{"id":"arxiv/2610.31110","url":"https://arxiv.org/abs/2610.31110","tags":["math.OC"],"score":0,"title":"On Modified Newton Methods for Unconstrained Optimization Based on Lanczos Tridiagonalization","domain":"math.OC","source":"arxiv","authors":["Aban Ansari-Önnestam","Anders Forsgren","Philip E. Gill"],"reviews":0,"subject":"math.OC","summary":"For a strongly convex unconstrained minimization problem, Newton's method with a line search offers rapid convergence to a minimizer by incorporating second-order information. However, for a general unconstrained problem the Hessian need not be positive definite, which can lead to the inability to find a search direction or to search directions that fail to produce descent. Modified Newton methods are designed to handle this situation. This paper considers modifications to a Newton-conjugate-gradient approach based on Lanczos tridiagonalization. Existing methods are extended using dynamic…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-22","reactions":{}}