pulsatrix 1.1.0
Loading...
Searching...
No Matches
Topological Data Analysis

Persistent homology of a layer's representations, as a typed pipeline: a point cloud built from activations, its distances, the filtration, persistence diagrams and their vectorizations. More...

Files

file  diagram_distance.hpp
 Distances between persistence diagrams, bottleneck and Wasserstein, and Certified<T>, a value with the error bounds behind it.
 
file  distance_matrix.hpp
 Pairwise Euclidean distances of a point cloud, computed with one GEMM on any backend, with a certified bound on their floating-point error.
 
file  h0.hpp
 H0 persistence, the clusters of a point cloud across scales, from its minimum spanning tree.
 
file  landmarks.hpp
 Farthest-point landmarks: a subset of a large point cloud that covers it, with a certified bound on what persistent homology loses by using the subset.
 
file  persistence_diagram.hpp
 A persistence diagram: the features of one homology dimension, each born at one scale and dying at a larger one.
 
file  point_cloud.hpp
 The input to topological data analysis: a layer's activations as a set of points, one row per input or per position, gathered across batches.
 
file  rips_persistence.hpp
 Persistent homology of the Vietoris-Rips filtration in every dimension up to a maximum: components (H0), loops (H1), voids (H2) and beyond, with each feature's critical edges and a representative cocycle.
 
file  rtd.hpp
 Representation Topology Divergence (RTD): how differently two representations of the same inputs are shaped, cluster by cluster and loop by loop, and CompareRepresentations(), one report with CKA, mutual-kNN alignment and RTD side by side.
 
file  significance.hpp
 Is a loop real? Significance tests for persistence diagrams: the universal null distribution, bootstrap confidence bands, and baseline comparisons against Gaussian clouds or a random network.
 
file  vectorization.hpp
 Turning a persistence diagram into numbers for probes, statistics and plots: Betti curves, persistence landscapes, persistence images and persistent entropy.
 

Detailed Description

Persistent homology of a layer's representations, as a typed pipeline: a point cloud built from activations, its distances, the filtration, persistence diagrams and their vectorizations.