When discussing the time taken by the Metropolis algorithm
to generate independent samples we considered a distribution with longest spatial length scale L being explored using a proposal distribution with step size . Another dimension that a MCMC method must explore is the range of possible values of the log probability ln P _(x).Assuming that the state x contains a number of independent random variables proportional to N, when samples are drawn from P(x), the `asymptotic equipartition" principle tell us that the value of ln P(x) is likely to be close to the entropy of x, varying either side with a standard deviation that scales as. Consider a Metropolis method with a symmetrical proposal density, that is, one that satisfiesAssuming that accepted jumps either increase ln P (x) by some amount or decrease it by a small amount, (is this a reasonable
assumption?), discuss how long it must take to generate roughly independent samples from P(x). Discuss whether Gibbs sampling has similar properties.
Already registered? Login
Not Account? Sign up
Enter your email address to reset your password
Back to Login? Click here