Integrate Slice Sampling: Covariance Matching MCMC

[lorcandelaney]

See #772

Includes:

1. Method for Covariance-Adaptive Slice Sampling: Covariance Matching.
2. Tests for Covariance Matching method.
3. TO DO: Notebooks.

[codecov]

Codecov Report

Merging #896 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff           @@
##           master   #896    +/-   ##
======================================
  Coverage     100%   100%            
======================================
  Files          66     67     +1     
  Lines        7052   7180   +128     
======================================
+ Hits         7052   7180   +128

Impacted Files	Coverage Δ
pints/_mcmc/_slice_covariance_matching.py	100% <100%> (ø)
pints/toy/_repressilator_model.py	100% <0%> (ø)	⬆️
pints/residuals_diagnostics.py	100% <0%> (ø)	⬆️
pints/_mcmc/_population.py	100% <0%> (ø)	⬆️
pints/toy/_sir_model.py	100% <0%> (ø)	⬆️
pints/_core.py	100% <0%> (ø)	⬆️
pints/toy/_sho_model.py	100% <0%> (ø)	⬆️
pints/_optimisers/_xnes.py	100% <0%> (ø)	⬆️
pints/_mcmc/_mala.py	100% <0%> (ø)	⬆️
pints/_mcmc/_rao_blackwell_ac.py	100% <0%> (ø)	⬆️
... and 57 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update fe3e6d7...dbc221e. Read the comment docs.

[ben18785]

Very readable code -- nice work! I've suggested a few (very minor) changes. I can't see where this code deviates from what is written. The only thing I'm unsure about is whether ask / tell returns the gradient or negative gradient and if F and R need to be transposed (as this is the last part of your chud function).

Can you add an example notebook here that shows the issue? I couldn't find one.

[MichaelClerx]

@lorcandelaney if the other PR is "correct" does that mean this one can go??

Sorry it's a different method isn't it?

[MichaelClerx]

Is this ready for re-review @ben18785 ?