NEWS.md

In rgudhi: An Interface to the GUDHI Library for Topological Data Analysis

rgudhi 0.2.0

In this minor release, I:

• Added the persistence_diagram_sample class;
• Implemented the Lagragian formulation to compute the Wasserstein barycenters of a sample of persistence diagrams;

• Fixed CRAN warnings:

  • Incoherence of autoplot method implementation following renaming of main argument x into object.
  • Rectify an invalid URL pointing to the paper for BIRCH clustering.

rgudhi 0.1.0

rgudhi v0.1.0 provides an almost full wrapper of the v3.7.1 of the GUDHI library for topological data analysis. Only the cover complex class is missing due to non-reproducibility issues with random number generators. With GUDHI accessible from R, rgudhi v0.1.0 features:

• data structure to encode simplicial complexes;
• computation of persistence diagrams;
• various usual preprocessing tools for persistence diagrams;
• a dedicated S3 class persistence_diagram for persistence diagram;
• plot() and ggplot2::autoplot() methods for persistence_diagram objects;
• vector and kernel representations of persistence diagrams;
• a number of metrics to quantify distances between persistence diagrams (Bottleneck, Persistence Fisher, Wasserstein, Slice-Wasserstein).
• functions to sample points from sphere (sphere()) and torus (torus());
• a persistence-based clustering algorithm coined Tomato.

The package also wraps all clustering algorithms from the sklearn.cluster module because they can be useful when using the Atol vectorization method for persistence diagram.

It also wraps all scalers classes from sklearn.preprocessing for use in various classes as well.

rgudhi documentation built on March 31, 2023, 11:38 p.m.