BFGS stands for Broyden–Fletcher–Goldfarb–Shanno (BFGS) method for solving unconstrained nonlinear optimization problems. More...
#include <pcl/registration/bfgs.h>
Classes | |
| struct | Parameters |
Public Types | |
| typedef FunctorType::Scalar | Scalar |
| typedef FunctorType::VectorType | FVectorType |
| typedef Eigen::DenseIndex | Index |
Public Member Functions | |
| BFGS (FunctorType &_functor) | |
| BFGSSpace::Status | minimize (FVectorType &x) |
| BFGSSpace::Status | minimizeInit (FVectorType &x) |
| BFGSSpace::Status | minimizeOneStep (FVectorType &x) |
| BFGSSpace::Status | testGradient (Scalar epsilon) |
| void | resetParameters (void) |
Public Attributes | |
| Parameters | parameters |
| Scalar | f |
| FVectorType | gradient |
BFGS stands for Broyden–Fletcher–Goldfarb–Shanno (BFGS) method for solving unconstrained nonlinear optimization problems.
For further details please visit: http://en.wikipedia.org/wiki/BFGS_method The method provided here is almost similar to the one provided by GSL. It reproduces Fletcher's original algorithm in Practical Methods of Optimization algorithms : 2.6.2 and 2.6.4 and uses the same politics in GSL with cubic interpolation whenever it is possible else falls to quadratic interpolation for alpha parameter.
Definition at line 114 of file bfgs.h.
| typedef FunctorType::VectorType BFGS< FunctorType >::FVectorType |
| BFGSSpace::Status BFGS< FunctorType >::minimize | ( | FVectorType & | x | ) | [inline] |
Definition at line 294 of file bfgs.h.
References BFGS< FunctorType >::Parameters::max_iters, BFGS< FunctorType >::minimizeInit(), BFGS< FunctorType >::minimizeOneStep(), BFGS< FunctorType >::parameters, and BFGSSpace::Success.
| BFGSSpace::Status BFGS< FunctorType >::minimizeInit | ( | FVectorType & | x | ) | [inline] |
Definition at line 305 of file bfgs.h.
References BFGS< FunctorType >::gradient, and BFGSSpace::NotStarted.
Referenced by pcl::GeneralizedIterativeClosestPoint< PointSource, PointTarget >::estimateRigidTransformationBFGS(), and BFGS< FunctorType >::minimize().
| BFGSSpace::Status BFGS< FunctorType >::minimizeOneStep | ( | FVectorType & | x | ) | [inline] |
Definition at line 332 of file bfgs.h.
References pcl::B, BFGS< FunctorType >::gradient, BFGSSpace::NoProgress, BFGS< FunctorType >::Parameters::order, BFGS< FunctorType >::parameters, BFGS< FunctorType >::Parameters::rho, BFGS< FunctorType >::Parameters::sigma, BFGS< FunctorType >::Parameters::step_size, BFGSSpace::Success, BFGS< FunctorType >::Parameters::tau1, BFGS< FunctorType >::Parameters::tau2, and BFGS< FunctorType >::Parameters::tau3.
Referenced by pcl::GeneralizedIterativeClosestPoint< PointSource, PointTarget >::estimateRigidTransformationBFGS(), and BFGS< FunctorType >::minimize().
| void BFGS< FunctorType >::resetParameters | ( | void | ) | [inline] |
Definition at line 153 of file bfgs.h.
References BFGS< FunctorType >::parameters.
| BFGSSpace::Status BFGS< FunctorType >::testGradient | ( | Scalar | epsilon | ) | [inline] |
Definition at line 412 of file bfgs.h.
References BFGS< FunctorType >::gradient, BFGSSpace::NegativeGradientEpsilon, BFGSSpace::Running, and BFGSSpace::Success.
Referenced by pcl::GeneralizedIterativeClosestPoint< PointSource, PointTarget >::estimateRigidTransformationBFGS().
| FVectorType BFGS< FunctorType >::gradient |
Definition at line 157 of file bfgs.h.
Referenced by BFGS< FunctorType >::minimizeInit(), BFGS< FunctorType >::minimizeOneStep(), and BFGS< FunctorType >::testGradient().
| Parameters BFGS< FunctorType >::parameters |
Definition at line 155 of file bfgs.h.
Referenced by pcl::GeneralizedIterativeClosestPoint< PointSource, PointTarget >::estimateRigidTransformationBFGS(), BFGS< FunctorType >::minimize(), BFGS< FunctorType >::minimizeOneStep(), and BFGS< FunctorType >::resetParameters().
