An incomplete listing of potentially useful Python
packages that perform Markov Chain Monte Carlo or otherwise do Monte Carlo sampling and/or evidence calculation:
Code | Algorithm | Bonus link |
---|---|---|
Cobaya |
accelerated Metropolis, nested sampling | |
dynesty |
nested sampling | |
emcee |
Goodman-Weare | paper |
kombine |
briefly tempered Metropolis-Hastings | paper |
lmc |
Metropolis-Hastings, Goodman-Weare | |
MultiNest |
nested sampling | papers 1 2 |
pymc |
Metropolis-Hastings | |
pymc3 |
Metropolis-Hastings, HMC | |
PyStan |
HMC | |
UltraNest |
nested sampling |
For R
users, there are numerous options on CRAN, but we list only a couple that we know or have heard of being used:
Code | Algorithm | Bonus link |
---|---|---|
rgw |
Goodman-Weare | CRAN |
LaplacesDemon |
many | CRAN |
RStan |
HMC |
Gibbs samplers that analyze models to determine conjugacy relations (some have Python
/R
interfaces):
Gibbs samplers specifically for linear regression:
Code | Language | multiple $x$'s | multiple $y$'s |
---|---|---|---|
linmix_err |
IDL | no | no |
mlinmix_err |
IDL | yes | no |
linmix |
Python | no | no |
lrgs |
R, Python | yes | yes |