Cybernetic optimization by simulated annealing: Accelerating convergence by parallel processing and probabilistic feedback control

The convergence of the simulated annealing algorithm is accelerated by a probabilistic feedback control scheme. This scheme uses two or more parallel processors to solve the same or related combinatorial optimization problems and these are coupled by a probabilistic measure of quality (PMQ). The PMQ is used to generate an error signal for use in feedback control. Control over the search process is achieved by using the error signal to modulate the temperature parameter. Other aspects of control theory, such as the system gain and its effects on system performance, are described. Theoretical and experimental results show that such a scheme increases the steady-state probability of the globally optimal solutions.