UltraNest support

[JohannesBuchner]

Would you be interested to add support for https://johannesbuchner.github.io/UltraNest/ ?
The interface should be very similar to dynesty.
UltraNest is a very reliable tuning-parameter-free algorithm. It can be faster than dynesty, especially if you take advantage of vectorization.
It also supports resuming from disk and MPI parallelisation, if that is useful to you.

[3fon3fonov]

Very Cool! I was not aware of UltraNest, thanks a lot! including another sampler to the Exostriker should be easy in the core, but a bit time-consuming on the GUI part. Yet, I will definitely have a look and experiment! Comments are likely to follow, so stay tuned.

[3fon3fonov]

After a long time on this request, I am ready to experiment! @JohannesBuchner I could not find out of the box whether is possible with UltraNest to control the number of Threads/CPUs to be used in a multiprocessing. For instance, with dynesty I use something like:

from pathos.pools import ProcessPool as Pool
thread = Pool(ncpus=N_threads)

sampler = dynesty.NestedSampler(partial_func, prior_transform, ndim, nlive=nwalkers, pool = thread,
                                    queue_size=threads, sample = dynesty_samp, bound = ns_bound)
sampler.run_nested(print_progress=print_progress,dlogz=stop_crit, 
maxiter = ns_maxiter, maxcall = ns_maxcall )

thread.close()
thread.join()
thread.clear()

Can you please comment/advice how can I pass a pool to UltraNest, similar to what Dynesty and emcee support?

Sorry if this was obvious and I didn't see in in your documentation!

[JohannesBuchner]

UltraNest can be transparently parallelized with MPI. Instead, the program is run with mpiexec (or mpirun). This runs N separate processes, which communicate with each other. No parallelisation code in exostriker is required. MPI is more powerful for distributing than python pools, which I recall often have the issue of serializing and sending the code to be executed to other cores.