softAbsMetric Class Reference

#include <softAbsMetric.h>

Inheritance diagram for softAbsMetric:

List of all members.

Public Member Functions
	softAbsMetric (int dim)
double	T ()
	Return the kinetic energy.
double	tau ()
	Return the quadratic form of the kinetic energy.
MatrixXd	Lambda ()
	Return the inverse SoftAbs metric.
void	bounceP (const VectorXd &normal)
SelfAdjointEigenSolver < MatrixXd > &	eigenDeco ()
	Return the eigendecomposition of the metric.
VectorXd &	lambdaDotP ()
	Calculate the product of the metric with the current momentum.
void	sampleP (Random &random)
void	setSoftAbsAlpha (double alpha)
	Set the SoftAbs regularization coefficient.
void	checkEvolution (const double epsilon=1e-6)
	Comparing the evolution implementations with finite differences.
void	displayState ()
	Display current state.
Protected Member Functions
virtual void	fComputeH ()=0
	Compute the Hessian at the current position.
virtual void	fComputeGradH (int i)=0
	Compute the ith component of the gradient of the Hessian.
void	fComputeMetric ()
void	fPrepareSpatialGradients ()
	Compute intermediate values necessary for the spatial gradients dTaudq and dPhidq.
void	fUpdateP ()
	Update any momenta-dependent auxiliary terms.
VectorXd &	dTaudp ()
	Gradient of the quadrtic form with respect to the momenta.
VectorXd &	dTaudq ()
	Gradient of the quadratic form with respect to the position.
VectorXd &	dPhidq ()
	Gradient of the psuedo-potential with respect to the momenta.
Protected Attributes
double	mSoftAbsAlpha
	Regularization coefficient.
MatrixXd	mH
	Hessian.
MatrixXd	mGradH
	Gradient of the Hessian.
MatrixXd	mPseudoJ
	Discrete "pseudo-Jacobian" of the SoftAbs transform.
MatrixXd	mAuxMatrixOne
	Auxiliary matrix for intermediate matrix calcs.
MatrixXd	mAuxMatrixTwo
	Auxiliary matrix for intermediate matrix calcs.
MatrixXd	mCacheMatrix
	Auxiliary matrix for caching spatial gradient terms.
SelfAdjointEigenSolver< MatrixXd >	mEigenDeco
	Eigendecomposition of the metric.
VectorXd	mSoftAbsLambda
	Transformed eigenvalues.
VectorXd	mSoftAbsLambdaInv
	Repciprocal of the transformed eigenvalues.
VectorXd	mQp
	Auxiliary vector for intermediate matrix calcs.
VectorXd	mLambdaQp
	Auxiliary vector for intermediate matrix calcs.

---

Detailed Description

Author:

Michael Betancourt

Abstract base class defining the interface for a Hamiltonian defined on a Riemannian manifold with a SoftAbs Metric

---

Constructor & Destructor Documentation

softAbsMetric::softAbsMetric

(

int

dim

)

[explicit]

Constructor

Parameters:

dim	Dimension of the target space

---

Member Function Documentation

void softAbsMetric::bounceP

(

const VectorXd &

normal

)

[virtual]

Evolve the momentum through a bounce off of a constraint surface

Parameters:

normal

Vector normal to constraint surface

Implements baseHamiltonian.

void softAbsMetric::checkEvolution

(

const double

epsilon = 1e-6

)

[virtual]

Comparing the evolution implementations with finite differences.

Compare the position and momentum evolution implementations with finite differences

Parameters:

epsilon

Size of finite difference step

Reimplemented from baseHamiltonian.

void softAbsMetric::fComputeMetric

(

)

[protected, virtual]

Compute the metric at the current position, performing an eigen decomposition and computing the log determinant

Implements dynamMetric.

void softAbsMetric::sampleP

(

Random &

random

)

[virtual]

Sample the momentum from the conditional distribution

Parameters:

random

External RandomLib generator

Implements baseHamiltonian.

Examples:

examples/main.cpp.

---

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

• base/softAbsMetric.h
• base/softAbsMetric.cpp