù ÚÛÜÝ ÞÜß
Ö×
36 Chapter 2 A New Basis for QED Bound States
Again, the binding energy effects appear as responsible for the bulk of the radiative corrections at order à. Note
å
that the origin of the slowness of the convergence of the corrections to áâ- ãä is clearly identi ed as coming from
the perturbative expansion of binding energy effects (see (2.26)). In other words, the four-point fermionic loop
with a Coulomb form factor is not well-behaved for æ ç è, and that limit appears as an inappropriate basis for
perturbation theory.
2.3.4 A Simple Ansatz
To describe the distortion of the spectrum as the binding energy varies, a simple ansatz can be used. Let us
modify the Ore-Powell amplitude modulus squared as
#
é õ ö÷øù øúû
üýþÿ
! "
êëìíîïðíñïëòó ô
ô
ô (2.28)
The reason for this form is the following. When only the vertical cut is taken into account, analytical problems
are generated by the virtual fermion propagators, which can blow up when the energy of one of the photons
is vanishing. A prescription to avoid this explosion
$%& ' ( # * $is# to consider that the external electrons have a mass of
. Then, in the static limit, we have ( ) )
+ 7 89 : 8;
, . / / 2 3 4 56
? @ A BC 6D<EF=>
# - -01
EF G
H I, the propagators cannot generate poles. The Ore-Powell
and similarly for the other propagator. As long as
result is obtained with J H KL, hence the virtual electron propagator is, in that limit
M NOP QR S QTO U VW
c_b d efg hipqrstu
j hk
XY Z [\ ] ^_ `a o f l mn
To replace those propagators by well-behaved ones, we multiply the Ore-Powell amplitude by the factor
vw x ys rs
vw x ys vz q { rs | uz qt}q
leading to the prescription (2.28).
The obvious question: what is the point in considering this ansatz, built on the vertical cuts only, since we
know that analyticity is restored by the oblique cuts? In other words, the vertical cut propagators do explode in
the true fermion loop, but any non-analytical term is cancelled by similar divergences in the oblique cuts. The
answer is that the ansatz can be simply integrated, and the ~nal form for the differential rate can be useful for
experimental ~ts. Another reason is that the analytical properties of the ansatz, as a function of and , are
very similar to the true fermion loop. Hence, the ansatz provides a valuable toy model, with which to study the
various pathological kinematical limits.
2.3 Orthopositronium Decay 37
The differential rate obtained from (2.28) for various binding energies is (the analytical expression is given
in the appendix A.2)
The limit , reproduces the Ore-Powell spectrum exactly, by construction. On the other hand, the limit
does not exactly reproduce the Euler-Heisenberg spectrum. Instead, we get
¡¢£¤
to be compared to
¥ «¬® ¶° ·¸¹¸ ½Â Å Æ ÇÈ
¼½¾¿ À Á ¾Ä¿ É ÊËËÌÍË Î ÍÏÌÐÐÑÒ Ó ÔÐÌÍÔÑÕÒÖ
¦§¨© ª ¯° ± ³´µ
² ¸ º » ü (2.29)
ª
More precisely, the ansatz corresponds to the spectrum obtained from an Euler-Heisenberg Lagrangian with the
scalar and pseudoscalar couplings of equal strengths
×ØÙÚÛÜÝÞÜßà á âãäåã äåæç è éêëì îí ïðñò
We do not have an explanation for this peculiar fact.
Between the two limits, it is very important to note that the ansatz spectrum is moving away from the Ore-
Powell spectrum at a much slower rate when ó increases than the true Coulomb spectrum. Consequently, the
ansatz cannot be used as it stands for ôts. Further, the evolution
ò of the total rate in terms of õ is smoother here,
leading to smaller binding energy corrections (of order õ instead of õ , see below).
Incompatible Limits and NRQED Expansion
any value of ó ö ÷, the spectrum behaves as øùú close to ø ú û ü. Explicitly, with the deônition
þõ òÿ ò,
ý ûFor
'() * '( * +,
! "#$ % & -(. /( * +01 2
R
3415 6789: 3;79>?<AF= >GA?@A=B>=HI >CDA = E? J K L MNOPQ
38 Chapter 2 A New Basis for QED Bound States
^ _ ` ab c dS T U V T U
Obviously, this series is highly singular for , i.e. . In the same way as for the decay - W XY Z [\ []^
,
efd
it appears that the limits and are incompatible.
It is instructive to expand the differential rate around , as is done in NRQED, and to analyze the
various terms. The lowest order is of course the Ore-Powell result (2.23), while higher orders are more and more
singular atgb f d h
kl pq
ijmnom } ~
rst u vxwyz{|
¡¢
® ¯° · ÁÂÃÄ
£¤¥ ¦ ©
§ ¨ª«¬ §¨±²³ ´ µ¶¸ ¹ ºº
¸ »¼ ½ ¾ ¿À
ÈÉ ÍÎ
ÅÆÇÊËÌÊ Ú ÛÜ Þ à á èé ô
ÏÐÑ Ò ÓÕÔÖ×ØÙ ÓÔÝÞß âã ä åæ ä çê ëì í î ï ðñ
ò ó
õö÷øøø æ
The integration over ðò gives the result
ûü
ùúýþÿý
! " + ,
-.
% ' ( )*
#%
$ & /.
01.2
345367 8 9 ::
=> >
In
A
the terminology of NRQED,
A
the Ore-Powell result ;<345367 is the lowest order, while the ;<?345367 is the @rst
relativistic correction . Even if the behavior of this @rst relativistic correction is highly unphysical near BC D E,
a IR @ nite answer can nevertheless be generated. This kind of correction is obtained, for example, in [99], [107].
A @nite result cannot be obtained beyond that order, and special techniques have to be designed to deal with the
divergences. Obviously, all these dif@culties simply originate in the badly chosen basis for perturbation theory.
To close this section, if the exact total rate is expanded around F D E, one @nds
U V fa ij j w {|
GH H ST W \ ns tuv q
IJ KL M NPO XY Z [ Z ] `a b cd e b h
QR ^ _ g k l m
no p q r x
m } | ~
a
qy z
Finite corrections at each order are obtained, the divergence encountered above is replaced by
,
which could be called, again in analogy with standard bound state terminology, a non-analytic term. Indeed, if
the binding energy is expressed in terms of the ne structure constant, we get corrections of the type
This illustrates some techniques used in NRQED, which amounts to extract such logarithmic corrections, with
the hope that the remaining corrections can be computed perturbatively (note well, however, that we do not claim
that all logarithmic corrections are of that type). All these complexities can be circumvented with the use of our
method .
2.3 Orthopositronium Decay 39
2.3.5 A Scalar and Pseudoscalar Resonance Model
We continue our study of analyticity in three-photon decay by taking a model where the initial vector particle rst
decays into a (pseudo)scalar plus a photon, with then the (pseudo)scalar decaying to two photons. Graphically
The couplings of the scalar and pseudoscalar to two photons are
¡ ¢£¤ ¥ ¦ § ¤ £ ¦¢¨
©ª «¬®¯°±
¯ ²° ³
ª µ ¶·
The couplings with a photon replaced by the orthopositronium are the same (with ´ ). The scalar or
pseudoscalar propagator is ¸¹
º» ¼ ½
º ¾ ¿ ÀÁ ¾
 ÃÄ
with ÅÁ the pseudoscalar and the scalar mass (we take them as equal here). Note that the constants ÆÇ È ÆÉ have
dimension of 1/mass.
The Photon Spectrum and Phase-Space Singularity
We will study the photon spectrum as the mass of the virtual resonance varies. The modulus squared of the
amplitude (there are three non-equivalent ordering of the photons) is the same for the scalar and pseudoscalar.
After integration, one gets the differential and total rates (the expressions are rather lengthy, and since the com-
putation is straightforward, will not be reproduced here). Of special interest is the low energy behavior of the
photon spectrum Ê
Ë Ñ Ô ÕÖ
Ê æç
å
× Ð ÎÏÛ Ò Ó Þß àá â ã ä
Ì Í ÎÏÖ
Ù Ú Ù ÜÝ
ÎØ
í
with èé ê ëíî ì ëï ì ð. As expected, the spectrum is in ñò near zero. The two singularities at ó ê ô and õ are
interesting. Explicitly, the differential rate at those points is
÷ ù
õ õ
í
ó êô ö
÷ ø
ñ
ê ú ñ û ü
ý þÿ þ
%& '
! #$ ! () * +, /0 1 2 30455
ø
! " - .
These behaviors are in contradiction with Low 6s theorem. This problem is due to our computation, which is
wrong when the scalar propagator singularity is inside the phase-space, i.e. for 78 between 9 and :. In fact,
;
the spectrum, and rate, can only be de ned for <8 = 9> <8 = :, where there is an analytical problem, and for
<8 ? :, where the spectrum is well-behaved.
40 Chapter 2 A New Basis for QED Bound States
Before turning to the resolution of the analytical problem, note that when @A B C, the normalized spectrum
becomes
F NS
@A B C D FE GH I JKHL M NOPQ R T PU (2.30)
G
which is nearly hthe Euler-Heisenberg one. In fact, this is the spectrum obtained from V W X
YZ[\ Z[]^
_ ` abcd fcdg or
e . The small difference between (2.30) and (2.29) is due to the contribution coming from the
interference between both terms of the Euler-Heisenberg Lagrangian.
Resummation and Finite Width Effects
i when the scalar (or pseudoscalar)
To correctly describe the process i mass is smaller than the positronium
one, we will need to sum an in nite class of diagrams. Again, we nd that analyticity requires some kind of
resummation for some values of the parameters of the theory. This is strictly similar to the Coulomb resummation
i
(or binding energy resummation) we advocate to cure NRQED in three-body decays. Here, the resummation is
even simpler to understand: if the (pseudo)scalar can decay into two photons, it has a nite width. Hence, its
propagator, after renormalization, should be of the form
j klm n h p qr o
l s t uvwxw y uz
Graphically, the summation is
Repeating the same computation as above with this new propagator
the differential rate behaves near zero as
{x w w y
{| } ~ ~
with ¡¢ £ ¤¥¦¥§¤¨© ª. For all values of «¥, the spectrum is now in agreement with Low ¬s theorem. Further,
the differential rate can now be de ned for all ®¯, and exhibit a standard Breit-Wigner shape, like for example
2.4 Application to Other Processes 41
for °± ² ³´µ and ¶± ² ³´³·:
The resonance appears at ¸ ² ¹ º°±, while a sharp increase appears for ¸ ² °±. For low energy, the spectrum
vanishes as ¸» . If ¼½ increases, the Breit-Wigner peak will decrease and spread, and in the limit ¼½ ¾ ¿, the
spectrum is again (2.30).
In the physical case of the positronium, the above treatment could apply to the computation of the para-
positronium contribution to the orthopositronium lifetime. Indeed, the mass of the parapositronium is less than
that of the orthopositronium. The parapositronium width has to be taken into account. Another application could
be the quantitative estimation of the scale anomaly contribution to the orthopositronium width. Indeed, this
anomaly should show up in the scalar channel. This analysis has not yet been done, but is an interesting perspec-
tive because it is obvious that the standard factorized approach will miss such contributions. We will encounter
again the scale anomaly in the next section.
À
Finally, let us emphasize that the present model is not designed to represent the physics of the analytical
problem of standard positronium computation. That problem has to do with the de nition of the initial state, and
needs a loop model with a Coulomb wavefunction form factor (i.e., containing a Coulomb photon resummation).
The possible recombination of the electron and positron into a parapositronium, after having emitted a soft
photon is a different problem, which should be treated using the technique of the present section. In other
words, if the Pirenne-Wheeler formula is used to compute the amplitude of the present resonance process, an
incorrect spectrum would be found for all values of ÁÂ and ÃÂ, since it is its treatment of the initial on-shell
electron-positron pair which is incorrect.
2.4 Application to Other Processes
À
In this nal section, we present several extensions of the method.
decay Ä-ÅÆ Ç ÈÉÊË and the weak decay Ì-ÍÎ Ï ÐÐ. Also, a connection to the scale
First, we present the application of the method involving the photon vacuum polarization. This includes the
Ñ
one-photon virtual
anomaly is brie y discussed.
ÒÓ ÔThen,
Õ Ö-×Ø
we accommodate the derivative formula (2.11) for radial excitations, by considering the pion decay
. This is also the Ùrst example of a production process.
The next application is a bit different, in that it concerns the spectroscopic observables. We compute the
hyperÙne splitting by taking double derivative of vacuum polarization graphs, and using the language of mass
renormalization.
The section ends with qualitative descriptions of other possible extensions, namely positronium, dimuonium
Ñ
high-energy formations and radiative transitions among positronium states. Finally, the extension of the method
to higher orders of perturbation theory is brie y discussed.
42 Chapter 2 A New Basis for QED Bound States
2.4.1 Orthodimuonium and Photon Vacuum Polarization
We now consider the decay Ú-ÛÜ Ý Þß Ý à
á àâ
The point-like amplitude is easily obtained in terms of the (divergent) photon vacuum polarization function
ãä åæ-çè é êëìíî ï ðñ òóô õóö÷øù ú ûøû ùüýþ ÿ
where
12345 7 6
' ( )* + ,- . /* + 0 = 89:;< D
"! $
# % & >?@ A B C (2.31)
@ E FGH G H
I , I the muon mass and J K L M N OPQRST U VWX YZ[\]^\ in dimensional regularization.
with
_ \ `The Coulomb form factor will be obtained from the derivative of the vacuum polarization, with respect to
ab : c
defg hijk l mnopqrstu v z {|}~ (2.32)
vwx y
¡¢ ¨ ¨ ²
£ ¤¥ ¦ §© « ¬ «
¦¯
« ¬ ±±±
§ ª ª® §© ª°
® °
This has the effect of removing the divergence, as it should. The decay rate is
³ ´µ · ¸ ¹º»¼½ ¾ ¿ ÅÆ ÍÎÏÐÑ ÒÓÔÕÖÖÔ
-¶ À ÁÂ Ã Ä Â
ÙÚ Á ÇÈ É ÊË ÌÌ ì öö
× Ø Û ÜÝ Þ ßà á â ãä å æç èè
éê ë
íï î ð ñó îò ööò
èè ë
ò ô õõõ
ú ÿ
To leading order in ÷ , we recover
the standard result ø ù -ûü ý þ þ when
. As a
function of the binding energy , the decay rate is a rapidly decreasing function (the plot is normalized to
one at )
2.4 Application to Other Processes 43
For electromagnetic bound states, the binding energy is related to the ne structure constant, hence the correc-
tions can be cast into
0 9 6
! " #$ -. / 36 7 8
- % & '( ) *+ , 12 345 : ; =<6 7 >>>;
AB
? @ C DE F GH Q R QSTUV W QSTXVY Z [[[\
I J KL M NO P
Compared to the binding energy corrections to the parapositronium two-photon decay rate, the present correc-
tions are much bigger. This is due to the high sensitivity of the form factor function near ] ^ _ (see picture
above). Note ` nally that the binding energy corrections obtained here, at order a, again account for a great deal
of the total corrections (computed using the standard factorized formalism), which are
tu
bcdefgh ij l m nonpq r s x y z{
-k v w | } ~
The last factor contains the radiative corrections to the intermediate photon propagator and to the nal
vertex, and is of no interest for our discussion.
Scale Anomaly
The decay rate is related to the derivative of the photon vacuum polarization function. In turn, this derivative
is related to the scale anomaly [207], [220]. This means that in principle, the decay rate receives a contribution
coming from the scale anomaly. As we will see, this contribution is subleading.
The anomalous Ward identity of interest to us is
¡ ¢£¤¥ ¦§¨ ©ª¨« ¬ ®¯¨
°±
where ²³ ³
´ ±µ¶±· ´¸¹º » ¼¸º¹½¾ ¿ À ÁÂÁÃÄÅÆÇ ÈÉ ÊË ÌÍÎÎ ÏÐÑ ÒÓ ÏÔÕ Ö× ØÙÕÚÛ ÙÜ
àáâã à
ä on the right-hand
and ÝÞ is the trace of the improved energy-momentum tensor. The additional term ß
Þ
side is the scale anomaly å it does not appear classically. Taking only that contribution into account, we get the
scale anomaly contribution to the Coulomb form factor
æçèéê
ë éìí îïðñ ò óôõö÷øùúû ýþ
ü ÿ
At the level of the decay rate, let us keep the dominant contribution coming from , and the anomaly
contribution as the only correction
!"-#$ % &'()* + ,- ./ 0 123/ :;<= AB=C HIJKI
> ? @D =EF G KK
NO . 45 678 99 f K
R S TU
L M P Q V W XY Z[\99 ] ab dc e ^ _ `
ij ]] ee
|
g h k lm n opq r st uvw xy z }{~ e
]
xx
x
of the complete binding energy correction.
Therefore, the scale anomaly accounts for
44 Chapter 2 A New Basis for QED Bound States
Note that if, for some reason, the anomalous contribution was to be omitted from binding energy corrections,
we would write
- ¹
¡¡¢£¤¥¦ §¨©ª «¬®¯° ±¯²³ ´µ¶·¸¸
¼½ Ô
À Á ÂÃ Ì Í Î
º » ¾ ¿ Ä Å ÆÇ È ÉÊ Ë Ï ÑÐ Ò ÓÓÓË
×Ø ËË ËË
Õ Ö Ù ÚÛ Ü ÝÞß à áâ ã äå æç è éêë ì íííî Ë
Ë
ææ îî
And this is exactly the result of the standard computation! The reason for this is rather obscure, probably
accidental.
Nevertheless, an interesting question arises: is the scale anomaly a true physical contribution to the decay
rate? In the derivative approach, the scale anomaly automatically shows up, while it is not clear how it could
arise in the NRQED standard approach. We take the point of view that the scale anomaly is not included in the
standard approaches. Furthermore, higher orders will never introduce it, so it is a truly new contribution. The
reason for this point of view is the following. The anomaly never shows up in the imaginary part of amplitudes.
The standard approach in the static limit amounts to consider the imaginary part at threshold (i.e. the process
is factorized), hence the anomaly will be missed at lowest order. Similarly, higher orders are computed by
introducing corrections to the wavefunction or scattering amplitudes, but the factorization is still used, i.e. one
is again working at the level of the imaginary parts. Since, at no stage in the NRQED process, the loop with its
Coulomb form factor is explicitly integrated, the scale anomaly will be missed.
ï
This does not answer the question, but just shows that our result is de nitely not attainable by NRQED
ï
analyses. At present, even if we do not see any compelling reason not to include the scale anomaly, none of the
two approaches is fully justi ed.
To conclude this section, note well that such scale anomaly contributions, if missed by current theoretical
computations,õcould introduce sizeable corrections in QCD. This is especially important since leptonic decays
like ðñ
ò ó ô ôö are very well measured, and have a close connection to the so-called ÷øù puzzle (see chapter
ú and û for additional details).
Orthopositronium Decay to Neutrinos
To explain the discrepancy between the experimental measurement and the theoretical value of the or-
thopositronium lifetime, one could invoke an invisible contribution to the width. Such undetected decay par-
ticles could be neutrinos, or new, beyond the standard model particles. Here we present the computation of the
decay width into an electron neutrino-antineutrino pair ü-ýþ ÿ , through a boson in the channel and a
in the channel
We will replace gauge boson propagators by and . Then, both point like amplitudes can be
expressed as
- ! "#$ %& '(( )' *+ ,-. / 0 ,1 2 /34 5 67849
2.4 Application to Other Processes 45
For the : boson exchange, this form emerges because the axial part of the : coupling does not contribute.
One essentially recovers the photon vacuum polarization as form factor, up to coupling constants
G I J K LMNO PQ R TU X YZ
;< ? @A B CD H VW
=> EF S [ \
` ab g h ijklmjn
with ]^ _ given by (2.31), cd ef and on p oq rst u n . For the v exchange, the loop
integration has to be done explicitly, using dimensional regularization (this is compulsory, in order to deal with
the quadratic divergences). After the loop integration, the Dirac structure wx yz { |}~ emerges and one can
factor out the form factor, again expressed in terms of the photon vacuum polarization
The Coulomb form factor is of course the same as before (see (2.32))
¡¢£¤¥¦
« ®¯°
¬ ¹ ¾¿ À Á¿ Ç
Ë
Á
§ · ¸º »¼ ½ ÂÁÃ Ä ÅÆ
Á Ç ÈÉÊ ÈÌ ÍÎ ÏÐ
± ² ³ ´µ¶
§¨© ª
Replacing ÑÒ ÓÔ by ÑÕÖ×Ø in ÑÙÒ Ú ÛÜÝ , summing both contributions, one get the lowest order positronium
decay amplitude Þ ßà-á â ã äåäåæ. From it, the decay rate is simply computed
ê
ç à ë ò ó ô õö÷ø ùú ûü ýþÿ
ß -á â ã äåäåæ è éïð ñ
ìíî ý
# % ) ) 34
&' ( ' 0 .
!"# $ * +, - . ,
+/ ,
* +1 - 222
( (
/ 1
At threshold, 5 6 78, we recover the result presented in the literature [108]. Obviously, this decay rate is very
small, of the order of 9:
;
-< = > ?@?@A NOPQ
B CD
-E F > GGGA H IJK L M
This is much too small to be of any interest.
For other neutrino Ravors, only the S exchange graph contributes, with the same result except for a minus
sign:
b efg v { }
T UV cd l m n opqr st uv w x ~
-W X Y Z[\] ^ [\] _ `a hij b k yz{ |
This amounts to a suppression of a factor of about relative to the electron neutrino case.
As a nal remark, we note that if the Fermi coupling had not been used, there is a delicate issue concerning
gauge invariance. Indeed, the general formalism states that to get the orthopositronium decay amplitude, one can
rst construct the equivalent process with a photon in place of the orthopositronium. For the exchange graph,
this is not the case since the orthopositronium cannot couple directly to the . Moreover, for the corresponding
photon graph, that coupling is important to preserve gauge invariance. Further studies should clarify the use of
the derivative method in weak interaction processes.
46 Chapter 2 A New Basis for QED Bound States
2.4.2 Pion Decay to Orthopositronium and Radial Excitations
The technique of taking the derivative of point-like amplitudes can be extended to other spherically symmetric
wavefunctions. The decay - is a good example [82]
To get a sensible theoretical prediction, one must sum the decay rates over the in nite tower of radial excitations
- in the nal state (- ¡ means th radial excitation of the -wave ¢ £ ¤ positronium state).
From the standard Schrödinger wavefunction for hydrogen states (see appendix C.1), we can write a general
expression for the form factor for radial excitations as
Í Î Ý
¥¦§¨
©ª« ¬¯® ° ± ²³´µ¶·¸¹¹º »¼½¼ ¾¿ À Á Ã ÄÅÆÇ
È É
Ï ÑÒ
È Ð ÓÔ ÑÕ Ò
ÙÚ
ÛÜ (2.34)
ÈÌ Ó Ö× Ø
ÉÊ Ë Õ
with Þ
ß ïð ñ òð õ
àáá â ã äåæçæ
èéê óô
æ ëì í î
üýþ
Where we have denoted ö÷ the mass of the ø-ù ú û state. The hypergeometric
functions are essentially the
well-known Laguerre polynomials ((2.34) is equally valid for parapositronium).
For the case at hand, ÿ is the photon vacuum polarization (2.31), where refers now to the electron mass.
The pion decay rate into orthopositronium states can be written as
- "# $%&'()*+ ,-/. 0 1
. 23 4 576 > ? @ A BC D
! $ 89 E
1 : ;< = FC GH IJ
OPQ
The \KDC LEFC GH M N corrections arise from the form factor for pion to two photons. The mass ratio RST UVWS X
YZ[
is also negligible compared to binding energy corrections, to which we now turn. Up to corrections of order
]^ ef
_ , we can write using ` _ a bcd
q
g w x yz {
hi jklmn o pr tu
s v {}
|
{ ~
(the are the numerical coef cients found by expanding ¡). Summing over ¢, we £nd
°
¤ µ Ã
¨ ©ª«¬ ® ¯± ² ³´ ³¶ · ¸¹ºº»¼ ½ ¸¹º¾¿ À ¸ÁÂ
¥ ¦§
Å ÆÇ È ÉÊ ÌÊÍ Î ÊÏÐÑ Ò ØÙ
where Ä Ë is obtained from the contribution of the Ó- Ô Õ Ö× only. For comparison, [82]
ÝÞßÙ
found the radiative correction to be Ö× Ú ÛÜ . The experimentally quoted branching fraction is [98]
à áâã ä åæ çè
ö ø ù úûüý þ ÿú
é êëì í îîï ð ÷
ðñòó ô õ
ð
ð
so the agreement is good. ð
What this little exercise shows is the power of our method as a mean to partially compute higher order
corrections. More importantly, it is by now apparent that other, more complicated wavefunctions can easily
be accommodated for. Any wavefunction that can be expressed, even approximately, from derivatives of the
2.4 Application to Other Processes 47
Coulomb one can t in our scheme. This may open the way to many applications in QCD, as we will see in
chapter 4.
2.4.3 Hyper ne Splitting and Mass Renormalization
We implement here the renormalization of the positronium state. The idea is to consider the bare positronium
mass as equal to twice the electron mass , and then to carry a ( nite) mass renormalization. The dia-
grams contributing, at lowest order, to this mass shift will be obtained as the second derivatives of the vacuum
polarization loops:
Let us rst obtain the mass renormalization equations, and then discuss the results numerically.
Mass Renormalization
For a pseudoscalar parapositronium state, the resummation of the Dyson series is trivial. The parapositron-
ium propagator is then
Of interest to us is the mass shift, de ned from the pole of
?@ABCDEFEG
#$%$ & '() * +,-.- /0123 4 56789:9 ; 5< = >89:9 HI
!" (2.35)
J
For the orthopositronium, the transverse part of the bare propagator is
U VWXY _
KNLM OPQR S T T Z[\]
`a b cad e\^fg
The self-energy of the vector positronium will be of the form
hij klmn o pqrstu v wtwux yz{|}z ~
Proceeding with the resummation of the Dyson series, we end up with
¡¡ ¢ £ ¤
Again, the propagator pole is at ¥¦ § ¨¦©, hence
¼½ ÁÂÃÄÅÆÇÄ ÍÎ
¨¦©ª¤«¬¤ ® ¯°± ² ³µ́¶·¸¹º· » ¾¿ À ÁÂÃÄÅÆÇÄ È ÁÄ É Ê ËÄÅÆÇÄ Ì ÏÐÑÒÓÔÑÕ (2.36)
48 Chapter 2 A New Basis for QED Bound States
The two equations (2.35) and (2.36) are self-consistent equations. By identifying Ö× Ø ÙÚ and ÖÛÜÝÞßÞà×ßáâ×ã Ø
ÖÝÞßÞà×ßáâ×ã, they are of the form
ä ïðñð ÷øùúûüûý
æ ì ë ê í îïðñð ò óô õ ö þÿ
å èéê
ç
å
%
!"# $ &!"# '
%() )*+,-.+/ by plugging the non-relativistic result 10 2
Our approach will be to Agenerate corrections to 0 !
345678 into 9:;<;=><?@> .
B
Second Derivatives and Hyper ne Splitting
The double derivatives of the vacuum polarization loops are
C < ><?@>A DEFG H JI KLJMNOPQRS UT V W Y [\]\^_^`a_bcad efgh
:; ;=
jk k WXY Z jl p
where a factor of i corrects for the factor appearing in the derivative of i m n opq , while the other
accounts for the double counting of Coulomb photon exchanges.
The computations are straightforward. For parapositronium, the point-like quadratic divergence disappears
and we rnd stu u tu u z { tu u
w x y v | } ~ v
v
§¥¦¦ ¨ ©
¡¢ £ ª¤«¬
¬ ® ¯°±²±³´µ´
½  ¾¾ÅÆÇÈ
¶ ·¸¹º» · ¼¸¹ ¾¿À Á ¸¹ÃÄ · ÉÊËÌ Í ÎÎÎ
Ð Ó Ô ÕÖ×ÖØÙÚ, ÛÜ Ý ÕÜ Þ ßÜàáâáãä and åàáâá æ çè é èêëìí in the right hand
Where we have used Ï ÑÒ
side. Proceeding similarly with the orthopositronium renormalization, we consider the second derivative of the
photon vacuum polarization function:
óôõö ÷ øùúûüù ý þÿ øùúûüù
îïðñòï
þ
ùúûüù (2.37)
" #$%&#' )(* 0 1234 756 :; 75<
! +,- . /
<75666 89
89 <
=>?>@ABC@
H L M P NQSN T
D EFG I J K NFGO E KFG QR J ONUV FG W XXX
Z K
At the order Y , there is also the annihilation diagram
\]]^_` fgh i jk lmnopq r sps qt uvwxyz {|}~~} as given by
whose contribution is [abcda e . Using
(2.32), the dominant contribution in the limit
is
¡ £¤¤ © ª«¬®« «¬®« ½
¥¦§¨¢ ¯ °±²² ³ª´µ¶·¸µ¹ º »¼ ¾ ¿ ÀÀÀ
2.4 Application to Other Processes 49
Á Ã
Hence the lowest loop contributions to the hyper ne splitting are, to order  :
Æ Ù éê ï ð ñòó ôõö÷øùø ú öööû
ÄÅÆÇ È ÉÊËÌÍ
ÎÏÐÑË Ò ÓÔÕÖ×ÔØ ã âá ä åæç èëì í îîî
ÛÏÜÑ ÚÚ Ý Þßàá
This is to be compared to the result (see chapter 1)
üýþÿ
"#$ &' 001 1 8 8;
!
#%
()*+,-
! . / 1)),- . 02 34 5 6 7
859 : 2
Q Z [ \ ] _`a
<=>
?> @ AB C DE FG E H I JK LB FG M N OP
RS T UV WX Y ^
b cde fghijkk l ghkmkkn lghogjknp qr n l ghkmojnp l hhhs
It is not surprising that our method does not reproduce exactly the above result, because we have neglected
for example), and
many diagrams (like electron self-energy
t u v w xyz{|insertions }~
as a basis. Taken individually,
because we used the lowest order non-
relativistic binding energy and wavefunction
the corrections of to the para- and orthopositronium masses are off by more than . On the other hand,
the difference between both corrections, giving the hyper ne splitting, is surprisingly good. Again, it seems that
contained in the lowest order loop (with all the ladder Coulomb photon exchanges in the form
all the effects
factor) suf ce to account for most of the radiative corrections. In conclusion, further studies of the application
of our method to hyper ne splitting appear as necessary.
2.4.4 Possible Extensions
We have seen that the parapositronium and orthopositronium are easily dealt with using the derivative formula
(2.11). It should not be dif cult to extend the formalism to transitions among positronium states
Usually, such transitions are computed using the quantum mechanical formalism for magnetic dipole transitions.
Here, we just have to compute a Feynman diagram, in a formally relativistic framework. In any case, the
transition - - is very small compared to the decay - .
Another simple extension is the description of positronium (or dimuonium) production processes, like for
example
50 Chapter 2 A New Basis for QED Bound States
The last graph is to be understood as a sub-process (see the production mechanism discussed in
the previous section). Those processes are interesting because they constitute a theoretical laboratory for the
quarkonium production case.
Still in connection with the corresponding QCD case, the description of higher excited states should be
made possible using our formalism. Naively, we can expect that their decay amplitude will be computed from
point-like amplitude with a well-chosen Dirac structure
£ ¡ ¢
¥¦¥ ¡§ ¤
¨¤©ª
«¦¬ §
«¦¥ § © ¤©ª
«¦® § ¨¤©¯ ° ¨¯© ¤ ° ±² ³´ µ¶·¸¹ º »¼»½
ÀÀÀ »¾ ¿
Á Á Å
These identi cations suf ce to enforce the correct selection rules (like for example ÂÃ Ã Ç ÈÈ). To be able
ÂÄ Â Æ
to extend the derivative formalism, we also need to know the wavefunctions. Those are given in appendix C.1,
but not in a very convenient form. In particular, it is not clear at present how to deal with the angular part of
those wavefunction (the É-state was especially simple, since spherically symmetric).
The Ê nal extension we want to mention is the hydrogen atom and the muonium. In other words, the case
of unequal constituent masses. The wavefunctions are the same as that of the positronium (except for the re-
ËÌÍÎÏÐmass).
duced
Ñ ÒÓÒÔAnyway, for unequal masses, the bound state does not usually annihilate. The only exception is
. Description of this process using the present formalism will be undertaken in the near future.
2.4.5 Higher Orders
At Õrst sight, the extension to higher orders is straightforward. It sufÕces to take derivative of two-loop point-like
QED amplitudes. This will regenerate the tree-level result, introduce all the corrections of order Ö, along with
corrections of higher orders. Logarithmic corrections will also be automatically generated from the threshold
singularities of the two-loop (or higher) graphs. For example, the following graphs are sufÕcient to get the
complete decay rate up to order Ö
the ×
ØÙÚlevel
The tree result is regenerated because of the Coulomb part of the binding graph (photon exchange between
). This is due to the iterative nature of the Bethe-Salpeter equation (see (1.11)). Note that the Coulomb
part of the binding graph also generates order Û corrections, as exposed in [87].
Unfortunately, things are less simple because precisely of the binding graph. To see what happens, let us
apply our derivative formula on the two-loop photon vacuum polarization graph (see appendix A.2), evaluated
2.4 Application to Other Processes 51
at the positronium mass. We Ünd the Coulomb form factor
ààá åæçè é êëì
ÝÞâßàãä î í ïð ñ òó ñõ
ô öó÷ øùú ó û ü ýþÿ
The rst term, 2, is the signal that something has gone wrong. We would have expected a instead. The problem
is simple to understand. When one simply takes a derivative, the form factor is introduced inside the cut, hence
we generate
This is not correct it is a double counting. Those diagrams are responsible for regenerating twice the lowest
order. Instead, we should consider only one of them, or proceed rst by inserting the Coulomb form factor inside
the imaginary part. Note that, if naively divided by two, the two-loop corrections become
)
!" # $ %&'(
78 XY
* + ,-- ./ * 01234 5 6 9 :; < => G H IJ H L PQ UVW Z[ \]
? @ AB C DE F K M JN O J R S T ^_
which agrees to order ] with the standard result. Future work should clarify the use of the method at higher
orders.
In any case, it should be obvious that the extension to higher order is possible ina principle. The present dif-
`
culty is only technical. Of course, we are still a long way from an exact order ] computation, necessary to
compete with other computation
b
schemes. This was not our objective. Instead, we think that we have success-
fully constructed a well-de ned perturbation theory for QED bound state computations.
52 Chapter 2 A New Basis for QED Bound States
2.5 Conclusion 53
2.5 Conclusion
In summary, we have analyzed the QED bound state decay formalism, using dispersion relations. The two prin-
cipal results were
1- The approximation done in standard approach to reduce the four-dimensional Bethe-Salpeter loop is
essentially a neglect of all but the vertical cut of the loop.
2- The four-dimensional loop amplitude, with the Coulomb form factor inserted at the level of the disper-
sive integral, and with a covariant description of the bound state spin, is obtained as the derivative of the
corresponding point-like bound state QED amplitude.
Since all the cuts are needed to preserve analyticity, and since a simple computational scheme is available,
we therefore conclude that QED bound state computations must be developed from a new perturbative basis.
Lowest order amplitudes constructed in this way have many properties
1- They are simple to compute, in a formally relativistic framework.
2- They are automatically gauge invariant.
c
3- They sum an in nite class of correction already at the lowest order.
4- They factorize the binding energy effects from the radiative corrections in the perturbative series, lead-
ing to enhanced convergence.
5- They predict correct photon spectra, i.e. spectra in agreement with Low ds theorem. This is impossible
using a perturbative NRQED approach.
e
While the main motivation for constructing a new basis was the con ict between NRQED and analyticity
in three-body decays, the method can be applied (or will be extended) to a variety of other processes (higher
c
excited states, hyper ne splitting, radiative transitions,...).
Also, our method apply to lowest order amplitudes, while present theoretical considerations in positronium
physics are at two loops and beyond. As said, our goal was to implement analyticity in bound state computations,
so it is natural to start with the lowest order. Anyway, to be competitive, our method should have to be applied
at two loops now. As we have seen in the previous section, there is no obstruction to this extension, at least in
principle.
The simplicity of the method may be accidental, but could also be the signal of some more fundamental
physics involving Coulomb bound states. For example, the fact that the derivative of the photon vacuum po-
larization contains all the necessary information about the positronium binding is very interesting. Further, the
appearance of the scale anomaly as a binding energy effect needs further study.
Chapter 3
Overview of Quarkonium Theory
This chapter presents the observables of quarkonium physics, as well as the theoretical description of quarko-
nium decays. It begins with a brief summary of the spectroscopy of heavy quark-antiquark bound states, and of
the annihilation modes of the quarkonium ground states. Then, the standard theoretical framework for the de-
scription of quarkonium annihilation is developed from the positronium annihilation one. Finally, the theoretical
predictions are compared to the experimental data, and the strong coupling extraction is performed. All the ma-
terial of this chapter is standard, and just serves to set the stage for the next chapter, where binding energy effects
will be introduced as a cure for the standard approach analytical defects.
This chapter is not intended to present the history of quarkonium physics. Only the notions of interest to us
are introduced. Also, the modern framework of quarkonium theory, i.e. non-relativistic QCD (NRQCD), will
not be discussed. This omission has no consequence, the simple computation scheme presented in this chapter
is a good illustration of more advanced techniques. In particular, the basic factorization hypothesis is central to
all models. Further, the formulas we will derive are still used in many practical applications. For reviews on
NRQCD, we refer to [93], [171], [176], [178], [179], [188], [189], [190], [192] and references cited there.
3.1 Quarkonium Physics Observables
3.1.1 Heavy Quarkonium Spectroscopy
gh
In the quark model, a meson is made of a quark
i
with an antiquark. Since the quarks are spin f fermions, their
association into
j
a meson can take only speci c quantum numbers. The following table summarizes the naming
scheme for un avored mesons [223]
p rp r tpp
q p sp o q oo rpp
kl m no
o. .. o oo
. ..
. .
uvpwxy z{x |}. |~y .
z{x y {x |}. |~y .
{x y
¡ ¢£
56 Chapter 3 Overview of Quarkonium Theory
Links between quantum numbers are ¤ ¥ ¦
©ª«¬
§¨
® ¯ ±²³´µ
°
¶ · ¸ º»¼½¾½¿
¹
ÃÃÄ
with À only relevant for neutral mesons. The ÁÁ Â mesons are collectively called charmonium (bottomonium
or, less often, beautyonium). Ë
Each ÅÆÇÈÉÊ state appears with all its radial excitations, as an in nite tower of states. The particles Ì names
receive an extension ÍÎÏÐ with Î the analog of the principal quantum number. For charmonium, the experimen-
tally observed states are [223]
ÑÒÓÔÕÖ Õ × ØÙ Ú ÛÜÝ ÛÜÞ Û Ü ßàáà âââ
Ô× æ çèéêëì íîïðïñ òè éîëì íóôïõñ
ã ä ä äå
ö÷ø ù ùúú ûüý þù÷ÿ ý þ
!"
#$ $ , -./ 01
% & '(( )*% + & &
# $2 2 $ , -.3 1
&(( )* + & &'
#$ 66 78 ; < =>??@A
4 5 4 9:
BC G
4 5 D EF
H IJ K K NOO P Q KR P
FF L M D L M D
HI NTUVW
S EFF LS M
XY UZZ
X
[ ]]
The identi cation of higher \ states as ^ _` radial excitations or a b` states is not well-established. The
ee i
occurrence of b-states around cd MeV is supported by the indications for the a bf state g f h bj, kll, at
mnmo mrrst
MeV, so that p q is probably the uvwt . Another possibility is that the xu vw and yuzw states are mixed
monot mrrs
into the mass eigenstates p q and p q . For the tower of u zw radial
} ~
excitations, a remarkable feature is
the equal spacing between {-levels (compare with the Balmer series, in | { ). Also apparent on the gure, the
ne and hyperne splitting are sizeable compared to the spacing between principal levels. Anyway, all energy
spacings are small compared to the bound state mass.
3.1 Quarkonium Physics Observables 57
For bottomonium, fewer states have already been observed [223]
¤¥¦ §¨©©¤ª ª¥¦ §¨©ª«« ¥¦ §¨©«®©
¡ ¢£ ¡ ¢£ ¡ ¬ ¢£ ¡ ¬ ¯¯¯
°± ¨ © ¨²³
°
´µ ¿ À ÁÂÃÄÅÆ Ç È¿ À Á ÅÈÉÈÆ
¶ · ¸¹¹ º»¼ ½¾ »¼ ½ ¾
Ê ¿Ë ÌÍÍ ÎÏ Ë ÐÌÑ Ò ÓÔÕÔÖ× ÎÏË ÐÖÑ Ò ÓÌØÖÙÙ×
Ú ÑÛ ÖÍÍ ÎÏÛ ÐÌÑ Ò ÓÔÔÌÜÝ Þß âã ä åæçâèéÝ
àá
æêê ìíîì ìïì ð
Some higher states have also been observed at ë and ëë MeV. As for charmonium, the identi cation
of those states as ñ òó or ô õó is not straightforward.
÷÷
For both charmonium and bottomonium, the states with quantum number ö are the most well-established.
ù ÷
This is simple to understand from the experimental setting used to detect such resonances, i.e. ø ø colliders.
A resonance with the quantum number of the photon is easily found in the ú-channel
The discovery of the û and ü family was made while studying such channels. The establishment of ý states
is relatively easy because of the þ ö transition from ô òó states. On the other hand, the singlet states are rather
ÿ
dif cult to observe, because the transition is very weak.
Potential Models
For QED bound states, a non-relativistic approach based on the Schrödinger equation with the Coulomb
potential gave rather good results for the spectrum. For quarkonium, we do not know the potential. One could
think of the analog of the QED Coulomb potential , which is for QCD
58 Chapter 3 Overview of Quarkonium Theory
with ! " #$%&' . This effective potential arises from the graph
(
(
Unfortunately, such a potential does not support con nement since there is a continuum of states (and, as we
(
have said, the Balmer series does not t well the observed spectrum).
(
To account for the con nement, various phenomenological potentials have been proposed, whose parameters
are tted to reproduce the observed spectrum7 (see [150], [154], [162], [170], [199]). The most popular one is
the Cornell potential
)*+,-.// 012 3 456 7 86 :9 <= > ? AB CDEF<G > HIJK LMN
O P QIHR@
Fit Parameters:
; S P QIJK LMN T (3.1)
U
At short distance, the interaction is described by a Coulomb potential, while the linear rise is felt at larger
U VW X YZ[
distances, and is responsible for con nement (note that if the Coulomb part is matched to the QCD Coulomb
\
]^_`a bcdec `a bcd
interaction, one nds , but the scale is not clear). The parameter is called the string tension. More
advanced forms have also been proposed, in which the Coulomb part is taken as , with the
running coupling.
Other forms have been proposed, with no connection to the perturbative QCD potential. Among them, let
us quote
yz }~
}
fghijk lmi nop q r s t nouovwx {
Fit Parameters:
z|
²³ ´ µ¶· ¸¹º» ²¼ ´ ½¶¾¿À ¸¹º
¡ ¢£¤ ¥ ¦ § ¨ ©ª« ¢£¬£® Fit Parameters: °¯ ÁË Ì´ÃÄÍÆÆ ÂÃÄÅÅÆÇ ÈÊ
ÈÉÊ É
±
where ÎÏ Ð ÑÒÓ
Ô ÕÖ is introduced for dimensional reasons.
The above potentials are quite successful in describing the spin-independent part of the spectrum. To de-
scribe the ×ne and hyper×ne structure, QCD inspired potentials have been constructed (accounting for tensor
forces, spin-orbit couplings,...). In general, the technique amounts to the non-relativistic reduction of amplitudes
to effective potentials. Various complicated theoretical issues are involved in those studies, and therefore lie out
of our main concerns (see for example [11], [18], [159], [187], [195], [196]).
3.1.2 Heavy Quarkonium Decay
Similarly to positronium, quarkonium decays either through the annihilation of its constituents or through tran-
sitions to lower states. In this section, the coarse pattern of experimental branching ratios is presented. Unfortu-
nately, this is only possible for charmonium, because there is not enough experimental information concerning
bottomonium. The coarse pattern should however be quite similar for both (the only new feature is the opening
of the ØÙØÚ channel in bottomonium decay).
Û Ü
In some cases, the observed decay rates, especially the leptonic modes, also serve as input for the ts.
3.1 Quarkonium Physics Observables 59
Ý
In the following, only -states will be analyzed
Þß àáâáãä å æãä Width
Mass (MeV)
æçãà å(MeV)
çãè
éêë çìíîïðñ ò ìïìó ìïìðñ ò ìïììô
õ ö÷øù úîðôïíî ò ìïìí ìï÷ññ ò ìïìúû
ýõ õ ö÷øù due to the Zweig rule, as will be discussed below. Note that the
The very small width of the ü or
þÿ width is not well-established, since recentisreported measurements range from
MeV [141] to
MeV [138].
The
Decay
The decay modes of interest to us are [223]
! "#$%& ' (
The ) is a pseudoscalar, like the parapositronium, and therefore can decay only into an even number of photons,
or into at least two gluons. The branchings can then be understood simply in terms of the processes
The emitted gluons, being colored, hadronize to form the *nal state hadrons. The relative strength of both
processes is in agreement with the rough theoretical expectation
+ ,-. / 001 89 @AB
2 345 / 661 7 :9; < = > ?
H
with CD E FG .
The I Decay
L
For JK , the main branchings are [223]
M N L O PQ P R
JK TUV WXYZ [ \]^\
Leptonic S
_`_a bcdd e fcgf
j p stut v wux
Hadronic hi klmn o qr
qy z {u|} ~ }
}
Transitions
60 Chapter 3 Overview of Quarkonium Theory
Note that the virtual photon transition to light hadrons (L.H.), as well as the transition to are included into the
total hadronic width.
As for the orthopositronium, the can decay into an odd number of photons. The simplest decay mode
is therefore the one virtual photon mode (both the photon and the have quantum numbers )
The leptonic modes, as well as roughly of the hadronic mode are of that type.
Since gluons are colored objects, the corresponding one-gluon process is forbidden (the is colorless).
This is the origin of the small width of the , and is called the Zweig rule. Indeed, if the one-gluon mode
were allowed, it would have a width of more than a thousand times ( ) the one-photon one. Instead, the
dominant strong decay mode proceeds through three gluons
We have also indicated the not yet observed three photon mode, to emphasize the similarity to the orthopositro-
nium decay. A rough estimate shows that the three-gluon decay branching is of the same order of magnitude as
the one-photon one
¡¢ £ ¤¤¤¥ ²
³ ¹
´µ ¶ ·¸
¦ ¡¢ £ § ¨ £ ©ª«¬ ® ¯° ±
¯
º »
where we have included a factor · · to account for the three-body phase-space suppression factor, like for
orthopositronium decay.
Finally, the radiative transition can be depicted
It is the analog of the magnetic transition ¼-½ ¾ ¿ À-½ ¾ Á Â, and is therefore quite small.
3.1 Quarkonium Physics Observables 61
ÅÆÇ
Finally, for the radial excitation Ã Ä [223]
È ÅÆÇ É ÊÇ Ê Ë
ÄÃ Ä Í ÑÒ Ó
Leptonic Ì ÌÎ ÏÐ ÏÐÏÔ
ÕÖ×Ø ÙÚÛ Ü ÝÞß
Hadronic
è
àáâãäåæ ç éê ëìÞí Ü ÝÞî
é ï ð ñòóô õö÷ø ùôú û üýþ
üý üýü
Transitions ÿ
! " #$%
&'() *+, - . +/
Again, the virtual photon mode and all the transitions are included into the hadronic mode. The transition
01 2 31
cannot proceed by <emission of one photon (Furry4s theorem). Instead, the bulk of it involves a
789 : ;
two-pion emission 5 6 5 ==. Those transitions proceed through two-gluon emission
Note also that the allowed transitions to >-states are roughly 30 times stronger than the forbidden one to ?@A.
3.1.3 Other Quarkonium Physics Observables
Exclusive Rates into Hadrons
F F GHIJ
Many exclusive decay modes of the BAC DE and into hadrons have been measured [223]. We do not
intend to give an account of them, because they are of no direct interest to us in the present work. Indeed, it is not
possible to relate exclusive decay modes to the decay modes into gluons described above, since a quantitative
IL GHJ
description of the hadronization process does not exist at present. Note however that using the Kavor or
IL GMJ
group, one can extract interesting information, such as for example the relative strength (as well as the
relative phase) of the one-photon electromagnetic and three-gluon strong decay modes contributing to a given
F N OPQRST U VWPXYSWQZ[ZT
multiplet of decay channels (like DE for example), see [169], [184]. Again, there
is no clear quantitative way to relate such information to the perturbative picture of the heavy quark annihilation
channels, and we will not detail these analyses.
Quarkonium Production
In proton-proton colliders, or heavy ion colliders, one can search for the formation of charmonium states.
There are many different mechanisms that can lead to the formation of a charm-anticharm pair (Drell-Yan, Gluon
fusion,...). The interest of charmonium production has been revived by data from HERA and the TEVATRON
P ^
on charmonium production in high energy V and VV collisions. Also, observation of \] suppression by NA38
62 Chapter 3 Overview of Quarkonium Theory
in nucleus-nucleus collisions raises the possibility to explore the phase structure of QCD. In particular, the
a e
suppression of _` production in b c d b c collision is sometimes interpreted as a signal of decon nement. In
a different context, the charmonium decay channels of the f mesons have many interesting properties for the
aij
study of gb violation (f h _` permits the extraction of the unitary triangle angle k in an especially clean
way (no FSI) [18]).
It is in describing quarkonium formation that NRQCD has found one of its greatest successes. Indeed,
the simple color singlet model (CSM) is in strong disagreement with experimental data (the predictions of the
CSM underestimate the production rates). Let us recall that the CSM model simply assumes that the process
of quarkonium formation factorizes as the creation of a free colorless ll pair (thus in a color singlet state)
followed by its binding into a quarkonium state. NRQCD, on the other hand, led to the so-called color octet
mechanism. The production mechanism receives a signimcant contribution from processes where the ll pair is
created in an octet state (typically, by a single gluon). In other words, one introduces dynamical gluons into the
quarkonium wavefunction. A further success of NRQCD is that, by invoking color octet contribution, one is able
to get rid of IR singularities in n -wave charmonium production, for which the CSM is rather non-predictive.
For further information on the quarkonium production models, we refer to the abundant recent literature (see
for example [167], [171], [198] and references cited there).
3.2 Quarkonium Annihilation Rates
In this section, we collect all the annihilation rates of heavy quarkonia ll into photons and gluons calculated
in the appendix B.2, and include the radiative corrections as obtained in the literature. The framework is that
of perturbative QCD. There are many references. Some early works are [144], [145], [146]. For more mod-
ern approaches, see for example [11], [147], [149], [151], [153], [164], [167], [191]. Others are listed in the
bibliography, or can be found in the bibliography of the cited papers.
The central hypothesis used to compute annihilation rates is essentially the static limit, i.e. the bound state
mass is taken to equal twice the constituent mass. The basic formula is then the old Pirenne-Wheeler positronium
one op w
| }~
n q r stu v xy z w { u{
adapted to the quarkonium case. One therefore assumes that the quarkonium wavefunction, in momentum space,
satis es
¡¢
up to a constant. We have discussed at length the limitation of that formula in the QED case. For instance,
oblique cuts are completely missed, and predicted spectra will be in contradiction with analyticity.
The raison d£être of this section is its status in current quarkonium physics (see for example the introductory
review in [223]). Our goal is to exhibit various problems occurring in the comparison to experimental data.
3.2.1 Decay Rate Formula and Selection Rules
The ¤rst thing to do is to translate the basic formula for decay rates to the case of quarkonium. Indeed, color
factors will appear. The difference with the pure QED case is the non-abelian nature of QCD, which introduces
multi-gluon vertices. Also, the basic selection rules are no longer valid. A pseudoscalar quarkonium can decay
into three gluons, even without any multi-gluon vertex. The fact that such a decay is not identically zero for
gluons is due to color degrees of freedom: there are eight different QCD gauge bosons. Indeed, the three-
3.2 Quarkonium Annihilation Rates 63
photon decay is impossible due to the compulsory symmetrization of the six possible permutations of photon
insertions in the quark current. For gluons, one can antisymmetrize such insertions, provided the color degrees
of freedom are also antisymmetrized. With these remarks in mind, we can proceed and write down the general
decay amplitude for a quarkonium.
For a quark-antiquark bound state, the state vector is expressed as
¥¦§¨©ª« ¬ ® ·¸ ¸
¥ ¯° ± · ¹ ¹º
²³´³µ¶
with » the number of colors. The amplitude is then calculated as
¼ ½¾¿ÀÁÂÃ Ä ÅÆ Ç È ÎÏÐ ÑÒ ÓÔÕ Ò ÓÔÕ Ö ×ØÙ
ÉÊ Ë Ì Í
with ÚÛÜ
ã
Ý Þ ßà á
äåâ æç è éêëì í î ï ð óôñò õ
î
and the static limit . The trace is carried over both the Dirac space and the color degrees of
freedom. The mass ï öis÷ñøùøùøùú
the mass of the constituents, i.e. the heavy quark mass. As we have said, the present
ô
formalism assumes ñò that the binding energy is negligible, i.e. û . For electromagnetic bound states
this was a good approximation (the difference being of order üý), but ö for
ñòquarkonia, the experimental spectrum
shows that binding energy can be a sizeable fraction of the mass þ. The philosophy we will follow is, whenever
possible, to express our results in terms of þ rather than ÿ , especially for the phase-space boundaries.
In conclusion, the master formulas for quarkonium annihilation rates are
,-.
! "#$% &'' ( )*+ -
/ 012 3445 6 785 9 : : DE B F GHI J OP QRST UVV W XYZ[\\]
;< = : >? AB
@ C C
KLM NN
`
with ^ _ a the phase-space integration.
Remarks
1) Remember to include a charge factor for each quark coupling to the photon.
2) Note that in each case, the bnal polarization of a gluon or photon can be replaced with more
complicated expressions. For example, the decay amplitude into cdd is obtained with the substitution
pqr
efghi jklm n ost x yz{ |}~
u vw
and so on. Of course, the trace in the expression of does not apply to those factors.
They are summed over when squaring the amplitude. Remember that if it is a three-gluon vertex which
replaces an external gluon polarization, the corresponding processes with ghost states must be included.
Such states need not be considered when external gluons are attached to the quark current since ghosts
do not couple to matter elds.
3) Decay rates into photons will always be expressible
in terms of the corresponding positronium
decay rates by a very simple substitution and a factor . For decays into gluons, some
amplitudes will collapse to the corresponding photon ones, but with more complicated color factors:
¡¢£ ¤¥¥ ¦ §¨©ª « ¬ ²³´ µ¶¶ · ¸¹º»
Color Factor ®¯° ±
Anyway, there is a lot of possible gluonic channels which do not exist for photons, since selection rules
64 Chapter 3 Overview of Quarkonium Theory
¼
are different and since QCD is not abelian. As a result, speci c decay amplitudes are very different in
general.
Inclusive Rate and Duality
The basic decay rate into gluons has somehow to be translated into observables. Gluons never appear as
asymptotic states, but hadrons are observed instead. One usually relies on duality concepts to relate the gluonic
decay rate to the inclusive hadronic rate. Such a procedure is only an approximation. One way to view such a
¼
change of basis is as follows. One can view the hadronization as a nal state interaction, so our presentation is
inspired from Watson ½s theorem. Let us analyze the decay ¾¿ into hadrons. The gluonic and hadronic basis are
related as
À ÃÃ Å ÔÕÖ× Ý á ää å æçèé êê ë ìíîï êê ë ìíîð ñ ñ ñ ó ö ùúûü ÿ
Á ÃÃÃ Æ Ò ØÙÚÛ Þ â êêê ë ìíîé êêê ë ìíîï êêê ë ìíîð ñ ñ ñ ô÷ ùúûý ô
Á
Á ÃÄÄ Æ ÊËÌÍÎÏÐ Ò
Ñ ØÙÚÜ Þ â êòò ë ìíîé êòò ë ìíîï êòò ë ìíîð ñ ñ ñ ô÷ ùúûþ ô
 ÇÆÈ É Ò
Ó ßÞà â
ã õô÷
ø õô
.. .. .. .. .. .. ..
. . . . . . .
(3.2)
which leads to
ö $
÷ !" %
÷
÷
!#
%
% (3.3)
ø
&
.. ..
. .
It is important to remark that in these formulas, the gluonic states are viewed as asymptotic. As soon as one
is considering asymptotic gluons, one can introduce real gluonic states in intermediate states (this is due to the
optical theorem). Therefore, we will make the additional assumption that an exclusive decay into hadrons is
factorized into a gluonic decay followed by a hadronization
' transition from gluons into hadrons. This means that
one is neglecting non-absorptive effects in a speci c exclusive decay channel.
The change of basis in (3.2) is made through ()*+,-./ , the transition 0-matrix from gluons to hadrons. Such
a matrix is taken as unitary, in 1rst approximation. In the context of 1nal state interactions, this hypothesis was
called elasticity, while in the present context, we will speak of duality. This denomination is justi1ed since from
(3.3), we directly obtain
234 5 66 27 8 9:; < === >? @ ABC D EFF AG H I I I J KLM N OPQR ST U VWX Y Z[\] ^] _ ^`a b cdef gh i j j j
Note that to leading order in kl , only the mrst term no p qq contributes.
3.2 Quarkonium Annihilation Rates 65
3.2.2 Pseudoscalar Quarkonium Decay
For pseudoscalar quarkonium decay, rates are [155]
Lowest Order Decay Widths Radiative Corrections
rst u vv wxyz{|}~
¢£ ¤
¡¥ ¤¡ ¦§
¨ ©ª
º ¼
«¬£ ®¯°¯± µ »
²³´¶·¸ ¹½ ¼¹ ¾ ¿ÀÁÂ not known
Ê ÐÑ Ò
ÃÄÅ Æ ÇÈÈ ÉË ÌÍÎ ÏÓ ÒÏ Ô ÕÖ×Ø is a correction to ÙÚÛ Ü ÝÝ
where Þß is the heavy quark charge, in units of Þ, à is the quarkonium mass and
óú
ì íì ò ó ô õ ù ù
þ
á âãä å æ é ê ë ð ûüýûü
ÿ
ö ÷
ø
çè íìî ïíìî ð ñ
with the heavy quark mass (
here) and is given below. Note that these rates are often given in
terms of the radial wavefunction, which is related to the wavefunction by the spherical harmonic normalization
Radiative Corrections to and Velocity Singularity
For all the decay modes, the quoted radiative corrections omit singular terms in !"# with !"# the relative
velocity of the quarks in the bound states. Such singularities cancel in ratio of decay modes. To be a bit more
explicit, let us consider the two-photon width. Including the radiative corrections, the rate is
$ %&'( ) **+ , -. /0 1 23 45 8 9 :;< ==
6 7
In? fact, this result is obtainedDfrom the Harris and Brown QED computation of the correction to the scattering
> >@ A BB with incoming C CE at rest (see chapter 1)
F GHI HJ K LLM N OP QRS RT U VVW XXY Z ]^_ [\ f g hi
` a b cd e j kk
m
This exhibits the well-known Coulomb singularity in l nopq of the binding graph. The Sommerfeld factor
r
prescription is to absorb such a divergence into the wavefunction. In QED, this is simple to understand since
the Schrödinger wavefunction is the in nite sum of ladder Coulomb exchange graphs. In QCD, we assume that
the same mechanism is at work, and absorb such singularities into the wavefunction [151]. Then, the QCD
correction is obtained with the substitution s t uv sw , since for each gluon loop insertion there is a factor
x }y{ {y~ ~} ~}
yz{ | |
66 Chapter 3 Overview of Quarkonium Theory
For example, the quark self-energy is in term of the electron self-energy
§ ® §
¡¢£¤¥ ¦¨ © ª «¬ ¡¢£¤¥ ¦¨ ©
The same can be said for all the other diagrams (except ¯, which is discussed below):
and the crossed ones. By the way, remember that it is the °rst diagram that generates the ±²³´µ¶ singularity.
Radiative Corrections to ·¸¹ º »» and Renormalization Scheme
It is interesting to discuss the radiative corrections to the strong decay. Since we work at the inclusive level,
these corrections include many different ¼nal states [151]
½ ¾¿ÀÁ  light hadronsÃ Ä Å ÆÇÈÉ Ê ËËÌ Í Î ÏÐÑÒ Ó ÔÔÔÕ Ö × ÛÜÝ Þ ßààá
Ø ÙÚ
åæ
â ã ä ç è ééê ëì í îï ðñ
òó
ô ï
The coef cient õ is given by (in the ö÷ scheme)
ï øùú û üý þÿ ù
õ
!"#$% &'()*+,-
45 8
where ./ 0 11 2 3 67 . This coef cient depends on the renormalization scale 9. To be able to compare the
8
:; <9=. This is the goal of the next few paragraphs.
expressions used in various papers, it is necessary to discuss the scale and scheme dependence of the coef cient
8 :; <9= comes from the scale dependence of the
Such a logarithmic scale dependence in the coef cient
8
strong coupling constant, hence there is no scale dependence for the electromagnetic decay radiative correction
coef cients. This can be simply understood from the diagram > above. In QED, it is simply taken into account
8
by a renormalization of the ne structure constant ?@ABC D EFGHI. If all internal photons are replaced by
gluons, the internal UV divergences and scheme dependences take care of themselves. On the other hand, if
the external photons are also replaced by gluons, the picture changes since external gluon vacuum polarization
graphs generate scheme dependences. In fact, the situation is a bit more complicated because of three-gluon
vertices like
J
but the main idea remains correct: most internal corrections combine into a nite correction, while the remaining
J
in nite corrections linked with the external lines are absorbed into EK L
by a renormalization at a scale , giving
3.2 Quarkonium Annihilation Rates 67
to both the correction
PQMN OPQ and the strong coupling RS OPQ
a scale (and scheme) dependence. The momentum
dependence of RS O is described in the appendix B.1, along with the relevant equations for the running.
Writing this scale dependence explicitly, the corrected decay rate is
T UVWX Y light hadronsZ [ \] c `a b b gh ijklm gh ne fgh
b de f f o p
\ ^_ _
This scale dependence is removed by redeqning both rs and tu at the common physical scale of the process,
here v w xy z{
|}~ light hadrons
Retaining terms up to order , equating the above two results (and generalizing to an arbitrary power of for
future use here, ) leads to
¡¢£¤¥ ¦§¨ ©ª«¬®¯ «¬ µ́¶·¸¹ º »¼ ½¾¿À¾ÁÂà ¿À¾Á
° ±² ³ Ä Å
ã å
Æ ÂÇÃ ¿ÈÁ ÉÊ ËÌÍÎ ÏÐÑ ÔÒÓÕ Ö× ØÙÚ çè
Û ÜÝ Þßà áâãäâåæ é êß ë ìí îïð ñò
øù
óô õö ÷ ú ûüý
(
þ ÿ
!"# $%&'
where we have used the one-loop running equation. Equating the )rst and last lines, the scale dependence of *+
is
- 0 7 8
*+ , +. / *+ , . 1 234
5 6 9:
In other words, the scale dependence of the correction is entirely determined by the scale dependence of ;< =8>,
as expected. The resulting picture is
DEF
?: G Q RSTU
=8> @ ABCH IJ LKM N O P V W
where X is scale independent. The numbers YZ and Y[ are \avor numbers, and as the notation suggests, YZ is
not necessarily equal to Y[. Indeed, their origins are different. The origin of YZ is in the number of \avors used
_
to run ]^ , while Y[ depends on the speci c radiative corrections to the process under consideration. From these
_
remarks, one can see that to give `a, there is a number of conventions to x: scheme, \avor numbers YZ and Y[
and scale. Unfortunately, there is no universal agreement. Let us review some conventions used in the literature.
1) The Grunberg scale: The more natural choice is b c de
fghij p q rstuvw
k l mno
z
since each gluon carries an energy roughly equal to xy. In this scheme, the coef cient {| depends
on }~ . Its value for charmonium is obtained with }~ as { while for bottomonium, with
}~ , { . This is the scheme we will choose in the present work8 .
2) The Barbieri scale: Another scale is used by Barbieri et al. [151] who take to get
Note that in their original paper, the scheme was employed. Furthermore, they used four avours in
The Grunberg scale is usually de ned so that the strong correction to some speci c ratio of decay modes vanishes [157]. In doing
so, is found to be very close to the natural scale ¡¢ .
