Numerical solution of fokker/planck/kramers equations

Some Fokker/Planck/Kramers equations of current interest are solved numerically for autocorrelation functions and spectra. It is demonstrated that uncritical use of these equations should be avoided because of the neglect of memory effects inherent in their make-up. Only in the case discussed by Evans (1976) does this type of equation produce realistic spectra, and then only over a limited range of temperature and viscosity. The way to proceed in problems involving molecular diffusion of this type is to use molecular dynamics simulation