a

r
X
i
v
:
m
a
t
h
-
p
h
/
0
0
0
5
0
2
0
v
1


1
6

M
a
y

2
0
0
0
On p-adic path integral
Branko Dragovich
Institute of Physics, [Link] 57, 11001 Belgrade, Yugoslavia
Dedicated to the memory
of [Link]
Abstract
Feynmans path integral is generalized to quantum mechanics on p-adic
space and time. Such p-adic path integral is analytically evaluated for quadratic
Lagrangians. Obtained result has the same form as that one in ordinary quan-
tum mechanics.
1. It is well known that dynamical evolution of any one-dimensional quantum-
mechanical system, described by a wave function (x, t), is given by
(x

, t

) =
_
K(x

, t

; x

, t

)(x

, t

)dx

, (1)
where K(x

, t

; x

, t

) is the kernel of the corresponding unitary operator acting as

follows:
(t

) = U(t

, t

)(t

). (2)
K(x

, t

; x

, t

) is also called Greens function, or the quantum-mechanical prop-

agator, and the probability amplitude to go a particle from a point (x

, t

) to a point
(x

, t

). One can easily deduce the following three general properties:

_
K(x

, t

; x, t)K(x, t; x

, t

)dx = K(x

, t

; x

), (3)
_
K

(x

, t

; x

, t

)K(x

, t

; x, t

)dx

= (x

x), (4)
K(x

, t

; x

, t

) = lim
t

t
K(x

, t

; x

, t

) = (x

). (5)
Since all information on quantum dynamics can be deduced from the propagator
K(x

, t

; x

, t

) it can be regarded as the basic ingredient of quantum theory. In

Feynmans formulation [1] of quantum mechanics, K(x

, t

; x

, t

) was postulated to
be the path integral
K(x

, t

; x

, t

) =
_
exp
_
2i
h
_
t

L( q, q, t)dt
_
Dq, (6)
where x

= q(t

) and x

= q(t

), and h is the Planck constant.

In its original form, the path integral (6) is the limit of the corresponding multiple
integral of n1 variables q
i
= q(t
i
), (i = 1, 2, ..., n1), when n . For the half
of century of its history, the path integral has been a subject of permanent interest
1
in theoretical and mathematical physics. At present days (see, e.g. [2]) it is one of
the most profound and promising approaches to foundations of quantum theory (in
particular, quantum eld theory and superstring theory). Feynmans path integral is
inevitable in formulation of p-adic [3] and adelic [4] quantum mechanics. It is worth
noting that just Feynmans path integral approach enables natural foundation of
quantum theory on p-adic and adelic spaces.
2. Recall that the set of rational numbers Q plays an important role in mathe-
matics as well as in physics. From algebraic point of view, Q is the simplest number
eld of characteristic 0. The usual absolute value and p-adic valuation (p is any of
prime numbers) exhaust all possible non-trivial norms on Q [5]. Completion of Q
with respect to metrics induced by these norms leads to the eld of real numbers
R and the elds of p-adic numbers Q
p
, (p = 2, 3, 5, ...). Thus Q is dense in R and
all Q
p
. From physical point of view, all numerical results of measurements are ra-
tional numbers. However, theoretical models of physical systems are traditionally
constructed using real and complex numbers. One can ask the following question:
Why real (and complex) numbers are so good in description of usual physical phe-
nomena, and, is there any aspect of physical reality which has to be described by
p-adic numbers. Construction of p-adic models and their appropriate interpretation
can gradually give answer to this question. Since 1987, there have been many pub-
lications (for a review, see, e.g. [6, 7, 8] ) on possible applications of p-adic numbers
in modern theoretical and mathematical physics. For a systematic approach to this
subject, p-adic [3] and adelic [4] quantum mechanics have been formulated.
Recall also that any p-adic number x Q
p
can be presented as the following
innite expansion
x = p

