Add DIMEMove
16 views
Skip to first unread message
Gregor Boehl
Oct 10, 2022, 5:22:49 AM10/10/22
to emcee users
Hi there,
I have been working on a move that is intended to provide fast burn-in for high-dimensional distributions and to perform well for multimodal distributions: the differential-independence mixture ensemble (DIME) MCMC. See here for an example: https://github.com/gboehl/emcwrap#tutorial
It seems to me that this goes well with the emcee vision, so I would be more than happy to contribute to this project.
Formally, the method is proposed here: https://gregorboehl.com/live/dime_mcmc_boehl.pdf
FYI, econ is very slow in terms of publishing. It is normal to wait a couple of years before something is out there. I know this sounds odd to people in other fields...
I also create a PR: https://github.com/dfm/emcee/pull/442
I am happy to generating some documentation, e.g. by adding the above tutorial to the emcee documentation. If this gets merged, I would remove the tutorial at emcwrap.
Looking forward to your thoughts!
Gregor
I have been working on a move that is intended to provide fast burn-in for high-dimensional distributions and to perform well for multimodal distributions: the differential-independence mixture ensemble (DIME) MCMC. See here for an example: https://github.com/gboehl/emcwrap#tutorial
It seems to me that this goes well with the emcee vision, so I would be more than happy to contribute to this project.
Formally, the method is proposed here: https://gregorboehl.com/live/dime_mcmc_boehl.pdf
FYI, econ is very slow in terms of publishing. It is normal to wait a couple of years before something is out there. I know this sounds odd to people in other fields...
I also create a PR: https://github.com/dfm/emcee/pull/442
I am happy to generating some documentation, e.g. by adding the above tutorial to the emcee documentation. If this gets merged, I would remove the tutorial at emcwrap.
Looking forward to your thoughts!
Gregor
Reply all
Reply to author
Forward
0 new messages