log probability interface for post-inference analysis

[YongchaoHuang]

extended methods for logprior, loglikelihood, logposterior for chains.

Close #450

[torfjelde]

Just so I'm following here, what's the difference between this PR and #437 ?

[YongchaoHuang]

Just so I'm following here, what's the difference between this PR and #437 ?

@torfjelde Thanks for following up. It's the same thing but this one is purely codes and aims to be the new API for logprior and logposterior. #437 is documentation. I really should only make one PR.
@devmotion made some very interesting comments in #437, which is linked to #166 and #170 I believe. @torfjelde Any suggestions on improving the functions here? Thanks ahead.

[yebai]

thanks @YongchaoHuang -- I left some minor comments below. In addition, can you move all helper functions for tests to the beginning of the model.jl file.

[YongchaoHuang]

thanks @YongchaoHuang -- I left some minor comments below. In addition, can you move all helper functions for tests to the beginning of the model.jl file.

Yes, all helper functions have been moved to test/test_utils.jl. Some of the helper functions can be removed in the future once #481 is merged.

[YongchaoHuang]

This PR depends on #481

[yebai]

@YongchaoHuang I have merged #481 into master. Do you want to remove your copy-and-pasted code so this PR can be merged?

[yebai]

If no one objects, I'll merge this PR in the next day or two.

[YongchaoHuang]

@YongchaoHuang I have merged #481 into master. Do you want to remove your copy-and-pasted code so this PR can be merged?

Many thanks. I am looking into it now.

[YongchaoHuang]

@torfjelde @yebai Maybe it's ready to merge?
Current failure with JuliaNightly is due to the overwriting of the rand method in Line 745 of src/test_utils.jl.

[yebai]

Let's get this PR merged. The code is self-contained and only adds a new feature. Further improvements can be introduced via separate PRs. Thanks @YongchaoHuang for the work, and @torfjelde @devmotion for the reviews.