We study rectangular dissections of an n × n lattice region into rectangles of area n, where n = 2k for an even integer k. We show that there is a natural edge-flipping Markov chain that connects the state space. A similar edge-flipping chain is also known to connect the state space when restricted to dyadic tilings, where each rectangle is required to have the form R = [s2u, (s + 1)2u] × [t2v, (t+1)2u], where s, t, u and v are nonnegative integers. The mixing time of these chains is open. We consider a weighted version of these Markov chains where, given a parameter λ > 0, we would like to generate each rectangular dissection (or dyadic tiling) a with probability proportional to λ|σ|, where |σ| is the total edge length. We show there is a phase transition in the dyadic setting: when λ < 1, the edge-flipping chain mixes in time O (n2 logn), and when λ > 1, the mixing time is exp (ω (n2)). Simulations suggest that the chain converges quickly when λ = 1, but this case remains open. The behavior for general rectangular dissections is more subtle, and even establishing ergodicity of the chain requires a careful inductive argument. As in the dyadic case, we show that the edge-flipping Markov chain for rectangular dissections requires exponential time when λ > 1. Surprisingly, the chain also requires exponential time when λ < 1, which we show using a different argument. Simulations suggest that the chain converges quickly at the isolated point λ = 1.
CITATION STYLE
Cannon, S., Miracle, S., & Randall, D. (2015). Phase transitions in random dyadic tilings and rectangular dissections. In Proceedings of the Annual ACM-SIAM Symposium on Discrete Algorithms (Vol. 2015-January, pp. 1573–1589). Association for Computing Machinery. https://doi.org/10.1137/1.9781611973730.104
Mendeley helps you to discover research relevant for your work.