Dead points and Truncation

[yallup]

Description

Feature request for an additional convenience alias for dead_points matching the live_points style, and a truncate method that gives the consistent union of the two allowing recalculation of mid run statistics

Checklist:

• I have performed a self-review of my own code
• My code is PEP8 compliant (flake8 anesthetic tests)
• My code contains compliant docstrings (pydocstyle --convention=numpy anesthetic)
• New and existing unit tests pass locally with my changes (python -m pytest)
• I have added tests that prove my fix is effective or that my feature works
• I have appropriately incremented the semantic version number in both README.rst and anesthetic/_version.py

[williamjameshandley]

This is a useful addition (I and @zixiao-h have certainly used this functionality)

Copying inline comments here:

@yallup: Currently tests are failing due to the second check here, is this defined correctly (should the logL_birth check be inclusive)?

@williamjameshandley: I think that the convention is consistent i.e. this version of the strictness ensures that the 'live points at iteration i' corresponds to 'live points at loglikelihood logL' in a sensible way.

I think the real issue is that samples.live_points() (i.e. the None functionality) is not returning the right thing. This should return 'the last set of live points', and at the moment returns the penultimate set.

I've pushed a new function 'contour' that aims to achieve this. Could you tidy up the docstrings with a new users eye?

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (1a035b7) 100.00% compared to head (20155b3) 100.00%.

Additional details and impacted files

@@            Coverage Diff            @@
##            master      #354   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           34        34           
  Lines         2973      2984   +11     
=========================================
+ Hits          2973      2984   +11     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[yallup]

This is a useful addition (I and @zixiao-h have certainly used this functionality)

Copying inline comments here:

@yallup: Currently tests are failing due to the second check here, is this defined correctly (should the logL_birth check be inclusive)?

@williamjameshandley: I think that the convention is consistent i.e. this version of the strictness ensures that the 'live points at iteration i' corresponds to 'live points at loglikelihood logL' in a sensible way.
I think the real issue is that samples.live_points() (i.e. the None functionality) is not returning the right thing. This should return 'the last set of live points', and at the moment returns the penultimate set.
I've pushed a new function 'contour' that aims to achieve this. Could you tidy up the docstrings with a new users eye?

Ah that makes sense and is consistent with what I was seeing, I added a small change to cater for if a float (implicitly a logL val) is passed but it happens to form a valid index (i.e 10.0). Docstrings seem sensible to me now, added a bit there an an explict contour test

[williamjameshandley]

This is not covered by the tests -- I agree that your additional float check is safer than what was there before. Can we think of anything which the try except covers (given that it is now never hit by the tests?) if not, remove the extra code, and just move to straight return statements from the if elif else block.

[williamjameshandley]

Please squash and merge!