torch_numopt.algorithms

Concrete optimizer implementations.

This package contains ready-to-use optimizers that inherit from the base classes in numerical_optimizer. They combine curvature estimators with step- selection strategies (line search or trust region) to provide complete optimization algorithms.

Available optimizers: - GradientDescent (and variants with line search / trust region) - ConjugateGradient (with line search) - Newton (exact Hessian, with line search, trust region, or CG) - GaussNewton (Gauss-Newton approximation) - LevenbergMarquardt (trust-region with adaptive damping) - LBFGS (limited-memory BFGS) - AdaHessian (diagonal Hessian with momentum)

Modules

adahessian	AdaHessian optimizer (diagonal Hessian with momentum).
conjugate_gradient	Non-linear conjugate gradient methods.
gauss_newton	Gauss-Newton optimization algorithms for least-squares problems.
gradient_descent	Gradient descent optimizers (first-order).
hutchinson_newton	Newton methods with a diagonal Hutchinson diagonal approximation.
lbfgs	Limited-memory BFGS (L-BFGS) optimizers.
levenberg_marquardt	Levenberg-Marquardt optimizer (trust-region variant).
newton	Newton-type methods using exact Hessian (full or block).