A New Finite-Difference Method for Nonlinear Absolute Value Equations

In this paper, we propose a new finite-difference method for nonconvex absolute value equations. The nonsmooth unconstrained optimization problem equivalent to the absolute value equations is considered. The finite-difference technique is considered to compose the linear programming subproblems for obtaining the search direction. The algorithm avoids the computation of gradients and Hessian matrices of problems. The new finite-difference parameter correction technique is considered to ensure the monotonic descent of the objective function. The convergence of the algorithm is analyzed, and numerical experiments are reported, indicating the effectiveness by comparison against a state-of-the-art absolute value equations.