PHYSICS · Multi-part · Fermi University Physics · Mathematical Methods for Physics · Computational Physics · Monte Carlo and lattice simulation of physical systems
Error, Chance and the Machine, Ch Markov-chain Monte Carlo
Problem
The double-well chain of Section 11.5, with U(x)=(x2−1)2, barrier height 1 and target p(x)∝e−βU(x), is run with the small proposal width 0.2. The eigenvalue calculation gives its relaxation time as trel=5.5×105 steps at β=12 and 2.1×107 steps at β=16.
((a)) Treat trel as AeβΔUeff with a constant A. Estimate the effective barrier ΔUeff from the two values and compare it with the true barrier.
((b)) Use the same form to predict trel at β=20, and compare the prediction with the eigenvalue result, 7.8×108 steps.
((c)) With the wider proposal width 0.5, runs of 106 steps changed sign about 24000 times at β=3 and 1200 times at β=8. Read as N(β)≈A′e−βΔUeff, what effective barrier do these counts give? Explain why they do not measure the barrier.