Experiments-Groups

We have the following datasets:

• dataset-random-graphs: The k-reduced-skew-spectra of Erdős–Rényi graphs.
• dataset-simple-graphs: some simple graphs downloaded from the link below
• dataset-all-graphs: all directed and undirected graph for small (very small :) ) n

We have the following files to create and process datasets:

• generate-random-dataset.py: generate the datasets in dataset-random-graphs directory and the dataset for `EXP-Atlas-undirected
• generate_all_graphs: folder with HPC code to generate in parallel(!) all directed graphs, eventually skipping some.
• fromsparse.py: convert g6 graphs into netnworkx->numpy and gerate k-reduced-skew-spectra

We have the following experiments

• EXP-on-graphs.ipynb: the experiments with all undirected graphs of n=7 nodes
• EXP-discriminating-important-graphs: our 3-reduced-skew-spectra can distringuish two important graphs :)
• EXP-new-QM7-regression.ipynb: Some examples on how the k-reduced-skew spectra improves the regression task on QM7-datasets

We have the following experimeents with Neural Networks:

• HGP-SL-extended: from https://github.com/cszhangzhen/HGP-SL

Useful links

• http://users.cecs.anu.edu.au/%7Ebdm/data/graphs.html

pip install torch_geometric==1.7.2