search

BAMresearch / bayem

Implementation and derivation of "Variational Bayesian inference for a nonlinear forward model." [Chappell et al. 2008] for arbitrary, user-defined model errors.
MIT License
2 stars 1 forks source link

Scipy prior #40

Closed TTitscher closed 2 years ago

TTitscher commented 3 years ago

Tackles #36

TTitscher commented 3 years ago

I get that the definition of the priors is part of the inference problem. However, I doubt a bit that the use of scipy.stats is really useful. In the end, unless the solver can directly use it, it has to be converted to another distribution (pyro.distribution, pymc3.Normal(...), vb.MVN, pytorch???) so the user would basically define it twice, like here. And I would totally get, if a user finds that impractical.

joergfunger commented 3 years ago

But that definition you are linking to would be in the library? So the user wouldn't see this either.