Abstract: Multi-scale quantum harmonic oscillator algorithm (MQHOA) is a meta-heuristic algorithm based on the theory of Quantum wave function. In the traditional MQHOA optimization process, the sampling scale of different individuals is not different. This mechanism limits the diversity of solutions. Aiming at the sampled individuals with different fitness levels, a scale adaptive strategy is proposed. This strategy reasonably expands the scale of individuals with poor sampling conditions and increases the diversity of sampling scales used by different individuals in the iterative process. In addition, a parallelization framework is proposed based on the scale difference. Seven groups of test functions are selected to test the improved algorithm (MQHOA-PS) and MQHOA on the Huawei Kunpeng 920 processor and AMD EPYC 7452 processor. The experiments show that the improved algorithm has higher accuracy and success rate and less time.

Key words: multiscale, adaptive, optimization algorithm, parallel computing, Huawei Kunpeng 920, AMD EPYC 7452