Answers are grounded in this paper and cite their sources.
The conjugate gradient method of Iiduka and Narushima constructs search directions that use objective function values in addition to gradient information, making it suitable for large-scale optimization. Its application to nonlinear systems of equations, however, remains unexplored. In this paper, we adapt this construction to nonlinear equations by using residual information in place of function values and combine it with the hyperplane projection technique of Solodov and Svaiter. The resulting class of methods requires no Jacobian information or matrix storage, making it well suited to large-scale and potentially nonsmooth problems. We establish global convergence under monotonicity and Lipschitz continuity assumptions. Numerical examples, including signal recovery problems, demonstrate that the proposed method is effective, efficient and competitive with existing approaches.
No takes yet. Share an insight, caveat, or question.
Ibrahim et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: