Anderson Localization – Looking Forward
Boris Altshuler
Physics Department, Columbia University
Collaborations: Igor Aleiner
Also Denis Basko, Gora Shlyapnikov,
Vincent Michal, Vladimir Kravtsov, …
Lecture1
September, 8, 2015
Outline
1. Introduction
2. Anderson Model; Anderson Metal and Anderson Insulator
3. Phonon-assisted hoping conductivity
4. Localization beyond the real space. Integrability and chaos.
5. Spectral Statistics and Localization
6. Many-Body Localization.
7. Many-Body Localization of the interacting fermions.
8. Many-Body localization of weakly interacting bosons.
9. Many-Body Localization and Ergodicity
>50 years of Anderson Localization
qp
…very few believed it
[localization] at the time,
and even fewer saw its
importance; among those
who failed to fully
understand it at first was
certainly its author…
Part 1.
Introduction and
History
Diffusion
D 0
2
r 2 Dt
Equation t
Diffusion
constant
r ,t
Can be density of particles or energy density.
It can also be the probability to find a
particle at a given point at a given time
Einstein theory
of Brownian
motion, 1905
The diffusion equation is valid for any random walk
provided that there is no memory (markovian process)
Diffusion
D 0
2
Equation t
Einstein-Sutherland Relation for electric conductivity
dn
e D
2
d Density of
states
If electrons would be degenerate
and form a classical ideal gas
ntot
William Sutherland
(1859-1911)
T
Will a fluctuation (wave packet) spread ?
t
r 2 t ? r 2 t
t
r Diffusion Constant
Einstein (1905): Random walk without memory
D 2 0 r 2 Dt
t
always diffusion
Anderson (1958): For quantum particles extended states r 2
t
Dt
not always localized states r
2
t
const
Basic Quantum Mechanics:
-spectrum
Potential well
Spectra
Continuous Discrete
Unbound states Bound states
r
2
L
O L d
r
2
r
O e
r
Extended states Localized states
L System size
d Number of the spatial dimensions
Localization of single-particle wave-functions.
Continuous limit:
Random potential
extended
localized
Experiment
Spin Diffusion
Feher, G., Phys. Rev. 114, 1219 (1959); Feher, G. & Gere, E. A., Phys. Rev. 114, 1245 (1959).
Microwave
Dalichaouch, R., Armstrong, J.P., Schultz, S.,Platzman, P.M. & McCall, S.L. “Microwave
localization by 2-dimensional random scattering”. Nature 354, 53-55, (1991).
Chabanov, A.A., Stoytchev, M. & Genack, A.Z. Statistical signatures of photon
localization. Nature 404, 850-853 (2000).
Pradhan, P., Sridar, S, “Correlations due to localization in quantum
eigenfunctions od disordered microwave cavities”, PRL 85, (2000)
f = 3.04 GHz f = 7.33 GHz
Localized Extended
Experiment
Localization of Ultrasound
Weaver, R.L. “Anderson localization of ultrasound”.
Wave Motion 12, 129-142 (1990).
H. Hu, A. Strybulevych, J. H. Page, S. E. Skipetrov & B. A. van
Tiggelen “Localization of ultrasound in a three-dimensional elastic
network” Nature Phys. 4, 945 (2008).
Localization of Light
D. Wiersma, Bartolini, P., Lagendijk, A. & Righini R. “Localization of light in a disordered medium”,
Nature 390, 671-673 (1997).
Scheffold, F., Lenke, R., Tweer, R. & Maret, G. “Localization or classical diffusion of light”,
Nature 398,206-270 (1999).
Schwartz, T., Bartal, G., Fishman, S. & Segev, M. “Transport and Anderson localization in disordered two
dimensional photonic lattices”. Nature 446, 52-55 (2007).
[Link], [Link], [Link], P. [Link], [Link], [Link] “Cavity Quantum Electrodynamics
with Anderson localized Modes” Science 327, 1352-1355, (2010)
Localization of cold atoms
Billy et al. “Direct observation of Anderson localization
of matter waves in a controlled disorder”. Nature 453, 891- 894 (2008).
87Rb
Roati et al. “Anderson localization of a non-interacting Bose-Einstein
condensate“. Nature 453, 895-898 (2008).
