Open jtlz2 opened 9 years ago
Patches welcome ;) I found DNest not as beneficial to the problems I applied it to, but if it is useful to you it should be easy to integrate it into jbopt.
Somehow a sane, general choice for perturb() has to be found. Possibly a Gaussian proposal on top of the prior transformation, but it probably needs to be tuned on-line.
@JohannesBuchner - any short-term plans to interface to DNest/PyDNest?