satyakisikdar / cnrg

Code release for the paper "Modeling Graphs with Vertex Replacement Grammars" by Sikdar et al.

Code repository for the paper Modelling graphs with Vertex Replacement Grammars by Satyaki Sikdar, Justus Hibshman, and Tim Weninger. Find the preprint here.

Run python3 -m pip install -r requirements.txt prior to running runner.py to install required packages.

Usage instructions (also visible by running python3 runner.py -h)

usage: runner.py [-h] [-g] [-c] [-b {full,part,no}] [-l LAMB] [-s] [-o OUTDIR]
                 [-n N]

optional arguments:
  -h, --help            Show this help message and exit
  -g, --graph           Name of the graph, looks in the ./src/tmp directory for the edge list (default: karate)
  -c, --clustering      Clustering method to use (default: louvain)
  -b {full,part,no}, --boundary {full,part,no}
                        Degree of boundary information to store (default: part)
  -l , --lamb LAMB      Size of RHS (lambda) (default: 5)
  -s , --selection      Selection strategy (default: level)
  -o , --outdir OUTDIR  Name of the output directory (default: output)
  -n N                  Number of graphs to generate (default: 5)

Edge lists of new undirected graphs can be added in the ./src/temp/ directory with extension .g.

cnrg's Issues

when you give the graph generator a target_n larger than the original

I have seen that the target_n larger than the original graph can go in to a while loop that never ends. Have you seen that?