What about charge transport ?
Problem: electrons interact with each other
Einstein Relation (1905) dn
e2 D
Conductivity d
Density of states
Diffusion Constant
t
D 2 0
t At zero temperature
t Extended states G µ Ld-2 ;
- metal
r s ¾L®¥
¾¾ ® const; D ¾L®¥
¾¾ ® const
æIö
G=ç ÷ Localized states
è V ø V =0 Conductance
- insulator: G µ e- L/x ; s = 0; D = 0
I L
V s =G Conductivity
A
Part 2.
Anderson Model;
Anderson Metal
and
Anderson Insulator
Anderson • Lattice - tight binding model
Model • Onsite energies ei - random
• Hopping matrix elements Iij
j i
Iij
I i and j are nearest
-W < ei <W
uniformly distributed
{
Iij =
0
neighbors
otherwise
Anderson Transition
I < Ic I > Ic
Insulator Metal
All eigenstates are localized There appear states extended
Localization length all over the whole system
One-dimensional Anderson Model
e1 I 0
ˆ
H I I
0 I eN
Why arbitrary
Q: weak hopping I is
not sufficient for
the existence of
the diffusion
? j
Iij
i
Einstein (1905): Marcovian (no memory)
process g diffusion
Quantum mechanics is not marcovian
There is memory in quantum propagation
A: Quantum Interference
!
Why ?
Quantum mechanics is not marcovian
Why ? A:
!
There is memory in quantum propagation
Quantum Interference
Memory!
Quantum mechanics is not marcovian
There is memory in quantum propagation
Why ? Quantum Interference
A:
!
Memory!
pdr The particle can go around
the loop in two directions
Phase accumulated
when traveling
along the loop O
Constructive interference probability of the return to the origin gets
enhanced quantum corrections reduce the diffusion constant.
Tendency towards localization
WEAK LOCALIZATION
Localization of single-particle wave-functions.
Continuous limit:
Random potential
extended
d=1; All states are localized
d=2; All states are localized
d >2; Anderson transition
localized
Two-site Two-well
I
Anderson = potential
model e2
e1
Hamiltonian
e I E 0
Hˆ 1
diagonalize ˆ
H 1
I e2 0 E2
E2 E1 e 2 e1 2
I 2
e I E 0
Hˆ 1
diagonalize ˆ
H 1
I e2 0 E2
e 2 e1 e 2 e1 I
E2 E1 e 2 e1 2
I 2
I e 2 e1 I
von Neumann & Wigner “noncrossing rule”
Level repulsion
v. Neumann J. & Wigner E. 1929 Phys. Zeit. v.30, p.467
Arnold V.I., Mathematical Methods of Classical Mechanics
(Springer-Verlag: New York), Appendix 10, 1989
In general, a multiple spectrum in
typical families of quadratic forms
is observed only for two or more E Ĥ x E x
parameters, while in one-parameter
families of general form the
spectrum is simple for all values of
the parameter. Under a change of
parameter in the typical one-
parameter family the eigenvalues
can approach closely, but when
they are sufficiently close, it is as if
they begin to repel one another.
The eigenvalues again diverge,
disappointing the person who
hoped, by changing the parameter x
to achieve a multiple spectrum.
e I E 0
Hˆ 1
diagonalize ˆ
H 1
I e2 0 E2
e 2 e1 e 2 e1 I
E2 E1 e 2 e1 2
I 2
I e 2 e1 I
von Neumann & Wigner “noncrossing rule”
Level repulsion
v. Neumann J. & Wigner E. 1929 Phys. Zeit. v.30, p.467
What about the eigenfunctions ?
e I e 2 e1 e 2 e1 I
ˆ
H 1
E2 E1 e 2 e1
2
I
2
I e2 I e 2 e1 I
What about the eigenfunctions ?
