MODIFYING CONSTRAINT-COMPLIANT POPULATIONS IN POPULATION-BASED OPTIMIZATION

摘要

Some embodiments are directed to determining a plurality of constraint compliant values for each of a plurality of constrained variables of an optimization problem, wherein a constraint condition mutually constrains possible values that can be used for the plurality of constrained variables, and wherein the plurality of constraint compliant values comply with the constraint condition. Some embodiments are further directed to generating a population of constraint compliant candidate solutions for a computer-based simulation that implements a population-based optimization algorithm for the optimization problem, wherein the constraint compliant candidate solutions use a subset of the plurality of constraint compliant values and each of the constraint compliant candidate solutions comply with the constraint condition. Some embodiments are further directed to, while running the computer-based simulation with the population of constraint compliant candidate solutions, determining that a mutated candidate solution created from mutating one of the constraint compliant candidate solutions fails to comply with the constraint condition. Some embodiments are further directed to modifying the mutated candidate solution to use at least one value randomly selected from the plurality of constraint compliant values for a corresponding one of the plurality of constrained variables resulting in a constraint compliant mutated candidate solution that complies with the constraint condition.