JuliaDynamics / ComplexityMeasures.jl

Estimators for probabilities, entropies, and other complexity measures derived from data in the context of nonlinear dynamics and complex systems

MIT License

54 stars 12 forks source link

Kaniadakis entropy #182

Closed HeineRugland closed 1 year ago

HeineRugland commented 1 year ago

Fixes #92

Implemented Kaniadakis entropy.
Added simple tests (no analytical tests were given in the book).
Added documentation example.
Fixed conflicting import of Entropies and Distribution packages that caused errors in documentation (no plots were generated).
The book did not state for which distribution Kaniadakis entropy is maximised, so entropy_maximum returns an error.

codecov[bot] commented 1 year ago

Codecov Report

Merging #182 (0f61496) into main (3c705f5) will decrease coverage by 0.01%. The diff coverage is 83.33%.

@@            Coverage Diff             @@
##             main     #182      +/-   ##
==========================================
- Coverage   84.94%   84.92%   -0.02%     
==========================================
  Files          37       38       +1     
  Lines         963      975      +12     
==========================================
+ Hits          818      828      +10     
- Misses        145      147       +2

Impacted Files	Coverage Δ
src/entropies/kaniadakis.jl	`83.33% <83.33%> (ø)`

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

kahaaga commented 1 year ago

Thanks, @HeineRugland! This looks good.

Could you please fix the following?

Line 34 in the Kaniadakis.jl file is not covered by the tests. You can fix this by including a test that computes the Kaniadakis entropy in a probability distribution containing a zero, i.e. entropy(Kaniadakis(), Probabilities([0, 1]).

We can implement entropy_maximum for Kaniadakis entropy later. It just means we can't do entropy_normalized with Kaniadakis yet. I'll open an issue for it.

kahaaga commented 1 year ago

@HeineRugland

The docstring for Kaniadakis isn't actually included in the documentation (i.e. it is not given to a @docs block anywhere). This needs to be fixed before merging.

There's a list of entropies in entropy.md, each under a level-two markdown heading (##). Kaniadakis should also be there, in its own @docs block.

kahaaga commented 1 year ago

@HeineRugland I'll just merge this and add the documentation entry as I rework the overall documentation. Thanks for the PR!