1 , e1; 2 , e 2 1 , E1; 2 , E2
e 2 e1 I e 2 e1 I
I
1,2 1,2 O 2,1
e 2 e1 1,2 1,2 2,1
Off-resonance Resonance
Eigenfunctions are In both bonding and anti-bonding
close to the original on- eigenstates the probability is
site wave functions equally shared between the sites
Anderson insulator Anderson metal
Few isolated resonances Many resonances and they overlap
Transition: Typically each site is in the
resonance with some other one
Anderson • Lattice - tight binding model
Model • Onsite energies ei - random
• Hopping matrix elements Iij
j i
Iij
I i and j are nearest
-W < ei <W
uniformly distributed
{
Iij =
0
neighbors
otherwise
Anderson Transition
I < Ic I > Ic
Insulator Metal
All eigenstates are localized There appear states extended
Localization length all over the whole system
Condition for
Localization:
energy mismatch
I< # of [Link]
energy e i e j typ W
mismatch
# of nearest
neighbors 2d
A bit more precise:
Logarithm is due to the resonances, which are not nearest neighbors
Condition for Localization:
Q:Is it correct ?
For low dimensions – NO. I for d 1, 2
A1:All states are localized. Reason – loop trajectories
c
A2:Works better for larger dimensions d 2
A3:Is exact on the Bethe lattice
Condition for Localization:
Q:Is it correct ?
For low dimensions – NO. I for d 1, 2
A1:All states are localized. Reason – loop trajectories
c
A2:Works better for larger dimensions d 2
A3:Is exact on the Bethe lattice
Rule At the localization At the localization transition the hoping
transition a typical matrix element is of the order of the
of the site is in resonance typical energy mismatch divided by the
thumb with another one number of nearest neighbors
Anderson’s recipe:
Consider an open system (Anderson Model).
Particle escape “broadening” of each eigenstate
E E i ; Im
- scape rate (inverse dwell time)
I 0 0
States localized inside the system – small
Extended states – large
Anderson’s recipe:
1. take descrete spectrum E of H0 insulator
2. Add an infinitesimal Im part i to E
3. Evaluate Im
1 2
4 1) N
limits 2) 0
4. take limit 0 but only after N metal
5. “What we really need to know is the
!
probability distribution of Im, not
its average…”
Probability Distribution of =Im
is an infinitesimal width (Im
part of the self-energy due to
metal a coupling with a bath) of
insulator
one-electron eigenstates
Look for:
V
Anderson Transition
I > Ic I < Ic
localized and extended
never coexist!
all states are
localized
extended
DoS DoS
- mobility edges
Temperature dependence of the conductivity
one-electron picture
Mobility Chemical
edge potential
DoS DoS DoS
Ec e F
T 0 0 T e T
T 0 T
there are extended states all states are localized
I Ic I Ic
Part 3.
Localization beyond
real space
Integrability and chaos
Localization in the angular momentum space
Quantum and Classical Dynamical Systems
Large number d 1 of the degrees of freedom
Conventional Boltzmann- Integrable Systems
Gibbs Statistical Physics
Equipartition Postulate
Ergodicity: time average =
space (ensemble) average
?
H H 0 V
d
d
degrees of freedom
integrals of motion
Ergodicity is violated
Invariant tori dimension d
Chaos
Hamiltonian H 0 pi , qi
Hamiltonian H pi , qi Energy shell, dimension 2d 1
Classical Dynamics
Ergodicity KAM region
Equipartition Fermi, Pasta, Ulam system Arnold diffusion
0
(connected nonlinear oscillators) Non-ergodic
Solar system
Quantum Dynamics ??? ...
Kolmogorov – Arnold – Moser (KAM) theory
A.N. Kolmogorov, Integrable classical Hamiltonian
Dokl. Akad. Nauk SSSR,
1954. Ĥ 0 d >1:
Proc. 1954 Int. Congress
of Mathematics, North- Separation of variables: d sets of
Holland, 1957
action-angle variables
I1 , 1 21t ;.., I 2 , 2 22t ;..
Quasiperiodic motion: set of the
frequencies, 1 , 2 ,.., d , which are in
general incommensurate. Actions I i are
Andrey integrals of motion I i t 0
Kolmogorov
1 2
I1 I2
…=>
tori
Vladimir Jurgen
Arnold Moser
Kolmogorov – Arnold – Moser (KAM) theory
A.N. Kolmogorov, Dokl. Akad. Nauk SSSR, 1954.
