Closed LRydin closed 3 years ago
Ha! Everyday one learns something: so what I think I'll do is optional features instead as explained here. This should work perfectly, allowing the library to remain light (solely depend on numpy
), and for EMD and other stuff (like the future plotters) check if EMD is install and throw a exception requesting an installation.
See #17. Should be sorted out. Closing.
I have been deliberating over the presence of the Empirical Mode Decomposition EMD in v0.3, i.e., the PyEMD by Dawid Laszuk. The EMD package is phenomenal, but comprises an overhead of ~700 kb (and usually some additional packages). My plan now is to create a separate repository (likely to be named MFDFA-EMD) which adds the EMD functionality included in v0.3.
The cause is simple: the original MFDFA is meant to be lightweight, i.e., depend only of
numpy
, so that one day maybe someone can make a parallel-isable version of it.[I'm adding this to the readme of the repository]