A methodology for generating sample functions for colored noise which has some relative advantages over existing ones is outlined. This involves simulating a stochastic process in terms of harmonic functions, the amplitude of which is related to the power spectral density of the noise. The computational advantages of the fast Fourier transform are utilized and the numerical integration of the noise equation is avoided. In light of this methodology, existing algorithms for colored-noise simulation are critiqued and compared with the present one. Numerical results for a particular case study, namely that of an overdamped particle in a bistable potential, are presented and the results are compared with existing theories.

• Received 1 August 1990

DOI:https://doi.org/10.1103/PhysRevA.42.7492