Astrophysics (Index)About

PyMC

(Python package for Bayesian statistical modeling)

PyMC is a Python library for modeling using Bayesian methods. It implements Markov chain Monte Carlo (MCMC), offering a number of algorithms, including the Metropolis algorithm, and newer algorithms such as No-U-Turn Sampler (NUTS). The name PyMC3 was used for PyMC's version 3, which was a major (total?) rewrite. Subsequent PyMC versions, 4, 5, etc., go by the name PyMC, and (at least initially) maintain PyMC3's modelling API. PyMC3 was built on a general Python library for such mathematical modeling, Theano (which, in turn was built on a Python library for numerical programming, NumPy). Theano's official development ceased and PyMC developers, for PyMCs use, maintained their own separate revised version called Aesara, and eventually a separate revised version of Aesara called PyTensor.


(software,statistics,Python)
Further reading:
https://en.wikipedia.org/wiki/PyMC
https://en.wikipedia.org/wiki/Theano_(software)
https://en.wikipedia.org/wiki/NumPy
https://en.wikipedia.org/wiki/Hamiltonian_Monte_Carlo#No_U-Turn_Sampler
https://pypi.org/project/pymc3/
https://www.pymc.io/
https://pypi.org/project/pytensor/
https://pypi.org/project/aesara/
https://pypi.org/project/Theano/

Index