Variational inference with PyMC

[arrjon]

I added a wrapper for variational inference with PyMC. The wrapper uses the PymcSampler as basis and directly supports all functions and methods from PyMC see here for details on VI.

To include this properly in pyPESTO one would need to add proper testing and write a Result Object to save the estimated parameters of the variational distribution.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 1.08696% with 91 lines in your changes are missing coverage. Please review.

Project coverage is 83.90%. Comparing base (7fe40ba) to head (bdd8873).

Files	Patch %	Lines
pypesto/variational/pymc.py	0.00%	53 Missing ⚠️
pypesto/variational/variational_inference.py	0.00%	35 Missing ⚠️
pypesto/variational/__init__.py	0.00%	2 Missing ⚠️
pypesto/store/save_to_hdf5.py	50.00%	1 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1306      +/-   ##
===========================================
- Coverage    84.48%   83.90%   -0.59%     
===========================================
  Files          157      160       +3     
  Lines        12959    13051      +92     
===========================================
+ Hits         10949    10950       +1     
- Misses        2010     2101      +91     

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

[PaulJonasJost]

Thanks a lot for putting this first draft together :)

Should discuss about further things needed aside from testing, before it can be merged.

[arrjon]

Mainly tests are missing, but otherwise ready

[Doresic]

All in all, very nice implementation! Thanks.
I left some questions/recommendations that would come up to my mind if I would try to use the module myself.