torch_numopt.numerical_optimizer

Base classes for numerical optimizers that use curvature information.

This module provides the core NumericalOptimizer abstract class and its subclasses for line-search and trust-region strategies. It handles parameter updates, learning-rate initialization, and direction computation using a curvature estimator.

Classes

LineSearchOptimizer(params, ...[, ...])	Numerical optimizer that uses a line-search algorithm to determine the step length.
NumericalOptimizer(params, curvature_estimator)	Base optimizer that uses a curvature estimator to compute a step direction.
TrustRegionOptimizer(params, trust_region[, ...])	Numerical optimizer that uses a trust-region strategy.