Physics Letters B 288 ( 1992 ) 129-139
North-Holland
PHYSICS LETTERS B
Stochastic quantization of bottomless systems
based on a kerneled Langevin equation
Satoshi Tanaka, Mikio Namiki, Ichiro Ohba, Masashi Mizutani,
Nobuyuki Komoike and Masahiko Kanenaga
Department of Physics, Waseda University, Tokyo169, Japan
Received 13 January 1992; revised manuscript received 4 May 1992
We propose a new method for quantizing systems with bottomless action functionals. Our method is based on the kerneled
Langevin equation of the stochastic quantization method. We prove that our kerneled Langevin equation describes a stochastic
process with a thermal equilibrium state. The proof makes the relation between our method and the conventional path-integral
quantization clear. We check numerically in a simple model that dynamical evolution leads to the desired equilibrium state. We
also calculateconvergent nonperturbative expectation values that are consistent with the perturbative results in the case of a small
coupling constant. We also confirm that our method gives the same perturbative results as the conventional path-integral quanti-
zation method. Finally, to augment understanding, we discuss a (d+ 1)-dimensional path-integral representation of our method.
I. Introduction
Since the stochastic quantization method (SQM) was first proposed [ 1 ], many authors have pointed out that
this method extends the territory of quantum theory. For example, Greensite and Halpern proposed a scheme
to apply the SQM to bottomless systems [2]. Their scheme is based on the (d+ 1 )-dimensional path-integral
representation of the SQM [3-6], and it was recently applied to two-dimensional gravity [7]. This scheme is
interesting but unsatisfactory, in that the relation to the conventional d-dimensional path-integral quantization
method is unclear. The key problem is that the stochastic process corresponding to the (d+ 1 )-dimensional
path-integral representation does not have an equilibrium state, reflecting the unboundedness of the action
functional, and so the Feynman path-integral measure is not given as an equilibrium Fokker-Planck distribution.
Here we propose a new stochastic quantization scheme to quantize bottomless systems based on a kerneled
Langevin equation [8] ~1. Our kerneled Langevin equation describes a stochastic process which has an equilib-
rium state, and the equilibrium Fokker-Planck distribution makes the relation between our method and the
conventional d-dimensional path-integral quantization method clear. The presence of an equilibrium state makes
it straightforward to apply our scheme to two-dimensional gravity, for example.
Let us briefly review the SQM based on a kerneled Langevin equation. A scalar field O characterized by an
action functional S[¢] is quantized by setting up the following kerneled Langevin equation,
0 f 8S ~ 8K(x,x';¢) f
Ot ¢ ( x ' t ) = - ddx'K(x'x';¢) 5¢(x',t~ + ddx' 5V(x',t) + ddx'G(x'x';¢)q(x"t)' (1.1a)
where t is the fictitious-time variable, and the kernel K(x, x'; 0) is a real functional depending only on ¢ at
fictitious time t. The gaussian white noise q(x, t) is characterized by the statistical properties,
(q(x,t))=O, (q(x,t)q(x',t'))=2da(x-x')3(t-t '). (1.1b)
~ Ref. [9 ] includes some arguments of stabilizingstochastic processes using a kernel.
0370-2693/92/$ 05.00 © 1992 ElsevierSciencePublishers B.V. All rights reserved. 129