Spectral Hypergraph Coarsening Code for
HyperSF: Spectral Hypergraph Coarsening via Flow-based Local Clustering https://ieeexplore.ieee.org/document/9643555/metrics#metrics
Ali Aghdaei, Zhiqiang Zhao, Zhuo Feng
This is the improved version of HyperSF incorporating effective resistance clustering method.
Julia Version: 1.5.3
Packages:
SparseArrays
LinearAlgebra
MatrixNetworks v1.0.1
RandomV06 v0.0.2
Run "Run_experiment.jl" to generate the coarsened hypergraph for the availble test cases.
The input format is in the format of hMetis that is every line corresponds to each hyperedge. The first line: #hyperedges, #nodes.
L: is an integer (L>0) to adjust the initial coarsening ratio before applying the flow-based technique
R: the ratio of selected clusters (low-quality clusters) for applying the flow-based technique (0<R<=1).
IDN: returns the assigned cluster to each node.
The code generate the coarsened hypergraph as "Out.hgr" as the same format of input hypergraph.