A multiplicative algorithm for Fermi-models of macrosystems

A range of interative multiplicative algorithms are considered that are employed in determining nonnegative solutions in conditional extremum problems which characterize Fermi-models of equilibrium macrosystem states. Conditions are established for convergence and estimation of the convergence constant of a multiplicative algorithm with a constant step. Situmations are identified where its rate exceeds that of a gradient constant step algorithm. The results of computing experiments are reported.