Skip to content

gauthierthurin/SphericalQuantiles

Repository files navigation

Spherical Quantiles

Codes used for the figures of the paper Monge-Kantorovich quantiles for spherical data, currently available on

https://arxiv.org/pdf/2407.02085

Codes available in python

Python 3.10.8

Packages required:

  • pyshtools==4.10
  • pot==0.9.0
  • geomstats==2.7.0
  • numpy=1.26

A full list is available in requirements.txt. You can also install dependencies with: pip install -r requirements.txt (but this may include unnecessary packages).

Notebooks for each experiment

The file functions.py contains the required functions.

Each notebook contains codes for some figures in the paper, as follows.

Tuto: to compute regularized quantile function, hence contours and sign curves. Also introduces the implementation of our stochastic algorithm, with checks of convergence.

RegVsUnreg: Comparison of regularized versus unregularized quantiles.

VolumesRegion: contains codes for volumes (and scale curves) of regularized quantile regions.

OtherQuantiles: Visual comparison of Mahalanobis/Spatial/MK quantiles.

Classif: codes related to depth-based classification.

About

Codes to compute regularized quantiles on the sphere via entropic optimal transport

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors