A new approach for treating fluctuations in noise theory

摘要

A new method, introduced in a previous article, is shown to overcome certain physical difficulties encountered in noise problems when treated by more common approaches. In particular, an internal inconsistency in the original Langevin approach for treating Brownian motion is explicitly presented in which problems arise because the statistical treatment is carried out at a too macroscopic level. As another example, a more realistic expression for the retrogression of a fluctuation is obtained in which the cusp at t=0, characteristic of Markov approaches, is avoided. The method is extended to treating systems of subsystems, and the concept of internal noise is introduced to enable a simpler, more macroscopic noise analysis. The method is also applied to determine statistical properties of fluctuations in irreversible thermodynamcis. We hope that our discussion will further illustrate the origin, nature, and treatment of fluctuations in linear classical systems.