The quantum canonical ensemble in phase space

TL;DR

This paper explores the quantum canonical ensemble in phase space, bridging high and low temperature regimes with semiclassical approximations and providing methods to evaluate thermal averages without trajectory searches.

Contribution

It introduces a semiclassical extension for quantum propagators at imaginary thermal time and a doubled phase space approach to simplify thermal average calculations.

Findings

Thermal Wigner function approaches classical distribution at high temperatures.

Quantum corrections to classical regimes are derived for all temperature ranges.

A doubled phase space method avoids trajectory searches in thermal average evaluations.

Abstract

The density operator for a quantum system in thermal equilibrium with its environment depends on Planck's constant, as well as the temperature. At high temperatures, the Weyl representation, that is, the thermal Wigner function, becomes indistinguishable from the corresponding classical distribution in phase space, whereas the low temperature limit singles out the quantum ground state of the system's Hamiltonian. In all regimes, thermal averages of arbitrary observables are evaluated by integrals, as if the thermal Wigner function were a classical distribution. The extension of the semiclassical approximation for quantum propagators to an imaginary thermal time, bridges the complex intervening region between the high and the low temperature limit. This leads to a simple quantum correction to the classical high temperature regime, irrespective of whether the motion is regular or chaotic. A variant of the full semiclassical approximation with a real thermal time, though in a doubled phase space, avoids any search for particular trajectories in the evaluation of thermal averages. The double Hamiltonian substitutes the stable minimum of the original system's Hamiltonian by a saddle, which eliminates local periodic orbits from the stationary phase evaluation of the integrals for the partition function and thermal averages.