A triangle process on graphs with given degree sequence

The triangle switch Markov chain is designed to generate random graphs with given degree sequence, but having more triangles than would appear under the uniform distribution. Transition probabilities of the chain depends on a parameter, called the activity, which is used to assign higher stationary probability to graphs with more triangles. In previous work we proved ergodicity of the triangle switch chain for regular graphs. Here we prove ergodicity for all sequences with minimum degree at least 3, and show rapid mixing of the chain when the activity and the maximum degree are not too large. As far as we are aware, this is the first rigorous analysis of a Markov chain algorithm for generating graphs from a a known non-uniform distribution.
Comment: 35 pages