Research Interests for Mark Huber

For high dimensional problems, Monte Carlo samples are a fast way to estimate integrals without the need to construct grids with exponentially many points. Within Monte Carlo simulation, my primary area of expertise is perfect sampling, algorithms that generate random variates from a variety of distributions that are interesting from either a theoretical or pratical point of view.

Keywords:

perfect simulation, Monte Carlo algorithms, mixing times

Current projects:

approximating the permanent

studying speed of covergence for parallel tempering

Markov chains for generating regular graphs

restoration of grayscale images

applications of the Randomness Recycler

Recent Publications

1. M. Huber, Perfect simulation for image restoration, Stochastic Models, vol. 23 no. 3 (August, 2007), pp. 475--487, Taylor and Francis [abs]
2. M. Huber, Perfect simulation with exponential tails, Random Structures and Algorithms (Accepted, June, 2007) [abs]
3. D. B. Woodard, S. Schmidler, and M. Huber, Conditions for Torpid Mixing of Parallel and Simulated Tempering on Multimodal Distributions, Stochastic Processes and their Applications (Submitted, 2007)
4. D. B. Woodard, S. C. Schmidler and M. Huber, Conditions for Rapid Mixing of Parallel and Simulated Tempering on Multimodal Distributions, Annals of Applied Probability (Submitted, 2007)
5. M. Huber and J. Law, Fast approximation of the permanent for very dense problems, in SODA 2008 (Accepted, 2007) [abs]