A Monte Carlo Learning/Biasing Experiment with Intelligent Random Numbers

A Monte Carlo learning and biasing technique that does its learning and biasing in the random number space rather than the physical phase space is described. The technique is probably applicable to all linear Monte Carlo problems, but no proof is provided here. Instead, the technique is illustrated with a simple Monte Carlo transport problem. Problems encountered, problems solved, and speculations about future progress are discussed.