torch_numopt.algorithms.newton

Newton-type methods using exact Hessian (full or block).

These optimizers compute the exact second-order derivatives and use them to form a quadratic model. They offer fast local convergence but may be expensive for large models.

Classes

Newton(params[, lr_init, lr_method, ...])	Newton method with exact Hessian (full or block) and fixed learning rate.
NewtonCG(params[, lr_init, lr_method, ...])	Newton-CG method (inexact Newton) using conjugate gradient to solve the linear system.
NewtonCGLS(params[, lr_init, lr_method, c1, ...])	Newton-CG with line search.
NewtonCGTR(params[, lr_init, damping, mu, ...])	Newton-CG with trust region (Steihaug-Toint).
NewtonLS(params[, lr_init, lr_method, c1, ...])	Newton method with exact Hessian and line search.
NewtonTR(params[, lr_init, ...])	Newton method with exact Hessian and trust region.