68 Chapter 3 Overview of Quarkonium Theory
the running of £¤ while only ¥ for the ¦§ dependent part of ¨©. They obtained
ß àáâ
¬®¯¬ ® ³´ ÕÖ Ù ãä çèéê
µ ¶· º »¼ ½ ¾ × Ø ÝÞ å æ
¨ª«
© ° ±² ¸¹ ¿ À Á Âà ÅÆ
Ä Ç È ÉÊ
Ë Ì ÎÏ
Í Ê ÐÑ Ò ÓÔ Ú ÛÜ
which translated into the ë ì scheme (simply drop the í and îï ðñ )
òóôõö÷øõ ÷ ýþ !"
ù ú û ÿü # $%&'(
For comparison, if )* + , everywhere, the correction is
J P QRSTU
12 FG
- V WW YZ
.
+ /3
0 45 6 > AB C D H
< = ?@
78 9 :; E IK L MN O X
;
[
]
3) The Brodsky-Lepage-Mackenzie Scale: The BLM scheme [211] rede nes \ at a scale such
that all ^_ dependence in \] are moved into the strong coupling. This leads to
efghi
\]
`ab c d
efilmhh no p qr stsuuv
with j k . Note, however, that it may be a problem to run the coupling down
v |}
to such a low scale (for w p x, y z {qq ).
There is a certain relevance in discussing such schemes. It is indeed necessary to have an unambiguous value
of the ~rst order coef~cient to
estimate the convergence of the perturbation series . In view of the above, the
coef~cient varies as respectively. With an of around , the Barbieri prescription
leads to
, which is not very convenient for a series expansion parameter. The BLM scheme is
sometimes advocated as an unambiguous scheme with which to check the perturbation series [211]. Its drawback
is that it is not applicable when three-gluon vertex loop diagrams are present.
Photon Spectrum in ¡ ¢ £¤ £¥
This is the ¦rst occurrence of analytical problems in quarkonium physics. Since we are using the simple
model of Pirenne and Wheeler, the photon spectrum is in contradiction with Low §s theorem. This is equally true
for the gluon energy spectrum in ¨©ª « ¬, but let us take the electromagnetic mode for illustration, so that
®
con nement does not enter the discussion.
In the present theory, the normalized differential rate is
¯ ¹ º»¼ º½½¾ ¿ ÀÁÂÁÃÄ Æ Í ÎÏ
Ð ÑÒÏ
ÓÔ
° ±²³ ±´´µ ¶ ··µ ¸ ÅÆ Ç É
ÈÊ Ë Ì ÕÏ
where the phase-space spectrum is
å
Ú
Û Ü ÝÞ ß Ûã ä
Ö ×Ø Ù Ú Ú á â ×Þ ß Ø
Þ ßØ à æç è éê
ä
í ð
with ä the reduced photon energy ä ë ì î ï .
Close to ñ ò ó, the spectrum is linear in ñ, while Low ôs theorem requires a cubic spectrum. Exactly the
same phenomenon was encountered in the discussion of õ -ö÷ ø ùúûúü, to which we refer for additional in-
formation.
3.2 Quarkonium Annihilation Rates 69
3.2.3 Vector Quarkonium Decays
For vector quarkonia, the results are [155], [18]
Lowest Order Decay Widths Radiative Corrections
ýþÿ
!"# $ %& $ '' ()* +,-./01203 4856 747 9 :;<= >?@@ ABC
DEF G HHH IJ KLMQ NOP RSTUVWX Y]Z[ \Y\ ^_`abcde
fgh i jjj klm noptu qrs vwx y}z{ |y| ~x
w
with
¡ ¢ £¤¥ § ¦¦ «¬ °®¯ for ® ³³ ®´
¨© ª ¯ ± ²
µ ´ and ¶¸¸¸
The strong radiative correction coef cients ¶·¸¸ ´ are scale dependent, as for the pseudoscalar
quarkonium two-gluon decays. The decays into three gluons and two-gluons plus a photon are corrected as
À ÃÄÅÆÇ Ã ÈÅÈÇÉÊ
´ ¹ º» ¼½ ¾¿ ÁÂ
¶¸¸¸
ËÌÍÍ
Î Ï ÐÑ ÒÓ ÕÖÔ ×ØÙÚÛ ×ÜÙÛÛÝÞ
which give, at the Grunberg scale ßà á âãä åæææ ææ
ç èé
ê ë ì í îï ë ð ñðòó ç
ñôòó
ê ë õ ö ÷ø ù ú ûúüý ûþüú
ÿ
As for the two-gluon mode of the , those corrections do contain different processes. For example,
accounts for
ÿ , [156].
Let us also note that the correction to the three-photon decay as quoted without justi cations in [155], [18]
is suspicious. Instead, we would expect it to be simply
times the QED orthopositronium correction,
i.e.
70 Chapter 3 Overview of Quarkonium Theory
Differential Rates
For the three vector particle decay modes, the present formalism is essentially the Ore-Powell one (see [149],
[156]). Therefore, the differential rates are
!"# $ %%%& +, /012
6 8 9 :;<=
'() * - .34 5 7 5
>? @ABC D EEEF KL RS T
GH C I JM NOP QU TQ V WXYZ
[\ ]^_` a bbcd J
ij t
m m rs
ef ` g hk lnoop qu tq v wxyz
with again
xyz { | w| } xyz | w } xyz xy
vw xy ~ w xyz ~ t
|}
¡¢ near ¡ £ ¤
This spectrum is in contradiction with Low ¥s theorem, hence cannot be correct. For further information, we refer
to the discussion on the orthopositronium decay.
The interest of the contradiction with Low ¥s theorem is greatly enhanced in quarkonium physics. Contrary
¨ © ª « ¬®¯°±²
to the orthopositronium case, both the spectrum for ¦§ or ³ ´ µ ¶ ·¸¹º»¼² have been
observed, and disagree with the Ore-Powell prediction. We will analyze that fact in the next chapter.
3.3 Analysis of Inclusive Charmonium Decay
In this section we use the theoretical decay rates reviewed in the previous section to extract information from the
experimental data for the charmonium, and, when applicable, for the bottomonium. The kind of information that
can be retrieved are the strong coupling ½¾ , ratios of wavefunctions, and predictions for yet unobserved modes.
The approach followed is inspired from [155], [167].
3.3.1 Pseudoscalar Charmonium Decay
The ratio of the two-gluon rate to the two-photon rate is, including radiative corrections taken at the scale
¿ À ÁÂ À ÃÄÅ, ÆÇ Ç
ÈÉÊ ËËÌ Í Ì ÜÝ
Ø ÙÚ ÛÞÜ ßà á âãäåæ
Light hadrons
Î ÏÐÑÒ ÏÓÓÔ Õ ÖÖ× ç è
Using the éê experimental values, we can obtain the strong coupling and the éê wavefunction as:
ñ ó
åæ ëìêí÷ î ïð ï ò ÷ïðï
ôõö ô ô ô üý
î óø ùö î úïðññ ò ïðûû þÿ
3.3 Analysis of Inclusive Charmonium Decay 71
The error quoted is purely experimental. The neglected terms of order give an idea of the theoretical
errors.
If one uses the scale of Barbieri et al. (but in ), the correction to the ratio is
"#$%
Light hadrons ! ' ( ')*)+ ,-
& . /
2 5 >>
which is a huge correction (in 0 1, the coef cient is 334 ). Solving for 67 839:; gives 67 839:; < = 43 .
HI
When run down to ?@ J , we get AB C?@D E FG . At the scale of the charm, the scale ambiguity generates a huge
ambiguity of order AK L MNO, which is the theoretical error quoted above.
Finally, a more recent experimental determination is [138]
P QRS T UUV W QX Y Z ]^_
N ONOV [ \
` abc d h i h j kl
hadronse f g
giving the same value of mn
pqr s tuvw
mn o
xy | } ~ |
z{ { z{
3.3.2 Vector Charmonium Decay
Vector Charmonium Radiative Decay into Hadrons
The rate for vector quarkonium decay into electron-positron can be used to estimate the rate for hadronic
decay via a photon. Including radiative corrections,
ª ª ª ² ³ ´µ
¡ ¢£ ¢¤¥ ¦ § ¨©« ¬ ® ¬ ¯° ± ¶ · ¸ ¹º»
A more precise determination of this ratio uses the experimental electron-positron annihilation cross section
ÊËÌÍ Î Ï ÐÑÒÓ Ô ÕÖ×ØÙÚÛÜ ô õö÷ø ùúúû ü ý ü þÿ
¼ ½¾¿À Á
Ã Ç ÈÝ
ÁÂÃÄÅÆÆ É Þßàßá â ã äåæç è éêéëì í
íîïðñï
í òò ó
í
í
!"# $
For the case of , and
% &'() * + * ,-./0123 4 5 ! 6% 789: ; <= <>? @ ABC D EF G
as quoted in [223].
72 Chapter 3 Overview of Quarkonium Theory
Vector Charmonium Hadronic Width
The ratios we analyze are, for HI J K and the scale L J MN
O PQRS T UUUV ] ^_` a bc
i jklm s t s v wx yz{|
O PQRS T WXWYZ [ \ def ghnopqo r u } ~ (3.4)
¡ ¢ (3.5)
where the radiative corrections to £££ include ¤£¥££¦¦, and those to §££ include §£££ and §£¦¦.
©ª« ¬ ® ©ª« ¬ ±®
To compare with experimental data, we would need the measured values of ¨ and ¯ ° .
©ª«
Since these are unavailable, we will proceed differently. First, we write the total width as
©ª« ¬ ®µ ©ª« ¬ ±®µ ¶ ¾¿À Á ÂÃÂÄÅÆ Ç ÐÑÒ Ó Ô Ó ÕÕÖ×Ø ÙÐÑÒ Ó ÚÛÜÝ
¯²³² ´ ¯ ° ¯°
·¸¹º» ¼ ½ Ì
ÈÉÊË ËÍ Î Ï
from which we extract the combination
Þ ßàáâ ã äääÝ å Þ ßàáâ ã ÜääÝ
êë
Þæçæ èé (3.6)
Solving equation (3.4), (3.5) and (3.6), we ìnd
í òóô õ ööö÷ ø ùúû
î òóô õ üöö÷ ø ùýþû
(3.7)
ï ðñ ÿ
ðñ
i.e. a quite small
value. Note that the branching is roughly a tenth of the branching.
One can also work at the Brodsky-Lepage-Mackenzie (BLM) scale . Remember that
!"#
is independent of (for , the BLM scale was
this scale is de ned such that ). The
ratios become $%
$ % &'( ) ***+ 6 789 : ;< A DEF J ST U
4 5
BC K L KMNO PQ R
&'( ) ,-./0 123 =>? @GHI H V W (3.8)
X RYZ[ \ ]^^_ @ ~
no v w xyxz {| }
h ij
` abcd \ ^^^_ efg k plmqrst u
o (3.9)
since the same scale renders
independent of . The interest of the BLM scheme is its supposed unambigu-
ous check of the perturbation expansion. The coef cient in the above formula shows that perturbative QCD
may be completely wrong when applied to quarkonium decays. Finally, remark that since those coef cients are
now independent of , the above formulas are equally valid for bottomonium.
We have chosen not to work in this scheme because the running down to using perturbative
QCD is questionable. As a manifestation of this breakdown, the system of (3.8), (3.9) and (3.6) does not admit
real solutions. The reason why below a certain scale the system becomes impossible is quite clear:
¦
become less than . For ¡¢ £ ¤¥ , the ratio (3.8) of branching fractions is negative!
Finally, let us also quote the Grunberg scale, i.e. the scale such that the correction to the ratio (3.4) vanishes
§¨ © ¦ ª « © ¤¥¬®¯
°± ² ³ ´ µ ² ¶·¸®¯
What is remarkable with this scale is that it is so close to the natural scale ®¯.
¼½ ¾ ¿ÀÀÁ
The branching fraction ¹ º» can be compared with experiment:
ÂÃÄÅ ÆÇÈÉ ¾ ¿ ÆÊ Ë ÌÍÎÏ Ð ÑÏ Ò ÓÔ ÍÕ Ö ×ÍØÏ Ù
(3.10)
3.3 Analysis of Inclusive Charmonium Decay 73
Ü ßà
where Ú stands forìÛí ÝîÞ áâ . On the theoretical side, from the prediction (3.7), we can obtain an estimate of
ã ä å æç ë ë
Þ Ú è éê by integrating the Ore-Powell spectrum between ï ð ñê and ï ð ò
ó ôõö÷ ø ù ô ëì ì ó ôõö÷ ø ùûû ì ü ýÿþ
ï ú ñê ûû ð
!"
This gives with (3.7), i.e. :
&'( ) * + , - ./011 0 2 3 456
#$ % %
in agreement with experiment (3.10). The reason for this rather good estimate can be found in the cancellation
of corrections (relativistic, binding energy,...) in the ratio 7889888 , since both involve the same kinematics. The
main problem is that the observed photon spectrum is much softer than the QCD predicted photon spectrum,
even if the integrated rate is in good agreement. This will be discussed in the next chapter.
<= >
The same type of analyzes@can
A
be carried out for : ;
> 2 D5 E 3> 6
decay channels. The main difference is the huge
<=>
branching fraction ? ;: ; 9: B C ; . The total width thus gives
FG G I J KL KMNO P QQQ O R J KL KMNO P SQQ O R T ] ^_`a b cdc ef g h q rst u v u wwx
H
UVWXY XZ [ \ i jklm ln o p p
y z ¡¢ £ ¤ £ ¥ ¦§¨ © ª «
{| }~
i.e. ¬
¨ ®¯« ° ±± ±« ©
¬
¨ ®¯« ° ²± ±«
¸¹
¬³ ³ µ ¶·
´
which leads to º
½ ¾¿ ¾ÀÁ à ÄÄ ÄÅ Æ ÇÈÉÊ
»
Ë ÌÍ ÌÉÎÅ Ã ÏÄÄ Å Æ ÐÈÇÑÊ
¼ ÒÓ ÔÕÖ × Ø Ù ÚÛÜ
ß äå æ
which is consistent with the ÝÞ results à it gives again a small value for áâ ã compared to the value extracted
from ç å decays.
Three-Photon Decay
ëëì í îîî
Finally, we can predict the width è é ê relative to the ï ïï one as
ð ñò óô ñõõö ÷ øøø ù
ú ûü ýþ ÿ
which implies
!
so that experimental sensitivity is nearing the theoretical prediction. Note that we have dropped the radiative
correction "# 4
$%& ' ((() * +, -. / .0123
56
7
The huge coef cient renders the perturbation series problematic. For instance, for 84 9 :;<= the rate becomes
negative.
74 Chapter 3 Overview of Quarkonium Theory
3.3.3 Ratios of Wavefunctions
Comparison of Vector and Scalar Wavefunctions
We can use the ratio
> ?@AB ?CCD E FFG bc
XY Z[\ ]^_\`aao p deb t u vw x y |
H IJKL MNNO P QRQST U VW Zfgh ijklmn qro s z {} ~
where we have included radiative corrections. Non-relativistically, the wavefunctions are the same. Furthermore,
the mass difference is negligible at this level, hence
¡ ¢£ ¤¥
For charmonium, we get with ¦
ª « ¬®
£ § ¨©
¯ °±² ³ ´´µ É
¯ °¶·¸ ³ ¹º ¹»¼ ½ ¾¿ ÀÁ  ÃÄ ÇÅÆ ÈÊ Ë Ì ÍÎÏ Ð ÑÎÍ
while the experimental values give
Ò ÓÔÕ Ö ××Ø ÙÚÛÜ ïðñ ò óðôõ öóñ÷ø
ù
Ý Þßàá â ãäåæç èéêëì í úûüý
î þÿ ÿ ÿ
The central value is too low by about a factor . This could be due to relativistic corrections (especially in the
difference of the wavefunctions, due for example to spin-spin forces), to binding energy effects,... Note !
however
that errors are quite big: the theoretical prediction is still in the experimental range, so that it is dif cult to draw
any conclusions about the equality of "#$ and %#" wavefunctions.
Radial Excitation Wavefunction
( ) *+*, /01 2 3456
With the experimental measurements of &' and - . , we can estimate the ratio of
wavefunctions of both states as
78 ; <=>> ?
@ F@ O PQ PRST U VWVXY efghg m n
opq r sptu v wsxy
9 : : G HIJK
?A > ?@ E LI Z [\ ]^ _ `a`bc d fijk l z{|}~
CD;
B : MNG
(3.11)
As expected, the wavefunction at zero separation decreases as increases. The three-gluon branchings extracted
before also give the ratio of wavefunctions
¦
³ ´µ¶ · ¸¸¸¹ º» ÐÑÒÑ ÓÙÚÛÜÝ ÞÜÛÛÝ
© ª«¬ ¼ ½¾¿À
ÚÙä
¦ § ¨
® « Á ÂÃ ÄÅ Æ ÇÇÇÈ ÉÊË ÐÓÔÕ Ö× ØÞÚÙßà Ý á ÞâÛÝ ã
¡¢£¤ ¥ ÌÍ Î Ï
¯°© ± ² ²
Obviously, the errors are still much too large to get a clear coherence test among determinations.
3.3 Analysis of Inclusive Charmonium Decay 75
3.3.4 Conclusion: Fit to the Charmonium Data
To close this analysis, it is interesting to proceed in the opposite way, i.e. to infer values for the various decay
å
modes from speci c values of æç èéêë
éê ìíîï ð ñó ôö ô ÷ ø
ò õ õ
ùú ûü ó÷
ó ÷ óøö ó
ýþ ÿõ ÿõ ÿõ ÿõ ÿÿ
obtained by running down the world average of ùú û þ , with the three-loop beta function (see
appendix B.1). To estimate the wavefunctions, we use the electromagnetic decay rates
! "#$ % &&' ( )*+,-./0 12 34 5 678 9:
; <
= >?@A BCDE F GH GIJ K LMNOP Q RNSTJ UVW
XYZ [\]^ _ `a`bc d efghijkl mn op q rs
t uv
w x
yz{|} ~
where we have denoted
This gives
¡¢£ ¤ £¢¥ ¦¢§ ¤ £¢¥ ¦¢¨§ ¤ £¢£ ¦¢©§ ¤ £¢£
ª« ¬®¯ ° ±²³´ µ ¶³· ±¶³¸ µ ¶³¸ ·³¸¹ º »¼½» ¾¼»¿ º »¼ÀÁ
Taking those values, we get ÂÃ Ä ÂÃ Ä ÂÃ Ä ÂÃ Ä Å
Ã Ç ÈÈ ÉÊ ¿¼Á ÉÊ ¿¼¹ ÉÊ ¿¼½ ÉÊ
Æ ËÌ ËÌ ËÌ ËÌ
keV
ÍÎ Ï ÐÑÒÑÓ ÔÕÖ× ÔÕÖ× ÔÕÖØ ÔÕÖØ keV
ÙÚ Û Ü ÝÞ ßà áâ áÞ MeV
ãäå Û æçèé êëìíî êëìíî êëìíî êëìíî
keV
ïðñ ò óóó ôõö ÷õö ÷õø ùõô
eV
úûü ò ý þõ÷ÿ þõ ÷ þõÿ÷ þõøô
MeV
úûü ò ó ý ÿ ø
keV Input
where stands for the totality of hadron states.
A recent measurement [138] of the width was twice the older one, at
which is in quite good agreement with the theory. On the side, the theoretical predictions do not t the
!
experimental values. The width for is greater than the total measured width of , "#$# % &' ( ) keV, by
,
) * + times !
This same fact was apparent before, in the extracted value of the strong coupling
-. / 01 2 ,34
567 8 9: ; <=>
67
The ?@ value is consistent with the world average, but not the A one.
As a Brst trivial explanation, it could be that the C quark mass may be too low for QCD perturbation theory.
If, however, we insist in taking seriously this Brst order DEFG analysis, we need a powerful suppression effect.
Usually, people invoke relativistic corrections like [167]
H I HJ KL M N OP
Q
PR
76 Chapter 3 Overview of Quarkonium Theory
with S the relative TU
T velocity and V W X. In charmonium, these relativistic corrections can be sizeable because
]^
SYZTY [ X\
In the next chapter, the observed suppression of vector quarkonium decays will be explained by binding energy
effects.
3.3.5 Vector Bottomonium Decay
For the __ bottomonium states, experimental data are quite reduced, `ab states have not been observed yet and
branching fractions for vector states (the upsilon family) are rather scarce. We will thus only present a limited
analysis, with emphasis on the strong coupling extraction at the c quark mass. Indeed, it is interesting to see
whether the extracted values of de fghi are in better agreement than de fgji with the world average value.
Leptonic Width
The leptonic widths for bottomonium are
k p q rsrtu v wxyzx { |}~
l
m o ¡¢£¤¥ ¦ §¨©ª« ¬
n
¯ ° ± ° ²³´µ¶·¸«
From these, we can estimate the width ® using
¯ ° ± ° ²³´µ¶·¸«
® Í
¯ ° ¹º»¼½ ¾ ¿ ÀÁÂà ÄÄÅÆÇÈÉÉ
Æ Ê ËÌ
®
Then ÎÏ
Ð Ñ Ò Ñ ÓÔÕÖ×ØÙÚ Û ÜÝÞ ßà áâ ã äåäæç è éêëì í îëïð ñ
Vector Bottomonium Hadronic Width
We proceed as for the charmonium. The ratios we analyze are (in the Grunberg scale)
ò éó ô õõõö ÿ
÷ øó ô ùúùûü ý þ
(3.12)
& )*+
. / 012 34
! "# $ %' ( , - 5 6 (3.13)
+
9 : ;;;< = 9 : >;;<
To further compare with experimental data, we extract the combination 7 8 78 from the
total rate as
D E FFFG H D E IFFG H J S T UV UWX Y Z d e f e ggh i j kd e lmfh
7?@? A BC BC
KLMNO NP Q R [\]^_ ^`^a b c
n j kd e oooh i j kd e fooh i kp i qh j rs t uvwxy
} z ~~~ } z ~~
z {| {|
| (3.14)
{
3.3 Analysis of Inclusive Charmonium Decay 77
where we have neglected . Solving the equations (3.12), (3.13) and (3.14), we nd
¡ ¢£ ¤ ¥¥¥¦ § ¨©ª
« ¢£ ¤ ¬¥¥¦ § ©®ª
¯° ±²³´ µ ¶·¸¹¹
Extrapolated to the º» mass, this gives a strong coupling of
¼° ±²³´ µ ¶·¸¹¹ ½ ¾¿ ÀÁÂÃ Ä ÅÆÇÅÈ
Ä ÅÆÇÊËÁ Ì . This does not change
i.e. a quite small value. Note that usually, people use the BLM scheme, i.e. É
ÐÑÒÓÔÕÌÖ × ÐÑØÓ
the conclusion: one obtains at the BLM scale a strong coupling of ÍÎ Ï , i.e. when run up to
ÙÚ ÐÑÒÛ
, ÍÎ ÏÙÚÖ × .
This closes our basic review of quarkonium annihilation modes as described in the standard factorized
approaches. In the next chapter, the binding energy effects will be introduced, as a cure for the analytical defects
of factorized models.
Chapter 4
Binding Energy Effects on Quarkonium
Annihilation Rates
In the standard schemes, quarkonium annihilation amplitudes are constructed assuming a factorization of the
bound state dynamics from the perturbative decay process. As discussed in the previous chapter, the basic model
is simply the extension to quarkonium of the Pirenne-Wheeler model
ê Ýëã ì íîïðñòó ôõöõ÷ ø ùúûû
Ü ÝÞ ß à áâã ä å
æç è
åé é
üýþÿ
As we have extensively discussed in the case of positronium, a modi cation of the quarkonium annihilation
description is unavoidable, since the predictions of any approach based on factorization are in contradiction with
analyticity. One has to introduce binding energy effects. Compared to positronium however, the introduction of
such effects is not just a matter of principle it can change the whole picture of quarkonium physics since the
typical quarkonium binding energies are huge
&'()*+
!" # $ % ,-)*+ . /012 (4.1)
1345 ;<=>
674 8 9 : ?>= @ 9ABC (4.2)
For these rough estimates, the constituent mass of the heavy quark is taken as the mass of the corresponding
lowest heavy Davored meson. We are forced to proceed in this way E since twice the running mass of the heavy
quark is less than the bound state mass (the origin of this fact is con nement). Another approach would be to use
the quark masses found in spectroscopic studies (see (3.1)). This leads to very similar estimates for the binding
energies.
nium physics: the FG extraction, the HI puzzle, the photon spectrum in inclusive radiative decays, and the ra-
We will review in this chapter the implications of binding energy effects for a number of problems in quarko-
E E
diative transitions between quarkonium states. The discussion will remain at a qualitative level in most cases.
E
The purpose is only to show that binding energy effects indeed lead to signi cant modi cations, and that those
modi cations go in the right direction, i.e. that binding energy effects can be invoked to explain the various
E
quarkonium puzzles. We are unable at present to do a quantitative analysis because a number of complicated is-
sues have to be resolved rst, like for example the precise form of the quarkonium wavefunctions.
80 Chapter 4 Binding Energy Effects on Quarkonium Annihilation Rates
4.1 Binding Energy Effects on Quarkonium Physics
The section begins with a discussion of the quarkonium wavefunction. Then, the various problems cited above
are systematically reviewed.
4.1.1 QCD Quarkonium Wavefunctions
To apply our method to the quarkonium, we need its wavefunction. No matter its form, it will always be possible
to express it as a combination of functions for which the derivative method applies. For example, the set of
functions built on the Coulomb wavefunction (see appendix C.1)
`
JK LMNO P QRS TUVYZ a [R\]^_ g
S TU W X V bc d ecf
can serve as a basis in which the quarkonium wavefunction can be expanded. All these functions are accommo-
dated in the derivative method through the operator
hg ijk l mnopq rstu vwt x yz{||}~
¡¢
£¤¥
´ Ä
¦ §¨© ª«¬®¯°±
«²«¬³
µ ¶· ¸ ¹º»¼ ½ ¾
¾¿ÀÁÂÃ
It remains to Ånd an expression for the quarkonium wavefunction. This is not an easy task. The form of the
QCD potential is not well-established (there are many alternatives). Further, in QCD, the parameter Æ cannot be
deÅ ned straightforwardly. For all these reasons, in the present section, we will simply take for quarkonium the
form factor of positronium, with the binding energy deÅned by the ÇÇ threshold. This will illustrate what can
be expected from binding energy effects on total and differential rates.
Further, because of conÅnement, the conÅguration space quarkonium wavefunctions should be less spread
than the Coulomb one. In turn, the momentum space wavefunctions are more spread out around È É Ê for
quarkonium than for positronium. But we know that the binding energy corrections are always proportional to
the spreading, in momentum space, of the wavefunction. Therefore, the binding energy corrections computed
with a Coulomb form factor will be lower estimates, the true corrections being more important.
Another approach would be to take each of the charmonium potentials and compute the wavefunctions by
solving the Schrödinger equation. Then, these solutions would be expressed in the basis of the Coulomb wave-
functions. Such studies are in project, and have not been completed yet. The present section is a Å rst step in that
direction.
4.1 Binding Energy Effects on Quarkonium Physics 81
4.1.2 Strong Coupling Extraction
The value of ËÌ as extracted from quarkonium physics does not Ít well with the values obtained from other
QCD observable measurements [223]
Average
Hadronic Jets
e + e - rates
+ -
e e event shapes
Fragmentation
Z width
Small x structure functions
ep event shapes
Polarized DIS
Deep Inelastic Scattering (DIS)
τ decays
Lattice
Heavy Quarks
0.1 0.12 0.14
(M )
We have seen in the previous chapter that the value of ËÌ extracted from charmonium and bottomonium inclusive
Ì ÎÏÐÑ ËÌ ÎÒÓ Ñ
decay rates are
ÔÕ Ë Ö×ØÙ Ö×ÙÙÚ
ÛÜÝ Þßàá ÞßÞáâ
ã Þßàä ÞßàÞÚ
where we have run the values up to åæ using three-loop running, and çè é ê between ëì and ëí, çè é î
between ëí and åæ (see appendix B.1). The value of ïð extracted from ñ ì ò óóôõö ò ÷÷ is in the world
ù ù
average range, but falls short of it when extracted from øô ò ÷÷÷ú÷÷óô øô ò ûüûý or þ ÿ þ ÿ
.
From the computation done in the positronium case, we know that binding energy effects will suppress the
decay rates. As a result, one will need a greater to
t to the observed rates. This is the general mechanism we
invoke to explain the discrepancy between the quarkonium determinations and the world average. In other
words, the smallness of the extracted in standard approaches is, in our view, the signal that the factorization
hypothesis is inappropriate, and that a loop model has to be considered.
Electromagnetic Decays
We begin our study with the electromagnetic processes (see chapter 3)
# $ %&' ()
! " * +
A C KM
, -./0 1 23 245 6 789 :;<=><? @ B @ GHI JK L
DC E F N OPQ R
] ^_` a b c m o
S TUVW X YYYZ [ \ g jk l n l r s rtuv wx
d efhi po q y z
82 Chapter 4 Binding Energy Effects on Quarkonium Annihilation Rates
and the corresponding { decays.
In the case of positronium, we saw that the bulk of the |rst order radiative corrections is accounted for by
binding energy effects. The same mechanism will be invoked for quarkonium, so that the strong corrections will
be replaced by a function of the binding energy. Of course, it will not be possible to directly test this hypothesis
since the quarkonium binding energy relation to the strong coupling is not known.
Let us start with the pseudoscalar charmonium electromagnetic decay
} ~
with ¯
¡ ¢ ¦§¨©ª «
¢¤ ¥ ¬ ®
¢ ®
¢£ ®
²
By taking the value of °± given in (4.1) and (4.2), we get the suppression ® factor. It is interesting to match this
·¸¹º» ¹
suppression to the perturbative correction ³ ´ µ¶ , to extract a value of ¼ :
½¾¾ ¹ ¿ » Ä Å ½¾¾ Æ ¹ ÇÈÉ
¼ ÀÁÂ Ã ¼ Ã
ÊË ÌÍÎÏ ÐÑ ÒÓËÉ Ô ÌÍÏÕ ÌÍÕÖ
Ê× ÌÍØÎ ÐÑ ÒÓ×É Ô ÌÍÙÚ ÌÍÕÏ
Note again that, at present, the matching with the perturbative QCD result has no grounding. Moreover, the
ÐÑ Ë
corresponding scale is not clear (it could be that this is at some low scale, between Û and Ó ). Finally,
ÐÜÝ Þß Ü
corrections are probably quite sizeable. In view of all these rough approximations involved (especially
à
in the de nition of the binding energy values (4.1) and (4.2)), the consistency of the values is surprisingly good á
the binding energy suppressions are of the order of the perturbative QCD corrections.
Let us proceed similarly with the leptonic decay of the vector quarkonium
ù û
â ãäåæ ç èéêëì í îïð ñòóôõö÷ ø ú ø
ü û ý þÿ
with
,-
" #$ #' (
! %&
)
*+
0
Then, for ./ and 1, we get 2345 6 7
;
< ? @AB C D E FG HIJ K
8 9 :
<= >
LMN OPQR FG HST K U OPV OPQW
X OPWY FG HSZ K U OPY OPQV
Again, the matching gives surprisingly good results.
The [nal decay \ ]
f ghi j kl x
v }~
^_` a bb b c d e st u w u
op
qr
m n y x z{ {{ |
is more complicated to handle since we do not have an analytical expression for . Numerically, we found
In this case, the matching gives , since it is around that value that perturbative QCD
corrections force the total rate to vanish. Obviously, this later fact is now well interpreted as a strong suppression
due to binding energy effects.
4.1 Binding Energy Effects on Quarkonium Physics 83
It is important to remark that, in all the previous cases, the factorization of the binding energy effects out
of the perturbation series will enhance their convergence. The same phenomenon was observed in positronium
physics. Here, it becomes vital to proceed along our approach, to avoid inconsistencies (such as corrections
greater than unity).
Strong Decays
Introducing binding energy corrections for strong decays is more complicated because of the issue of double
counting. In the electromagnetic modes, the nal state particles do not experience the strong interaction, hence
the whole correction comes from initial state radiative corrections. This is the same situation as in the
positronium case. Here, the situation is much more complicated because, rst, there are strong corrections to
the nal state also, and second, because we are considering inclusive rates, so that decays to many gluons, quark
pairs,... also contribute.
