LEARNING MODE OPTIMIZATION

摘要

Given a database of performance values-point associations that grows every time optimization is run on that same circuit (but with different performance specifications), the optimizer gets progressively more efficient by evaluating a starting point (215) and evaluating a within cost question (215). If the cost is within a parameter chosen the results are stored (270). If the cost is not within (215) the parameters then an iteration is performed (220-250). It is more efficient because the probalility of beginning optimization with a good starting point close to the optimum solution also increases with increased information in the database. This invention therefore enhances the speed of the class of optimizers that uses search-based algorithm.