[WIP] Update examples to work with latest CES code #92
Conversation
Codecov Report
@@ Coverage Diff @@
## master #92 +/- ##
=======================================
Coverage 85.19% 85.19%
=======================================
Files 7 7
Lines 547 547
=======================================
Hits 466 466
Misses 81 81
Continue to review full report at Codecov.
|
There was a problem hiding this comment.
Looks good! I think the heart of my comments are that we have built an interface with ParameterDistribution s, and so we should use always use it. That is, rather than directly accessing the (effectively private) members e.g posterior.distribution[i] we should be instead "getting" them. i.e call get_distribution(posterior) and then using the returned objects. This is typically easier to maintain down the line if everything interacts with the distributions in the same way. If the right getter functionality isn't there we should add the getter functions over bypassing them.
Use getters (Ollie's comments), additional explanations, beautification, etc.
|
Superseded by #94 (messed up the squash and rebase :-)) |
The goal of this PR is to get all examples synced up with the latest changes in the code base (in particular, PRs #88 and #89)
Cloudy_example.jllearn_noise.jlplot_GP.jlIn addition,
get_distribution()(inParameterDistribution.jl) has been modified to return the array of samples when called forSamples( rather than the message "Contains samples only").get_distributionnow returns aDictwith the parameter names as keys and the corresponding distribution (in the case ofParameterizeddistributions, such as Normal(0.0, 1.0)) or the corresponding samples (in the case of parameters represented bySamples) as a parameter_dimension x n_samples array.