Instead of searching for a denite prescription to deal ¥
with all those decays, let us simply consider the two-
gluon mode of the ¡¢, and the three gluon mode of the £¤ , to illustrate what can happen. The binding energy
corrections to the two-photon and two-gluon modes of the ¡¢ are the same. Hence, the ratio of the two modes
is not corrected, and this explains why a value of consistent with the World average can be extracted from
them. For the three-gluon mode, we write
ÌÍ Î
¦ § ¥ ¨ © ©©ª ¶ ·¸¹ º »¼ Á
Ê Ê Ê ËÏ Î Ð
£¤  ÃÄÅÆ
« ¬®¯ ¨ °± °²³ ´ µ ½¾¿ À ÇÈ
Å È É ÒÔ ÕÖ× Ø
ÑÒ
Ó ÙØ Ú
À
Þßà á âãäå
We have found previously that ÛÜ Ý . Now, the binding energy correction factor corrects this value
as
è ïð ñ
è ëì ëí îò
ñó
Þßà á âãäå æ ç
è þÿ
ÛÜ Ý éê ÷õ ø
ô
õ õ õ öù ø ú û üý
When run to , we get , in rough agreement with the world average.
We expect that the same mechanism üý will work for the other decay processes, i.e. that the necessary com-
pensation of the binding energy suppression by the strong coupling will lead to an overall enhancement of .
4.1.3 Rho-pi Puzzle
þ
The rho-pi puzzle is the apparent violation of the rule in exclusive hadronic decay channels of the and
, especially ([165], [175], [177], [181], [183], [186], [193], [194], [200]) . The
rule follows simply from the identity
!"## ) %& # *+, -. -/011 *2 3456789
2
$ %& # '() ( !"## :;< =>?@ABC ?DEFGC
'( ( UV UWX
H IJKKLMLQRMP LNOP^T_`_ab
LNOPP SYLMZ[\]
IJ h ijkk
c delmffg gnohk
g gpq r stu v wsx r stu
and then from the additional assumption that the exclusive decay branchings follow the same rule (i.e., that the
hadronization has no effect).
84 Chapter 4 Binding Energy Effects on Quarkonium Annihilation Rates
Many exclusive hadronic decay modes of the y z{|} and ~y that have been measured. Their ratios
are[137], [139]
~yy z{|}
The rule is more or less valid, except for a few channels like , which fall short
of the rule by as much as a factor of . No convincing explanation of this fact has been proposed yet (the
most popular explanation is the presence of a glueball [160], [163], [173]).
Here, what we just want to point out, is that the rule will not survive binding energy effects. For gluonic
modes, a prefactor appears
¡¢£ ¤¥¦§¨ ¯´± µ ´³¶
©ªªª «¥¦§¨©¬¬® ¯° ¯±²³³ (4.3)
But the electromagnetic part is unaffected by such effects:
· ¯° ¯±²³ ¸ ¹ ¸ º»¼½¾¿ÀÁ  ÃÆ ÃÌÍÁ Ç ÎÏÐÑÒ ßàáßâ ßáãÜä
 ÃÄÅÆ Ç È Ç Éʼ½¾¿ÀÁ Ë Ó ÔÕÖ× Ø ÙÚÙÛÜ Ý Þ
This could explain why the åæ decay exhibits the violation of the
ßàä rule, while not the purely electromagnetic
çæ decay (which is isospin violating).
Unfortunately, with our parametrization of the binding energy in terms of the è meson mass, the prefactor
in (4.3) enhances the é Þ
ãêÜ three gluon mode compared to the ëìé, because the ííí mode is enhanced close to
the threshold for èè creation. In our view, this is the signal that the constituent mass chosen is inappropriate to
correctly describe the dynamics. For instance, if one takes the Lagrangian î quark mass of
ßáã ï ßáð ñòó , it is
clear that the ëìé is very close to the threshold, while now it is the é Þ
ãêÜ ô ííí which is strongly suppressed.
The treatment of boundõ states above the constituent mass is clearly complicated because such a phenomenon is
÷ø
a manifestation of con nement. Further studies are necessary, but in any case, it should be clear that binding
energy effects introduce an important violation of the naive ö rule.
Scale Anomaly
We have seen that the binding energy corrections to the decay - ù úû ü ýþÿ receive a contribution from
same probably happens for the leptonic decay
ÿ ÿþ
the QED scale anomaly, a contribution which is apparently missed by standard factorized computations. The
. This would be far reaching: the leptonic mode has
4.1 Binding Energy Effects on Quarkonium Physics 85
and is used to extract the wavefunction. This, in turn, will affect both the
been considered as clean,
always
rule, the extraction of and so on.
A complete study of all those effects has yet to be done, and is left for future work.
4.1.4 Photon Spectrum in Quarkonium Inclusive Radiative Decays
The next aspect of quarkonium physics to be discussed in some details is the photon spectrum in
and , which is described in terms of the underlying processes !" ##. It
$
is indeed in spectra that the modi cations due to the binding energy may be directly observed. As we have seen
when discussing the orthopositronium decay, the photon spectrum found using the Pirenne-Wheeler formula
is in contradiction with Low %s theorem. By introducing binding energy effects, the spectra should move from
the wrong Ore-Powell prediction towards the Euler-Heisenberg one, i.e. softer at both the low energy and high
( ) * + ,-./012
energy end points. This behavior is what is observed for the &' [134], [135]
and for the 3 4 5 6 789:012 [136], [140]
In both cases, the increase at low energy is due to bremsstrahlung processes ([168], [180], [185], [197])
86 Chapter 4 Binding Energy Effects on Quarkonium Annihilation Rates
; ;
and the Ore-Powell prediction has been corrected for the ef ciency. For example, in the < case, the ef ciency
varies as [140]
?
The suppression of the spectrum at its high end point is greater for the => than for the <, as expected since
?
the binding energy is greater for => . To explain such a strong suppression purely in terms of binding energy
;
effects is not possible though @ even the Euler-Heisenberg spectrum is unable to t the observed spectrum.
Field As Model
;
The fact that even the Euler-Heisenberg point-like result seems to be insuf ciently suppressed is rather
surprising, but understandable from the point of view of Field Bs model [161]. This model introduces some
distortion of the photon energy spectrum because of gluon dynamics. It can be viewed as a tentative introduction
of hadronization effects. Because gluons can radiate additional gluons, the CpromptC gluons emitted from the
quark line acquire an invariant mass
;
This means that the high end point of the photon spectrum must be modi ed, since in presence of such massive
gluons, the maximum energy the photon can take away is reduced. The result of Field Bs Monte Carlo simulation
is
where the plain line is the Ore-Powell spectrum, and the dotted line is the Field Bs model spectrum.
Obviously, Field Bs model is in contradiction with analyticity; (it is built on the Ore-Powell lowest order
result). It should therefore be quite clear that it has to be modi ed somehow. In fact, it would be interesting
to repeat Field Bs computation, but taking the binding energy corrected spectrum as a lowest order basis. This
analysis should lead to correct spectra (i.e. consistent with Low Bs theorem), and in agreement with experiments.
It is left for future studies.
4.1 Binding Energy Effects on Quarkonium Physics 87
4.1.5 Transitions among Quarkonium States
There are many possible transitions among quarkonium states, like for example
D
EFG H IJK LMN O P MQ
G RSTU V WXY Z[S\ ] Z[Z^
_ RSTU V `ab c defg hi j k
b lmno p qrst uvw x yz{
| }~
¡ ¢£¤¥ ¦§¨¨ © ¦§ª«
¬® ¯ °±²³ ´µ¶· ¸ ¹¶º
»¼½ ¾ ¿ÀÁÂ ÃÄÅÆ Ç ÃÅÃ
Most of these transitions involve the È states, for which a treatment is still lacking in our model. The hadronic
 ÉÊËÌ Í ÎÏÐÑÒÓÔÕ
transition is generally dealt with using some low energy techniquesÏlike chiral perturbation
Î Ð Í ØÙÚ
theory, and will not be considered here. There remains the Ö × (suppressed) transition
ÝÞ . This decay,
Ú
as a Ûrst approach, can be described assuming a point-like structure for the Ü and the ß . In other words,
we assume that binding energy is so important that the structure ofÝ the bound states is no longer relevant. Such
Þàà Úàà
processes are calculated using a loop model, with the coupling Ü and ß given by áâãäåæ and áçèéê ,
respectively. Then, it is easy to ënd
ö þÿ
ì íîï ð ññò ó ô õ÷øù
úûüý
3 4 EFG J G G
F J
G
#$ ABC D H I KLM N O H I PQ R
!" + 5 6 79:
8
%&'( ), * ,/01 2 STUV
-. 7;<=
8 > ?@
with W
Z[ \ ] ^_`a^bc
XX Y efdg
d
h m n op q r stuvswx | } ~~x
ijk l z{y| zy|
y y
The second term appearing for
comes from the contribution of oblique cuts (for , there is only vertical
cuts).
We now take the and couplings identical (this is valid to leading order, in the non-relativistic approx-
imation). Using the measured value for the branching , we nd
¡ ¢ ¨ ©ª«
¤ ¥ ¦§
¢¢ ¡ ¢ £
¬¤ ¥ § ©ª«
®
Plugging these values into the two-photon width,
º»¼ ½ ¼»¾ ÁÂÃ Ä ÅÆ Ú ÛÜÝ
¯°± ²³ ¶ ¿À Ù
¹ÈÉÊ Ë ÌÉÍÎ Ï ÊÌÐÑ Ò ÓÔ
¤ ´ µµ¶ · ¸ Ç ÕÖ
Þß × Ø á ÛÜÝ
Ù
×à
to be compared with the experimental value
â ãä ß å ææç è éê ë ìç í îïðñ ò óô
In conclusion, the point-like approximation, with in addition the assumption of equal couplings, is consistent
with the (not yet very precise) experimental data.
88 Chapter 4 Binding Energy Effects on Quarkonium Annihilation Rates
4.2 Conclusion
We have seen in this chapter that binding energy effects can play an important role in quarkonium physics.
While their introduction was motivated by analyticity, and in particular photon spectra, they lead to important
corrections in all aspects of quarkonium physics. They can explain, at least in part, the õö puzzle, the smallness
of ÷ø as extracted from vector quarkonia and the softness of the photon spectra in inclusive vector quarkonium
radiative decays.
To close this chapter, we just want to point towards other applications, which are currently active research
subjects in the literature.
One obvious extension is the description of ù -wave quarkonium states, the úûüý, and in particular, the two-
photon decay modes of the úûþ and úûÿ which have been studied experimentally [142], [143]. Such a study, of
course, will rely to some extent on the corresponding results for QED ù -wave positronium states. Therefore, we
are not yet in a position to say much.
Once the above extension is established, it can be interesting to push the formalism further down in energy,
close to or less than GeV, in order to describe particles such as the , or scalars þ , tensors ÿ , and so on.
For instance, in the case of the , it is established that a description in terms of a loop is adequate ([216],
[221]), with a point-like coupling of the to the . The matching with the corresponding quark loop, with a
non-constant form factor, can lead to interesting information about QCD. Concerning scalar states, the prediction
of their two-photon decay rates is important for instance in the identi cation of the quark model multiplets (see
for example [166], [172]).
Another aspect that would need to be studied is quarkonium production (see for instance [167], [171], [198]).
Examples of production mechanisms are the Drell-Yan process, gluon fusion, photon to vector quarkonia transi-
tions,... In rst approximation, it is always assumed that some kind of factorization holds, in order to separate the
perturbative creation of a heavy quark pair from their non-perturbative hadronization into a quarkonium state. As
mentioned before, in the NRQCD context, the quark pair is not necessarily in a color singlet state. This would
have to be incorporated in the present loop model context. Anyway, it should be clear that binding energy effects
are also important in production mechanisms.
89
Conclusions and Perspectives
In this thesis, we have presented a method to deal with bound state annihilation through Lorentz invariant
form factors. Typically, a decay amplitude is built as a loop amplitude, with a form factor inserted to account for
the non-trivial structure of the bound state:
This is to be compared to the standard factorized method, symbolically:
For QED bound states, we have shown how to relate that form factor to the bound state Schrödinger wavefunc-
tion, so that total decay rates agree with the ones computed using a factorized approach (at least at lowest order).
From that analysis, it emerges that one can simply take the derivative of point-like QED amplitudes to account
for the bound state structure
Concerning differential rates, our form factor method is de nitly superior to factorized approaches like NRQED,
since it does not contradict Lows theorem. This property of our method follows from its non-perturbative
treatment of the binding energy, and from the four-dimensional loop structure of the amplitude (taking into
account oblique cuts). For a review of the properties of the method, we refer to section 2.5.
For QCD bound states, the stage is set for more elaborate studies. Effective form factors can be retrieved
from quarkonium phenomenological potentials. For simplicity, in the present work, we have used a phenom-
enological form factor obtained from the Coulomb wavefunction, with an ad-hoc parametrization of the binding
energy (to be short, the and constituent masses have been taken as the masses of the lightest and mesons,
respectively). The analysis was therefore qualitative, but already at that level, interesting conclusions have been
obtained.
90 Conclusions and Perspectives
The following table reviews the main processes studied in the thesis. Also indicated are the main properties
or consequences of the form factor approach obtained in each case:
QED QCD
Electromagnetic Strong
Increased Convergence BEE on the extraction of
Low s Theorem
Low s Theorem "
BEE ! Softening of the photon spectrum
Increased Convergence
#
% Extraction of '(
BEE $ )* puzzle
&
(no 14% rule)
+ , Increased Convergence
Scale Anomaly
BEE = Binding Enegy Effect
91
The method presented here may be extended to a great variety of processes. The following table summarizes
some of them:
QED QCD
Higher excited state
decays,...
Decays (P,D,...-wave) -.
/ 345 6 789
0
0 :8 6 ;<=9
1
Transitions
= >?@A 6 ;<=BB
between states 0
0
2 CCC
I
E J Kproduction
ILM
Production rates D
F GH
NNN
GH
(for example)
O
Fine and Hyper ne Quarkonium Spectrum
Splittings (mixing ?)
(+ other diagrams)
Unequal constituent
mesons,...
masses PQ
92 Conclusions and Perspectives
Some questions which remain open in para- and orthopositronium physics, or concerning the application of
the method to quarkonium physics are
Positronium
1. How to deal with higher order corrections,
,. . .
compulsory to compete with other techniques ?
2. What is the exact meaning of Rrenorma-
S
lizationR in the hyper ne splitting computation ?
(i.e., is the bare Ps mass really TUV ?)
3. Is the Scale Anomaly a missed
contribution to W-X Y Z [[[,
thereby explaining the \-] ^ lifetime puzzle ?
b e
c f
4. Why is the derivative technique so simple ? c f
c f
(i.e., is there a fundamental property of c f
] a c f
Coulomb bound states behind this ?) _ ` c d
f
g
`
Quarkonium
5. What are the quarkonium wavefunctions ?
(or alternatively, which of the proposed j k lmn
charmonium potential is giving the best
h
t to the decay rates ?) i
6. Is it necessary to extend (and if yes, how)
the form factor approach to account
for all the color octet model contributions ?
7. Can the method be extended to light
pqrqn
mesons such as the o l , uuu
and what can be learned about duality ? s t
Positronium and Quarkonium
8. Why is there a kind of duality between
binding energy corrections and vrst order
radiative corrections for lowest lying w states
(what we called xincreased convergencex) ?
93
The third question is probably the most interesting one. Indeed, the scale anomaly could provide a solution
to the orthopositronium lifetime puzzle within the framework of QED. This is to be contrasted to recent attempts
invoking the mirror world [129], extra dimensions [133] or new light particles [115]. It would also be a nice
evidence of a non-trivial quantum yeld effect at the level of atomic physics.
Concerning quarkonium physics, the present method may provide the most promising discriminating test
among phenomenoligical potentials, because decay rates get sensitive to the whole wavefunction, and not only
its value at zero separation. This is however not as simple as it seems, because one should be sure to keep the
sizeable strong radiative corrections under control (i.e., to avoid double counting). Also, NRQCD analyses have
shown that a color octet contribution has to be introduced to yt the data. From our point of view, it is not clear
whether this is an artefact of the NRQCD method, or a true physical contribution that needs to be introduced in
the present form factor approach. In conclusion, the main limitation of the form factor approach in quarkonium
physics is simply the lack of a reliable computation technique from yrst principle. Using a mixture of QCD and
effective potential arguments is rather suspicious, and in any cases difycult to control.
Finally, the last question needs comments. We have repeatidly argued in the text that the increased con-
vergence of the perturbation series, when binding energy corrections are factored out, comes from the fact that
binding energy corrections already account for a great deal of radiative and relativistic corrections. This is cer-
tainly true, but is not the whole story. For instance, binding energy corrections and radiative corrections do not
scale similarly with the principal quantum number z (as z increases, the binding energy corrections decrease
while the radiative corrections stay the same). In other words, it appears that it is only for the lowest lying {-
states that binding energy corrections and radiative corrections are roughly the same. More suprisingly,
we have
presented some indications that the same rough equality holds for the lowest lying {-states |}~ and |~ .
This could be the signal of some kind of sum rule that needs to be discovered.
In conclusion, the form factor approach to bound system introduced in this thesis offers interesting perspec-
tives for a number of theoretical problems. While bound states are the most common states of matter encountered
in Nature, they are also very difycult to deal with theoretically. Any new information should therefore be seen
as a great opportunity to reach a better understanding.
95
Appendix A
Resources for QED Bound states
A.1 Parapositronium
A.1.1 Details of the Dispersion Study
Let us prove the assertion that
Dispersion relations constructed on the vertical cut imaginary parts
of the loop model reproduce standard convolution-type amplitudes.
We will particularize the discussion to the parapositronium decay into
, for which there are vertical cut
contributions only. Let us emphasize that the whole discussion of this section is readily extended to any para- or
orthopositronium vertical cut contributions to the decay amplitude.
We rst compute the imaginary part of (2.3) for an arbitrary initial squared mass . Considering the two
possible vertical cuts, we obtain by replacing the two propagators on each side of by delta functions
- ¡¢ £¤¥¦§ ¤̈©
® ¸ ¾ ¿À Å
¹ º »¼ ½ ÁÂ Ã Ä Æ Ç È
ª « ¬ É ÊË Ì ÍÎÏ
¯ °¯±²³ ´µ¶ ·
×Ú ãæ é êëì ì
ÐÑ Ò ÓÔÕ Ö×Ø Ù ãä å å è
íîïðñ
Û Ü ÝÞ ßàá â ç
ùú û üýþ
After a straightforward integration over òó and ôõ ô, with ö ÷ ø , we reach
'* ./01 234 5 36 5 89:;< <
ÿ
!" # $% &' ( ) + ,- 7 =>?@A
In the course of the derivation, the delta functions forced BC D E and FG F D H
IJKL MNJ. In other words, the
electron momenta are O
UVW \ m n op
P I Q R S T X YZ[ with ]^_` a bcd e fgh i jkl (A.1)
96 Appendix A Resources for QED Bound states
q rs t
This kinematics is to be understood in the trace evaluation. The angular dependence arises from the relative
uv w xy z {|y
orientations of and the photon momentum . Note also that the relation (A.1) cannot be satis ed for the
physical value . This is obvious since the loop cannot have an imaginary part for the physical
bound states, its constituents being always off-shell. From the kinematics (A.1) one can prove that the factors on
both sides of}~ are true projectors, which serve to enforce gauge invariance in the expression
Indeed, those two projectors play exactly the same role as external spinors when establishing Ward identities.
The real part will now be calculated using an unsubtracted dispersion relation
¨
¢£ ¨ ©§ ª« ¬® ¯° ± ²³´
¡¤¥¦ (A.2)
µ³ ¶ · ¸ ³ ¹ º»¼
where it is understood that should be replaced by everywhere, i.e. scalar products that will appear when
¿ÀÁ Â Ã Ä Å ÆÇÈÉ ¶
evaluating the trace should be expressed with the kinematics de ned for an initial energy . Since ,
½¾
the principal part can be omitted and . Now let us write the form factor in the general
form ÊË Ì ÍÎ
ÏÐ ÑÒÓÔ ÕÒÓ Ö ×ØÙ Ú ÛÜÝÞ ßàáâ ãäåæ ç èé ê ëìí îï (A.3)
÷
with ð ì ñ òó ô õóö and øù the bound state wavefunction at zero separation. Then (A.2) can be written as
ú ûüýþ ÿ
!"
#
+ . 2 345 6 7: =>?@ @
$%& '() *+, - / 0 1 78 9 ; 9 < ABCDE
Let us transform the F integral back into a GH G integral, keeping in mind the constraints obtained when
extracting the imaginary part. Using HI J KL
M NOP, QK R S TUVW VUV, the decay amplitude dispersion integral is
de mno |
X YZ[\ ] ^ yz {|
} ~
_ `a b fcghij pkqrl stuvw x
where, as the notation suggests, it is understood that any appearing in the amplitude must be replaced by
¡¢ ¡£ ¤ ¥¦. In particular, §¨ ¦ ©ª« can be replaced by ¬® with ® ¯ °±² ±³ ´ µ¶. This amounts to consider
the scattering amplitude with incoming on-shell electron-positron having momenta ·¸¶ ¹ º»¼ ½ ¾¿À Á ÂÀ (since
ÃÄ Á Å). Note the fact that ÆÇ È ÉÊË, namely that apparently the energy is not conserved. This is not surprising
since the present formula is a dispersion integral, done along the cut where ÌÍ ÎÏÐ Ñ ÒÓÔ. Finally, in view of
the kinematics, we introduce Õ Ö Ô× Ø ÙÏÐ Ú Û and ÕÜ Ý ßÞ à áâã ä å (hence æç è éçê è éë and ì í îìï ð ñ)
to write the amplitude simply as
üý
ò óôõö ÷ ø
!"# $%& ' (#) *+,-./+0
ù ú þÿ û ÿ
(A.4)
5657689: >: ? >@:
where 123 4 is the amplitude for on-shell ;< = H I
C = scattering into AB . Gauge invariance isMNO
; JH present
P
due to the two projectors, well de ned since DE F DG . This ends our demonstration, and K L
Q RS
- T U V WWX.
A.1 Parapositronium 97
A.1.2 Sommerfeld Factor and Form Factors
To reach a better understanding of the parapositronium two-photon decay rate, we now repeat the derivation
using the language of dispersion relations. This is necessary because we will want to try different forms of form
factor YZ , and thereby get information on the physical content of the binding energy corrections.
^ _`
The loop integral [\ ] has an imaginary part obtained by cutting the propagators
no kquv wxy z |{ } ~
abcd ef gh i j
k l pkmqrn s t
Using an unsubtracted dispersion relation and inserting the Coulomb form factor, we reach the two equivalent
forms for the loop form factor
Æ Ç ÈÆ É ÊËÌ Ö
¡¢£¤ ¥¦§¨ © ª«¬ ÀÁ Â
º » ¼½ ¾ ¿ Ã Ä Í
¯®°± ²µ¶·
³´ ¸¹ ¹ Å × (A.5a)
Î Ï ÐÑ Ï ÒÓÔ
Õ
ßà
Û
Ü Ý áÞâãäß å æçèé ëì ìî ï ðîêï ñ
Ø ÙÚ (A.5b)
í òóôõö ÷ø
where we have set ùú û üýþÿ (see 2.10). The rst expression is obtained by integration of the
second one over the angular variable
, and by de ning
.
Decay Rate in the Static Limit
Let us rst analyze the static limit, i.e. for the form factor. To compute the decay rate in that limit,
we do not need to specify . We just need to know that the function is normalized to unity and behaves
as a delta function of the momentum in the limit of vanishing binding energy
! "#$
+,- . /0 23 > A BCAD =E@
%&'(# ) * 1 67 89:; < = ?@
45 (A.6)
M P QRPS LTO
By setting F GHIJ K L NO in (A.5b), we must recover exactly the lowest order decay rate
U LV `a
- W X Y ZZ[ \ ^] _ (A.7)
In some sense, the static limit serves as a boundary condition for the behavior of b . We get
yz
cdefghij jlm nopq r stu {|}~y y {z~
k |}~y
v w {x
Importantly, this result is independent of the binding energy : the loop does not introduce any corrections in the
static limit. The decay rate in that limit is therefore:
§ ÂÃ ÃÆ
¤¥¦§ ¨ «¬®¯°±² ²´ · ¸¹ º¸ » ¼ ¿À Á
- ¡ ¢ £ © ª ³ µ ¶ º ½¾ ÄÅ
ª
È Ã Ê ËÌÍÌÎÏ Ð. It remains to match Ñ such that purely kinematical corrections vanish (Ò factors in
with Ç É Ç
ÔÕ Ö Ô×Ø Ê × ÎÙ ÛÙÜ
the above formula arise from products like Ó Ò and from the Ú decay width factor, while Ý
à äå
comes from electron propagators in the loop and from the wavefunction Þß ). With the de nition á â ã æ,
98 Appendix A Resources for QED Bound states
êë
the decay rate is exactly èç é ë words, the value for ì obtained by matching (2.8) and (2.2)
as it should. In other
is such that no correction arises from factors íî inï the static limit.
To conclude, let us repeat that we have not speci
è ñ ed
ò the form factor. This means that any form ôõö
factor which
÷
has a three-dimensional delta function limit for ð gives the correct lowest order decay rate ó . In the
following, we shall present three forms, all built on the Schrödinger momentum wavefunction.
Schrödinger Form Factors and Binding Energy Corrections
Using either the formula (A.5a) or the derivative approach (2.11), we will now go through different calcula-
tions of ø ùú - û ü ý þþÿ, obtained for speci c choices of (or equivalently, ). Namely:
(A.8a)
!"
#$% & '%() (A.8b)
2 34 3
*+++ ,-./ 0 13 3 567859 ?@A
4 : ; <= > (A.8c)
IJ I
where BC D EC F GCH ECKCH . As discussed previously, LM isOjust the Schrödinger momentum wave-
function for the bound state. The NMM form factor has no clear signi cation and only serves the purpose of
illustration. Finally, we will show that NMMM can be viewed as a Coulomb binding corrected Schrödinger form
factor, in the spirit of the Sommerfeld factor. Of course, we know that the form factor to be used must be re-
O
lated to the Bethe-Salpeter wavefunction, and that it is the rst form factor NM that emerges from the reduction
of the Barbieri-Remiddi wavefunction through dispersion relations (2.8). Anyway, from the analysis of NMMM , we
will clearly identify the physical content of the binding energy corrections due to NM , and thereby avoid double
counting.
All these form factors satisfy the delta limit property (A.6), but are differently peaked near PQ P R S. Graphi-
^
cally, TUV WXYZ, as function of XY with [ \ ] (_` plain, _`` dashed and _``` small dashed lines) is
30
25
20
15
10
0.5 1 1.5 2
As a result, we expect that all of them will reproduce the lowest order decay rate (A.7), but will introduce some
a dependent corrections proportional to their spreading around bcd e f. Such corrections can then be expressed
in terms of the gne-structure constant, yielding a corrected lowest order rate of
h ij tu v {||
-k l m nno p rq s
q w xy z
The physics of each form factor is discussed below. If we naively add the present corrections }y to the
radiative corrections up to order ~ ~
-
A.1 Parapositronium 99
the theoretical value is modi ed as
Lowest-order Corrections Decay Rate, with
Form
£¤¥¤¦§factor
¨¦ ¦¤
© ª¡¢ Radiative
«¬®¯° Corrections
´ µ¶
± ²³ ·¸¹ plus ¡¢
º» ¼½¾¿ÀÁ Á¾ÃÄÅÁ Æ Ç½È ÉÊËÌÍ
ÎÏÏ ÐÑÒÓÔÕÖ ×ØÙÚÙÛ Ü ÝÞß àáâãä
åæææ çè êëìí ê îïïê ð èñò óôõö÷
é é
ùúûûüû ý þÿþþ þ
Experiments ø
where the experimental measurement (1.3) is also included for comparison.
Discussion
If the
is naively added to the order radiative corrections, one ends up with excessive corrections,
quite far from the experimental value. Our goal is to show that one should not add them, because it would amount
to a double counting. Obviously, rst order radiative corrections do contain a Coulomb photon exchange contri-
bution. In the language of NRQED, one should consider the exact integration result (2.15) as a resummation of
some higher order effects.
As stated in the text, the standard approach regarding is simply to discard it, hoping that it will
be generated by radiative corrections. This is inappropriate, but let us nevertheless continue with the idea, and
expose one of the technique used to get rid of the exact form factor integration corrections.
The basic idea is to absorb the Coulomb correction into the wavefunction. To this end, consider the third
form factor. We view it as a representation of the modi cation of the intermediate state due to Coulomb
photon exchange. When dispersion relations are used, one considers intermediate states as asymptotic states,
and integrate over the corresponding phase-space. This is the application of the optical theorem for absorptive
parts. Let us assume that, due to the modi cation of the asymptotic Hilbert space by the long-range Coulomb
interactions, a correction factor must be introduced in
& '() * +),- ./ 0 .3 4
12 56
- ! " ##$ % 7 58
9: ; <<= (A.9)
and that this factor is >? DE D
@A B CD <D FGHIFJ
E K (A.10)
In some sense, one could think of L MEN as the Sommerfeld factor of Harris and Brown [69]. Due to long range
Coulomb interactions, the wavefunction at contact OP is QrenormalizedQ by
RS RT U VWX VT YZ[\] i b kl m no
P ^ _ `abcdef g h j h p q (A.11)
by using p rr o, i.e. st s uu v, with w the center-of-mass velocity. Correspondingly, x ytz for small binding
energy is
v
x ytz { |t { |} ~
s s w
~ tz
The factor of arises because x y is dened at the amplitude level, while the Sommerfeld factor is at the decay
rate level.
The same arctangent factor arises in the work of [84], [87] on radiative corrections, by a detailed
analysis of the binding graph and its Coulombic part (which is, after all, an initial state
Coulombic interaction).
In conclusion, if one is interested in the lowest approximation (i.e. the ) the static limit is just
ne. If one wants the lowest approximation to accuracy, one must consider it as arising from the
binding graph, as demonstrated by Adkins [87]. In other words, by taking only diagrams for the scattering
100 Appendix A Resources for QED Bound states
amplitude , one nds
¡¢ £ ¤ § £ ¥¨
- ¥ ¦ © ªª « ¬ ®¯°
³
The ±² in the above expression is the lowest approximation to ´ µ¶· accuracy, arising from the ´ µ¶· amplitude.
To reproduce it using our effective form factor method, the Coulomb corrected Schrödinger form factor ¸¹¹¹ is
to be used, as in [87].
What we gained using our method is a better understanding of its relation to the Sommerfeld factor, i.e.
asymptotic Coulomb interactions. º¹¹¹ being more peaked than º¹, the suppression of the intermediate phase-
space is interpreted as a manifestation of the iterative structure of Bethe-Salpeter construction of the wavefunc-
tion. Therefore, the method of [87] is nothing but the rephrasing of that of Harris and Brown [69]. However, the
method of [87] or the present integration of º¹¹¹ is surely
¼ more appropriate, because one does not rely on static
limits, hence avoids the well-known static divergence » ½¾¿À of radiative corrections to ÁÂÃÄ Å ÆÆ.
A.1.3 Radiative Corrections to Parapositronium Decay
In this section, the computation of the radiative correction to the two-photon decay rate is presented. We will use
the standard technique throughout, i.e. compute the corrections in the static limit, and then use a Sommerfeld
factor to extract the Çnal IR divergence. This is a bit different from the Ë Harris and Brown computation [69],
since their work was simply to reduce the one-loop computation of È ÉÊ ÊÌ Í ÎÎÏ of Brown and Feynman
[65]. Here, we will do the complete computation.
The tree-level width is
×ØÙ Ü
Ð ÑÒ Ù Þ ß âÚ
-Ó Ô Í ÎÎÏ Õ ÖÚÙ Û Ý Û àá
Ü ã
with Ý the con guration space Schrödinger wavefunction of the positronium at zero separation. We intend
ã
to compute the radiative correction to order ä. To get a nite result, the associated bremsstrahlung processes
parapositronium å æææ also have to be calculated. Fortunately, thanks to selection rules (charge conjugation),
this decay is not allowed. Therefore, even if it could appear as quite exotic, this decay channel is maybe the
simplest realistic example of radiative correction calculation.
For the purpose of comparison, we present the computation using multiplicative renormalization and BPHZ
renormalization. Details can be found at the end of this section.
