slp::detail::AdjointExpressionGraph< Scalar > Class Template Reference

#include <sleipnir/autodiff/adjoint_expression_graph.hpp>

Public Member Functions
	AdjointExpressionGraph (const Variable< Scalar > &root)
void	update_values ()
VariableMatrix< Scalar >	generate_gradient_tree (const VariableMatrix< Scalar > &wrt) const
void	append_gradient_triplets (gch::small_vector< Eigen::Triplet< Scalar > > &triplets, int row, const VariableMatrix< Scalar > &wrt) const

Detailed Description

template<typename Scalar>
class slp::detail::AdjointExpressionGraph< Scalar >

This class is an adaptor type that performs value updates of an expression's adjoint graph.

Template Parameters

Scalar

Scalar type.

Constructor & Destructor Documentation

AdjointExpressionGraph()

template<typename Scalar >

slp::detail::AdjointExpressionGraph< Scalar >::AdjointExpressionGraph ( const Variable< Scalar > &  root )

inlineexplicit

Generates the adjoint graph for the given expression.

Parameters

root	The root node of the expression.

Member Function Documentation

append_gradient_triplets()

template<typename Scalar >

void slp::detail::AdjointExpressionGraph< Scalar >::append_gradient_triplets ( gch::small_vector< Eigen::Triplet< Scalar > > &  triplets, int  row, const VariableMatrix< Scalar > &  wrt  ) const

inline

Updates the adjoints in the expression graph (computes the gradient) then appends the adjoints of wrt to the sparse matrix triplets.

Parameters

triplets	The sparse matrix triplets.
row	The row of wrt.
wrt	Vector of variables with respect to which to compute the Jacobian.

generate_gradient_tree()

template<typename Scalar >

VariableMatrix< Scalar > slp::detail::AdjointExpressionGraph< Scalar >::generate_gradient_tree ( const VariableMatrix< Scalar > &  wrt ) const

inline

Returns the variable's gradient tree.

This function lazily allocates variables, so elements of the returned VariableMatrix will be empty if the corresponding element of wrt had no adjoint. Ensure Variable::expr isn't nullptr before calling member functions.

Parameters

wrt	Variables with respect to which to compute the gradient.

Returns

The variable's gradient tree.

update_values()

template<typename Scalar >

void slp::detail::AdjointExpressionGraph< Scalar >::update_values ( )

inline

Update the values of all nodes in this adjoint graph based on the values of their dependent nodes.

---

The documentation for this class was generated from the following file:

• include/sleipnir/autodiff/adjoint_expression_graph.hpp