A simulator for detecting potential bias in an MxN districting grid
The only package required for running the simulator is numpy. It is installable via pip:
$ pip install numpy
The simulator takes a JSON file as input detailing the districting, see example3x3.json for an example 3x3 districting.
For generating 10 5-step simulations on the 3x3 grid defined in example3x3.json, the command would be the following:
$ python simulation.py --steps 5 --num-simulations 10 --file example3x3.json
The command line args can be viewed with the -h flag:
$ python simulation.py -h
usage: simulation.py [-h] -s STEPS -n NUM_SIMULATIONS -f FILE
Run the gerrymandering simulator on an nxn grid
optional arguments:
-h, --help show this help message and exit
-s STEPS, --steps STEPS
Number of steps to take for the random walk
-n NUM_SIMULATIONS, --num-simulations NUM_SIMULATIONS
Number of simulations to run
-f FILE, --file FILE JSON file representing the grid to simulate on
There are two units of output for the simulation. one is a file of the form XXXXXX-results which prints out all of the districtings that were achieved during the simulations to a file.
There is also a graph printed showing the distribution of the Efficiency Gap values for the simulation like so:
