SearcharxivSearch

arXiv subjects

Michael Howes

Publications and source records attributed to Michael Howes.

3 recordsLinked to original sources

Limit profiles and cutoff for the Burnside process on Sylow double cosets

This article gives sharp estimates for the mixing time of the Burnside process for Sylow $p$-double cosets in the symmetric group $S_n$. This process is a Markov chain on $S_n$ which can be used to uniformly sample Sylow $p$-double cosets. The analysis applies when $n = pk$ with $p$ prime and $k < p$. The main result describes the limit profile of the distance to the stationary distribution as $p$ goes to infinity. From the limit profile, we get the following two corollaries. First, if $k$ remains fixed as $p \to \infty$, then order $p$ steps are necessary and sufficient for mixing and cut-off does not occur. Second, if $k \to \infty$ as $p \to \infty$, then cut-off occurs at $p \log k$ with a window of size $p$. The limit profile is derived from explicit upper and lower bounds on the distance between the Burnside process and its stationary distribution. These non-asymptotic bounds give very accurate approximations even for $p$ as small as 11.

math.PR

Random sampling of contingency tables and partitions: Two practical examples of the Burnside process

This paper gives new, efficient algorithms for approximate uniform sampling of contingency tables and integer partitions. The algorithms use the Burnside process, a general algorithm for sampling a uniform orbit of a finite group acting on a finite set. We show that a technique called `lumping' can be used to derive efficient implementations of the Burnside process. For both contingency tables and partitions, the lumped processes have far lower per step complexity than the original Markov chains. We also define a second Markov chain for partitions called the reflected Burnside process. The reflected Burnside process maintains the computational advantages of the lumped process but empirically converges to the uniform distribution much more rapidly. By using the reflected Burnside process we can easily sample uniform partitions of size $10^{10}$.

stat.CO

Markov Chain Monte Carlo Significance Tests

Monte Carlo significance tests are a general tool that produce p-values by generating samples from the null distribution. However, Monte Carlo tests are limited to null hypothesis which we can exactly sample from. Markov chain Monte Carlo (MCMC) significance tests are a way to produce statistical valid p-values for null hypothesis we can only approximately sample from. These methods were first introduced by Besag and Clifford in 1989 and make no assumptions on the mixing time of the MCMC procedure. Here we review the two methods of Besag and Clifford and introduce a new method that unifies the existing procedures. We use simple examples to highlight the difference between MCMC significance tests and standard Monte Carlo tests based on exact sampling. We also survey a range of contemporary applications in the literature including goodness-of-fit testing for the Rasch model, tests for detecting gerrymandering [8] and a permutation based test of conditional independence [3].

stat.ME