通信学报 ›› 2016, Vol. 37 ›› Issue (7): 79-86.doi: 10.11959/j.issn.1000-436x.2016136

• 学术论文 • 上一篇    下一篇

具有能级稳定过程的MQHOA优化算法

王鹏1,黄焱2   

  1. 1 西南民族大学计算机科学与技术学院,四川 成都 610225
    2 淮阴师范学院计算机科学与技术学院,江苏 淮安 223300
  • 出版日期:2016-07-25 发布日期:2016-07-28
  • 基金资助:
    国家自然科学基金资助项目;模式识别与智能信息处理四川省高校重点实验室开放基金资助项目

MQHOA algorithm with energy level stabilizing process

Peng WANG1,Yan HUANG2   

  1. 1 School of Computer Science and Technology, Southwest University for Nationalities, Chengdu 610225, China
    2 School of Computer Science and Technology, Huaiyin Normal University, Huaian 223300, China
  • Online:2016-07-25 Published:2016-07-28
  • Supported by:
    The National Natural Science Foundation of China;Sichuan Key Laboratory Open Foundation of Pattern Recognition and Intelligent Information Processing

摘要:

在量子模型下将优化问题转化为求解约束态的基态波函数问题,通过泰勒近似采用谐振子势阱对目标函数进行逼近,类比量子谐振子的波函数图像提出了一种改进的多尺度量子谐振子优化算法。算法包括3个基本迭代收敛过程:能级稳定过程、能级降低过程和尺度降低过程,算法的收敛过程与物理模型基本吻合。改进算法将主观控制参数减少为1个,同时参照量子模型定义了算法的波函数和零点能。实验结果表明,改进算法的复杂函数优化性能优于多种常见优化算法,对于Ackley、Griewank、Sphere、Sum Squares、Zakharov等高维标准测试函数均能以100%的概率获得全局最优解。

关键词: 优化算法, 函数优化, 多尺度量子谐振子算法, 波函数, 基态

Abstract:

An improved multi-scale quantum harmonic oscillator algorithm (MQHOA) with energy level stabilizing process was proposed analogizing to quantum harmonic oscillator's wave function. Inspired by quantum model, the op-timization problem was transformed to finding ground state wave function of bound state. Harmonic oscillator potential well was used to approach objective function under the condition of Taylor approximation. Energy level stabilization, en-ergy level reduction, scale reduction were the basic iterative convergence processes of MQHOA, coinciding with its physical model. Only one subjective control parameter was needed in MQHOA whose wave function and zero-point en-ergy were defined with reference to quantum model. Experimental results show that MQHOA's performance is superior to several other common optimization algorithms. For high dimensional testing functions including Ackley、Griewank、Sphere、Sum Squares、Zakharov, etc, the global optimums can be obtained precisely with 100% probability.

Key words: optimization algorithm, function optimization, MQHOA, wave function, ground state

No Suggested Reading articles found!