Module optimization

Expand description

Optimization primitives over ndarray vectors.

Structs§

AdamConfig: Configuration for Adam optimizer.
BFGSConfig: Configuration for BFGS quasi-Newton optimization.
LineSearchConfig: Configuration for backtracking line search.
MomentumConfig: Configuration for momentum gradient descent.
ProjectedGradientConfig: Configuration for projected gradient descent with box constraints.
RMSPropConfig: Configuration for RMSProp optimizer.
SGDConfig: Configuration for SGD.

adam: Minimize objective with Adam.
adam_complex: Minimize objective with complex Adam updates.
backtracking_line_search: Perform Armijo backtracking line search.
backtracking_line_search_complex: Perform Armijo backtracking line search for complex vectors.
bfgs: Minimize objective with BFGS quasi-Newton updates.
bfgs_complex: Minimize objective with complex BFGS quasi-Newton updates.
gradient_descent: Minimize objective with fixed-step gradient descent.
gradient_descent_complex: Minimize objective with fixed-step complex gradient descent.
momentum_descent: Minimize objective with momentum gradient descent.
momentum_descent_complex: Minimize objective with complex momentum gradient descent.
projected_gradient_descent_box: Minimize objective with projected gradient descent under box constraints.
projected_gradient_descent_box_complex: Minimize objective with complex projected gradient descent under box constraints.
rmsprop: Minimize objective with RMSProp.
rmsprop_complex: Minimize objective with complex RMSProp.
stochastic_gradient_descent: Minimize objective with stochastic gradient descent.
stochastic_gradient_descent_complex: Minimize objective with complex stochastic gradient descent.