Proc. 1954 Int. Congress of Mathematics, North-Holland, 1957
1 2 Given the set of the integrals
I1
I2
…=>
of motion I all trajectories
belong to a torus
Will an arbitrary weak perturbation Vˆ of an
integrable Hamiltonian Ĥ 0 destroy the tori and
Q: make the motion ergodic (when each point at the
energy shell will be reached sooner or later) ?
Most of the tori survive weak
A: and smooth enough perturbations
KAM
theorem
Classical Dynamics: Quantum Dynamics:
from KAM to Chaos
H H 0 V Many – Body Localization
Classical, d>>1 degrees of freedom Quantum, d>>1 degrees of freedom
Integrals of motion are
I1, I2, . . ., Id quantized – quantum I2
integrals of motion
I2 numbers
Form a “lattice”.
Energy Sites of the “lattice” –
Shell eigenstate of the
Space of the 0
integrals of motion
integrable system 0
I1 I1
Rectangular billiard
Lx I2 0
Ly
I1
Two integrals of motion nx ny
px ; py
I1 px ; I1 p y Lx Ly
Energy
px2 p y2 2mE
shell:
Classical Dynamics: Quantum Dynamics:
from KAM to Chaos
H H 0 V Many – Body Localization
Classical, d>>1 degrees of freedom Quantum, d>>1 degrees of freedom
Integrals of motion are
I1, I2, . . ., Id quantized – quantum I2
integrals of motion
I2 numbers
Form a “lattice”.
Energy Sites of the “lattice” –
Shell eigenstate of the
Space of the 0
integrals of motion
integrable system 0
KAM Perturbation – coupling
of the different sites
Most of the tori of the “lattice” - bonds
survive weak and
smooth
perturbation I1 I1
Classical Dynamics: Quantum Dynamics:
from KAM to Chaos
H H 0 V Many – Body Localization
Classical, d>>1 degrees of freedom Quantum, d>>1 degrees of freedom
Integrals of motion
I1, I2, . . ., Id are quantized – I2 Finite
integrals of motion
I2 quantum numbers motion
Form a “lattice”.
Energy Sites of the “lattice”
Space of the Shell –eigenstate of the
0 integrable system 0
integrals of motion
KAM Perturbation – coupling
Most of the tori of the different sites
survive weak and of the “lattice” -
smooth bonds
perturbation I1 I1
Wave functions
localized in the space
of quantum numbers
Classical Dynamics: Quantum Dynamics:
from KAM to Chaos
H H 0 V Many – Body Localization
Classical, d>>1 degrees of freedom Quantum, d>>1 degrees of freedom
Integrals of motion are
I1, I2, . . ., Id quantized – quantum I2 Finite
integrals of motion
I2 numbers motion
Form a “lattice”.
Energy Sites of the “lattice” –
Shell eigenstate of the
Space of the 0
integrals of motion
integrable system 0
KAM Perturbation – coupling
of the different sites
Most of the tori of the “lattice” - bonds
survive weak and
smooth
perturbation I1 Quantum random walk I1
Localization?
Many – Body Localization is an analog of
the Anderson Localization in a finite-
dimensional space of a quantum particle
subject to a random potential
KAM Most of the tori survive weak and
theorem: smooth enough perturbations
I2 I2
Vˆ 0
I1 I1
0
I
I I1 ,..., I d I2 0 0
c 0
The integrals of
motion are quantized
Energy shell I1
px2 p y2
E
2m
Vˆ ,
Matrix element of
the perturbation
0
I
I I1 ,..., I d Anderson Model !
AL hops are local – one can distinguish “near” and “far”
KAM perturbation is smooth enough
Pradhan
& Sridar, Sinai
PRL, 2000 billiard
Square Disordered
billiard localized
Disordered
Localized extended Localized
momentum space extended real space
Glossary
Classical Quantum
Integrable Integrable
Hˆ 0 E , I
H 0 H 0 I ; I t 0
Perturbation Perturbation
V ; I t 0 Vˆ V ,
,
KAM Localized
Ergodic (chaotic) Extended ?
Question:
What is the reason to speak about localization if we in
general do not know the space in which the system is localized ?
Need an invariant (basis independent) criterion of the localization