A.1 Parapositronium 101
i. Multiplicative Renormalization
The diagrams are
with in addition the crossed processes. The çrst one (è ) is the tree-level. For the others, the calculations are at
the end of this section, and we quote
î ê ï ð ñòó ê ï õö÷
êéë ìí í Oô
O
SE
ûþ ÿ üù
Vø+Vù ûúü ý û
WF +WF
O 1
+
B !"# $ %&' ( ) ,* ( ) O -(./0
@
P 2+P3 89
546 7 : ; < O =>?
where A B CDE FGD is an IR cutoff (G the electron mass, GE the photon mass used as an IR regulator). Let us
make some comments about the calculation. First, the self-energy insertion correction is computed from the full
propagator HIJ
W
KL M NOPQRSTUYX Z X[ Z\ Z VVW ]^ _` a bcde f ghijjkl
But this is more than is needed at order m, so we truncate the full propagator
n
op qrs q t upv w po q rs q yz x{ |z } ~
}
{ }
~ } |z }
{
The rst term generates the already computed lowest order correction, and can be discarded. The result quoted
in the table is then obtained with
(A.12)
102 Appendix A Resources for QED Bound states
The and corrections are a bit special. They arise from the LSZ formula. Indeed, the decay amplitude is
related to the -matrix element through factors, introducing the corrections
¡¢£¤ ¥¦-§¨ © ªª« ¬ ® ¯ °±²³´ µ¶-·¸ ¹ ºº» ¼ ½¾¿À Á¶-·¸ ¹ ºº»Â
à ÄÅÆÇ È ÉÉÊË Ì ÍÎ Ï ÐÊÑ Ì ÍÎÎÒÓÔÕ Ï ÒÓÖÕ Ï ×××
where
ÒÓÖÕ contains ØÙÚÛÜÝ corrections (strictly speaking, the Þgures are wrong because external lines are
amputated ß instead, they should be understood as reminder of the LSZ external line factors). Therefore, the
decay rate is, to order à
á âã-äå æ ççè é êëìíîïð é êëìíð ñ ò óóô ÷ ø ùúû ÷÷üýþÿ
øü
ë õö
õö
The
rst term is the lowest order (tree-level), while the rest constitutes the corrections.
The corrections are then computed from
%
$ & $ ' ()* +,
!" #$ (A.13)
while - corrections are .
56
/ 01234 2 0 7
89 :
The only effect of this last correction is to change the bare charges at the vertex of the annihilation ;<;= > ??
C G
to the renormalized one @ A B D EF . Speci cally,
V V
XYU XY c
HIFJ KL T V W T V W \]^ _
- M N O PPQ R S U S V R S U S V [ `
Z Z ab
Hence de t
fg vw
j r t u z{| } { |
h i k l m nno p q s q
- xt y ~
Therefore, as soon as the charge is taken as renormalized, we can forget about the corrections. Combining
all the other corrections, we get
¦ ¾
¨© ®± µ
£ ¦ § ° · º »¼½
¡ ¢ ¤ ¥
- ª¦ «¬ ®
¯
¬ ² ³ ¶
´· ¸ O ¹
with the renormalized coupling constant ¿À .
1Á Note that the combination Â Ã Ä Ã ÅÆ is IR Çnite, because the intermediate electron is off-shell.
On the other hand, the external lines introduce IR divergence trough È É and Ê. In fact, the result
we found is theÌÍresult of Harris and Brown, except that the velocity divergence is replaced by an IR
divergence in Ë Î . So this bad divergence is really a Coulombic divergence, due to the soft photon
exchange between the initial
Ð ÑÒ
state particles. Taking into account the Coulombic long range interaction
among the initial state Ï by introducing a Sommerfeld factor, the diverging term is seen to cancel
exactly, leaving ÓÔ
å
ÕÖ çè íð ó
Ú â å æ ï
×ØØ Ù Û Ü Ý ÞÞ ß à á ã ä
- ë ì ë
éå ê í î ñ ò
2ô The õö factors are seen to renormalizethe
electric charge at the vertex of the scattering ÷ø÷ù ú
ûû þ
. One could wonder how the ü of ý ÿ ý get renormalized. In fact, the standard view
concerning this is Hydrogen-like. Recall that in hydrogen atom energy level computations, one is intro-
ducing vacuum polarization
as a very small correction the Lamb shift. Usually, it is argued that all such
corrections are
.
A.1 Parapositronium 103
ii. BPHZ Renormalization
Now, the diagrams are
with again the corresponding crossed channels. First, note that there is no external line correction in the coun-
terterm method, because the elds have been rescaled. The various results are
$
"#
V +V O !
)* + , + /0 ,1
SE % '(
&
(- -.
D
5 6 7 89: ; < => < ; ABC
B ,2& 34 ?; O@
GH
I
CT E FJ
OP Q RS
CTK +CTL MN
UV Q RS W X
Note that the T part of the mass counterterm does not contribute (because an odd number of Dirac
matrix insertions into the fermion current vanishes). Also, the vertex counterterms are trivial to write down.
UY Z UV
Then, since , the CT amplitudes are (using (A.12) and (A.13))
\]
Y [ ^ b cd e fgh i j kl tu v w v q xyz {|
CT _ ` a m n o qp
r s
CT} +CT~ }
When combined with their respective amplitudes, we get
V +V +CT +CT
£ ¦ § ¨©ª £ ¦ £ ¨©ª «¬
SE+CT ¡ £¢¤ ¥
¸ ¾
¯ ° ± ²³´ µ ¶ · ¶ µ »¼½
B £¢¤ ® ¹µ Oº
All UV divergences have disappeared. The combination gives the ¿nal expression for the corrected decay rate
ÓÔÑ éê
ÀÁÂÃ Ç Ï Ñ Ò Ü Ý Þß Ý áÞ ã æ çè
ÄÅÅ Æ Ð Ø
-È É Ê ËËÌ Í Î Î ÕÖ × Ú ÛÙ à âã ä O å
104 Appendix A Resources for QED Bound states
No electric charge renormalization was necessary, since the lowest order is immediately computed with the renor-
malized charge as coupling. Note that the Ward identity ëì í ëî used here was also present in multiplicative
renormalization, in the form ï í ðëñ ïò .
Both the mass and the vertex counterterms are IR divergent. In turn, all the UV ónite corrections appear
IR divergent: the external lines divergences spill along the whole õfermion line. This is quite undesirable on
physical grounds, since it is really the radiations from the initial ô ôö state that create IR divergences. This
was especially clear in the method of multiplicative renormalization. On the other hand, the present method is
rather straightforward from a computational stand. The general philosophy regarding the BPHZ method is that
of a very powerful book-keeping, ideally suited to the systematics of perturbative expansion. Indeed, the ÷ø
insertions in multiplicative renormalization required an expansion of the full propagator. Beyond one-loop, this
kind of procedure quickly becomes too cumbersome.
The two methods are therefore complementary: multiplicative renormalization has a straightforward phys-
ical interpretation, especially concerning the origin of both IR and UV divergences, while the BPHZ renormal-
ization is in most practical computations (beyond one-loop) the only manageable computation method.
Details for the Computation of Radiative Corrections
The parapositronium decay rate is computed from the formula
* +, - +0 1 6
ù úû
!" #$% $& ' (() ./ 234 5
-ü ý þ ÿÿ 5
5
5
where 7 is the positroniumB mass, 8 the electron mass, 9:; the two-photon phase-space (with the associated
C DEFEGH
<=> @ I J
? A?
factor in front) andPQR the con guration-space Schrödinger wavefunction at zero separation.
O S TTU
The amplitude KLM N is theU scattering amplitude in the static limit, i.e. with both electron and
positron having momenta V W XYZ[Z [Z ^
[
. Finally, at the present level of precision, \ W ]Y.^
The scattering amplitude is modi ed by radiative corrections. We intend to compute the rst order correc-
tions to the positronium decay rate from the one-loop or order _ corrections to the scattering amplitude
fg h dei
`ab c deab abg h jjj
From this, the krst order correction to the width is obtained as the interference between krst order and leading
order scattering amplitudes
lmno pq } { ¡ ¢ £¤ ¥¦§¨©ª ±
- r s t uuv w xz
y x{ | ~ | «¬ ® ¯° °
´µ
Therefore, we will calculate the contribution to ²³ of each correction in turn. We will proceed according to
multiplicative renormalization (see the ¶gures above).
iii. Tree-Level Decay Rate
º
To tree level, the scattering amplitude ²·¸ is extracted from the scattering amplitude ¹ ¹» ¼ ½½ in the static
limit ¾ ¿ À Á
ÎÏÐÑ Ò Ó ÔÕÖ× Ø Ù Ú Þ ãå çè
æ é ê èé çè æ å ôõ
Å þÿ
ÂÃ ÄÆ Ç ÈÉ ÊËÌÍ ß à áâ ãä
ÛÜ Ý ë ìç í îï ö÷ ø ùú ûüý
ð ñ òóò þ ÿ
Note that this amplitude is manifestly gauge invariant. Squaring and summing over polarization,
! "#$ %& "#'$ () * +,-. )/)
(
(
A.1 Parapositronium 105
The width is then simply
J P
0 12 ?@ABCDEFE GHI GJ K LMN GHI GJ K N O
9 = >
-3 4 5 667 8 :9 ;< OJ Q
iv. One-Loop Corrections : Self Energy Insertion
There are two terms RST
[\ ]^ _
U V WXYZ ` `c b d e fg h ij gh
ab l likh m no p qrsttu
v
wyz
ox { | }|~
with . The self energy function is to one-loop
¤¥
® ± ¶ · ¸¹º» ¼ ½³¾¿
¡ ¢ £ ¨ ©ª «¬ ¬ ¯° ¬ ª° ²³ ´ µ
¦§ ¡
Æ ÍÉ ÎÏ Ï äåæ
À ÁÂÃ Ä Å ÇÈÉ ÐÑÒÓ Ô ×
Õ ÖÕ ÖÕ Ó ØÙ Ù
ÊÉ ËÌ ÚÙ Û ÜÝ Þ ßà Úá Ù âã
(A.14)
éê ëçìisí an
with IR cutoff, è is the dimensional regularization dimension parameter. The renormalization constant
îïð is calculated from (A.14) (an IR cutoff is not necessary, éê is IR ñnite)
ó üýþÿ
#$ % & % ' &
òó ôõö ÷ ø
ó û
!" "(
ôùúö ÿ
Now, computing the traces we get
8
)* + -, % &./012
" 3 4 5 6 97 8 :; <=> ?@ABCDECFD >G HIJ KLMNO
PQ RSTU V WXYZ[\] e
^_` a bc_ fgh d \]ijkilj
e eno vno v v v
using m a f p q rs t u p wrs t wu and r t x. Therefore, the correction to the decay width is
yz
obtained as
}{| ~-
¡¢ £¤-¥¦ § ¨¨© ª «¬ ® ¯³ ´
°±²
Since the self-energy function µ ¶·¸¹º ¼ ¿ »¹ ¼ ½. It
is not evaluated on-shell, it is IR convergent and we can set
¾
remains to work on the divergent integral. We have, expanding around
Ù
ÀÁ ¿ÃÄ ¾ÅÆ ÇÈÉÊË Ì ÍÎ ÏÐÑ ÒÐÓÔÕ Õ ÚÛÜÝÞßà áâ Ûâ ÛÞ ãäååæçèé ê ëì íî ï ðñ ò ñóôõ ñö÷ø ùúû
ÖÕ ×Ø
Therefore, the results are
üýþÿ -
*+ 7
!" -# $ % &&' ( ) , -/.0 123 4 56
106 Appendix A Resources for QED Bound states
Combining both self energy corrections, we get the 8nal result
HI
;< @
9:=> WX
? -A B C DDE F G J KMLN OPQ P M R S TUV M
Note that this result is free of any IR divergences. There remains a UV divergence in the result. This is expected.
Z
This divergence can viewed as the Y [ factors that will combine with the vertex corrections, ensuring the
\
niteness of the combination.
v. One-Loop Corrections : Vertex Corrections
\
There are four diagrams : the two photon permutations with a correction to the rst vertex, and the two corre-
\
sponding to the second vertex. Let us consider the rst two corrections:
{
]^_
` a b cdefgh ijk no l n un o pv w x n u p on v
m mp o q rs t s mp o nm yo q rz t
with |} ~ . The vertex function
is a quite complicated function of the momenta. A general
decomposition for it is
with a divergent quantity and the nite remainder, while stand for a complicated tensor and
Dirac structure. Let us write
¯ ° ³ ´µ°¶ ÄÅÆ
¥ ¦ §¨ ©¥ ª ¦ ª § « ¬ ©® « ¤
¡¢ £ ¤ ¤ ¤ ° ± ·¸ ² ¶¹º»¼» ½¿À
¾ ÁÂÃ
Ü Û ²
Ç ÈÉ ÊÉËÌ Í Î Ï Ô Õ Ö× ØÔ Ù Õ Ù Ö Ú Û
ÐÑ Ò Ó Ó Ó ÝÞ
One feature of this decomposition is that ß and à are the same for both diagrams, since
áÞ âãäã å æç è áÞ âã ä æ åãç è éê ëì âì å íî ï ð ñð ï íî ò ñí ï óî ò óôõ
ô ö ÷øùúø
with . Then
û ü ýþÿ ú !"#$
"%#
The resulting contribution is
56 KL
&'*(+) ,- 7 89: ; <=> ?@> A BC A DEF GH I J R R
-. / 0 112 3 4 7 M ONP Q S T UV
It remains to work on the Feynman parameter integrals. These can be written as
O U ]^[ _ ` k
RS W X PY abcd e f i e fj n op q rst ou ou q vs q w ow q vs x ov q ys x yz s{|}~}
Z[ \ g h l m
¡ ¢£ ¤ ¥ ¦ §¨©
³ ¶ ¼
ª ± ¶ ·¸ ¹ ¸ º »
¼ ½¼ ¾ ¿À ¾ Á ½Á ¾ ¿À ½¿ ¾ ÂÀ ÂÃ
£ « ¬ ¯ ® ° ´ ®² µ º º
Ä ¾ ÊË Ì
» º ÅÆ Ç È É
A.1 Parapositronium 107
Note that again, there is no IR divergence in those integrals. Finally, the vertex correction is
ÛÜ ìí
ÍÎÑÏÐ ÒÓ àã ä áå æ
- Ô Õ Ö ××Ø Ù Ú Ý Þàßá â æ ç è éêë
where we have multiplied by two to account for the (identical) corrections to the second vertex.
vi. One-Loop Corrections : Wavefunction Renormalization
As explained in the main text, we only need to compute the electron wavefunction renormalization constant
îï î
. In the on-shell renormalization scheme that we are using, it is given by ï ð ñ ò óôõ ö÷ø. This is can be
evaluated from (A.14)
ùõô ö÷ø ú
!"
û þ ÿ
üý
" # $% ! & '
?@A B CDE
) , -. /01 23 0 45 67 0 3 0 89: 4;< = > H
( *+ NO
CE F G B CD I J K L M
Q TUV U W U X YZ[ \]
P RS
To get the correction to the decay rate, iwe consider the four diagrams with external electron line renormalization.
_
There is a factor ^ ` a b c d̀ ef̀ gh on external lines, i.e.
jkl m qr s x
mn st x y | }~
direct o p Crossed u v w
v z{
The leading order reproduces the leading order amplitude for . The nal correction induced by the
wavefunction renormalization is
£¤
- ¡ ¢ ¥ ¥¢¦ §¨© ¨ ª ¨ ¥ «¬ ®¯
vii. One-Loop Corrections : Binding Correction
°
We now turn to the most dif cult part of the calculation: the binding graph. The amplitude is
¸¹ À Þßàá
±²³ º
´ µ ¶ º»·¼½¸ ¾ ¿ÀÁÂÃÄ ÅÆ ÇÅ È É ÊËÌÍ ÏÐÑ ÏÒÎ Ó Ô ÕÖ×ØÙÚÛÜ âãÝ ä åã
¿
with æç è é ê éë and
ò ò
ìíî ï ðñòóôõö òó ù òó òó
÷ ÷ü û ý þû ôÿö þû ôÿö ÷ø û ú÷ ù ÷ü û ý þû ôõö
÷øù úû
where the two terms stand for the direct and crossed diagrams. The decay amplitude is constructed as before.
can simplify the calculation by noting that the crossed diagram denominator is equal to the direct one if
ø We
.
This motivates us to change the sign of in the crossed diagram. The trace then gives
&'()*+,-*./01 230456: 7/187/8
"! $# 9% 38 : 05647/8 7./01
90;<=>?@A?B@ C<=>?@A?D@ EFGH I70/JHK ML
70/ 23. 0564
N O PQR STUV SWX YZ[ \] ^_`a bcad efg hijk lmkn opq rsqt
with the denominator
108 Appendix A Resources for QED Bound states
u
From here, the procedure is standard: introduction of Feynman parameters, loop momentum shift, and loop
momentum integration in -dimension. We get
vw xyz{ |}~ ¡
¢
¤£¦ §¨© ª «¬ ª ® ª «¯ ° ±² ±³´ µ¶ ·² ±³´ Æ ÇÈ ÓÔ
£¥ ¸µ¶ ±³´¹ º »¼ º½¾ ¼ ¿ÀÁÂ Ã Ä Í Å
ÉÊ ÇËÌ Ç ÉÆ ÇÎÌÆ ÏÐÑ ÒÕÓ
Ö
After the (quite dif cult) integration,
×Ø ÙÚÜÛ ÝÞßàáâãäåæ ç èéê ãäíîï ó ô õ ô óù ô ý
ëê ì íðî ñò ö÷ø úø O ûøü
Finally, the contribution to the width is obtained as
þÿ
-
O
With this nal correction, we can now combine all the results. This is done in the main text.
A.1.4 Dispersion Relations and Kaon Decay
#
The amplitude for !" "$ is computed at lowest order in a pion loop model. The pion is treated as
a point-like charged particle, which allows simple scalar QED treatment. Special attention is paid to the low
energy behavior of the amplitude.
Decay Amplitude and Loop Integration
The amplitude for kaon decay into %&'&$ is given by
VW
J K H L MN KOPQ
( )*+ 0 56789 :;< GH I F F abac d eef
, - ./ /12 3 4 = > ?@?AB CD
EF RS T UZ[\ ]^_ `
XY
A.1 Parapositronium 109
with g h i ji k h l m n. The opqr s tt amplitude arises from the one-photon and two-photon (the so-called
seagull graph) coupling
as uvw x
yzy{ | }}~
¡¢ £ ª«
¢¤ ¥¦§ ¨ ©
®¯
with © the pion mass. The amplitude ¬ ° ± ²³²´µ, taken as a constant, has been factored out. The
integration is done using dimensional regularization to preserve gauge invariance and get a ¶nite result. We
obtain
· ¸¹º Ì Ø ÝÞß
ÄÅÄÆ Ç ÈÈÉ Ê Ë Ñ Ò Ó Ö Ò Ó× Û Ü âãä åæ ç èé ê è ã æ äë ì í îï ð ñíò
»¼½¾¿ ÀÁ à ÍÎÏÐ à á ó ôõö÷øù
Ô Õ Ù Ú
The denominator function is ú û üø ýþ ÿ ÿ ÿ with the de nitions
and the kaon mass. In the kaon rest-frame,
with the reduced photon energy
. When integrating over Feynman parameters, the last term vanishes. Therefore, the gauge invariance of
the amplitude becomes manifest
+ - X Y Z[ \ Z X
! " # $%&%'( ) * , 56 AB CD EFG HIJ KL HM J NOP HMQRS TUV W Z] U V
./012 3 4 7898:; =<> ? @ (A.15)
where the Feynman parameter integral is
k pq
^ _`a bc d e f i e fj n o v
r s u s vw pq x q q s rw x yz
g h l m ot o t
The prescription yz gives the sign of the imaginary part.
Feynman Parameter Integration via Dispersion Relations
vw
To calculate { tu| , we shall use dispersion relations (see for example [6], [219]). A direct integration is
possible, but we shall gain insight into the dynamics of the process by using dispersion techniques.
110 Appendix A Resources for QED Bound states
viii. Absorptive Part Extraction
A }rst integration over the Feynman parameters gives
~
¡ ¢ ¡ ¢ £
£ ¤ (A.16)
² ³ ² ² ³
where ¥¦ § ¨©ª « ¥¦ ¨¬ ®© ¨¯ ° ± ° ´µ . The imaginary part of ¶ ·¸¹º comes from the values of for
which the °´». Integrating on the
³ argument of a logarithm is negative. There the logarithm imaginary part is
relevant values, we ¼nd
ÈÉ ÊËÌ Í ËÌ Î ÏÐ Ñ ÒÓÔÕÖ
½¾ ¿ ÀÁÂÃÄ Å ÆÇ Ô × Ù
Ø ÙÚ Û ÜÝÞ ß Ú Û àÝ
ãä åæç è æç é êë ì íîïðï ñò
Ûá â ô
ó ôõ ö ÷øù ú ÷ ö ûø (A.17)
or, recalling the deünition of ý and þ (see [208])
$
ý þ
()* 0 1 234 7 ;< =>?@A
ÿ % & '+ * , -.* / 5 4 6
#!"" . 89 :
O UVW j nop
D FGHI QR SQ T XWYZ[\] v w { xyz}~
B CJ IKL MN P^ _ `^ _ abc k l mq p r st p u z|
(A.18)
E dcefghi t
Obviously, the rst line in (A.18)corresponds to the vertical cuts (which contribute only if the decaying
particle s mass is larger than ) while the second line corresponds to the oblique cuts (which contribute
only if the virtual photon energy is larger than )
A.1 Parapositronium 111
¤ ¤
Further, one can see that in the soft photon limit (or ), ¡¢£ behaves as a constant.
Indeed, applying L Hospital s rule twice, we get
¥¦§ ¨©ª ª ¯° ±²³ ¸ · ¹º»¼
«¬ª ® «¬ª ´ µ¶ ½ ¼ ¿ ¾ ÃÄÅ Ã (A.19)
¾ ÀÁÂ
ÄÇ ÊÃ Ë ÆÄÌ Ã. We will come back to this
while individually the contribution of each cut is divergent as Æ È
point later.
ÀÉ
ix. Dispersive Part Integral
Ñ ÐÕÔ Ö Õ×ØÙÚ Ó ÐÕÔ Ö Õ ÐÛ ÜÝÞßàáâ ÍÎ Ï ÐÑÒÓÔ as a function of the avail-
In order to write down the unsubtracted dispersion integral, let us express
able energy through and . Then
ãäå æç æèéê ëì íîéêê ï ð î
ó ÷ø ù úû úõü ýþ úõüü
ñ ò î ôõö (A.20)
with õö ÿ , such that we can omit the principal part. For such a kinematics
In the next subsection we shall analytically continue to the physical value . The result of (A.20) is
easily obtained in terms of the integrals
+
#$%&'( ) * , - . 4 1 2 5 6 1 6 27 8 9:;<=>? @ 9
! " +, / +,01 2 3 3 AB
J
valid for AB C D (see [6]). We can write E FGHI with K L G L D and K L I L D as
b
GHIJMN O O [ \] ^ [ _]] ` _ a \] ^ a _]]
EF P QR N ST U QR N STV WX
YZ Z Z Z Z Z (A.21)
no x
in terms of c def g hijklm p and q rps t uvw
x yz{|}~ .
x. Analytic Continuation
The and , are obtained by analytic continuation of (A.21). The analytic continuation of
results for
and are
yz{|}~
¶
¡ ¢£¤ ¥ ¦§ ¨ © «¬®¯ ° ·¸ ¹º» ¼ º» ½ ¾¿ ½ ¾
© ± ²³ ´ µ ÀÁÂÃ
© ª
³
The equalities hold for Ä Å ÆÇ ÈÉ Ä Ê Ë. The right hand sides deÌne the analytical continuation for Í Î Ï.
Hence the result is conveniently expressed as (A.21) with
Û ÜÝÞß à áÞ áâ
Ð Ñ Ò Ó Ô ÕÖ×ØÙÚ ï òóô
Í å ç ç ì í î ì
ã äæ ä è é è ê ë ï ðñ
(A.22)
û
õ ö÷ø ù ú
!"#
ü ýþÿ
112 Appendix A Resources for QED Bound states
Had we chosen to analytically continue $ and % in the lower half complex plane, their imaginary parts would
have had the opposite -sign. One can verify from (A.22) that the upper half complex plane analytic continuation
reproduces &' ( )*+ , as given in (A.17). This result corresponds to [212] where a sign mistake has to be
corrected, and to [218], [216].
Decay Width, Differential Width and Low .s Theorem
0 123 E F GHI
The differential decay width is given by / 4 5 67689: ; => < ?@ABC D D . After a straight-
forward integration over the electron-positron phase-space, we get the differential rate in terms of the reduced
photon energy J:
h v
GK LMN WX hi j k q
O P QRQS TU V h lm n op
rs t u
u
Y Z[\ fg wxy z{ | }~ u
] ^ _`_ab c de
e
with and . The bounds on the integration are ¡¢£ ¯ ¤ and ¡¥¦
§ ¨©. A numerical integration of the differential rate gives the prediction for the rate ª«¬ ® ®°± relative to
²³́
µ ¶·¶¸ ¹
ÀÁ Å
Â Ã Ä ÄÆÇÈ Ö× Ø Ù× ÑÚ
º»º ¼ ½ ¾É ¿ÊËÌ (A.23)
Í Î ÏÐÏ ÑÒ Ó ÔÕ
ã éê ë ìíîï ò õö÷
äå æ contribute for çè
where the real and imaginary part of Û ÜÞÝß à áâ and ðñ ð ó ô , respectively.
Let us turn
to the soft photon behavior of the differential width. We have to analyze the loop integral function
ø ùú ÿ
ûü ý þ . This function tends to a constant for very low :
!" # (A.24)
(note that for $" % &, the imaginary part of (A.24) is the same as in (A.19)). Let us emphasize the strong cancel-
lation between the two cuts in the soft photon limit. Individually, each cut diverges as '(, but their combination
/ 0 12 3 14567
is convergent. Since the remainder of the amplitude tends to zero like ) (due to the *+,- . factor),
9
Low 8s theorem is veri ed: the amplitude behaves as : close to : ; <. The resulting spectrum is thus in := (:>
from the squared amplitude and a : from phase space).@ In other words, had we forgotten one cut, the resulting
spectrum behavior would have been E divergent as ? : instead of vanishing like :=. In fact, this := spectrum is
exactly the same as in the AB C D DFG differential decay rate.
xi. Comparison with the Two-Photon Decay Mode
From the amplitude (A.15), we readily obtain the amplitude for HIJ K LL by removing the electron current, the
photon propagator and by taking the limit M K N:
Y _à b cdcef
O PQRS T UUV W X Z mn op mrn st kq m u vwx y vwz{|} ~
[\ ] ^ gh ijk l ql l
with
i.e.
·¸¹º»¼½¾ ÙÚÛÙÜÝÞß
¡ ¢£ ¨ © ª«¬ ¿¾ àß
¤¥ £ ¦§ °± ´µ ¶ÀÁÂÃÄÅÆÇ
® ¯²³ Ì Í Ó Ó áâãäåæçè
ÈÇ É Ê Ë ÎÏË Ð ÑÒ
(A.25)
Ò Ø éè êëë
Ò Ô ÕÒ Ö×
A.2 Orthopositronium 113
which corresponds to the result given in [6], [218], [214]. The relative decay rate is :
ì íîïð ñ òòó
ô õö÷ø ù úûúüý þ ÿ
!" # $! % & $'()
while the experimental value for this ratio is [223].
We can now compare the two electromagnetic modes :
/0
* 1 2 343567 D
++ , - .
8 9:;
< = >>? @ ABAC
in accordance with [208]. This ratio EFF is similar to [223]
G HIJ K LM LNOP
\
Q RST U VVW X YZY[
The small difference is due to the phase-space factor. The similarity between a constant coupling model like for
] T U ^_^`a
and the present loop model can be understood from the behavior of the photon energy spectrum.
Indeed, the pure phase-space spectrum is very strongly peaked at high b, and therefore the decay rate is quite
k o x
insensitive to the details of the c dfeg h ij
lm n function. Speci cally, if we replace p qsrt u vw
yz { by its value for
| } ~
(A.25) we nd the ratio
, very close to (A.23).
The discussion of the more general case of non-constant form factor, i.e. if the amplitude
is not taken as a constant, can be found in [29], [30] and references cited there.
A.2 Orthopositronium
A.2.1 Phase-space Structure
The goal of this section is to set up the kinematics of three-photon decays, along with numerical methods for
phase-space. Suppose that we are considering a process , a
parametrizing the associated three-body
vector particle, with coupling constant . When there are no angular dependences in the amplitude, the only
relevant variable are the energy of two of the photons (see C.3)
® «¬ ®« ®«
® ´ ¬ ´
¡ ¢ £¤¥¦§ ¨§ £© ª «¬ ª « ª «®¯ °® °® °® ° ° ®
±²³ ²´ ¬ ±²³ ²´ ±²³ ²´ ±
¯ ¯ ¯ ® µ ¶ ²³ ¯
¬
The amplitude is cast into a function of only · µ ¸ , ´ and ´. This is achieved by ¹rst removing one of
the photon momenta using the energy conservation º» ¼ ½ ¾ ¿À ¾ ¿ÎÁ . Then, there remains only scalar products
such as  à ¿Ä Å ÆÇÈ since the É particle is at rest Ê Å ËÆ Ì ÍÌÍÌ Í , and
ÏÐ Ñ ÒÓ Ô ÕÖÕÓ × ØÙÚ Û ÛÙÜ Û ÝÞß àáâ
Momentum conservation can serve to ãx this angle
â é êëìí é êëìî ï êìíìî
äåæ àáâ ç è ê ðñò ó óñ ó
ô
114 Appendix A Resources for QED Bound states
ø
Hence õö ÷ õø ù ú û ö ý øþ ÿ ú
ü ü
Finally, if we dene the reduced photon momenta as , we get
!"#
$ $ 34 2 9 : ;<=>?@? AB
#% & ' )( *+ ,-. / -0 1 2 5 67 8 ; A
@ E F GHI EJ GH DKI EJ
The decay rate is then obtained by integration over C D and . The Lnal variable we will use
S
is the reduced mass of the particles running in the loop M N cOPQR Q . In terms of all these variables, the
UV b
differential decay probability is T WXY WQ Y WZ Y M[ \ ]^_` a a . Then, the decay rate is expressed as
defe gh i jjjk l m m
no pqrstuv w xyz |{ }~ z
z ¡
{
Note how we factored the coupling constant from the amplitude.
Of utmost interest to us will be the differential rate (i.e. the photon energy spectrum)
¢ £ ¤ ¥¥¥¦ µ ¶ Ç
² ³ ´ »¼ ½¾ ¿ÀÁÂÀ¼ ÂÃ Ä ÀÁ Ä À¼Â ÅÆ
§¨© ª ¬
« ®¯°± ¶·¸¹ º
Ê
ËÌË ÍÎÏÍÐÑ.
and the normalized differential rate ÈÉ
Numerical Approximation for the Phase-Space Integrations
The amplitudes we have to manipulate are very complicated, and, in general, numerical methods are the
only alternative. Further, to save a substantial amount of computer time, it is a good idea to fully exploit the
symmetry of phase-space. The method we present here is borrowed from Adkins [87], and will be used in all our
computations (except for the simple Ore-Powell approximation, because it can be straightforwardly integrated
analytically).
Energy-momentum conservation implies that the photon reduced energies lies within the plane ÐÑ Ò ÐÓ Ò
ÐÔ Õ Ö, and since Ð× Ø ÙÚÛ ÜÝ, phase-space is restricted to a triangular area
x3
1
0 x2
x1
1
A.2 Orthopositronium 115
The differential probability Þß àáâãáäãáå ã æç is fully symmetric under permutations of áâãáä ãáå . This means
that only a sixth of phase-space needs to be computed
èéê èéê ã èéêç
where the center-point corresponds to à ã . We choose the darker sixth of the triangle in which
áä ë ìí ë ìî.
ðñ
Now, to get an estimate for the phase-space probability distribution, we have to evaluate ï ìîò ìó òìí òôõ
at various points inside the dark triangle. To this end, we can introduce auxiliary variables as a coordinate system
for the dark triangle: ö
÷ ÿ
ýþ
ùú û ü ÿ
ø
When
runs between and , the
spanûthe darküarea. Then, we divide the triangle into
cells,
or bins (here with and
):
ü
For ,
, the point at the center of each bin has coordinate:
!
