The objective of this experiment is to demonstrate different annealing strategies in the solutions to optimization problems using a Hopfield model. This is illustrated using the weighted matching problem as a case study.

The optimization is guided by the activation dynamics of a Hopfield network. The activation dynamics is guided by the energy surface defined by the activation states of the units in the Hopfield network. The three relaxation strategies studied are: (a) Deterministic relaxation (b) Stochastic relaxation (c) Mean-field approximation