something's up with PyMC3's diagnostics...


above is their geweke statistic versus my version & that found in CODA.

Not sure what’s up, but it’s not good…

imported from: yetanothergeographer