求解非线性互补问题的非精确光滑化牛顿算法OA
Inexact Smoothing Newton Algorithm for Nonlinear Complementarity Problems
传统牛顿法求解大规模非线性互补问题时,精确求解线性子系统会显著增加计算复杂度,矩阵求逆与存储需求也成为制约算法效率的主要瓶颈.为此,提出一种基于光滑对称扰动函数的非精确光滑化牛顿算法,通过引入非精确解策略和自适应参数调整机制,在保证全局超线性收敛(局部二次收敛)的同时,显著降低了计算成本.数值实验表明,本文算法在求解千维以上问题时,迭代次数较经典方法减少40%~60%.
When using traditional Newton's method to solve large-scale nonlinear comple-mentarity problems,accurately solving linear subsystems significantly increases computa-tional complexity,while matrix inversion and storage requirements also become the main bottlenecks restricting algorithm efficiency.Therefore,an inexact smoothing Newton algorithm based on smooth symmetric perturbation functions was proposed.By introdu-cing an inexact solution strategy and an adaptive parameter adjustment mechanism,the computational cost is significantly reduced while ensuring global superlinear convergence(local quadratic convergence).Numerical experiments show that the algorithm proposed reduces the number of iterations by 40%to 60%compared to classical methods when sol-ving problems with dimensions over a thousand.
朱子旋;芮绍平
淮北理工学院教育学院,淮北 235000淮北师范大学数学与统计学院,淮北 235000
数理科学
非线性互补问题光滑化方法非精确光滑化牛顿法自适应参数局部二次收敛
nonlinear complementarity problemsmoothing methodinexact smoothing Newton algorithmadaptive parameterslocal quadratic convergence
《青岛大学学报(自然科学版)》 2026 (3)
3-8,6
安徽省教育厅自然科学研究项目(批准号:2023AH050348,2023ZK027,2023ZK030)资助.
评论