"# $ % !%
We can then construct a matrix out of the differential probability (for a given value of &)
'( )*+, -. / 012 34 5 67:89 ;< 5 89= >: 5 67? ;@A
It is the computation of that matrix that takes so much computer time. Of course, the values of B and C to be
F FF H
used to get a reliable answer depend on the assumed smoothness of the differential probability. As we will see,
in most cases, a coarse grained mesh B DC of E to G points is suf cient to get a total rate at the percent level
J
of precision, because of the high redundancy of phase-space (i.e. I points in the dark triangle means K LI
J
H
points for the whole phase-space). Note nally that the chosen parametrization has one technical advantage in
116 Appendix A Resources for QED Bound states
that evaluations of the differential probability on phase-space boundaries never occur. This is welcome since a
numerical evaluation there is highly unstable (if not divergent), especially numerical integrations.
Once the matrix is found, we use some Mathematica tools to build interpolation functions. Namely, we
construct an interpolation like
MNOP QRSTU V WXYZ[\]^_Y`]X abcd befg hi j klm no p klqr qr wwxy
s n o p t us pklm nv z{
Then, an interpolation to the whole phase-space is built by expressing back |}~ in terms of reduced photon
energies, as
x
¦ § ¤ £¡¢¡¢¨
¡¢ £ ¤¥ ©¡
£ ¤ª
¶ ½ ¼ º »Á
«¬ «® «¯ ° ±²³´ µ ·¸ ¹ º» ¼ ½¾ º¸
¿ À¹ º» ¼ ½Ã
ÄÄÄ
Note that at this step, we rely on Mathematica to extrapolate the interpolation beyond its range, since boundaries
are now included (if one suspect that the differential probability is not smooth, sizeable loss of precision could
occur. Increasing the number of bins is then compulsory). From this interpolation of phase-space, the differential
rate, spectrum and total rate can be obtained simply by numerical integrations.
A.2.2 Euler-Heisenberg Lagrangian
The Euler-Heisenberg Lagrangian [201] is the one-loop QED effective action. It can be computed using the
Schwinger Proper time technique [10], [18], [28], [203]. For a constant Åeld strength, the result is expressed in
closed form to all orders as
ÆÇÊÈËËÉ Ì Í Ï Î Ö ÚÛÜ ÝÞßàáâ ãäåæçè éßãàá åæç éßäàá
ÐÑ ÒÔ Ó ÕÖ× ØÙ çêëè éßãàáçêë éßäàá ì í î ïð
ù úûù
with ñòó solutions of ñô õ öô ÷ ø and üý þ ÿ . This expression was obtained in 1951 by Schwinger
[203].
To get the various couplings, we expand the term in square brackets as a series in . Each order corresponds
to a new term in the effective Lagrangian
with
"
&'(
" # $%
!
, =
)*+ 45 6 758 9 : @AB
- . / 10
2 3 ; <= >?
K
F O R STK K R TO U
CDE Q Q
G H IJ
LMNO P
^_`a
Y
VWX hij k lmnop q lmp on q rmst
Z [ \ bcd
]
ef g
|}~
x
uvw
y z {
¡¢£¤¥ ¦§
¨°± ² ³ ´´µ¶·¸ ¹ ººµ» ·¼ ³ ººµ¼ ·» ¹ ´´µ¸ ·¶ ¹ ½¾ µ¿² À
¨©ª« ® ¯
¬
A.2 Orthopositronium 117
ÂÃ Ä Å ÆÇÈ É ÊÈË ÍÎ Ð Ñ ÒÓ Ô ÕÖ
Transcribed into ÁÂÃ Á and ÌÍÎ ÌÏ :
Ú ç
×ØÙ ëìí
Û Ü ÝßÞ
àá â ãå ä æ
ç è éê
ñ
îïð
ò ó ô ö÷
õ úû ü ý
ø ùúû ù þ ÿ
"#$
!
89;< = >? 89; B
%&'( 5 6789: :89: C C DEGHI
) * + /01
,-.
23 4 @ A DE F
S ]^`a b cc d ]^`e f hikl m n
JKLM UZ [\]^_ _]^_ ghi gj o pq u rsvwx
N O PQR
TUVWX rs t
Q Y
¡ ¦§¨ © ¯°
yz{|
¢£ ¤¥¦§¥ ª ª «¬®
{} ~
«¬
·¸
The ±rst term is just an in±nite energy shift, and can be set to zero ² ³ ´. To compute the UV divergent µ¶¹
term, a regularization procedure must be introduced. Using a simple UV cutoff
Ä Ï
·¸ ÇÈÉ Ê ËÌÍ Î
º
µ¶ » ¼ ½ ÀÁ ÐÏ Ñ Ò
¿ ¾ ÃÄ ÅÆ
And one can recognize the photon wavefunction renormalization constant, up to irrelevant Ónite terms.
Feynman Rules for the Four-photon Vertex
Instead of directly give the Feynman rules associated with the Euler-Heisenberg Lagrangian, we consider
the pieces ÔÕ
Ú
à áâäå
Ö× Ø Ù
ÛÜÝÞ ß áâã
è
æçèé ê ë ñ óõ öò÷
ìíîï ð òóô ôõö ô
þ
øùúû
!"
ü ý
ÿ !
where #
$ %& ' ( %&+,$
)* +,
and the identity in the third line is easily established from contractions of Levi-Civita symbols.
For each term, the Feynman rules are (with the momentum -ow . / 0 1 2 1 3, i.e. . in-going, 04243
out-going)
GHIJ
56789
:;< =>? @? AB CD E F
KLMN OP
e
QRSTU bcd i k
VWX YZ[ \[ ]^ _` a fg j l
h lmn o
pqrst
¡¢ £¤
uvw xyz {z |} ~
118 Appendix A Resources for QED Bound states
where
¥¦ § ¨¨©ª«¬ ®¯°±²³ ´µ¶·¸ ® ¶¸ ·µ³ ¹ ´µ¯±¸ ®¯¸±µ³ ´²¶·° ® ¶° ·²³ ¹ ´·¯± ¶ ®¯¶ ±·³ ´²¸ µ° ®¸° µ²³³
º» ¼ ½¾¿ÀÁÂÃÄ Å ½Ä¿ÀÁ¾Ã Š½Â¿¾ÁÀÃÄ Å ½¾¿ÄÁÀàŠ½Â¿ÄÁ¾ÃÀ Å ½Ä¿¾ÁÂÃÀ
ÆÇ ¼ È¾Ä Éʽ¿ËÁÂÃÀ Ì ÍÎÏÐÑÒÓÔ Ì ÍÎÑÐÏÔÓÒ Õ ÍÏÑÐÓÒ ÎÔ Ì ÍÓÏÐÑÒ ÎÔ Ì ÍÓÑÐÏÔ ÎÒÖ
×Ø Ù ÚÛÜ ÝÞßàáâãäå æ çèéêëìíî æ çèíêéìëî ï çéíêèìëî æ çëéêèìíî æ çíëêéìèîð
ñò ó ôõö ÷øùúûüýþÿ
!" !" # ! " ! " ! " $
%& ' ()* +,-./0123 4 56789:;< 4 5798;: 6< = 56987<;: 4 56;89: 7< 4 59;87: 6<>
?@ A BCD EFGHIJKLM N OPQRSTUV N OUQRPTSV W OPURST QV N OUSRPT QV N OPSRQTUVX
YZ [ \]^\_` abcdebfge h bdgebcfei h \]` \_^ abcge bdfe h bcde bfgei h \]_\`^ abdfebcge h bdge bcfei
The Euler-Heisenberg Feynman rules can then be reconstructed since
u
jklm n qrop t vwx yz{|z{|}~
s
£¤¥¡¢ ¦ §
£¨¥¡¢
In the text, we prefer to use the Lagrangian under the form
©ª«¬ °±®¯ µ¶·¸¹ ·¸º» ¼ ¾½ ¿¹·¸ ÁÀ ÂÃÄÅÆ
²³ ´
Ç ÈÉÊËÌ ÐÑÒÓÒÔÕÖ× Ø ÙÉÊ
ÍÎÏ ÚÛÜËÌ Ý ßÞ àáâãä
ÚåÜ
to exhibit explicitly the scalar and pseudoscalar currents. The Feynman rules remain of course identical.
A.2.3 The Photon Two-point Function
The photon vacuum polarization reads
ðñ
æáâ çèéê ë ìíé î òóï ö ÷ø ùúû ý ü ý ý ü
ôõ þ ÿ ü þ ÿ ü
Using dimensional regularization techniques, which preserve gauge invariance, we nd
:;<=
! ' " #% +,%-. %$ & /
()* & 12 34 560 347829
Expanding in terms of > ? @ A B, with CDE F
GHI J KLMNOP Q RORPST UVWX
T UVWX Y Z \[ Z _[ a d d ed f ghijk lmn op qr opstn
] ^ ] b` c un v w x yz{
| } ~
z
A.2 Orthopositronium 119
Note that we may also integrate over the non-expanded expression, with the result expressed in terms of
hypergeometric functions
¡ ¥¦§ ¨© ª «¬ ¬ ¬® ±¯°
¢ £¤ ²
°³
´
A completely integrated expression for the nite part can be obtained simply as
µ¶ ·¸¹º » ¼½ À à à ÄÅ ÆÃÇÈÉÊ ÄÅ Æà ÄÅ ÆÃÇËÇ
¾ Á¿ Â (A.26)
with Ë Ì ÍÎÏÐÎ. A direct integration gives (see for example [9])
ÑÒ ÓÔÕÖ × ØÙ ÜÝ Þ àáß Þ âà ãä å æç è ãäé êçè ë ìíîï
ÚÛ
íî
The function ë ì is
ñ õ ýþÿ û
òò ö÷øù ú û ü ü
ò
íî ð óò
ëì
òò
òô $ % &$ ' !"#(
% )*+ , -.
"#
where each term is obtain by analytic continuation. In fact, the /rst two lines are identical, due to the properties
of arctangent functions. The last equality is valid in the upper complex plane
0 5 CD VW
789 E F OPQ R S TR S U
1.23 4 5 6 6: ;<=> ? @ AB GHI J KLM N VW X YZ[
R X TR S U
_`a `
The real part of \] ^ , for all b , is therefore
n
cd efgh i jk j uvw x y zy { |vw} ~
l mo p rs
q qr t
Two-loop Vacuum Polarization
The photon two-loop vacuum polarization function is ([7], [217]):
«¬ ° « ±µ ±µ ¼ ½ ¸¹ ¼ ½ ¼¾¸¹ À
¡¢ £ ¤ ¥¦ ² ® ¸¹ º » · · ¸¹
§¨© ® ±¯ ²³´ ±´ ¯ ¶· ¿¸ ¶ ·
¦ ª
¼ º ¸¹ ÍÎ
½» · » Â Ã ÄÅ ÆÂ Ç ÅÈ É
¿¸ Á Å Ê Ë ÍÌ Ê
É
120 Appendix A Resources for QED Bound states
with ÏÐ
ÐÑÒ ÓÔÕ Ö×Ð Õ
ÕØ Ù Ú
Û ÑÒ Ó Ü ÝÑ Þß à áâãä åæç ß à èâãä åç ßé æ ê ëèâãì åæç ß à âãì åç ßé íî ï
øù
ðñ òñ íî óô õ ïö õ íî óô ð ïö÷ íî
ù ú ûü ý þÿ ý þ
þ ý þÿ þ
A.2.4 The Photon Four-point Function
We are interested in the four-point function only for a speci c kinematics, i.e. that of the decay of a massive
photon into three real photons. We can pictorially represent our process as
Where the is an arbitrary coupling constant.
The corresponding amplitude is given in terms of the vector to three-photon current [113], [205]
!" #$%& * 56 >?@ ABCD H
) -./01 2 3 789
' ( +, 4 : 9; <= E F IJ
G KLMNO
T [`
PQRSTUV WX YZ[ Y\] ^abc
_ d ce f
ghij k
g n opq rs n otq rs n ouq p t u
where l m f . The v correspond to helicity amplitudes (indexed by w w w ), whose
yz o{ |}~ | }
arguments x are short-hand for , and also . This expression must be summed over all
polarization states . Because of numerous identities among the helicity amplitudes, the modulus squared
of the amplitude can be expressed in terms of only four of them. Conventionally, we choose
¢£¤¥
¦ § ¨ ©ª« § § º»
¾¿ÀÁ
Â
¡¡
¬ ®¯° ¯ª ¯± ² ³´ µ ·
¶ ¸¼½½
¹
ÅÆ
The dimensionless Ã Ä are given in the next subsection. In term of these, the modulus squared of the current
is Ç
Î ÊÉÊË ÊÌ ÎÐÑÒÑÓ ÑÔÕÖ× Ø ÙÚÛÜÝ Þß àáÚâã ä ß àÚáâã ä ß àâáÚãå
Ï (A.27)
ÈÉÊËÊÌ Í
with
á óô óô ô ó
ß àáÚâã Ø ç èéê ïðñ ò òõ öó÷ ïðñ ò î òõ öú÷ ïðñ õ öú÷ ðïñ ò
â æ ëìì íî ò
ò
ò øøø íî
ò
ò
ò
ó
ùú òò øøø íî øøø íî ò
ò
æ
ôæ ùú -
!" &'( ) % !" '&( ,
### $%
ùó ùû üý þ ÿ
%
) *+ ### $% ,
,
,
A.2 Orthopositronium 121
The ./s for other arguments are obtained by permutation of the 01. Note that there is no restriction in the
contraction (A.27) since current conservation (Ward identity) implies
2 8 K L
9 545657 9;<=<> <? X Y Z X Y Z\
: @ C EFGH I JMN
AB D J O P Q
3456 57 RSRT RU VW W W V[W W W
where the last factor is a scalar function, and where ] ^ _` a _b a _c . Therefore, the longitudinal part of the
vector boson polarization sum never contributes.
The amplitude is then simply
d ef g hhhi j k k rs t } yxyz y{ }
lm no p q u v ~
wxyzy{ |
The three-body phase space is
and therefore
¥¦ ¥§ ¨ ¯«ª«¬ « ¯±²³ ²´ ²µ¶·¸
¡ ¢ £ ¤ ¤ °
©ª«¬ « ®
½¾¿
À
¹ º ÂÃÄ Å Æ ÇÈÉ ÇÈÊ ËÌ ÍÎÏÐÑ Ò Ì ÍÏÎÐÑ Ò Ì ÍÐÎÏÑÓ
»¼
Á
Helicity Amplitudes
The amplitude and squared amplitude can be expressed using helicity amplitudes. Let us note
Ô Õ Ö × ØÙ Ú Û Õ Ö × ØÜ ÚÝ Õ Ö × ØÞ
122 Appendix A Resources for QED Bound states
ø
à áâãäå íî
Then, we have the four independent amplitudes êëì ñòó ôòõ ô òö÷ :
ß æçç è éïðð
óû ÿ
ùú
üüü ý
þ
% &'()* + % &,()*- . /0 0 789:
! " # $ 1 2 345 6
@ G K K KG K K
; <=> ? = AB RSTU
C CH I LJ D L G D NL M BH D C D
D EF GOP Q
I I I
F
X YZ [ \] `\ d d\ d Fd d
VW [_ [ f [ [ [ klmn
^ ^a b e c[ ` ge [ `]a ^ ` \hi j
q rk s tu t
op s wx s { z s z
v vy st v|} ~
ª «¬®¯°®¯ ±®²
¡¢ £ ¤¥¡ £ ¦§¡ ¨ ©
¶ » » É ÍÇ Ì ÎË
³´µ¸¸ ÄÅ ÄÅ
¹ º¼ ½ ¾ ¿ ÀÁ ¼û Á ÂÆû Á ¾ û ½ Á ÂÇ
Ä ÈÉÊË Ì ÔÕÖ× ØÕÖ× ÙÕÖÚ
· ÏÐ ÑÒ Ó
Ö ÜÙ Ý Þß
Û Ò ÓØÕÖ×ÞÕÖ× ÙÕÖÚ
ØÔ à
for the árst type, and
éê ê üý ý
äå
âãæææ ç
ô ö÷
è ë ì î í êï ðñ òó õ ñ òø õ
ô öù ú û ý
þ ÿ ý ÿ
ì ÿ
5 6; 6
#$ & # ,-./ 0 1234 2 8 9: 2 2 < 5 ; 2 =< ABCD
!" % (% 67 6 2 6 >? @
# '$ # )* + 4 9:
GHA N O PQ N!"O PQ
EF O O YZ[\
IJ K L M P K R M ST NO
R S NO R T K URP VW X
[_ [ bY c de xy x }~ { x
]^ ] m nopqr spqr tpqu v w
`a fg f dk l
h ij z{ | zy |
¥ ¦§¨©ª« ¨©ª ¬¨©
¡
¢ £ ¤
®¯°±³³ ¸ ¸ ¸ Â Â Â Ã Ä ÅÆÇÈÉÊ
² ´ µ¶ µ ·¹ º » º ¼ ½ ¾¿ À¹Á¸ º ¿ À»Á¸ º ¿ À¼ Á¸
È ÈÎ Ü Üß è èï
ËÌ Ë
Í Ó ÔÏÕÖ× ÐÕÖ× ØÕÖÙ Ú Û ä åæçèéêçèé ëçèì í î í ñ ò óô õðö÷ø ñ ö÷ø ùö÷ú
ÏÐÑ Ò Ý Þ àá â ã æ ð
for the second type. The form factors appearing in the amplitudes are
!"#$ %&
û õüý þ ÿ
( . 1 1 6 789
& ' ( ) *( + ,( * & - / 0 2 3 2 4 5
:;
I J KL J M NI J OPQR b
`a cd ef
< =>? @ A B E FG H y
C D S T UV W X fY Z
gh jkl mno p no q rs t u v
[ \i ]^_
wx
A.2 Orthopositronium 123
The constant z (with a limit z { | understood) appearing in the argument of the logarithm is necessary to
unambiguously select the correct sheet, i.e. to give a sign to the imaginary part. The result for } ~ is
then obtained by analytical continuation, with the convention z transcribing as a lower half complex plane
continuation (a z would lead to an upper half plane continuation). Note well that both form factor have to be
de ned with the same conventions. The last form factor is more complicated:
where
£ µ ¶· ¸ ¹º ¸ » ¼½ ¾ ¿ÀÁÂ
¡ ¢ © ³´ È
¤¥ ¦ § ¨ ¦ª«¬ ® ¯± ° ² À
à ÄÅ ÆÇ
As for É and Ê , an explicit form for Ë can be given above and below threshold. For ÌÍÎ Ï Ð,
Ò Þß àá
â ã äå
Þß àÜ
ã
ç ê ëì í î
ï û
ð ô õö ÷øù ú ü ý þ
Ë ÑÓ ÔÕÖ×Ø Ù ÛÚÜ Ý äå æ Ü ç Úèé î
ï ð ñòó ÿ
where the angle is
Denoting ! " #
$ % '& , the analytic continuation to ()* + , is
8 ?@ A BC
9
9 <= > BC D <= >E A F G JKL M N OP QR STW X
UV
9
9 E G HI VY
- .( : ^ _ ` a b b ^ _ ` a b
* /0123 4 657 9 XZ[ \] ` a cd a e fg ` h ij
9 p qr p wx
; klmn op q st q uvn oy z {| z }~
9
A.2.5 The Ansatz
Here we collect the expressions for the differential rate and total rate using the ansatz (2.28).
The differential rate can be expressed as
¤¥ ¦§¨© ¦®¯ °±²
¡ ¢ £ ª© « ¬ ³² ´
with µ¶ Í Î Ï Ð ÑÒ
·¸¹º» ¼ ¶ ° Á ÂÃÄÅ Æ ÇÄÈ ÉÊË Ì Ó Ô Õ Ö×
º ½ ¾¿À
where
ØÒ Ù Ú ÛÜ Ý Þßà Ý Þáàâã Ý äÞ Ý åæç èé ë ìèíé ë ì îêï ë ðñæ ë ìòóôõ
ÞÝà êëæ êëæ ö÷ ø ùú ø ù
úü ú
û ù ú û ÿ û
ýþ ø ø ÿ
( ) *+ ) ,+-.
! " ! & '
/01 2 34 2 356
#$ % ! % ! &
124 Appendix A Resources for QED Bound states
The total rate is given by
DE DJK
7 89 : ;;;< = > CF
?@ ABC G GF (A.29)
F HI FH
with
Q QM eh
MN O P S Wc d c d f
R S QUTTQ T MNW X Z[ \Z Y ^_` a b efg e f hg ih f jgkl
IL SV X]
L
no p qrs x t usy } z} | ~y
fm t vw
q t us z{ | }~y z | {~y
¡ ¢¡£¤
¥¦ ¡§¨ ©ª
« ¼½
¬® ¯° ± ¸¹ º » ½ ÀÁ¹ º » ÅÆ ÇÉ Ê
È ÍÎ Ï Ð Ñ Ò
² ³ ´µ ¶ · ¼½¾ ¼½¾
»¿ »¿ Â Ã Ä ÈË Ì Ð Ñ ÓÒËÔ
125
Appendix B
Resources for QCD Bound States
B.1 Running of the Strong Coupling Constant
In this section, we will shortly review the basic formulas for the running of the strong coupling constant ÕÖ.
It is based on [11], [209], [210], [213], [222], [223]. The basic formula is the renormalization group evolution
×Ú
equation ×Ø ×ÚÚ
× ÛÜ
ØÙ
ÕÖ
ÙÕÖ Ù (B.1)
Ü
In perturbation theory, the function can be calculated from loop corrections as
×ÚÚ ÜÞ
Ü ÛÝ
ÙÕÖ Ù ßà áâã äåæ ç êèëìé ï ðñò ó ö÷ôõ ûü ýþÿ
íî øù ú
with
#$%
! " $& '(!
The next term )* is also known, but too complicated to be given here. Note that only )+ and ), are scheme
independent.
B.1.1 One-Loop Running
To compare the values of -. obtained in various processes, one has to use the renormalization group equation
to /run/ 01 2345 up to 67, which is the conventional comparison point. The world average for 89 :
;<= is
>? :;<= @ ABCCDCE ABAAF
To lowest order, the coupling constant running is
=
GH>?G:G @ IJK O
LM NP QRS T UP QRS V YZ [\ ]^_`WXa
H _bcde (B.2)
h
where fgbcd is an integration constant. For speci c runnings, we may cast this formula in several useful forms.
126 Appendix B Resources for QCD Bound States
1i Taking the difference at two scales, we get
m nop q rst
jkl u vwx y |z{} ~
(B.3)
Since , the coupling constant decreases as the energy increases, a phenomenon called asymptotic
freedom.
2 To lowest order in ¡ , one sometimes uses
¡ ¢£¤ ¥ ¡ ¢¦¤ §¨ © ª« ¬® ¯° ´ µ¶ ¼½
¾ ¿ÀÁÂÃ
±² ³ · ¸ ¹ º » (B.4)
3Ä Forgetting about Å, we can transcribe the running as an equation in terms of ÆÇÈÉ :
ÊË ÌÍÎ Ï ÒÓ Ô ÐØÙÚØ Ñ è
with ßÛÜÝ à á âãä åæ éêëçì íîïð
Õ Ö× ÛÜÝÞ (B.5)
ù
Hence once ñòóô is known, so is õö ÷ø at any scale ø.
The quantity úû is a function of üý . This dependence comes about through vacuum polarization diagrams
with üý quark þavors in the loop. Quarks much heavier than the typical energy in the process must decouple,
and at a given energy scale ÿ , only üý with ý lighter than ÿ must considered. The running is therefore a
function of the number of quark þavors used. For example, taking
,
0.35
0.3
0.25
0.2
0.15
0.1
1 2 5 10 20 50 100 200
(plain) from top to bottom. This
" (i.e. #$is %an&):illustration of asymptotic freedom since
as . This is true as long as !running
with
0.25
0.2
0.15
0.1
0.05
1 5 10 50 100
for '( ) *+,+-. and *&.
Running Down to Beauty, Charm Scales from the Weak Scale (I)
As an example, we will take the running of /0 down to the charmed quark mass, which we take at 12 )
3*456 . This example
-=> 7
is interesting since one has to de ne how to treat the 8 quark threshold, 9: ; < between
DE G
and ?@ A BC F , and HI J B between KL and KM. Of course, the threshold is not a step function, and
B.1 Running of the Strong Coupling Constant 127
N
the detailed evolution close to it is mass-dependent. Anyway, as a rst approximation, such dependences are
O
ignored.
1 The exact running using either (B.3) or integrating numerically the differential RG equation (B.2), gives
a strong coupling at PQ of
RS TUVW X YZ[[\ ]^_`
a bc defg h ijkllm nopq r st uvwx y z{|}~} (B.6)
(note that we are taking too many signi cative numbers for the purpose of comparison). Ignoring the threshold
gives the values
¡¢ £¤¥¦ § ¨©ª«¬
®¯°
± ²³ ´µ¶· ¸ ¹º»¹¼
½
2 The approximate formula (B.4) gives
ÍÌ ÎÏ ÐÑÒÓ Ô ÕÖ×ØÕ ÙÚÛÜ
¾¿ ÀÁÂÃ Ä ÅÆÇÇÈ ÉÊË Ý Þß àáâã ä åæçèé
which is quite far from the exact one-loop running (B.6). The reason why this approximation is so poor is
because ê
î ô
ë ìí
ïðí ñ òó
The formula (B.4) neglects terms of order
õö÷ øùúûüýþ ÿ
(in case we would use (B.4) to go upwards
, this becomes , really bad indeed). For
the last step
!"# $ %&%'
( )*+,
In conclusion, the approximate running formula should be used only to run between closely related scales, which
- ./01
are not too low (i.e. ).
3 For the last scheme, we take the value of according to
2/01 34567 8 9: ;<= >?BCD @AGHIJ KLMO N PQPRRSTU
EF
Then, the value of VW XYZ[ is B.5
\W XYZ[ ] `a bc d ^_ ij s r tuvwwx
efg h klmn opq
As before. To continue down to yz, we have to change {|}~ as
{|}~
¡¢£
and ¤¥ ¦§¨© ª «¬®¯®, exactly the same result as before.
128 Appendix B Resources for QCD Bound States
B.1.2 Running Beyond One-Loop
The general RG equation (B.1) is integrated as
° ±² ° ±´
²³ µ ¶´·
If we expand
µ ¶´· up to µ¸, we have
¹ ÆÇ ÈÉ
º »¼½ À¾¿ Ø Ù ãä ë
ÚÛ á â åæçèéê
¿ Á Â ÃÄÅñ ÊËÌÍîõÎÏÐö Ñ ÔÕÖ×
ÒÓ îïîù ÜÝÞßà
úûü ù ûü ù
ý ú ü ü ýüù ü ü ù
ò ÷øó
ì îïí ð ó ô îï ô þÿ üù ý
ÿ ùü
To go from the rst to the second line, we have integrated, then Taylor expanded the result around
and nally moved constant terms into . This expression must now be inverted to get as a function
of . This is not an easy task in
general, and we proceed iteratively. Let us introduce the convenient
notations !
"#$ '%& *+, -
& ( )
The .rst order is simply /2
/0 1 34 567 89 : ;< =>? @ CDE
AB
)
as before. The next order is found from
FD GJ FM NOP K T
GE H I K L FD Q UVRS
W X Y Z[\]^_`a defghijk
bc lm
RS
xyz{ y { x
n op q tu
rs
v w y|} ~ y } y} ¡ ¢£¤¥¦
The constant § is chosen as ±
§ ¨ ©ª« ®¯ °
ª¬ ª¬
so as to eliminate the constants in the logarithm. Then, solving for ²³´ and expanding the result around µ ¶ ·:
¸¹´ »¼½
¶ º¼¾¿ µ¾ ÀÁ µ
i.e. Ó ØÙ ÚÛ ÚÛ Þ ÜÝ í
¹´ ÃÄÅ Æ È Õ Ö × Ýßàá òóôõ ö÷
Ç
ÉÊ ËÌ ÍÎ Ô âãä åæ çè î ï ð ñ ø÷ùúû ü (B.7)
ÏÎÐÑÒ é êëì
è
Note that the integration constant ý is now interpreted as ýþÿ (i.e., ýþÿ is de ned by this expression).
Obviously, this method can be pursued to higher order. The result to three loops usually quoted is
4 7 89
! ! " # %&' $ ( )*) + , - /00
. 1 2 3 56 (B.8)
'
)'
Let us emphasize that contrary to =>the one-loop result, this expression for the running is not exact, since
we have made an expansion in :;< . Anyway, in general, ?@AB is estimated to be a few hundreds of
CDE DE
is getting better as the scale increases. Around FG , we can estimate
HIJK L, so
M that
NOPQRS TUV W XY Wparameter
the expansion
Z, so that the above formula should be quite precise (at least in the perturbative
B.1 Running of the Strong Coupling Constant 129
regime, remember the initial [ expansion of the integrated RG equation, and that the value of \ is calculated in
perturbation theory).
_`a def
Using the world average value of [] ^ , we can get the value of bcghi using jkl- and m-loop running:
pq
norst
1 loop
uv wxy 2~
loops 3 loops
~
l z {| } z
z
|
¡¢ ©
¦¥§ ¨¨ ¤ª«
££¤¥¥ ¦ª¬ ¨ £¤ª¦ª®
° ²³´
¹· determination comes from the logarithmic dependence of ¯ on ± . Note that we have
¶· imprecise
The rather
given ±µ²³´¸ ±µ²³´ just for the purpose of illustration, since one usually considers º» ¼ ½ close to ¾¿ .
Running Down to Beauty, Charm Scales from the Weak Scale (II)
Now that the two and three loop running formulas have been presented, we can use them to get the value of
the strong coupling at the ÀÁ and ÀÂ mass scales.
As a Ãrst exercise, it is instructive to compare the exact running (i.e. integrating numerically the differential
equation) to the approximate formula (B.7) and (B.8)
ÄÅÆ Ç ÈÉ
2-Loop 3-Loop
ÊË ÄÌÍÉ ÎÏ
(B.7) (B.1) (B.8) (B.1)
ÐÐÑÒ ÓÔÕÕÖÖ ÓÔÕÕÖ× ÓÔÕÕÑØ
âããä
The convergence is manifest (with one loop, we obtained ÙÚ ÛÜÝÞ ß àá ).
Integrating årst with æç è é to êë, and then with ìç è í to êî, we get
1-Loop 2-Loop 3-Loop
ïð ñêîò õö÷ õø õùø
óô óô ó óô
To compare the result with one, two and three loops, we draw the plot of ïð úûüý (note þÿ was kept
throughout the plot range)
0.4
0.3
0.2
0.1
2 5 10 20 50 100 200
The plain line is the -loop result. One can see that in the low energy range, the -loop (bottom line) is down
by more than . Since the series is alternating, we can expect the exact curve to lie between the -loop and
-loop results (in fact, the -loop result would be indistinguishable from the -loop one).
130 Appendix B Resources for QCD Bound States
Finally, to illustrate the effect of the quark threshold, let us plot the running with
xed at and with
changing to at
0.45
0.4
0.35
0.3
0.25
0.2
0.15
2 5 10 20 50
One can see that the two curves are identical down to
, and then, the exact curve (plain) with is
accelerating upwards.
B.2 Standard Annihilation Rate Computations
This appendix contains the computation of quarkonium decay rates in the static limit, i.e. assuming .
B.2.1 Pseudoscalar Quarkonium Decays
Pseudoscalar Quarkonium Electromagnetic Decays
The transcription of positronium results is straightforward. The two-photon decay is given by
% ' ( . / 5 6 78 5 ;<=> BC D
!"# $ & $ 00 1 2234 9: ?@ AE DA (B.9)
)*+,) -
If we ignore the binding energy and express
E as FGH (as is required in the present approach), we arrive at
W Z
I JKL MNNO P QQO R STU VXYZ ^_ `
[\ ] ]
The rate into abcbd (or any pair of light leptons) is equally straightforward to obtain
ef ghi gjjk l mno npq v y z{|}
rs t uw x ~ (B.10)
where and the phase space spectrum is
¦
¡ ¢ £ ¤¥ § «¬
¨©ª
B.2 Standard Annihilation Rate Computations 131
The total width ( ranges from ® to ¯ ° ±):
Î
² ³´µ ¶··¸ ¹ º»¼»½¾ ¿ ÀÂÁ ÃÄ ÅÆÇÈÉÊ ËÌÏÍ ÎË Ð ÑÒÓ (B.11)
with ã
Ð ÑÒÓ Ô ÕÖ×Ø Ù Ú ÛÚ ÙÜÝ Þ ßàá â Þ äå æ ç
å æ äå æ çè
The ratio of the two electromagnetic modes is
éê ëìí îïïð ñ òóôóõö ÷øù
ú ûüý þÿÿ
Except for the , this is standard. Integrated, we get
*
! "" # %$& ' ()
Pseudoscalar Quarkonium Decays into Two Gluons
Let us calculate the amplitude
+ ,-. /001 2 331 4 5 ;<=>? @AA B CCDE
67 8 9 :
HH I JJK
The scattering amplitude F G has three contributions: the L and M channel (intermediate virtual quark)
and the N channel (virtual gluon, with three-gluon vertex).