(x
0
+ x
1
p + x
2
p
2
+ ), Z,
where x
i
= 0, 1, ..., p 1 are digits. We will use the Gauss integral [7]
_
Qp

p
(x
2
+ x)dx =
p
() | 2 |

1
2
p
p
_

2
4
_
, = 0,
where
p
(a) = exp(2i{a}
p
) is the additive character, and {a}
p
is the fractional
part of a Q
p
.
p
(x) is a complex-valued arithmetic function (for a denition, see
[7]) with the following properties:

p
(0) = 1,
p
(a
2
x) =
p
(x),
p
(x)
p
(y) =
p
(x + y)
p
(x
1
+ y
1
),

p
(x)
p
(x) = 1.
3. p-Adic quantum mechanics, we are interested in, contains complex-valued
functions of p-adic arguments. There is not the corresponding Schr odinger equa-
tion, but Feynmans path integral approach seems to be quite natural. Feynmans
path integral for p-adic propagator K
p
(x

, t

; x

, t

), where K
p
is complex-valued and
x

, x

, t

, t

are p-adic variables, is a direct p-adic generalization of (6), i.e.

K
p
(x

, t

; x

, t

) =
_

p
_

1
h
_
t

L( q, q, t)dt
_
Dq, (7)
2
where
p
(a) is p-adic additive character. The Planck constant h in (6) and (7)
is the same rational number. Integral
_
t

t
L( q, q, t)dt we consider as the dierence
of antiderivative (without pseudoconstants) of L( q, q, t) in nal (t

) and initial (t

)
times. Dq =

n1
i=1
dq(t
i
), where dq(t
i
) is the p-adic Haar measure. Thus, p-adic path
integral is the limit of the multiple Haar integral when n . To calculate (7) in
this way one has to introduce some order on t Q
p
, and it is successfully done in
Ref. [9]. On previous investigations of p-adic path integral one can see [10]. Our
main task here is derivation of the exact result for p-adic Feynmans path integral
(7) for the general case of Lagrangians L( q, q, t), which are quadratic polynomials
in q and q, without making time discretization.
A general quadratic Lagrangian can be written as follows:
L( q, q, t) =
1
2

2
L
0
q
2
q
2
+
L
0
q
q +

2
L
0
qq
qq + L
0
+
L
0
q
q +
1
2

2
L
0
q
2
q
2
, (8)
where index 0 denotes that the Taylor expansion of L( q, q, t) is around q = q = 0.
The Euler-Lagrange equation of motion is

2
L
0
q
2
q +
d
dt
_

2
L
0
q
2
_
q +
_
d
dt
_

2
L
0
qq
_

2
L
0
q
2
_
q =
L
0
q

d
dt
_
L
0
q
_
. (9)
General solution of (9) is
q x(t) = C
1
x
1
(t) + C
2
x
2
(t) + w(t), (10)
where x
1
(t) and x
2
(t) are two linearly independent solutions of the corresponding
homogeneous equation, and w(t) is a particular solution of the complete equation (9).
Note that x(t) denotes the classical trajectory. Imposing the boundary conditions
x

= x(t

) and x

= x(t

), constants of integration C
1
and C
2
become:
C
1
C
1
(t

, t

) =
(x

)x

2
(x

)x

2
x

2
x

1
x

1
x

2
, (11a)
C
2
C
2
(t

, t

) =
(x

)x

1
(x

)x

1
x

2
x

1
x

1
x

2
. (11b)
Since C
1
(t

, t

) and C
2
(t

, t

) are linear in x

and x

, the corresponding classical

action

S(x

, t

; x

, t

) =
_
t

t
L( x, x, t)dt is quadratic in x

and x

. Note that the

above expressions have the same form in R and Q
p
.
Quantum uctuations lead to deviations of classical trajectory and any quantum
path may be presented as q(t) = x(t)+y(t), where y

= y(t

) = 0 and y

= y(t

) = 0.
The corresponding Taylor expansion of S[q] around classical path x(t) is
S[q] = S[x +y] = S[x] +
1
2!

