Generates a topological order of a directed acyclic graph which is good for locality/data reuse, meaning the non-zero entries of the adjacency matrix are closer to the diagonal.
It works by recursively bi-sectioning using a direction incetivised spectral bi-partitioning algorithm.
With a dot file:
python ./SpectralTopologicalOrdering.py ../test_data/test_graph.dotWith an upper-triangular matrix in MatrixMarket File Format:
python ./SpectralTopologicalOrdering.py <Adjacency Matrix>.mtxWith a lower-triangular matrix in MatrixMarket File Format:
python ./SpectralTopologicalOrdering.py ../test_data/test_graph.mtx --lowTBA