\]^_ ` a _ ` cb d_^ v w x yz{|}
OPQR STT U VVW X YVZ [
e fg h ijk lm no p qr st uy ~z
¡¢ £¤¥¦§¨§©ª« ®¯° ±²³ ´ µ¶·¸ ¹ º»¸ ¼½¾ ¿ ÀÁ ¿ ÃÄ ÅÀ ¿ ÆÇÁÈÉ ÊË ÅÆÇÁ ÊÌ ÅƾÁ
¬ Â È
where À Í Æ¾ ¿ ÆÇ Í Î ¿ ÎÏ and ÐÑ Ò Ð ÏÑ Ò ÓÑ. However, the Ô-channel amplitude vanishes in the trace,
Ö×ØÙ
because only one (traceless) Gell-Mann matrix Õ Ú is inserted in the quark current. The color trace for the
other contributing amplitude gives
Û Ü ÝÖÞÖß à á â ã äåæåç è á é êãë ìíî ï ð ìíî
ñ (B.12)
Then the two-gluon amplitude can be written as (external projectors have been used)
ò óôô õ öö÷ ø ùúû ü
ýþ
ÿ
Now that the color structure is factored out, we recover the QED amplitude, which is gauge invariant when
! " !#
. Contrary to the scattering case, the three gluon vertex is not needed to ensure gauge invariance. Then,
using known results $
$
A
% &'( )**+ , --+../ 0 1 / ;< <
1 56 7 89: = > = ? @B CDEFGFH
2 56 34
7 89: /
Note that in the gluon polarization sum, we simply use IJK , since transverse states do not couple to the quark
current.
132 Appendix B Resources for QCD Bound States
The decay width is
l
L MNO MPPQ R SST U V V ^_ `a bcd effg h iijkl mno ml p qrst z{ |
Z
W XY[ \ ] ` k w
uvx y y (B.13)
Pseudoscalar Quarkonium Decay into Gluon-Quark-Antiquark
This is the analog of the decay into a photon and an electron-positron pair. The color structure of the
amplitude is seen to factorize, and
}~ ¡¢£
}
¢
The ¤nal formula is then ¥ ¨©
¦ § §ªª« ¬ ®®¯ ´ º» ¼ ÀÁ ÂÃ
°± ² ³µ ¶·¸ ¹½ ¼¹ ¾ ¿ÀÄ
Ä ÊÄ
where  ŠÆÇÈÉ with ÇÈ the Ënal quark mass. Typically, the spectrum function Ì ÍÎÏÐÑ is (for Ð Ò ÓÔÓÕ)
Ö ×x,aØ
100
80
60
40
20
x
0.5 0.6 0.7 0.8 0.9 1
The total width ÙÚ
ÛÜ ÝÞÞß à áââß ã äå ëì í
æ çè î íê ï ðñò
é ê (B.14)
î øù
For example, with óô õ ö÷ MeV, the function ú ûüý is
þÿ û ý
ú
Pseudoscalar Quarkonium Decay into Three Gluons
There are two topologically distinct contributions to this decay. One is obtained with the three gluon coming
from the quark line, and the other from the two-gluon decay amplitude, with one of the gluon going into two
!""# $
gluon by way of the three-gluon vertex. All these amplitude are of order . As discussed in the text, the
%&'
inclusive rate light hadrons is related via duality to the rate into all possible gluonic states. To
leading order, it is the two-gluon rate. To order , one has to consider the rate into ((( ())
, and strong
((
radiative corrections to the rate into . Such a computation is not very complicated, but rather lengthy, and will
not be pursued further.
B.2 Standard Annihilation Rate Computations 133
B.2.2 Vector Quarkonium Decays
Vector Quarkonium Electromagnetic Decays
* (=3) and substitute + , -./ +. This gives for
As for the para-state, we just have to include a color factor
01 1 >?@A C FGH IJ M
23 445 6 7778 9 :<; =BC
three photon decays:
DE KL (B.15)
NO PQR PSST U VVVT Z[ ^_`a g h i
Y \ ]bc def d jkl i h i
with the differential width
WX (B.16)
and the usual three-photon positronium decay spectrum (note that contrary to m jnol jkl
phase space, which is pjk here)
, is not the pure
h ijkl q p ip rstu wxy rstust |}
xy (B.17)
st v xw rstuz v{ rstu ~
Plotted, this is
d dx1
2
1.5
0.5
x1
¡.
0.2 0.4 0.6 0.8 1
and
¢
For lepton pair production, we nd
³
£ ¤¥¦ §¨¨© ª «¬«® ¯ °±²³´µ³ ¶·¹¸¶³ º» ¼ ÀÁ ½¾¿ ÅÆ ÇÈÉ Ê
Â Ã Ä ËÌ
where ÍÎ originates from the amplitude and
Ë from phase space. Since we are assuming Ë Ï ÐÍÎ:
à
Ñ ÒÓÔ ÒÕÕÖ × ØÙØÚÛ Ü ÝÞßàáâà ãæäåãà è é íêëîì òó ÷ø ôõö
á ç ïð ñ ù
and when ÷ö úú ÷ù ûü ü
ýþ ÿÿ
(B.18)
Vector Quarkonium Decay into Three Gluons
Now we turn to the amplitude
"#$%&'() .. / 00012 34 789 :; 7= 9 :> 7@ 9
! *+, - 56 <6 ?6 (B.19)
134 Appendix B Resources for QCD Bound States
with
o
ABCD B J ^_^`^ab c d yzy{y|}
EFG H I KLMN OP N Q R S UV W UXYTW Z [\ ] \ fgh i jkei l m n pq r pst r u vw x
~ ¯°¯±¯²³
w
¡ ¢ £¤ ¦§ ¨ ¦©ª¥¨ « ¬ ®
Á Ü
´µ ¶ ËÌËÍËÎÏ Ð Ñ æçæèæéê
¸¹º » ¼½·» ¾ ¿À ÂÃ Ä ÂÅÆ Ä Ç ÈÉ Ê É ÓÔÕ Ö ×ØÒÖ Ù ÚÛ ÞÝ ß Ýàá ß â ã ä å
where ë ì íî ï íð ï íñ ì ò ï òó and ôõ ö ôóõ ö ÷õ . There is no three gluon vertex since we must
have three gluon vertex insertions into the quark current to get a non-vanishing trace in (B.19) (also, a diagram
øø ù ú ù úú ù úúú
, of the same order, does not contribute because of the tracelessness of Gell-Mann
matrices). Since all gluons emerge from the quark current, no ghost amplitude need to be considered and gluon
polarization sums can be taken as úûü.
This amplitude is again expressible in terms of the three photon decay amplitude. There is two non equivalent
permutations of color matrices to be traced : ý þ ÿ and ý þ ÿ . For this, we use the identities (B.12)
and
We can then work out the traces as
!" # $
% & #!" # $
Let us consider the 'rst pair of amplitudes:
4 j
()*+ JKJLJMN O P efegehi
, -./0 1 2 3 56 89:; 8<7 ; = >? AB C AD @C F GH I H RS Q \] ^ \_ ^ a bc d
E T U VW U X Y Z `[
The Dirac trace will be the same for both terms, hence we arrive at
k hgf p q r stutvtw x twtvtuy z r {|} ~
no
clm
¡¢ £ ¢ ¤ ¥¦ ¨© ª «¨ ¬§ª ®¯°
The same manipulations apply for the two other pairings, and we are left with the three-photon amplitude
¶
multiplied by the color factor ±²³´µ . This can be understood since the ·nal gluons should be symmetrized,
hence only the symmetric tensor ¸¹º», and not ¼¹º», contributes. We thus arrive at
ÊËÌ ÏÑ
ÒÓ ×Ø Ù ÚÛÜÝ â ã Þ Ý ÜÝ â ã Þ ß ÜÝ
½¾ ¿ÀÁ ¿ÂÂÃ Ä ÅÅÅÃ ÆÇ È ÉÊËÌÍ
½ Æ Ý Ý Ý Ý Ý Ý
ÎÏÐ ÔÕ Ö Þ Ý Þß à á Þ ÛÞß à á Þ ÛÞ Ý ä
which equals, up to a color factor, the modulus squared of the three-photon decay amplitude. Using
ë
å çè åæçè é ê
æ ì
and integrating over phase space, we arrive at
í îï ð îññò ó ôôôò õ ö ö
ú ü
û
÷ ÷ø ùý üù þ ÿ
(B.20)
B.2 Standard Annihilation Rate Computations 135
with the differential rate given by
!" #$% #&&' ( )))* ./0 67 8
+, - 1. 234 59 85 : ;<=> (B.21)
Vector Quarkonium Decay into Two Gluons and a Photon
This time the amplitude is
J IKL RSTUVWXY ]] ^ _``ab cd ghi jk gm i j go i
? @AB @CCD E FFGD H IM
N OP Q Z[ \ ef lf nf
with
peln e tuvw txsw yz{ }~ } | ¡ ¢£
dk q r
Ê
¤¥ §¨©ª «¬¦ª ®¯ ±² ³ ±´ °³ ¶·¸ ¹º»º¼½ ¾ ¿ ¸ Á ÆÀ Ì
Ë Í Ë Î Ï ÍÐÑÒ ÓÔÕÔÖ×
µ ÃÄ Å ÄÇÈÉ
å ïðïñò ó ô ÿ
ØÙÚ ÜÝ Û í ö÷øùúûõù üýþ
Þß àá ß âãä æç è æéê è ëìí î
where and . The color trace is just yielding a delta function (B.12) and
there is no subtlety associated with symmetrization. We thus recover again the three-photon amplitude, with the
() *-./0 /0 345 6 789:
; : 9: ;< 9:
result
! ""# $$% & '() ' 1
*+ , 20 ;::;:< = >?;@:8;:< = >?;@:8;:: A
and then
R U Z^ _``a b ccdaef g hij kl f m no q q vw q
B CDE CFFG H IIJK L M M S T YZ [\ ] e
N OP QV U W X [ n prsst ux qu
with again the photon spectrum
yz {|} {~~
Other Vector Quarkonium Decay Channels
Following the same line of reasoning as in the pseudoscalar case, any other decay channel into gluons and
quarks will be considered as radiative corrections to the inclusive decay rate to hadrons. Other exclusive and
semi-inclusive decay channels therefore necessarily involve photons and lepton pairs. For example:
¡ ¢ £ ¤¥¦ §¨¨© ª «¬®¯° ± ²
(due to color)
³ ´µ¶ ·¸¸¹ º »»¼½¼¾¿ À
see [152]
Á ÂÃÄ ÅÆÆÇ È ÉÉÊËÌÍÎ Ï
should be obtainable from the above
Other
ÐÑ ÒÓÓelectromagnetic
Ô Õ ÖÖÖ modes are strongly suppressed (recall that the 2001 upper limit on the branching ratio for
ÛÜÝ
is ×Ø× Ù Ú ).
137
Appendix C
Useful formulae
C.1 Wavefunction for Hydrogen-like Atoms
The complete wave function for the hydrogen atom can be written
Þßà ãäåæ ç è
áâ éêë ìíîïîðñ ò ôó õö÷
The corresponding energy spectrum is degenerate for ø and ù, with
úû ü ýþ ü
ÿ
This is the well-know Balmer series. The time-independent part is
! " #$ ! % !
The radial wavefunction is expressible in terms of Laguerre polynomials
'()* 1 2 3 2 456 O
#$ ! " & BCD EFGHFIJ
+,-. / 01 7 3568=9:;< > ?@ A KLF LJ MPQ
N RS
0
while the TVU are standard spherical harmonics. The complete wave functions can also be written in terms of the
hypergeometric functions representation of the Laguerre polynomials
Vab k l mno w q r q xst
W Z[ \ [ ]^ _ `
~
XV Vade Vafgh pk l pno m q rst uv r y wst z |{ }
UY c i j
v
j j
Explicitly, the rst wavefunctions are
§¨ ©¨
Shell ¡ ¢£¤¥ ¦ ¢ª
« ¨¬ ® ¯ ¯ ®° ±
²³´µ ¶·¹̧
º»¼ ½ ¾ ¿ ¿ ¾À Á Ê ÍÎ Ï
ÂÃÄÅ Æ ÇÈ ÉË Ì ÐÑ Ò
Ó Ô ÕÖ× Ø âã ä åæç è
Ù ÚÛ Þ ßà á
î ÜÝ
é êé ëìíî ï øù úûüý þÿ
ð ñòóô õö ÷
! " #$% 6 9 : ;<= >
34 5 78
'()*& +,- ./ 12
0 S TU V
? @? ABCD E GHF K LM ONP QR XWY Z [ \]^ _
` a bcd F IJ t uv |
f eh k lm onp qr s wx y z {
} ~{ eg ij
®¯
©ª «¬ ¬ °±²¬ ³
¡¢£¤¥ ¦§ ¨
138 Appendix C Useful formulae
with ´
µ º»
µ ¶ ·¸¹
¼½» ¶ ¼¿¾
The wave function at the origin can also be calculated. Only for À states are the values non vanishing
ÁÂÃÄÄ ÅÆÇÈÉÊË Ì ÍÍÎ ÒÓ Ô Õ Ö×ØÓÙÚ Û Ü à äåæç è é è ïìðì
ÏÐ Ñ Õ Ñ Ù ÝÞ ß áâ ã êëìíîì êîì
÷ø ù ú
since ñòó ôõ ö and û
üýþ ÿ
C.1.1 Momentum Space Wavefunctions
The general expression of the momentum space wavefunction was found by Podolsky and Pauling [58]
34 4
)*+,- 56 7 89 6 : ;<=
? <I
!" # $# %& ' ( .
/0 1 2 > ?@ A ;<= BDC EFG H
NOPQNOP XY ZY [ [ \]
J KLMR ZY ^V ] W
STU VW a ijk jl mno
r p qo
V W _` o s qo t u
`fh
v _b c dbe`f b g
It is found by taking the Fourier transform of the con guration space
y wavefunctions (the integration is not
straightforward though). The wyx are Legendre polynomials, the z{ are Gegenbauer functions, and
|} ~ }
The rst few are
Shell
¡
¢£¤ ¥ ¦ § § ¦¨ «¬ ®¯° ±² °³
©ª ´
µ¶·¸¶¹º
» ¼ ½¾¿ À Æ ÁÂÃÄÅ Ê ¾ ËÌÍ Î
Ç ÅÈÄ ÅÉ
Ï ÐÏ ÑÒÓÔ Ô× ØÙ ÚÛÜ ãäåæ çèé ê
ÕÖ Ý
Þ ÜßÛ Üàá â
ëì í î ï ð ð ïñ ÷øùúûüýþúÿ ü
ò óôõ ö
! " # $ %&' (
) *) +,-. /0 12314567 8 CDE
;<=>?=@6956:
A B F GHI J
K L MNO RSTUVW a bcd ef g hi
P XQYZ[VZ\ ] ^_ `
stuvwxytxz
e jh aklm no pqrp{
|}~} ¦§¨© § «¬ ® ¯
¡¢ £¤ ¥ ª
³ ¶·¸ ¹º»¼½
with ° ± ² ´µ .
C.2 Loop Integrals, Gamma Functions and Related 139
C.2 Loop Integrals, Gamma Functions and Related
C.2.1 Dirac Matrix Algebra
¾¿¾
¿ À Á
¾¿¾Â¾ Ä Å Æ¾Â
¿ À Ã Á Å Æ
¾¿¾Â¾Ç¾ ÂÇ Ê ¾Â¾Ç
¿ À ÈÉ ÃÁ È
¾¿¾Â¾Ç¾Ë¾ ÅÄ¾Ë¾Ç¾Â Ê Å Æ ¾Â¾Ç¾Ë
¿ À ÃÈ Á
C.2.2 Feynman Parameters
To combine denominators :
Ì á âã ä àåæ
ÍÎÏÏÏÐÑ Ò ÓÕ Ô Ö× ×ÙÚ ÛÜ ÝÞ ß à ç î
Ö èéêé ë ììì ë íîêîï
ÔØØØ
Note that derivatives like for example ð
öû
ñò ó ô õö üý÷ øù ú
ê ö öû þûÿ
ø øù ú
are easily obtained from the general equation. We will only quote, since it is used in the text, the formula
! " #"$%&
C.2.3 Complex logarithms
The convention for the logarithm cut is along the negative real axis' The rules for the logarithm of a product are
then ( ( (
) *+,- . ) *+- / ) *,- / 0 *+1 ,-
0 *+1 ,- . 234 56 789: ;< 6 789: =< 6 79: ;=< 8 6 79: ;< 6 79: =< 6 789: ;=<>
Important consequences are ?@ ?@ ?@
?@ E7;=< A 7;< B 7=<
If 9: ; 9: = C D
F M N OP QRS R WX
G H I JK LF If TU V TU
140 Appendix C Useful formulae
Also, with YZ[ real : \ \ \
] ^Y[ _ `ab c ] ^Y _ `ab d ] ^[ _ `aeYb
C.2.4 D-Dimensional Integrals
All the formulas we need are
f ghi pvrs w x
jklmh jio pnqrs t xuyz{|}~
ÁÂÃ
¡¢£ ¤ ² ³ ´µ¶ ³ ·¸ · ¿À ÀÄ
¥¦ « ¬® ¯ ° ± ¹ º»¸ ¼½¾
§¨©ª
ÐÖÒÓ×Ø Ù âãä øùú û
Å ÆÇÈ ÈÍÈÎ é ê ëìí ê îï î ö÷ ÷
ÉÊËÌÇ ÉÈÏ Ð ÑÒÓ Ô Õ ÚÛ à á å æ ç è ð ñòï óôõ
ÜÝÞß
ü ýþÿ ÿ #$ +,-./ ,/
!"#
þ
% & $
') (*
ghijk hk
0 123 38393:3; =C?@ D KLMKNO P QRSQ TU P QRUQTS Z [ \]^ [ ^_
45672 43< = >?@ A BE ?GHI J V W X Y ` ab_ ce
df
F
B
Coherence among integrals can be checked using the fact that for symmetric integration :
lmln o p lrstu
q
vt v u vwvx y z |
{|{ } ~
To correct for the change of dimension in the measure, one usually introduces an arbitrary mass scale as
Of special interest are expansions near . Using :
®
¡¢ £ ¤ ¨ ©ª«¨ ¬ ® ¯ ©ª« ¬ ¯ ³´
¥¦§ § ° §± ² ± µ ¶ ···
and : ¸¹
º» ¼ ½ Ä Ë
Â Ã¿Ä Å ÆÇ È Å ÉÊ ÄÌ
º ¾¿ À Á
one gets, with Í Î Ï Ð Ñ :
Ö ÖØ
Ýâãäå æ ç ÿ
ôõö
ÒÓ Ô Õ Ð Ñ× íîïðñòó ÷øùú û ü
ÙÚ ØÛÜÝ Þàá
ß èéêëå ì ýþ
Õ
"#
ü ! $ % & )* +,- . / 012
'(
: BCD K
3 4 8 9 ?@ A ?@ F IJ
567 2 1 ;< = > ED ; > ED = G H
C.2 Loop Integrals, Gamma Functions and Related 141
Note that in this last formula, we have introduced another arbitrary scale LM. This is a common procedure. The
N
usual divergent quantity is de ned as OP Q
Z[\
WX Y
R ST UV ]\ (C.1)
where it is customary to take ] as the electron mass in QED. Therefore,
a b cdae jopqr s t
[^ _ ` hij k l r y
z { |}~
fg e mn uvwx (C.2)
`
C.2.5 Digamma, Polylogarithms and Riemann Functions
The digamma function is de ned as the logarithmic derivative of the gamma function
Generalizing, the -th digamma function is
Special values are
¡¢
£¤¥¦ §¨© ª «¬
Note that the Euler constant can also be written as the limit of the series
±
µ
¢ ª ®¯° ½¾ ¿À
±²³ ´ º »¼
¶·¸ ¹
Á
The Riemann zeta function is de ned from the series
 ÃÄÅ Æ È
Ç Î
ÉÊË ÌÍ
This function may be related to digamma function as
Ð ×ØÙÚÛ
Ï ÐÑ Ò ÓÔ Õ Ö àá âãä
ÜÝ Þß
å
The de nition of the polylogarithms is æ
çè éêë ì îí
ôõ
ïðñ òó
÷ø ù
Of special interest in Feynman parameter integration is the Spence function or dilogarithms ö , to which
we now turn. òú
142 Appendix C Useful formulae
Spence Functions
û
These functions are de ned through the above series, or through the integral
&'( )*
üýþ ÿ
" # $ % $
! + -. ,/ 01
,
0 / -45
A useful representation is obtained with the change of variable 2 3
@ABCDE
678 9:; <
= >? M
F HI GJ KL
RST UV Y ] d ejk
N O PQ P ] ^_` a e lmin opq r
W XW \
Z[ ^ b ef
cg h r
rstu vwxyuz{
r
| r
~
with the coef cient } being the Bernouilly numbers,
} }
......
The numbers not appearing in the table are also absent in the series expansion. Some basic properties are :
(Euler, 1768)
¡¢ ¡¢
£ ¤ ¥¦§ ¨©ª«
¬§ ¨ª« ® ¬§ ¨©ª« ¯ ¬
¤° § ±²³´
¸
with ² µ ¶²µ · ¹ º¹» not on the logarithm cuts. Also
¼½ ÇÈ É Ê Ö ÙÚÛ ÜÝÞ ß àáâ ÜãÞ ß äåæ Üç ß ÝÞ äåæ Üç ß ãÞ
¿ Á ÃÀ
» ¾· À Â Ë Ì Í Î Ï ÐÑÒ ÓÕ Ö
Ô ×Ø
ÂÄ Å Æ ê
ïð ñò ó ô ö ùúû üý
àáâ ÜãÞ ß àáâ ÜÝÞ è àáâ é
ëì í î õ ö÷ ø þÿ
# $ %&' ()* + ), 34
5 <= >?
D O P QR S
O
T UVWX YZ[
! " + ) *- . /01 2 5 66 77 89 : ; @ A BEFBC
G HIJ KL MN
due respectively to Spence (1809), Schaeffer (1846) and Kummer (1840). The second identity becomes, for
complex arguments :
\] YZ[ ^ \] Y_[ ` \] b gh ijm kl
no p nqrs tu z
X X X acd e f v w x y x }~
x{ |
ª
¡ ¢ £¤ ¥¦ §¨©
¬ ²³ ´ µ ¶· À ÁÂ Ê
§£¤ « ® ¯ ¬ ° ± µ ¸¹ ¶·º» ¼ ½¾ ¿ À Ã Ä Å Æ ÇÈ ÁÂÉ
®¯
C.2 Loop Integrals, Gamma Functions and Related 143
The discontinuity of the dilogarithm function is the same as the discontinuity of ËÌ Í
Î Ï ÐÑÒ Ñ Ó ÔÕÖÎ×
for
Ð Ó ØÙÚ Û Ü the argument of ÝÞ ßÜ à Úáâ becomes negative. The discontinuity propertyforis expressed, hence
along
the cut as
Ýãåæäç èéêë ìí î êïð ñ éêë ìí ñ êïðò ó ôõê ö÷ ìíð
for í ø ù and ú û ü. Put differently, we may write
ýþÿ
as can be seen from the de ning integral. Finally, some values of the function are
!"# $ % & /0
' ( )* +,- .
and graphically, the real and imaginary parts are
-4 -2 2 4 6 8 10
-2
-4
-6
/45 6 78
Note that the real part becomes 12 at around 3 .
C.2.6 Hypergeometric Functions
0 ? @AB CDE FE GE H I
.:; <; =; >
The hypergeometric function 9 is deJned as
CGI P Q
@AB CDE FE GE H I K \]^ ]_ `a b cd efg
L T TUVQ WX Y Z[
CFI CF M NO
R S
L L
This function is a solution to the differential equation
h ij k h l m nn o pq r st o u o vw xy z n r tuz { |
and it has the series expansion
}
}~
¡
with . From this series, it appear that
¡« ¡
¢£¤ ¥¦§ ¨§ ©§ ª ¢ £¤ ¥¨§ ¦§ ©§ ª
144 Appendix C Useful formulae
Some evaluation properties are
³µ ³ ¸ ° ¸ ²µ
µ ¶ ·¯ ·¯
¬® ¯°± ²± ³± ´
³ ¸ °µ ¯³ ¸ ²µ
·¯ ·
¹ °²
°± ²± ³± º µ ¶
¬® ¯° » ¼½ ¾ » ¼½ ¿ » ¼½ À Á
¹º ¬® ¯ ³
Shifting relations are often useful:
Æ Ê Ì ÍÎÏ ÐÑ Ñ Ê Ò ÓÑ Ô Ì ÍÎÏ ÐÑ Ò ÓÑ Ê Ò ÓÑ ÔÌ Õ Ö
¿ Â ÃÄ Å ½ ¾½ ¿½ ÇÁ È É È Ë É Ë ÈË É Ë
Ê Í ÎÏ ÐÑ Ñ ÊÑ Ô Ì Ê Ð Ì ÍÎÏ ÐÑ Ñ Ê Ò ÓÑ Ô Ì Ð ÍÎÏ Ð Ò ÓÑ Ñ Ê Ò ÓÑ ÔÌ Õ Ö
É Ë ÈÉ È É Ë È É Ë
and especially
Ê Í ÎÏ ÐÑ Ñ ÊÑ Ô Ì Ê ÐÌ Í ÎÏ ÐÑ Ò ÓÑ Ê Ò ÓÑ Ô Ì Ð Ó ÔÌ ÍÎÏ Ð Ò ÓÑ Ò ÓÑ Ê Ò ÓÑ ÔÌ Õ Ö
É Ë ÈÉ È É Ë È É È É Ë
Finally, one has sometimes to consider de×nite integrals involving hypergeometric function. Those can be done
Û
with ØÙÚ Û
Û Ûßá
ôõöõ÷ øù
ÜÝ ÞÝ ßÝ àá â ã ì ìíî ÷ úûü ýü þ ü ÿ
äá ß å æç èê é ë
ã ã é ïð ñ òó
ûü ýü
All Hypergeometric functions with de nite values for may be expressed in terms of simple functions
of . Some trivial cases are
if or
With or less than zero, Hypergeometric functions are often simple polynomials in !
*
"#$ % & '() *) +, - * ' (.+ / ( . 0* / ., +1
/ -4 ./ -4
since obviously, the series expansion will terminate at the order 2 such that 3 2 or 2 . Up to now,
all the functions are purely real. Some more complicated cases, which may develop imaginary parts, are
A FG HI
:; 8 < BCDE
156 789 89 8 9<= > JK L IMNOP
8?< @
PQR STU TU VUWX \a
Y Z \[ ]^_ ` Z
[
bcd e f gf hf \i j k l m p qrs tu k mvi
mn o
[
To have logarithms, it is necessary that wx y z {. Derivatives can also be obtained. For example
| |
} }
|} ~ x x x
|} ~ x x x
Finally, some hypergeometric functions with half-integer argument give elliptic functions. In general, orthogonal
polynomials can also be expressed in terms of .
C.3 Cross Section, Phase Space 145
C.3 Cross Section, Phase Space
The two-body to -body cross section is expressed as
© ª« ¬ ® ¯° ±¯² ± ³³³´µ²
¡ ¢ £¤¥¦§¤ ¨
¶·¸¹º » ¼½ ¾¿ À Á¿Â Á¿Ã ÄÅ
with Æ
Ó Õ Ö× ØÙÚ Û ÙÜ Ý Þ ß Ý ààà Ý Þáâ ãå
Ç ÈÉÊ Ë ÉÌ ÍÉÎ Ï ÐÐÐÏÉ ÇÑ Ò È Ô Ñ ä
íîïðñ îòó ô
Å æçè éêëì
(this section is based on [23], [223])
C.3.1 Two-Body Phase Space
The invariant two-body phase space integral is
õö÷ øù úû ü ýû÷þ ÿ ø
!"#
$%&'( %)# %)*
in the center-of-mass frame, where + , -, with + . /0 1 /2 hence /2 1 ./0. The remaining delta can be
converted to a constraint over the energy 34 as
56 5 7 89: 8 ; <= <= = =
4 = > ? @ A? B A=
hence C
] ^ _ `abcd bceb
D= EF GH ?I H =J K Q
> E LM NO P O R ST R USVT R WVT X YZ Z
f ghijk d
Z[ \ a
s v |}
Since lmn o p nm qr t
, u t uw uv t u x yz{yzand { z x ~ ~ ~ ~
. Now let us integrate over
. The delta
function gives
and since we are using the relation ¡¢£,
the correction to the measure is ¤
¥¦
¤¥¦ §¨ ¸
©ª ª «¬® ¯ °® ± ²³
³´ µ ¶ ¹º ·
Hence »¼
ÅÆ Æ
º½ ºÂ Ã Ä Ç È ÉÇÊËÌ
Í ÎÏÐÑÒ ÎÓÒ Ô ÓÌ Ñ
· ¾¿ÀÁ¿
which can be written in terms of invariants as
ÕËÌ æ æ æ æ
ÕÖÒ Î× ØÙ ÌÚÙ Ò Ñ Û Þ ß àáâã ä åæç ä é êå ë ì åæç
ÜÏÐ Ò Ý èæ è
æï ð æ ï
with è í î ê
ë ì ð æ ç . As a special case of interest, if å ë åæ,
ø
ñòó ôõ ö÷øù÷óú û üý ó
þÿ
146 Appendix C Useful formulae
C.3.2 Three-Body Phase Space
The invariant two-body phase space integral is
% ( % %
% ' % "&( !" ' % "&
!"#$ "& !"#$ #$ %
145 :4=
) !"#$ * !+, - ./ - .0 - .12 67 3 : 7;5 67 < : 7;= 7;>
89 89 :
=
; ;
with : constrained as : ? @AB C DEF and DF G H I DJ I DE. Then, in the center-of-mass frame,
Y [ \] ] ]
DF G I KDJ L DEM N OPQ R SQP T UVW X VY Z ^ _ ` a _ `] b c defd deg dhij k
Now the integral can be written
lmn o p qw
qrstu v x y z{ y z| y z}~
In the case of an unpolarized cross section, the amplitude squared is invariant under global rotations in the c.m.
¡ ¢ £¤¥ ¦§¨¥which we choose to be the angle between
frame. So only one out of the four angular variables is non-trivial,
and . Also, since and , we arrive at
Æ ÇÈÉ ÊËÌÉ ÊÍÎ ÊËÌÎË ÏÐÑ Ò
§©ª « ¬ ª ° ±²³ ´ µ¶ ´ µ· ´ ¸¹·º » ¼½¾ » ¼½ ¼¾ ¼½ ÃÄ Å ÓÔÕ
¬®¯ ½ ¿ À Á Á Á ÁÂ
and integrating over ÏÐÑ Ò with the delta function :
Ø Ø ØÕ Ø ÝÔ ÝÔ ÔÔ
ÏÐÑ Ò Ö ×Ù Ú ×Ø Û × Ú ÜÝ Û ÞÜ Ù Û ÞÜ Ø Ú Þ Ù Ø
ÛÞ ßàá ß ßàâ ß
The correction to the measure is
ã ýþÿ
é î
ã äåæ ç è ê ë ìí ë ìî ë ïðñ ò óôõ ò óôô ö ÷ øó õ ø øóô øùúû ü ü
Hence
(C.3)
The bounds for the two last integrals can be found by constraining, for a given value of say, the value of
such that the exists.
C.3.3 Recursive Calculation of Phase-Space
In some case, we will compute the phase space integrals recursively, using
!"#$% &&&%# ' ()*+ ,-. / -01 23-4563 7773-89 2;
<:=>?
with @ A BC D EEE D BF .
147
Bibliography
Standard Books on Quantum Field Theory
[1] P. A. M. Dirac, GThe Principles of Quantum MechanicsG, Oxford.
[2] N. Bogolioubov and D. Chirkov, GIntroduction à la Théorie Quantique des ChampsG, 1960, Dunod.
