Simplifies some of the analytic 1D likelihoods

@colm.talbot @matthew-pitkin @gregory.ashton I created a superclass for the analytic 1D likelihoods that @matthew-pitkin recently introduced. It reduces some code redundancies. I called the superclass _Analytic1DLikelihood. It is private because the class because users should never instantiate it.

added 3 commits

• 81dcc883 - Renamed self.function -> self.func
• 6f28d5bd - Refactored N -> n (PEP8)
• 62cc3229 - Made the tests run properly

Compare with previous version

added 25 commits

• d6feb0a3 - Made Analytical1DLikelihood public again
• 071fcbdd - Removed an alias and refactored a key
• 21ecf70c - Reformatted file
• d2990851 - Deleted redundant tests
• 5cbdd850 - Split a test into two
• e359092c - Added tests for Analytical1DLikelihood
• 2be27fac - Extracted two private methods
• 964e86b8 - Refactored code some more in GaussianLikelihood
• 7e7795f7 - Refactored code some more
• dde1ee41 - Shifted y setting logic into a property/setter structure
• 2ac8f384 - Added check that x, y in Analytical1DLikelihoods are always numpy arrays
• 53871d3f - Created two properties for the rate and the sumlogfactorial respectively.…
• e59da276 - Renamed a property
• d6127a83 - Fixed a bug I just introduced
• 42a042c2 - Introduced a mu property and simplified ExponentialLikelihood
• 8d106493 - Some more reformatting
• 280fd8fb - Moved the calculation of the residual into the parent class
• 3bfac960 - Refactored some more
• 3e43896d - Removed the residual aliases and replaced them accordingly
• 6fb97e69 - Added the possibility of x or y being a single float
• 130ddc5c - Added some docstrings
• a4c837a9 - Added some more docstrings and refactored some more
• 17d04bf1 - Fixed errors I introduced myself
• 251472b9 - Removed some of the slower and unnecessary properties
• faf745d3 - Renamed function -> func

Compare with previous version

added 10 commits

• dcd3be7b - Added tests for Likelihood base class
• 21f2372c - Added tests for setting x and y
• a40c6ea5 - Made logic a bit safer for obscure data types
• 75705732 - Simplified test syntax
• ed41d915 - Tests cases of `PoissonLikelihood`
• 7ca8c69c - Corrected an incorrect cast
• 61cc31c2 - Tests cases of `ExponentialLikelihood`
• f833301b - Added test for default case of `ExponentialLikelihood.log_likelihood`
• 06261851 - Changed a comment
• 38cc1332 - Added some more tests for 100 % coverage.

Compare with previous version

added 1 commit

• cbe52f8f - Made a check python 2 friendly

Compare with previous version

@matthew-pitkin I added some tests so that we achieve 100 % coverage for the core.likelihood module. I changed quite a bit but I think it should be alright, could you have a look and see that especially the log_likelihood evaluations are still fine?

unmarked as a Work In Progress

mentioned in merge request !153 (merged)

approved this merge request

mentioned in commit d52b5aff

merged