Parallel Search Algorithm for Geometric Constraints Solving

摘要

We propose a hybrid algorithm - (Parallel Search Algorithm) between PSO and simplex methods to approximate optimal solution for the Geometric Constraint problems. Locally, simplex is extended to reduce the number of in-feasible solutions while solution quality is improved with an operation order search. Globally, PSO is employed to gain parallelization while solution diversity is maintained. Performance results on Geometric Constraint problems show that Parallel Search Algorithm outperforms existing techniques.