绝对值方程Ax-|x|=b是一类不可微的NP-hard问题.假设矩阵A满足一定条件,将转轴法融入到遗传算法中,设计了一种新的求解绝对值方程的方法.混合后的方法能够有效克服转轴法对初始点要求敏感和遗传算法在迭代后期计算效率下降的缺陷.数值实验显示整合后的算法在求解精度和收敛速度上有所提升.
Absolute value equations Ax =-|x |=b is a non-differentiable NP-hard problem.Under the suitable condition of A,a kind of new method for solving absolute value equations is proposed by combining rosenbrock algorithm with genetic algorithm.This method effectively avoid the defects which rosenbrock method is require accurate of the initial point and genetic algorithm in the late iteration calculation efficiency decline.Numerical experiments show that the consolidated algorithm improved on precision and convergence speed.