dynesty

dynesty copied to clipboard

Do not compute n_effective when limit is infinite

15

Currently the maximum number of effective samples `n_effective` in `Sampler.sample()` defaults to `np.inf`, implying no limit, yet `self.n_effective` is still computed (and then compared with `np.inf`), which takes time. The...

edbennett

dynesty 1.2.3 issue changes values of weights compared to 0.9.5.3

3

**Describe the bug** I recently upgraded my laptop and moved from dynesty 0.9.5.3 to 1.2.3. I'm finding different results using dynesty on the same data after the change---in particular, some...

mmfausnaugh

Making it easier to restart interrupted runs

4

This is issue is track/discuss the development of functionality that would allow an easier restarting of the dynesty sampler in ther case of the crash (timeout in HPC environment). (see...

segasai

Samples near the edge of the box

1

Over last few days I was trying to debug the issue in sampling slowing down to a crawl, and it was happening with all the samples sitting exactly at the...

segasai

Saving derived quantities during sampling

2

This is the issue to discuss the way of saving the quantities derived during the likelihood function evaluations. Sometimes the likelihood function does some heavy calculations and compute some derived...

segasai

Allow variable mcmc length

9

This PR adds a new option that allows the length of the MCMC exploration for the `rwalk` method to vary during the analysis such that a target number of proposed...

ColmTalbot

Add additional rwalk proposals

3

This PR adds a bunch of different methods for proposing points with the `rwalk` sampling. These are a combination of different scale distributions for the existing method and ensemble moves....

ColmTalbot

Bounding issues/diagnostics

15

Hi, I am looking at the results of the sampling of my complicated problem (with results different from multinest gives me) and I was looking at diagnosing what's going on,...

segasai

Thoughts on reusing/improving the bounding information

3

Hi Josh, This is definitely not a bug, but just some thoughts on a sampling issues I had recently. The problem that I've been struggling over last few months features...

segasai

Add an autocorrelation time estimate to rwalk and rstagger

6

- Adds two new tuning parameters to the rwalk and rstagger methods: maxmcmc and nact - maxmcmc controls the maximum number of steps - nact controls the number of autocorrelation...

GregoryAshton