当前位置: X-MOL 学术Quantum Inf. Process. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Spiking neural network dynamic system modeling for computation of quantum annealing and its convergence analysis
Quantum Information Processing ( IF 2.2 ) Pub Date : 2021-02-15 , DOI: 10.1007/s11128-021-03003-5
Chenhui Zhao , Zenan Huang , Donghui Guo

Quantum annealing algorithm is a classical natural computing method for skeuomorphs, and its algorithm design and application research have achieved fruitful results, so it is widely integrated into the research of modern intelligent optimization algorithm. This paper attempts to use the spiking neural network (SNN) dynamic system model to simulate the operation mechanism and convergence of the quantum annealing algorithm, and compares the process of searching the optimal solution to the elastic motion in the quantum tunneling field, and the change of function value during the operation of the algorithm is the simple harmonic vibration or damped vibration of quantum. Spiking neural network dynamic system model simulates the human brain by incorporating synaptic state and time components into their operational models, which represents the process of quantum fluctuations. The local convergence in the early stage and the global convergence in the late stage of the algorithm are proved by using the qualitative theory of ordinary differential equations to solve and analyze the dynamic system model, and a reasonable theoretical explanation is given for its operation mechanism. Several typical test problems are selected for experimental verification. The experimental results show that the numerical convergence curve is consistent with the convergence conclusion of theoretical analysis. Finally, both theoretical and experimental analyses show that the SNN dynamic system model established in this paper is suitable to describe the quantum annealing algorithm for optimization.



中文翻译:

量子退火计算的尖峰神经网络动态系统建模及其收敛性分析

量子退火算法是一种经典的拟态化自然计算方法,其算法设计和应用研究取得了丰硕的成果,因此已广泛地融入现代智能优化算法的研究中。本文尝试使用尖峰神经网络(SNN)动态系统模型来模拟量子退火算法的运行机理和收敛性,并比较在量子隧穿场中寻找弹性运动的最佳解和变化的过程。该算法在运算过程中的函数值是简单的谐波振动或量子的阻尼振动。尖峰神经网络动态系统模型通过将突触状态和时间成分纳入其操作模型来模拟人脑,这代表了量子涨落的过程。利用常微分方程的定性理论对动态系统模型进行求解和分析,证明了算法的早期局部收敛和后期全局收敛,并给出了合理的理论解释。选择了几个典型的测试问题进行实验验证。实验结果表明,数值收敛曲线与理论分析的收敛结论相吻合。最后,理论和实验分析均表明,本文建立的SNN动态系统模型适合描述优化的量子退火算法。利用常微分方程的定性理论对动态系统模型进行求解和分析,证明了算法的早期局部收敛和后期全局收敛,并给出了合理的理论解释。选择了几个典型的测试问题进行实验验证。实验结果表明,数值收敛曲线与理论分析的收敛结论相吻合。最后,理论和实验分析均表明,本文建立的SNN动态系统模型适合描述优化的量子退火算法。利用常微分方程的定性理论对动态系统模型进行求解和分析,证明了算法的早期局部收敛和后期全局收敛,并给出了合理的理论解释。选择了几个典型的测试问题进行实验验证。实验结果表明,数值收敛曲线与理论分析的收敛结论相吻合。最后,理论和实验分析均表明,本文建立的SNN动态系统模型适合描述优化的量子退火算法。

更新日期:2021-02-15
down
wechat
bug