polyfem::solver::Scaling Class Reference

#include <Parametrizations.hpp>

Inheritance diagram for polyfem::solver::Scaling:

Collaboration diagram for polyfem::solver::Scaling:

Public Member Functions
	Scaling (const double scale, const int from=-1, const int to=-1)
int	size (const int x_size) const override
	Compute DOF of y given DOF of x.
Eigen::VectorXd	inverse_eval (const Eigen::VectorXd &y) override
	Eval x = f^-1 (y).
Eigen::VectorXd	eval (const Eigen::VectorXd &x) const override
	Eval y = f(x).
Eigen::VectorXd	apply_jacobian (const Eigen::VectorXd &grad, const Eigen::VectorXd &x) const override
	Apply jacobian for chain rule.
Public Member Functions inherited from polyfem::solver::Parametrization
virtual	~Parametrization ()=default

Private Attributes
const int	from_
const int	to_
const double	scale_

Detailed Description

Definition at line 33 of file Parametrizations.hpp.

Constructor & Destructor Documentation

Scaling()

polyfem::solver::Scaling::Scaling	(	const double	scale,
		const int	from = -1,
		const int	to = -1
	)

Definition at line 52 of file Parametrizations.cpp.

References from_, scale_, and to_.

Member Function Documentation

apply_jacobian()

Eigen::VectorXd polyfem::solver::Scaling::apply_jacobian ( const Eigen::VectorXd &  grad_full, const Eigen::VectorXd &  x  ) const

overridevirtual

Apply jacobian for chain rule.

Let g(y) = g(f(x)). Given ∂g/∂y, compute ∂g/∂x = ∂g/∂y * ∂y/∂x.

Parameters

[in]	grad_full	∂g/∂y.
[in]	x	Where ∂g/∂x is evaluated.

Returns

∂g/∂x.

Implements polyfem::solver::Parametrization.

Definition at line 83 of file Parametrizations.cpp.

References from_, scale_, and to_.

eval()

Eigen::VectorXd polyfem::solver::Scaling::eval ( const Eigen::VectorXd &  x ) const

overridevirtual

Eval y = f(x).

Parameters

[in]

x

x.

Returns

y.

Implements polyfem::solver::Parametrization.

Definition at line 71 of file Parametrizations.cpp.

References from_, scale_, to_, x, and y.

inverse_eval()

Eigen::VectorXd polyfem::solver::Scaling::inverse_eval ( const Eigen::VectorXd &  y )

overridevirtual

Eval x = f^-1 (y).

This is not a strict inverse in mathematical sense, one may choose "reasonable" x even if f is not one-to-one.

Parameters

[in]

y

y.

Returns

x.

Exceptions

std::runtime_error

Throw if not implemented.

Reimplemented from polyfem::solver::Parametrization.

Definition at line 59 of file Parametrizations.cpp.

References from_, scale_, to_, and y.

size()

int polyfem::solver::Scaling::size ( const int  x_size ) const

inlineoverridevirtual

Compute DOF of y given DOF of x.

Parameters

[in]

x_size

The DOF of x.

Returns

DOF of y.

Implements polyfem::solver::Parametrization.

Definition at line 38 of file Parametrizations.hpp.

Member Data Documentation

from_

const int polyfem::solver::Scaling::from_

private

Definition at line 44 of file Parametrizations.hpp.

Referenced by apply_jacobian(), eval(), inverse_eval(), and Scaling().

scale_

const double polyfem::solver::Scaling::scale_

private

Definition at line 45 of file Parametrizations.hpp.

Referenced by apply_jacobian(), eval(), inverse_eval(), and Scaling().

to_

const int polyfem::solver::Scaling::to_

private

Definition at line 44 of file Parametrizations.hpp.

Referenced by apply_jacobian(), eval(), inverse_eval(), and Scaling().

---

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

• src/polyfem/optimization/parametrization/Parametrizations.hpp
• src/polyfem/optimization/parametrization/Parametrizations.cpp