Awesome
A Unified Framework for Optimization-Based Graph Coarsening
The implementation of FGC, GC, Two-stage and FGCR algorithms along with all comparison metrics that are mentioned in the paper and application on clustering can be found in FGC_experiment.ipynb. It also carries code for rest of the experiments like effect of hyperparameters and to evaluate plots of eigen values, loss curve and epsilon similarity. State of the art algorithm's implementation are put in Local variation.ipynb along with experiments to evaluate REE, DE, HE, RE using LVE, LVN, Kron reduction and HEM. It also contains supporting library files used in Local variation.ipynb. The code for local variation, Kron reduction and HEM algorithms is taken from "Graph reduction with spectral and cut guarantees" paper. Also, the code for Node classfication task using FGC with sparsity written in cuda using pytorch can be found in Node_classification_FGC_with_sparsity.ipynb.