STATIONARY FUNCTIONALS AND MONTE CARLO.

Use of stationary functionals from variational theory in Monte Carlo computations for source problems is discussed. A set of minimum variance estimators based on the optimal linear combination of the terms constituting these functionals is suggested, particularly, a Modified Nakache-Kalos-Goldstein Method. However, first order cancellation of errors is not implied. A two-stage sequential procedure using these estimators is used. Relationship to previous work is discussed, and the results of numerical application to discrete random walks for estimation of inner products over linear systems are exposed. Data are tabulated.