I am organizing the tutorial "Software for tensor-based analysis of general hypergraphs" with Sinan Aksoy for SIAM MDS24.
Abstract:
Hypergraphs enable faithful modelling of group interaction data, which are ubiquitous across multiple domains. Encoding general hypergraphs using multilinear algebra (e.g. as adjacency tensors) presents both theoretical and computational barriers. This tutorial focuses on recently proposed tensor times same vector (TTSV) algorithms developed specifically for hypergraph tensors. TTSV is a fundamental subroutine in many tensor algorithms, including those used to compute hypergraph centrality and clustering. We first provide the necessary background on TTSV for hypergraphs. Then we present a tutorial on the accompanying software packages, in both Python and Julia, which are compatible with the XGI, HNX, HGX, and SimpleHypergraphs.jl hypergraph libraries. Lastly, we offer participants several examples to work though, and encourage them to apply, as well as to extend, these methods to their domains of interest.
Files:
Please download files for the tutorial here.