2
S[x] = S[x] +
1
2
_
t

_
y

q
+ y

q
_
(2)
L( q, q, t)dt, (12)
3
where we used S[x] = 0. We have now
K
p
(x

, t

; x

, t

) =
p
_

1
h
S[x]
__

p
_
_

1
2h
_
t

_
y

q
+ y

q
_
(2)
L( q, q, t)dt
_
_
Dy
(13)
with y

= y

= 0 and S[x] =

S(x

, t

; x

, t

).
Note that K
p
(x

, t

; x

, t

) has the form

K
p
(x

, t

; x

, t

) = N
p
(t

, t

)
p
_

1
h

S(x

, t

; x

, t

)
_
, (14)
where N
p
(t

, t

) does not depend on end points x

and x

. To calculate N
p
(t

, t

)
we use conditions (3) and (4). Substituting K
p
(x

, t

; x

, t

) into (4) we obtain (for

details, see [10]):
N
p
(t

, t

) =

1
h

2

S
x

(x

, t

; x

, t

1
2
p
A
p
(t

, t

), (15)
where |A
p
(t

, t

)|

= 1, (| |
p
and | |

denote p-adic and absolute value, respec-

tively). Replacing (15) in equation (3) we get conditions:
A
p
(t

, t)A
p
(t, t

)
p
() = A
p
(t

, t

), (16)

1
h

2

S
x

x
(x

, t

; x, t)

1
2
p

1
h

2

S
xx

(x, t; x

, t

1
2
p
|2|

1
2
p =

1
h

2

S
x

(x

, t

; x

, t

1
2
p
,
(17)
where
=
1
2h
_

2

S
x
2
(x

, t

; x, t) +

2

S
x
2
(x, t; x

, t

)
_
. (18)
Analysing the above formulae we obtain [10]
A
p
(t

, t

) =
p
_

1
2h

2

S
x

(x

, t

; x

, t

)
_
. (19)
For details of a quite rigorous derivation of (19), see [11].
As the nal result we have
K
p
(x

, t

; x

, t

) =
p
_

1
2h

2

S
x

1
h

2

S
x

1
2
p

p
_

1
h

S(x

, t

; x

, t

)
_
(20)
which is the p-adic Feynman path integral for quadratic Lagrangians. The corre-
sponding path integral of ordinary quantum mechanics [10] can be transformed into
the same form as (20), i.e. in such case index p is replaced by index . This sup-
ports Volovichs conjecture [12] that fundamental physical laws should be invariant
under interchange of number elds Q
p
and R.
Acknowledgement. Author wishes to thank the organizers of the Bogolyubov
Conference: Problems of Theoretical and Mathematical Physics, for invitation to
participate in Moscow and Dubna parts of the Conference.
4
References
[1] [Link], Rev. Mod. Phys. 20 (1948)367.
[2] [Link] the 6
th
Int. Conference on Path Integrals from peV to TeV: 50 Years
after Feynmans Paper, Eds. [Link] et al. (World Scientic, Singapore,
1999).
[3] [Link] and [Link], Commun. Math. Phys. 123 (1989)659.
[4] [Link], Teor. Mat. Fiz. 101 (1994)349; Int. J. Mod. Phys. A10
(1995)2349.
[5] [Link], Ultrametric Calculus (Cambridge Univ. Press, 1984).
[6] [Link] and [Link], Phys. Reports 233 (1993)1.
[7] [Link], [Link] and [Link], p-Adic Analysis and Mathemat-
ical Physics (World Scientic, Singapore, 1994).
[8] [Link], p-Adic Valued Distributions in Mathematical Physics (Kluwer,
1994).
[9] [Link], J. Math. Phys. 32 (1991)147.
[10] [Link] and [Link], Mod. Phys. Lett. A12 (1997)1455.
[11] [Link], [Link] and [Link], p-Adic Generalization of Feynmans
Path Integral, in preparation.
[12] [Link], Number Theory as the Ultimate Physical Theory, preprint CERN-
Th. 4781/87, July 1987.
5