search

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

Register Complexity.jl once v2 is released #188

Closed kahaaga closed 1 year ago

kahaaga commented 1 year ago

I've finished migration of the complexity measure to Complexity.jl. This is a reminder to register the package once we release v2.0 here.

Datseris commented 1 year ago

I want to talk about this to you at some point soon, because I have rethought things and I am not sure yet whether Complexity.jl is needed...

Datseris commented 1 year ago

What was the reason again to not have a single package which we could call eg. ComplexityMeasures.jl ? That extracts an entropic/complexity quantity from a single input?

kahaaga commented 1 year ago

The idea was that the scope of the package was increasing a lot (i.e. not actually computing just probabilities and entropies), and therefore we should split into two separate packages. But I actually don't see any particular reason why we can't just have everything in one package. But, then a name like ComplexityMeasures.jl or something would be much more fitting.

It feels kind of artificial to have to separately document i.e. multiscale in both Entropies.jl AND Complexity.jl. It leads to a lot of redundancy, and the user has to navigate back and forth between package documentations for conceptually disparate, but related concepts.

Datseris commented 1 year ago

Not only that, but the line of what is a "complexity measure" and what is an "entropy" i think is rather blurred. Not only it depends on ones background, and even the exact university they studied (i.e., who was their teacher), but it also depends on the fact that we have so many different entropy measures that in the literature many of them could just be used as a complexity measure and noone would challenge the nomenclature.

Datseris commented 1 year ago

furthermore I'd say the more applied the researcher, the more the line is blurred. Someone that works on classifying measured ECG signals using complexity measures, without a rigorous mathematical background, will most likely not care about or know the difference. And honestly good for them, because for their purposes it doesn't really matter if a metric is rigorously an entropy or not.

Datseris commented 1 year ago

so all in all I vote for one ComplexityMeasures.jl package. In the meantime I'll check out whether this rename can be done in this repo so that we don't lose commit history.

kahaaga commented 1 year ago

so all in all I vote for one ComplexityMeasures.jl package.

I'm on board with that.

In the meantime I'll check out whether this rename can be done in this repo so that we don't lose commit history.

I think this should be possible to do, simply by renaming the repo and getting a new URL, and then re-registering the package under a different name. But some care must be taken, so that package installation doesn't break for users of the old package name. I think it requires a manual PR to the general registry. This is relevant.

Datseris commented 1 year ago

Okay rename sounds rartther straightforward. But lets do it once #190 is finished first?

kahaaga commented 1 year ago

Okay rename sounds rartther straightforward. But lets do it once https://github.com/JuliaDynamics/Entropies.jl/issues/190 is finished first?

yes, this is something we could time with the release of the new package.