try to make the minimum autocorrellation estimation lower

Currently, the act estimate seems to be almost always the same as walks.

This is due to an artefact of the translation from CPNest.

This MR attempts to fix this.

added Sampling label

added 1 commit

• f6737045 - remove print statement

Compare with previous version

added 3 commits

• f6737045...bd7abac5 - 2 commits from branch master
• 8e013a2d - Merge branch 'master' into reduce-minmcmc

Compare with previous version

@colm.talbot have you done a quick run to test this out? @shanika.galaudage is asking for a release: is it worth getting this in to make things smoother?

It would probably be easier/faster to reduce walks in the default online settings, bearing in mind that the actual minimum number of steps will be ~nact*walks. I think a few tens is probably as high as we should go while we use the current method.

(I can't directly help with anything until tomorrow.)

approved this merge request

changed milestone to %0.6.4

approved this merge request

merged

mentioned in commit 0ca9c1d8