This simulation investigates whether multi-member districts tend to produce fairer representation that single-member districts. In particular, we use an MCMC algorithm to sample from the space of all possible district plans in a toy model and plot the relationship between voter ideology and elected official ideology for that sample.
Mostly functional (MCMC works, as do other social choice functions, and the general set up). At this point, we need to add some more social choice functions, and fix some bugs in instant run-off voting, and produce and analyze the data. Example analysis: varying the number of members per district and studying the slope of the line fitted to the relationship between voter and elected ideology, or the range of elected ideology for fixed voter ideology, etc. A more ambitious step would be to incorporate real data into this, such as the data from Australia included at the github, or to build in more accurate models of voter ideology. (Currently ideology of an individual voter is modeled as a uniformly random number between 0 and 1, and they vote based on which representative has the nearest ideology.)