On a randomized version of exhaustive local search
Matthias Löwe · Stochastic Models · 1996
We introduce a stochastic version of the parallel local search algorithm. Using techniques similar to those used to establish conver-gence of the simulated annealing algorithm we are able to give a cooling schedule for which the process converges and to analyze the speed of convergence by spectral gap methods