MCMC takes a long time

[darthoctopus]

In my test runs it takes ~ 6 hr for a full MCMC run to complete for the dipole mode parameters. Aside from shrinking the prior volume, what would accelerate this?

[nielsenmb]

I'll have to take a look at the code, but Jax might be a solution.

Otherwise, there's a "new" nested sampler called JaxNS, which seems to function in a similar fashion to Dynesty, but where all the backend stuff is Jaxed. In principle it should be faster, but I haven't really tried it that much.

[grd349]

I would leave JaxNS alone for now - I don't think it is mature enough.