MCNP外文教案.pptVIP

Mathematical FoundationsofMarkov Chain Monte Carlo Algorithms Based on lectures given by Alistair Sinclair Computer Science Division U.C. Berkeley Overview Random Sampling The Markov Chain Monte-Carlo Paradigm Mixing Time Coupling Flow Geometry Random Sampling ? - “very large” sample set. ? - probability distribution over ?. The Probability Distribution Typically, Application 1 : Card Shuffling ? - all 52! permutations of a deck of cards. ? - uniform distribution [?x w(x)=1]. Application 2 : Counting How many ways can we tile some given pattern with dominos? Application 2 : Counting (cont.) Sample tilings uniformly at random. Let P1 = proportion of sample of type 1. Compute estimate N1* of N1 recursively. output N* = N1* / P1. Application 3 : Volume Integration [Dyer\Frieze\Kannan] ? : a convex body in Rd (d large) Problem: estimate vol(?) Application 4 : Statistical Physics ? - set of configurations of a physical system ? - Gibbs distribution ?(x)=Pr[ system in config. x]=w(x)/Z where w(x)=e-H(x)/KT The Ising Model n atomic magnets configuration: x?{-,+}n H(x)= -(#aligned neighbors) Why Sampling? statistics of “typical” configurations. mean energy (E?[H(x)]), specific heat, … estimate of “partition function” Z:=Z(T)=?x??w(x) Estimating the Partition Function Let ?=e-1/KT ? Z:=Z(?)=?x?? ?-H(x). Define 1=?0 ?1 … ?r=?. Application 5 : Optimization ? - set of feasible solutions to an optimization problem f(x) - value of solution x. Goal: maximize f(x). Idea: sample solutions where w(x)=?f(x). Application 5 : Optimization Idea: sample solutions where w(x)=?f(x). Application 6 : Hypothesis Verification in Statistical Models ? - set of hypotheses X - observed data Application 6 : Hypothesis Verification in Statistical Models (cont.) Sampling from ?(?)=P(?/X) gives: Statistical estimate of hypotheses ?. Prediction: Model comparison: normalization factor = P(X) = Prob[ model generated X ] Markov Chains Sample space ? Random variables (r.v) over ? X