ksmirnov - Kolmogorov-Smirnov test for MCMC convergence¶

 ksmirnov Kolmogorov-Smirnov test of similarity between the empirical distribution at the start and at the end of the chain.

Kolmogorov-Smirnov test for MCMC convergence.

Use the K-S tests to compare the distribution of values at the front of the chain to that at the end of the chain. If the distributions are significantly different, then the MCMC chain has not converged.

bumps.dream.ksmirnov.ksmirnov(seq, portion=0.25, filter_order=15)[source]

Kolmogorov-Smirnov test of similarity between the empirical distribution at the start and at the end of the chain. Apply a median filter (filter=15) on neighbouring K-S values to reduce variation in the test statistic value.