A practice implementation of Newton’s method for local convergence and optimization, covering both univariate and multivariate cases; extending the first and second derivatives in the 1D case to the Jacobian matrix (gradient vector) for finding roots and the Hessian matrix for finding extrema, respectively, in the multivariate case.
cost levenberg-marquardt gradient-descent jacobian hessian taylor-series quasi-newton saddle-point interior-point quadratic-convergence non-convexity ill-conditioning
-
Updated
Aug 26, 2026 - Python