search

CamDavidsonPilon / Probabilistic-Programming-and-Bayesian-Methods-for-Hackers

aka "Bayesian Methods for Hackers": An introduction to Bayesian methods + probabilistic programming with a computation/understanding-first, mathematics-second point of view. All in pure Python ;)
http://camdavidsonpilon.github.io/Probabilistic-Programming-and-Bayesian-Methods-for-Hackers/
MIT License
26.72k stars 7.88k forks source link

Questions about content and TF #451

Open bluesky314 opened 5 years ago

bluesky314 commented 5 years ago

Is the content exactly the same for the TF version as the original. I'm guessing if something cannot yet be done in TF it could have been skipped. There seems to be nothing present in the official description.

Thanks

CamDavidsonPilon commented 5 years ago

AFAIK there are no differences in content, except for the syntax and semantics of the implementation language.

Allgoerithm commented 5 years ago

Actually, there seem to be differences. I noticed while reading chapter 2 ("More PyMC/TFP"), so my examples will be drawn from there. I noticed this by accident, so there may be more differences than I stumbled upon.

Not surprisingly, some "Pro Tips" for PyMC-specific syntax are missing in the TFP version as they don't have a direct analogue in TFP. But there is at least one difference that goes a bit deeper: In the part that is entitled "An algorithm for human deceit", there are two alternative modelling approaches being presented. In the PyMC2 and the PyMC3 version of the book, both are solved using a Metropolis-Hastings-sampler, while in the TFP-port, Hamilton Monte Carlo is used for the second version of the model ("Alternative PyMC2/PyMC3/TFP Model").