AutocorrError, 'The chain is too short to reliably estimate the autocorrelation time'

I’ve been using emcee without any problems for a while. Recently I tried running my program in Windows10, but I get an error with the autocorr (that I didn’t get before). Using the quickstart.py example I get the same problem but I can’t figure out whats wrong whatsoever. I hope you can help me figuring out whats wrong . This is the log I get running quickstart.py :

In [1]: run quickstart.py
Mean acceptance fraction: 0.191032
---------------------------------------------------------------------------
AutocorrError                             Traceback (most recent call last)
C:\Users\anibal\Documents\GitHub\emcee\examples\quickstart.py in <module>()
     58 # Estimate the integrated autocorrelation time for the time series in each
     59 # parameter.
---> 60 print("Autocorrelation time:", sampler.get_autocorr_time())
     61
     62 # Finally, you can plot the projected histograms of the samples using

C:\Users\anibal\Anaconda2\lib\site-packages\emcee-2.2.1-py2.7.egg\emcee\ensemble.pyc in get_autocorr_time(self, low, high, step, c, fast)
    501         return autocorr.integrated_time(np.mean(self.chain, axis=0), axis=0,
    502                                         low=low, high=high, step=step, c=c,
--> 503                                         fast=fast)
    504
    505

C:\Users\anibal\Anaconda2\lib\site-packages\emcee-2.2.1-py2.7.egg\emcee\autocorr.pyc in integrated_time(x, low, high, step, c, full_output, axis, fast)
    116             break
    117
--> 118     raise AutocorrError("The chain is too short to reliably estimate "
    119                         "the autocorrelation time")
    120

AutocorrError: The chain is too short to reliably estimate the autocorrelation time

[dfm]

The most recent version of emcee has a more robust autocorrelation time estimator so I reduced the dimensionality of the quickstart example and ran more steps of MCMC. This error won’t be thrown for the example now and, if you’re finding it with your own code, that means that your chains probably haven’t converged! If you must, you can reduce the c parameter to a smaller value but the estimate won’t be as reliable so be careful.

[dfm]

@deltasata: you need to update emcee (to >3) or follow the old docs. Python 2.7 is no longer officially supported for emcee v3 however.