Configure the benchmarks you want in the file get_problem_directories:
  it is possible to adjust the graph size (number of nodes), the probability
  of an edge (Bollobás), and others depending on the domain.
Then, run 'setup'. Generating the PDDL problem
  with the default settings may take a while.
After this step, the folder 'problems' will contain the generated PDDL problems.

To solve these, the script solve_all may be useful: it takes the horizon windows
  limits into account when calling M to solve the problems.

The scripts bash_results.py and results.py (the latter receives a folder as argument)
    may be useful for analyzing the solutions.

The directory 'chromatic' has some of our benchmarks for determining the
  chromatic number of a graph using our k-colorability domain.