It is pointed out that a method recently described by Hillery and Zubairy can be generalized to yield the multitime correlation function in any order for an interacting quantum field in the form of a phase-space average. The correlations are expressible as
c-number integrals with the help of the coherent-state propagator, but the associated phase-space density is complex.
Using the recently developed phase-space techniques for the treatment of quantum-mechanical problems, we set up a procedure for calculating multitime-correlation functions in terms of the joint distribution functions. The correlation functions are then expressed simply as integrals over the associated phase space. Explicit expressions are given for these joint distribution functions, in terms of Green's functions of the
c-number equations of motion for the phase-space equivalent of the density operator. Using these joint distribution functions, an exact regression theorem is rederived, and the connection with the multitime correspondence between classical and quantum stochastic processes is discussed.
We discuss a procedure for the calculation of operator correlation functions that arise frequently in quantum-optics problems. The starting point of our method is the tradiational quantum regression theorem, and its implementation requires only the solution of the master equation for the density operator of the system of interest. An attractive feature of this scheme of calculation is that it offers some distinctive advantages over the conventional Langevin-equations approach or other procedures that rely on the mapping of the density operator into appropriate phase-space distribution functions. Although the practical implementation of this method is likely to require numerical computations for virtually every model of physical interest, the procedure is sufficiently general to allow consideration of both field and atomic correlation functions, as well as multitime expectation values of mixed products of atomic and field operators. In this paper we provide a fairly detailed derivation of the formulas of interest, and illustrate our results with specific examples and tests of the procedure.