[3] E. Merzbacher, GQuantum MechanicsG, 1961, Wiley.
[4] J. D. Jackson, GClassical ElectrodynamicsG, 1962, Wiley.
[5] J. Bjorken, S. Drell, GRelativistic Quantum MechanicsG, GRelativistic Quantum FieldsG, 1964, McGraw-
Hill.
[6] K. Nishijima, GFields and Particles : Field Theory and Dispersion RelationsG, 1969, Benjamin N.Y.
[7] J. Schwinger, GParticles, Sources, and FieldsG Volume I, II and III, 1973, Perseus Books.
[8] C. Cohen-Tannoudji, B. Diu, F. Laloë, GMécanique QuantiqueG, Volume I and II, 1973, Hermann.
[9] J. Jauch and F. Röhrlich, GThe Theory of Photons and ElectronsG, 1980, Springer.
[10] C. Itzykson, J.-B. Zuber, GQuantum Field TheoryG, 1980, McGraw-Hill.
[11] F.J. Yndurain, GThe Theory of Quark and Gluon InteractionsG, 1983, Springer.
[12] T.-P. Cheng and L.-F. Li, GGauge Theory of Elementary Particle PhysicsG,1984, Oxford.
[13] R. Rivers, GPath Integral Methods in Quantum Field TheoryG, 1987, Cambridge.
[14] G. Jones, D. Singerman, GComplex Functions: an Algebraic and Geometric ViewpointG, 1987, Cam-
bridge.
[15] W. Greiner, A. Schäfer, GQuantum ChromodynamicsG, 1989, Springer.
H
[16] T. Kinoshita (Editor), GQuantum ElectrodynamicsG, 1990, World Scienti c.
[17] O. Nachtmann, GElementary Particle Physics : Concepts and PhenomenaG, 1990, Springer-Verlag.
[18] J. Donoghue, E. Golowich, B. Holstein, GDynamics of the Standard ModelG, 1992, Cambridge.
[19] F. Gross, GRelativistic Quantum Mechanics and Field TheoryG, 1993, Wiley.
[20] M. Kaku, GQuantum Field TheoryG, 1993, Oxford University Press.
[21] W. Greiner, J. Reinhardt, GField QuantizationG, 1993, Springer.
[22] W. Greiner, J. Reinhardt, GQuantum ElectrodynamicsG, 1994, Springer.
H
[23] A. Di Giacomo, G. Paffuti, P. Rossi, GSelected Problems in Theoretical PhysicsG, 1994, World Scienti c.
[24] M. Peskin and D. Schoeder, GAn Introduction to Quantum Field TheoryG, 1995, Addison-Wesley.
[25] S. Weinberg, GThe Quantum Theory of Fields (I, II and III)G, 1996, Cambridge.
[26] F.J. Yndurain, GRelativistic Quantum Mechanics and Introduction to Field TheoryG, 1996, Springer.
[27] J. Zinn-Justin, GQuantum Field Theory and Critical PhenomenaG, 1996, Clarendon Press - Oxford.
[28] A. Dobado, A. Gomez-Nicola, A. Maroto, J. Pelaez, GEffective Lagrangian for the Standard ModelG,
1997, Springer.
Thesis
[29] G. Lopez Castro, J. Pestieau and C. Smith, hep-ph/0004209, G. Lopez Castro, J. Pestieau, C. Smith and
S. Trine, hep-ph/0006016 and hep-ph/0006018.
[30] J. Pestieau, C. Smith and S. Trine, hep-ph/0105034, Int. J. Mod. Phys. A17, 1355 (2002).
[31] J. Pestieau and C. Smith, hep-ph/0111264, Phys. Lett. B524, 395 (2002).
[32] J. Pestieau and C. Smith, hep-ph/0111380, submitted to Eur. Phys. J. C.
148 Bibliography
Positronium Experiments
[33] A.E. Ruark, Phys. Rev. 68, 278 (1945).
[34] M. Deutsch, Phys. Rev. 82, 455 (1951) IM. Deutsch, Phys. Rev. 84, 866 (1951) I M. Deutsch, E. Dulit,
Phys. Rev. 84, 601 (1952) I M. Deutsch, S. Brown, Phys. Rev. 85, 1047 (1952).
[35] V. W. Hughes, S. Marder, C. S. Wu, Phys. Rev. 106, 934 (1957).
[36] E. Theriot, R. Beers, V. Hughes, Phys. Rev. Lett. 18, 767 (1967).
[37] V. W. Hughes, JPhysics of One- and Two-electrons AtomsJ, North-Holland (1969), pp 407-429.
[38] E. Theriot, R. Beers, V. Hughes, K. Ziock, Phys. Rev. A2, 707 (1970).
[39] A.P. Mills, G.H. Bearman, Phys. Rev. Lett. 34, 246 (1974).
[40] A. P. Mills, Jr., S. Berko, and K. F. Canter, Phys. Rev. Lett. 34, 1541 (1975).
K
[41] T.C. Grif th, G.R. Heyland, K.S. Lines, T.R. Twomey, J. Phys. B11, L743 (1978).
[42] D. Gidley, A. Rich, E. Sweetman, D. West, Phys. Rev. Lett. 49, 525 (1982).
[43] M.W. Ritter, P.O. Egan, V.W. Hughes, K.A. Woodle, Phys. Rev. A30, 1331 (1984).
[44] R.S. Conti, S. Hatamian, A. Rich, Phys. Rev. A33, 3495 (1986).
[45] S. Hatamian, R. S. Conti, and A. Rich, Phys. Rev. Lett. 58, 1833 (1987).
[46] P. Hasbach, G. Hilkert, E. Klempt, G. Werth, Nuovo Cim. A97, 419 (1987).
[47] C. Westbrook, D. Gidley, R. Conti, A. Rich, Phys. Rev. Lett. 58, 1328 (1987).
[48] K. Danzmann, M. S. Fee, and S. Chu, Phys. Rev. A39, 6072 (1989).
[49] C. Westbrook, D. Gidley, R. Conti, A. Rich, Phys. Rev. A40, 5489 (1989).
[50] J. Nico, D. Gidley, A. Rich, P. Zitzewitz, Phys. Rev. Lett. 65, 1344 (1990).
[51] D. Gidley, J. Nico, M. Skalsey, Phys. Rev. Lett. 66, 1302 (1991).
[52] M. S. Fee, A. P. Mills, Jr., S. Chu, E. D. Shaw, K. Danzmann, R. J. Chicherster, and D. M. Zuckerman,
Phys. Rev. Lett. 70, 1397 (1993) I M. S. Fee, S. Chu, A. P. Mills, Jr., R. J. Chicherster, D. M. Zuckerman,
and E. D. Shaw, Phys. Rev. A48, 192 (1993).
[53] E. W. Hagena, R. Ley, D. Weil, G. Werth, W. Arnold, and H. Schneider, Phys. Rev. Lett. 71, 2887 (1993).
[54] R. S. Conti, S. Hatamian, L. Lapidus, A. Rich, and M. Skalsey, Phys. Lett. A177, 43 (1993).
K
[55] R. Ley, D. Hagena, D. Weil, G. Werth, W. Arnold, and H. Schneider, Hyper ne Interact. 89, 327 (1994).
[56] A. Al-Ramadhan, D. Gidley, Phys. Rev. Lett. 72, 1632 (1994).
[57] S. Asai, S. Orito, N. Shinohara, Phys. Lett. B357, 475 (1995) I O. Jinnouchi, S. Asai, T. Kobayashi,
hep-ex/0011011.
Positronium Theory
[58] B. Podolsky, L. Pauling, Phys. Rev. 34, 109 (1929).
[59] S. Mohorovicic, Astron. Nachr., 94 (1934).
[60] J. A. Wheeler, Ann. N. Y. Acad. Sci. 48, 219 (1946).
[61] J. Pirenne, Arch. Sci. Phys. Nat. 28, 233 (1946) I 29, 121, 207 & 265 (1947).
[62] V.B. Berestetski, L.D. Landau, Zhur. Exsptl. i Teort. Fiz. 19, 673 (1949) I V.B. Berestetski, Zhur. Exsptl.
i Teort. Fiz. 19, 1130 (1949).
[63] A. Ore and J. L. Powell, Phys. Rev. 75, 1696 (1949).
[64] R. Ferrel, Phys. Rev. 84, 858 (1951).
[65] L. M. Brown, R. P. Feynman, Phys. Rev. 85, 231 (1952).
[66] L. Wolfenstein, D. G. Ravenhall, Phys. Rev. 88, 279 (1952).
[67] R. Karplus, A. Klein, Phys. Rev. 87, 848 (1952).
[68] T. Fulton, P. Martin, Phys. Rev. 95, 811 (1954).
[69] I. Harris and L. Brown, Phys. Rev. 105, 1656 (1957).
Bibliography 149
[70] R. Barbieri, P. Christillin, E. Remiddi, Phys. Rev. A8, 2266 (1973).
[71] M. Stroscio, Phys. Rept. C22, 215 (1975).
[72] A. Barut, J. Kraus, Phys. Lett. A58, 361 (1976).
[73] R. Barbieri, E. Remiddi, Phys. Lett. B65, 258 (1976).
[74] G. P. Lepage, [Link]. A16, 863 (1977).
[75] W. E. Caswell, G. P. Lepage, J. Sapirstein, Phys. Rev. Lett. 38, 488 (1977).
[76] W. E. Caswell and G.P. Lepage, Phys. Rev. A18, 810 (1978).
[77] W. E. Caswell and G.P. Lepage, Phys. Rev. Lett. 41, 1092 (1978).
[78] G. Bodwin and D. Yennie, Phys. Rept. 43, 267 (1978).
[79] R. Barbieri and E. Remiddi, Nucl. Phys. B141, 413 (1978).
[80] W. E. Caswell and G. P. Lepage, Phys. Rev. A20, 36 (1979).
[81] D. Yennie, CLNS-79/436, 1979 (1979 Cargèse Summer School).
[82] M.I. Vysotsky, [Link]. 29, 845 (1979).
[83] W. Buchmüller, E. Remiddi, Nuovo Cim. A60, 109 (1980) L Nucl. Phys. B162, 250 (1980).
[84] Y. Tomozawa, Ann. Phys. 128, 463 (1980).
[85] A. Rich, Rev. Mod. Phys. 53, 126 (1981).
[86] M. Stroscio, Phys. Rev. Lett. 48, 571 (1982).
[87] G. Adkins, Ann. Phys. 146, 78 (1983).
[88] G. Adkins, F. Brown, Phys. Rev. A28, 1164 (1983).
[89] G. Lepage, P. Mackenzie, K. Streng, P. Zerwas, Phys. Rev. A28, 3090 (1983).
[90] J.R. Sapirstein, E.A. Terray, D.R. Yennie, Phys. Rev. D29, 2290 (1984).
[91] G. Adkins, Phys. Rev. A31, 1250 (1985).
[92] G.T. Bodwin, D.R. Yennie, M.A. Gregorio, Rev. Mod. Phys. 57, 723 (1985).
[93] W. E. Caswell and G.P. Lepage, Phys. Lett. B167, 437 (1986).
[94] H. Olsen, Phys. Rev. D33, 2033 (1986).
[95] J. Malenfant, Phys. Rev. D36, 863 (1987).
[96] G. Adkins, M. Bui, D. Zhu, Phys. Rev. A37, 4071 (1988).
[97] I.B. Khirplovich, A.S. Yelkhovsky, Phys. Lett. B246, 520 (1990).
[98] L.G. Afanasev et al, Phys. Lett. B236, 116 (1990).
[99] E.A. Kuraev, T.V. Kukhto, Z.K. Silagadze, Sov. J. Nucl. Phys. 51, 1036 (1990).
[100] P. Labelle, MRST Meeting 92 (CLNS-92-1161).
[101] V. Dvoeglazov, R. Faustov, Y. Tyukhtyaev, Mod. Phys. Lett. A8, 3263 (1993).
[102] A. Pivovarov, [Link]. D47, 5183 (1993).
[103] V. Dvoeglazov, R. Faustov, Y. Tyukhtyaev, Mod. Phys. Lett. A8, 3263 (1993).
[104] P. Labelle, G. Lepage, U. Magnea, Phys. Rev. Lett. 72, 2006 (1994).
[105] P. Labelle, hep-ph/9407233.
[106] I.B. Khriplovich, A.I. Milstein, J. Exp. Theor. Phys. 79, 379 (1994) L I.B. Khirplovich, A.I. Milstein,
hep-ph/9607374.
[107] R.N. Faustov, A.P. Martynenko, V.A. Saleev, Phys. Rev. A51, 4520 (1995).
[108] J. Govaerts and M. Van Caillie, Phys. Lett. B381, 451 (1996).
[109] T. Kinoshita, M. Nio, Phys. Rev. D53, 4909 (1996) L Phys. Rev. D55, 767 (1997).
[110] P. Labelle, S.M. Zebarjad, C.P. Burgess, Phys. Rev. 56, 8053 (1997).
[111] U. Jentschura, G. Soff, V. Ivanov and S. Karshenboim, Phys. Rev. A56, 4483 (1997) L U. Jentschura,
G. Soff, V. Ivanov, S. Karshenboim, hep-ph/9706401 L I.F. Ginzburg, U. Jentschura, S. Karshenboim, F.
Krauss, V.G. Serbo, G. Soff, Phys. Rev. C58, 3565 (1998).
[112] T. Kinoshita, Int. Work. on Hadronic Atoms and Positronium in the Standard Model, Dubna, Russia,
26-31 May 1998.
150 Bibliography
[113] V. Antonelli, V. Ivanchenko, E. Kuraev, V. Laliena, Eur. Phys. J. C5, 535 (1998), V. Antonelli, Int. Work.
on Hadronic Atoms and Positronium in the S.M., Dubna, 26-31 May 1998.
[114] A. Vairo, Found. Phys 28, 829 (1998).
[115] A. Czarnecki, Acta Phys. Polon. B30, 3837 (1999).
[116] A. Czarnecki and G. Kardhenboim, 14th Int. Work. on HEP and QFT, Moscow 1999, hep-ph/9911410.
[117] A. Czarnecki, K. Melnikov, A. Yelkhovsky, Phys. Rev. A59, 4316 (1999) M Phys. Rev. Lett. 82, 311 (1999).
[118] G. Adkins, K. Melnikov and A. Yelkhovsky, hep-ph/9905553.
[119] K. Melnikov, A. Yelkhovsky, Phys. Lett. B458, 143 (1999).
[120] A. Pineda, J. Soto, Phys. Rev. D59, 016005 (1999).
[121] A. Czarneski, K. Melnikov, A. Yelkhovsky, Phys. Rev. Lett. 83, 1135 (1999) M hep-ph/9910488 M Phys.
Rev. A61, 052502 (2000).
[122] R. Faustov, A. Martynenko, hep-ph/0002281.
[123] R. Hill, MRST Meeting, Rochester, May 2000.
[124] G. Adkins, R. Fell, J. Sapirstein, Phys. Rev. Lett. 84, 5086 (2000).
[125] A.H. Hoang, P. Labelle, S.M. Zebarjad, Phys. Rev. A62, 012109 (2000).
[126] A. Burichenko, hep-ph/0004063.
[127] R. Hill, G. P. Lepage, Phys. Rev. D62, 111301 (2000).
[128] A. Manohar, I. Stewart, Phys. Rev. Lett. 85, 2248 (2000).
[129] By R. Foot, S. Gninenko, Phys. Lett. B480, 171 (2000).
[130] B. Kniehl, A. Penin, Phys. Rev. Lett. 85, 1210 (2000) M ibid. 5094 (2000).
[131] K. Melnikov, A. Yelkhovsky, Phys. Rev. D62, 116003 (2000) M Phys. Rev. Lett. 86, 1498 (2001).
[132] S. Karshenboim, hep-ph/0201241.
[133] By S.N. Gninenko, N.V. Krasnikov, A. Rubbia, hep-ph/0205056.
Quarkonium Experiment
[134] G. Abrams et al, Phys. Rev. Lett. 44, 114 (1980).
[135] [Link] et al, Phys. Rev. D23, 43 (1981).
[136] Crystal Ball Collaboration (DESY), Phys. Lett. B267, 286 (1991).
[137] F. Harris, American Physical Society (APS) Meeting of the Division of Particles and Fields (DPF 99),
Los Angeles, CA, 5-9 Jan. 1999.
[138] CLEO Collaboration (G. Brandenburg et al), Phys. Rev. Lett. 85, 3095 (2000).
[139] Xu Guofa (for the BES Collaboration), 31st International Symposium on Multiparticle Dynamics (ISMD
2001), Datong, China, 1-7 Sep. 2001
[140] CLEO Collaboration (B. Nemati et al.), Phys. Rev. D55, 5273 (1997).
[141] BES Collaboration, (J.Z. Bai et al.), Phys. Rev. D62, 072001 (2000).
[142] M. Stancari (for the E835 Collaboration), Nucl. Phys. (Proc. Suppl.) B82, 306 (2000).
[143] CLEO Collaboration (B. Eisenstein et al), Phys. Rev. Lett. 87, 061801 (2001).
Quarkonium Theory
[144] T. Applequist, H. Politzer, Phys. Rev. Lett. 34, 43 (1975).
[145] A. De Rujula, S.L. Glashow, Phys. Rev. Lett. 34, 46 (1975).
[146] P. Freund, Y. Nambu, Phys. Rev. Lett. 34, 1645 (1975).
[147] V. Novikov, L.B. Okun, M. Shifman, A. Vainshtein, M. Voloshin, V. Zakharov, Phys. Rept. C41, 1 (1978).
[148] K. Koller and T. Walsh, Nucl. Phys. B140, 449 (1978).
Bibliography 151
[149] S. Brodsky, D. Coyne, T. DeGrand and R. Horgan, Phys. Lett. B73, 203 (1978).
[150] C. Quigg, FERMILAB-Conf-78/82-THY (1978) N C. Quigg, FERMILAB-Conf-79/74-THY (1979).
[151] R. Barbieri, E. d OEmilio, G. Curci, E. Remiddi, Nucl. Phys. B154, 535 (1979).
[152] J.P. Leveille and D.M. Scott, Phys. Lett. B95, 96 (1980).
[153] I.I.Y. Bigi, Nucl. Phys. B164, 239 (1980).
[154] W. Buchmüller, S.-H. Tye, Phys. Rev. D24, 132 (1981).
[155] W. Kwong, P. Mackenzie, R. Rosenfeld and J. Rosner, Phys. Rev. D37, 3210 (1981).
[156] P. Mackenzie, G. Lepage, Phys. Rev. Lett. 47, 1244 (1981).
[157] G. Grunberg, [Link]. B114, 271 (1982).
[158] J. Körner and J. Kühn, M. Krammer, H. Schneider, Nucl. Phys. B229, 115 (1983).
[159] P. Moxhay, J. Rosner, Phys. Rev. D28, 1132 (1983).
[160] W. S. Hou and A. Soni, Phys. Rev. Lett. 50, 569 (1983).
[161] R.D. Field, Phys. Lett. B133, 248 (1983).
[162] W. Kwong, J. Rosner, C. Quigg, Ann. Rev. Nucl. Part. Sci 37, 325 (1987).
[163] S. Brodsky, P. Lepage and S.F. Tuan, Phys. Rev. Lett. 59, 621 (1987).
[164] A. Bizzeti (Crystal Ball Collaboration), Proc. XXIV Int. Conf. on HEP (1989).
[165] S. Pinsky, Phys. Lett. B236, 479 (1990).
[166] E. Sima Oan Ackleh, thesis (University of Tennessee), 1992.
[167] G. Schuler, (thesis) CERN-TH.7170/94.
[168] S. Catani and F. Hautmann, Nucl. Phys. Proc. Suppl. 39BC, 359 (1995).
[169] G. Lopez Castro, J.L. Lucio, J. Pestieau, AIP Conf. Proc. 342, 441 (1995).
[170] E. Eichten, C. Quigg, Phys. Rev. D52, 1726 (1995).
[171] G. Bodwin, E. Braaten, G. Lepage, Phys. Rev. D51, 1125 (1995).
[172] T. Barnes, F.E. Close, P.R. Page, [Link], Phys. Rev. D55, 4157 (1997).
[173] W.S. Hou, Phys. Rev. D55, 6952 (1997).
[174] K. Chao, H. Huang, J. Liu, J. Tang, Phys. Rev. D56, 368 (1997).
[175] S. Olsen, J. Mod. Phys. A256, 4069 (1997).
[176] E. Braaten, Third Int. Work. on Particle Physics Phenomenology, Taipei, November 1997.
[177] S. Brodsky and M. Karliner, Phys. Rev. Lett. 78, 4682 (1997).
[178] A. Manohar, Phys. Rev. D56, 230 (1997).
[179] P. Ko, PPPP Workshop, Seoul, Korea, October 1997.
[180] F. Hautmann, proceedings of PInternational Conference on the Structure and the Interactions of the
Photon (Photon 97) P Netherlands, 10-15 May 1997 (hep-ph/9708496).
[181] Y.-Q. Chen and E. Braaten, Phys. Rev. Lett. 80, 5060 (1998).
[182] A. Petrelli, M. Cacciari, M. Greco, F. Maltoni, M. Mangano, Nucl. Phys. B514, 245 (1998).
[183] M. Suzuki, Phys. Rev D57, 5717 (1998).
[184] M. Suzuki, Phys. Rev. D57, 5717 (1998) N Phys. Rev. D60, 051501 (1999).
[185] M. Krämer, proceedings of P4th International Symposium on Radiative Corrections (RADCOR 98) P,
Barcelona, 8-12 Sep 1998 (hep-ph/9901448).
[186] J.-M. Gérard, J. Weyers, Phys. Lett. B462, 324 (1999).
[187] A.M. Badalian, V.L. Morgunov, Phys. Rev. D60, 116008 (1999).
[188] B. Grinstein, Int. J. Mod. Phys. A15, 461 (2000)
[189] N. Brambilla, XXIII Int. Work. on the Fundamental Problems of High Energy Physics, Protvino (Russia),
June 2000.
[190] I. Stewart, 7th Conf. on the Intersections of PArticle and Nuclear Physics, Quebec City, Canada, May
2000.
[191] Y.G. Gu and S.F. Tuan, Nucl. Phys. A675, 404c (2000).
152 Bibliography
[192] M.E. Like, A.V. Manohar, I.Z. Rothstein, Phys. Rev. D61, 074025 (2000).
[193] S. F. Tuan, Commun. Theor. Phys. 33, 285 (2000).
[194] Y. F. Gu and X. H. Li, Phys. Rev. D63, 114019 (2001).
[195] F.J. Yndurain, Nucl. Phys. Proc. Suppl. 93, 196 (2001).
[196] A. Penin, Nucl. Phys. Proc. Suppl. 96, 418 (2001).
[197] S. Wolf, Phys. Rev. D63, 074020 (2001).
[198] A. Leibovich, Nucl. Phys. Proc. Suppl. 93, 182 (2001).
[199] W.Q. Chao, C.S. Ju, hep-ph/0111325.
[200] Y.F. Gu, X.H. Li, Phys. Rev. D63, 114019 (2001).
General Field Theory
[201] H. Euler, W. Heisenberg, Z. Phys. 98, 714 (1936).
[202] C. Yang, Phys. Rev. 77, 242 (1950).
[203] J. Schwinger, Phys. Rev. 82, 664 (1951).
[204] F. E. Low, Phys. Rev. 110, 974 (1958).
[205] B. De Tollis, Nuovo Cim. 32, 757 (1964), ibid. 35, 1182 (1965) Q V. Constantini, B. De Tollis, G. Pistoni,
Nuovo Cim. 2A, 733 (1971) Q M. Laursen, M. Samuel, G. Tupper, A. Sen, Phys. Rev. D27, 196 (1983).
[206] H. Chew, Phys. Rev. 123, 377 (1961) Q J. Pestieau, Phys. Rev. 160, 1555 (1967).
[207] S. Coleman, R. Jackiw, Annals Phys. 67, 552 (1971) Q M. Chanowitz, J. Ellis, Phys. Rev. D7, 2490 (1973) Q
S. Adler, J. Collins, A. Duncan, Phys. Rev. D15, 1712 (1977).
[208] L.M. Sehgal, Phys. Rev. D7, 3303 (1973).
[209] W. Caswell, Phys. Rev. Lett. 33, 244 (1974).
[210] E. Egorian, O. Tarasov, Theor. Math. Phys. 41, 863 (1979).
[211] S. Brodsky, G. Lepage, P. Mackenzie, Phys. Rev. D28, 228 (1983).
[212] L. Bergström, G. Hulth, Nucl. Phys. B259, 137 (1985).
[213] D.W. Duke, R.G. Roberts, Phys. Rept. 120, 275 (1985).
[214] L. L. Chau, H. Y. Cheng, Phys. Lett. B195, 275 (1987).
[215] G. Ecker, A. Pich, E. de Rafael, Nucl. Phys. B303, 665 (1988).
[216] J. L. Lucio M., J. Pestieau, Phys. Rev. D42, 3253 (1990).
[217] D.J. Broadhurst, J. Fleischer, O.V. Tarasov, Z. Phys. C60, 287 (1993).
[218] J. Bijnens, G. Ecker and J. Gasser, Nucl. Phys. B396, 81 (1993) Q G. D RAmbrosio, G. Ecker, G. Isidori,
H. Neufeld, 2ST DAPHNE Physics Handbook:253-313 (hep-ph/9411439).
[219] B. Kniehl, Acta Phys. Polon. B27, 3631 (1996).
[220] J. Horejsi, M. Schnabl, Z. Phys. C76, 561 (1997).
[221] J.A. Oller, Phys. Lett. B426, 7 (1998).
[222] S. Bethke, J. Phys. G26, R27 (2000).
[223] D.E. Groom et al, Eur. Phys. J. C15, 1 (2000).
Vous aimerez peut-être aussi
- Notes sur la physique de la matière condenséePas encore d'évaluationNotes sur la physique de la matière condensée69 pages
- Physique des Interactions Fortes ULBPas encore d'évaluationPhysique des Interactions Fortes ULB134 pages
- Introduction à la théorie quantique des champsPas encore d'évaluationIntroduction à la théorie quantique des champs455 pages
- More Related Content: Home - Search - Pacs & MSC - Journals - About - Contact UsPas encore d'évaluationMore Related Content: Home - Search - Pacs & MSC - Journals - About - Contact Us8 pages
- Introduction à la théorie quantiquePas encore d'évaluationIntroduction à la théorie quantique542 pages
- Introduction à la mécanique quantiquePas encore d'évaluationIntroduction à la mécanique quantique81 pages
- Potentiels scalaires et gravité faiblePas encore d'évaluationPotentiels scalaires et gravité faible181 pages
- Naissance de la mécanique quantiquePas encore d'évaluationNaissance de la mécanique quantique3 pages
- Interaction forte en physique des particulesPas encore d'évaluationInteraction forte en physique des particules353 pages
- Impact of Decoherence on Logic GatesPas encore d'évaluationImpact of Decoherence on Logic Gates67 pages
- Nouvelle physique à basse énergie et EFTPas encore d'évaluationNouvelle physique à basse énergie et EFT90 pages
- Modèle Standard de La Physique Des Particules - WikipédiaPas encore d'évaluationModèle Standard de La Physique Des Particules - Wikipédia13 pages
- Propriétés électroniques des matériauxPas encore d'évaluationPropriétés électroniques des matériaux70 pages
- Introduction à la mécanique quantiquePas encore d'évaluationIntroduction à la mécanique quantique123 pages
- Introduction à la mécanique quantiquePas encore d'évaluationIntroduction à la mécanique quantique13 pages
- Résolution de l'équation de SchrödingerPas encore d'évaluationRésolution de l'équation de Schrödinger9 pages
- Introduction à la physique des particulesPas encore d'évaluationIntroduction à la physique des particules436 pages
- Urophysics Etters: Uropean Hysical OcietyPas encore d'évaluationUrophysics Etters: Uropean Hysical Ociety9 pages
- Introduction à la physique nucléairePas encore d'évaluationIntroduction à la physique nucléaire64 pages
- Vente de sonde 86b en physique des particules100% (1)Vente de sonde 86b en physique des particules427 pages
- Mesure des couplages du boson de HiggsPas encore d'évaluationMesure des couplages du boson de Higgs37 pages
- Amplitudes Yang-Mills et Gravité en QFTPas encore d'évaluationAmplitudes Yang-Mills et Gravité en QFT6 pages
- Introduction à la Physique NucléairePas encore d'évaluationIntroduction à la Physique Nucléaire95 pages
- Fonctions de distribution dans un gaz FermiPas encore d'évaluationFonctions de distribution dans un gaz Fermi55 pages
- Diffusion en ondes S et équation de Klein-GordonPas encore d'évaluationDiffusion en ondes S et équation de Klein-Gordon45 pages
- Angle et symétrie en mécanique quantiquePas encore d'évaluationAngle et symétrie en mécanique quantique674 pages
- Théorie DFT et matériaux thermoélectriquesPas encore d'évaluationThéorie DFT et matériaux thermoélectriques83 pages
- Propagation de la lumière en milieux électromagnétiquesPas encore d'évaluationPropagation de la lumière en milieux électromagnétiques147 pages
- Fondements de la physique quantique et statistiquePas encore d'évaluationFondements de la physique quantique et statistique18 pages
- Modèle nucléaire et découvertes clésPas encore d'évaluationModèle nucléaire et découvertes clés43 pages
- Introduction à la Mécanique QuantiquePas encore d'évaluationIntroduction à la Mécanique Quantique21 pages
- Symétries de l'espace-temps en physiquePas encore d'évaluationSymétries de l'espace-temps en physique26 pages
- Naissance des particules élémentairesPas encore d'évaluationNaissance des particules élémentaires5 pages
- Gaz Parfaits Quantiques et PropriétésPas encore d'évaluationGaz Parfaits Quantiques et Propriétés119 pages
- Introduction à la statistique des gaz quantiquesPas encore d'évaluationIntroduction à la statistique des gaz quantiques33 pages
- Systèmes de particules en physique quantiquePas encore d'évaluationSystèmes de particules en physique quantique13 pages
- Introduction à l'électrodynamique quantiquePas encore d'évaluationIntroduction à l'électrodynamique quantique4 pages
- Introduction à la méthode Hartree-FockPas encore d'évaluationIntroduction à la méthode Hartree-Fock32 pages
- Analyse fonctionnelle et espaces vectorielsPas encore d'évaluationAnalyse fonctionnelle et espaces vectoriels43 pages
- Musique et rituel : identités et significations100% (2)Musique et rituel : identités et significations30 pages
- Dynamiques de groupe et prise de décisionPas encore d'évaluationDynamiques de groupe et prise de décision11 pages
- 12-Annexe - VIII-2 - E3 LV - Grille - Fiche - Aide - Évaluation - BTS - GTLA - 2025Pas encore d'évaluation12-Annexe - VIII-2 - E3 LV - Grille - Fiche - Aide - Évaluation - BTS - GTLA - 20256 pages
- Guide complet sur la librairie PandasPas encore d'évaluationGuide complet sur la librairie Pandas19 pages
- Rapport de visite à la chaufferie de MérignacPas encore d'évaluationRapport de visite à la chaufferie de Mérignac5 pages
- Contrôle PEB1 : Étude thermique bâtimentPas encore d'évaluationContrôle PEB1 : Étude thermique bâtiment11 pages
- Réalisations du Gouvernement EstiméPas encore d'évaluationRéalisations du Gouvernement Estimé7 pages
- Guide pratique des clubs d'investissementPas encore d'évaluationGuide pratique des clubs d'investissement36 pages
- Liste de vérification pré-départ véhiculePas encore d'évaluationListe de vérification pré-départ véhicule1 page
- Essai virtuel : Révolution mode à GomaPas encore d'évaluationEssai virtuel : Révolution mode à Goma4 pages
- Téléchargement Livre Nacho 01 GratuitPas encore d'évaluationTéléchargement Livre Nacho 01 Gratuit5 pages
- Évaluation sur l'identité numérique et sécuritéPas encore d'évaluationÉvaluation sur l'identité numérique et sécurité4 pages
- Effets des traitements sur la dureté des aciersPas encore d'évaluationEffets des traitements sur la dureté des aciers6 pages
- Exercices d'équilibre des forces en physiquePas encore d'évaluationExercices d'équilibre des forces en physique3 pages
- Signification des Nombres et VibrationsPas encore d'évaluationSignification des Nombres et Vibrations4 pages