ReCom in Action

What does MUM_TX do? We use Markov Chain Monte Carlo methods to generate ensembles of district maps so that we can understand what an unbiased map should look like.

We move between maps with a technique called ReCom, in which two adjacent districts are combined and then split randomly, but in a way that preserves the legal rules governing the map (for example, that each district has to have roughly equal population).

Sorry, our animation was too memory-intensive and got kicked off! We will work on one that is slimmer for you. In the meantime, look at slides 40-42 in this talk to see an example of two consecutive steps in the chain.

The following animation shows ReCom in action. We began with the 36 current US Congressional districts. We added two "mini-districts" to reflect the fact that Texas will now have 38 districts because of its population growth in the past decade. We then used ReCom to drive the map towards population balance. The last 20 maps of the sequence are within 0.5% (check on #) of population balance.