Multicanonical Monte Carlo scheme for finding rare growth factors

A multicanonical Monte Carlo (MMC) scheme developed to approximate the tails of growth factor probab

The script gf_mmc_driver.m runs multiple independent instances of a MMC method in parallel with MATLAB's Distributed Computing Toolbox. The outer loop in script gf_mmc_driver.m represents the basic MMC iteration, which initializes and runs a Markov chain and performs a Berg update. In each of the independent MMC instances, a different initialization strategy was used for the Markov chain. For more information, see T. A. Driscoll and K. Maki in the Education section of SIAM Review, to appear 2006-2007.

To use simply unzip file and first edit the job manager name in jm=findResource() found on line 37 of gf_mmc_driver.m. Then run the driver script gf_mmc_driver.m from command line. Requires functions mcmc1 and mcmc2 provided in gf_mmc.zip file.