libcolgraph 0.0.5

this library provides support to construct graphs and their coloring graphs. a coloring graph is a metagraph representing all the valid colorings of a graph. each vertex of a coloring graph represents a coloring of the base graph.

coloring-graphs

a coloring graphs library written in C++ for speedy computation and wrapped in Python for ease of development and extension!

what

this library provides support to construct graphs, their coloring graphs, and biconnected component metagraphs. a coloring graph is a graph representing all the valid colorings of a graph. each vertex of a coloring graph represents a coloring of the base graph. two colorings are considered adjacent if they differ in only one vertex's color.

in this project, we represent a coloring as an integer, which, when converted to base k (for a k-coloring), is a number of length |V| and represents the vertex-wise colors [0,k) for each vertex.

the library is under development, being written using Python and C/C++. the web application uses the library to provide useful GUI for quick drawing and graph construction. in the future, we plan to develop a supplemental subsection of the library containing useful graph algorithms and ability to run simulations to test structural graph theoretic conjectures about graph coloring and coloring graphs. for documentation, feel free to take a look inside libcolgraph/ and read the docstrings. for questions, reach out.

screenshot of the web GUI

[Clockwise] A 7 vertex BaseGraph that is a hexagon with a central vertex and a missing 'spoke' leads to a quite complex ColoringGraph with k=4 colors. You can see the formation of polyps at the edges. The structure of the resultant ColoringGraph is shown in the Meta ColoringGraph produced by a run of modified Tarjan's algorithm for biconnected components. The central 'mothership' can be seen, adjacent to which there are cut vertices, and finally the stray singular coloring vertices at the tips of polyps.

for a static demo, go to the project's gitlab pages.

usage

• as a module

  plot a BaseGraph, its ColoringGraph, and the Meta ColoringGraph generated by Tarjans. you would need to have a graph formatted in an adjacency matrix file. opens plots in new browser windows.

  basic usage:

    colgraphplot [INPUT_GRAPH]
    ---
      usage: colgraphplot [-h] [-k COLORS] [-v] [--no-bg] [--no-cg] [--no-mbg]
                     [--no-mcg]
                     INPUT_GRAPH
  
      positional arguments:
        INPUT_GRAPH           read in BaseGraph from adjacency matrix file
  
      optional arguments:
        -h, --help            show this help message and exit
        -k COLORS, --colors COLORS
                              number of colors to use to create ColoringGraph [default: 3]
        -v, --verbosity       set output verbosity
        --no-bg               hide BaseGraph?
        --no-cg               hide ColoringGraph?
        --no-mbg              hide meta BaseGraph?
        --no-mcg              hide meta ColoringGraph?
  

  launch web GUI:

    colgraphweb
    ---
      usage: colgraphweb [-h] [-i INPUT_FILE] [-k COLORS] [-n] [-v] [-p PORT]
  
      optional arguments:
      -h, --help            show this help message and exit
      -i INPUT_FILE, --input-file INPUT_FILE
                            read in BaseGraph from adjacency matrix file
      -k COLORS, --colors COLORS
                            number of colors to use to create ColoringGraph [default: 3]
      -n, --new             open a blank canvas?
      -v, --verbosity       set output verbosity
      -p PORT, --port PORT  port on localhost to launch GUI on [default: 5000]
  

• as a library

  import libcolgraph
  
  bg = libcolgraph.BaseGraph()
  bg.load_txt('./in/hexmod.in')
  
  cg = bg.build_coloring_graph(4)
  
  print('bg {} led to a cg {}'.format(bg, cg))
  
  for v in cg.get_vertices():
      # do something
      pass
  

installation

• for installation from source refer to detailed install instructions

• pypi

  python3 -m pip install libcolgraph [--user] [--upgrade]
  

  things to note:

  • currently a binary wheel is available only for manylinux distributions e.g. centOS, Debian family, RedHat family, etc.
  • if your distribution is not manylinux-supported, then pip will need to compile locally using swig and setuptools. in that case, make sure you have setuptools and swig installed, as they will be needed for compilation.
  • in the future we will release wheels for MacOS as well. these might not be as frequently maintained, however, so your best bet would be to compile locally using swig.

contribute

see contributing guide

help

full documentation coming soon

who

Coloring Graphs lab, University of Richmond. Multiple contributors. (C) 2017-2019