# mcmc - Problems Primarily of two types: Integration and...

This preview shows pages 1–6. Sign up to view the full content.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Problems Primarily of two types: Integration and Optimization Bayesian inference and learning Computing normalization in Bayesian methods p ( y | x ) = p ( y ) p ( x | y ) y p ( y ) p ( x | y ) dy Marginalization: p ( y | x ) = z p ( y , z | x ) dz Expectation: E y | x [ f ( y )] = y f ( y ) p ( y | x ) dy Statistical mechanics: Computing the partition function Z = s exp- E ( s ) kT Optimization, Model Selection, etc. Monte Carlo Principle Target density p ( x ) on a high-dimensional space Draw i.i.d. samples { x i } n i =1 from p ( x ) Construct empirical point mass function p n ( x ) = 1 n n i =1 x i ( x ) One can approximate integrals/sums by I n ( f ) = 1 n n i =1 f ( x i ) a . s .--- n I ( f ) = x f ( x ) p ( x ) dx Unbiased estimate I n ( f ) converges by strong law For finite 2 f , central limit theorem implies n ( I n ( f )- I ( f )) = n N (0 , 2 f ) Rejection Sampling Target density p ( x ) is known, but hard to sample Use an easy to sample proposal distribution q ( x ) q ( x ) satisfies p ( x ) Mq ( x ) , M &lt; Algorithm: For i = 1 , , n 1 Sample x i q ( x ) and u U (0 , 1) 2 If u &lt; p ( x i ) Mq ( x i ) , accept x i , else reject Issues: Tricky to bound p ( x ) / q ( x ) with a reasonable constant M If M is too large, acceptance probability is small Rejection Sampling (Contd.) Markov Chains Use a Markov chain to explore the state space Markov chain in a discrete space is a process with p ( x i | x i- 1 , . . . , x 1 ) = T ( x i | x i- 1 ) A chain is homogenous if T is invariant for all i MC will stabilize into an invariant distribution if 1 Irreducible, transition graph is connected 2 Aperiodic, does not get trapped in cycles Su ffi cient condition to ensure p ( x ) is the invariant distribution...
View Full Document

## mcmc - Problems Primarily of two types: Integration and...

This preview shows document pages 1 - 6. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online