Module automatic_differentiation

Automatic differentiation for exact gradient and Hessian computation

This module provides automatic differentiation capabilities for optimization, supporting both forward-mode and reverse-mode AD for efficient and exact derivative computation.

Re-exports

pub use dual_numbers::Dual;

pub use dual_numbers::DualNumber;

pub use forward_mode::forward_gradient;

pub use forward_mode::forward_hessian_diagonal;

pub use forward_mode::ForwardADOptions;

pub use reverse_mode::reverse_gradient;

pub use reverse_mode::reverse_hessian;

pub use reverse_mode::ReverseADOptions;

pub use tape::ComputationTape;

pub use tape::TapeNode;

pub use tape::Variable;

Modules

dual_numbers

Dual numbers for forward-mode automatic differentiation

forward_mode

Forward-mode automatic differentiation

reverse_mode

Reverse-mode automatic differentiation (backpropagation)

tape

Computational tape for reverse-mode automatic differentiation

Structs

ADResult

Result of automatic differentiation computation

AutoDiffOptions

Options for automatic differentiation

FunctionWrapper

Wrapper for regular functions to make them compatible with AD

Enums

ADMode

Automatic differentiation mode selection

Traits

AutoDiffFunction

Function trait for automatic differentiation

Functions

autodiff

Main automatic differentiation function

create_ad_gradient

Create a gradient function using automatic differentiation

create_ad_hessian

Create a Hessian function using automatic differentiation

optimize_ad_mode

Optimize AD mode selection based on problem characteristics