Switch default dynestty sampling arguments to use rslice
The default dynesty sampler arguments used bilby's default number of walkers and ACT calculation. When testing these settings (using bilby 1.1.5 and dynesty 1.0.1) for an analysis using the whole O1 data set it can take of order 2 hours to perform a standard 4-parameter PE job. I think this is due to the walks taking so long to converge towards the bulk of the likelihood when there is a broad h0 prior. However, using the dynesty rslice sampler this takes on order of minutes. I therefore plan to switch the default sampling method to be rslice
. Users can change this if they want.
Edited by Matthew Pitkin