Computational Stochastics: available software

Acor, estimating the mean of a time series: statistical error bars

Markov chain Monte Carlo produces a time series of correlated values that are to be used to estimate the expected value of something. A statistical error bar is an estimate of how far the mean of the time series is from the actual mean of the process -- what you would get if the series were infinitely long. Closely related to the statistical error bar is the autocorrelation time. The statistical error bar of the mean is the standard deviation of the time series divided by the effective sample size. The effective sample size is L/tau, where L is the length of the time series and tau is the autocorrelation time. This program is very simple, but just complicated enough that you might prefer to download rather than writing it from scratch.