A Goal Directed Heuristic Optimisation Algorithm for the MOODS Synthesis System

The optimisation strategy used in MOODS is based on a simulated annealing approach1, 2, 3, 4. The simulated annealing algorithm is shown in Figure 1a. A cost function is used to evaluate the design and determine its position in the design space. Graph transformations are randomly selected and their effect on the design evaluated with respect to the user specified cost function. Transformations will be applied to the design if an improvement results ormay be if a degradation results. The probability of accepting a degradation is based on a ”temperature” parameter which is reduced exponentially by Tstep after selecting a given number of transformations, Istep , thus less degradation transforms are applied to the design as optimisation progresses. Occasional degradation of the design is necessary to avoid local optima (Figure 1b) and achieve the global optimum for the cost function specified.