elxAdaptiveStochasticLBFGS.h

Go to the documentation of this file.

/*=========================================================================
 *
 *  Copyright UMC Utrecht and contributors
 *
 *  Licensed under the Apache License, Version 2.0 (the "License");
 *  you may not use this file except in compliance with the License.
 *  You may obtain a copy of the License at
 *
 *        http://www.apache.org/licenses/LICENSE-2.0.txt
 *
 *  Unless required by applicable law or agreed to in writing, software
 *  distributed under the License is distributed on an "AS IS" BASIS,
 *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 *  See the License for the specific language governing permissions and
 *  limitations under the License.
 *
 *=========================================================================*/
#ifndef elxAdaptiveStochasticLBFGS_h
#define elxAdaptiveStochasticLBFGS_h
 
#include "elxIncludes.h" // include first to avoid MSVS warning
#include "itkAdaptiveStochasticLBFGSOptimizer.h"
 
#include "elxProgressCommand.h"
#include "itkAdvancedTransform.h"
#include "itkMersenneTwisterRandomVariateGenerator.h"
#include "itkComputeJacobianTerms.h"
#include "itkComputeDisplacementDistribution.h"
#include "itkMultiThreaderBase.h"
#include "itkImageRandomSampler.h"
#include "itkLineSearchOptimizer.h"
#include "itkMoreThuenteLineSearchOptimizer.h"
 
 
namespace elastix
{
template <class TElastix>
class ITK_TEMPLATE_EXPORT AdaptiveStochasticLBFGS
  : public itk::AdaptiveStochasticLBFGSOptimizer
  , public OptimizerBase<TElastix>
{
public:
  ITK_DISALLOW_COPY_AND_MOVE(AdaptiveStochasticLBFGS);
 
  using Self = AdaptiveStochasticLBFGS;
  using Superclass1 = AdaptiveStochasticLBFGSOptimizer;
  using Superclass2 = OptimizerBase<TElastix>;
  using Pointer = itk::SmartPointer<Self>;
  using ConstPointer = itk::SmartPointer<const Self>;
 
  itkNewMacro(Self);
 
  itkTypeMacro(AdaptiveStochasticLBFGS, AdaptiveStochasticLBFGSOptimizer);
 
  elxClassNameMacro("AdaptiveStochasticLBFGS");
 
  using Superclass1::CostFunctionType;
  using Superclass1::CostFunctionPointer;
  using Superclass1::StopConditionType;
 
  using typename Superclass2::ElastixType;
  using typename Superclass2::RegistrationType;
  using ITKBaseType = typename Superclass2::ITKBaseType;
  using SizeValueType = itk::SizeValueType;
 
  //  using LineSearchOptimizerType = LineSearchOptimizer          ;
 
  //  using LineSearchOptimizerPointer = LineSearchOptimizerType::Pointer;
 
  using typename Superclass1::ParametersType;
 
  using LineSearchOptimizerType = itk::LineSearchOptimizer;
 
  using LineSearchOptimizerPointer = LineSearchOptimizerType::Pointer;
  using LineOptimizerType = itk::MoreThuenteLineSearchOptimizer;
  using LineOptimizerPointer = LineOptimizerType::Pointer;
 
  void
  BeforeRegistration() override;
  void
  BeforeEachResolution() override;
  void
  AfterEachResolution() override;
  void
  AfterEachIteration() override;
  void
  AfterRegistration() override;
 
  void
  StartOptimization() override;
 
  virtual void
  LBFGSUpdate();
 
  void
  AdvanceOneStep() override;
 
  void
  ResumeOptimization() override;
 
  void
  MetricErrorResponse(itk::ExceptionObject & err) override;
 
  void
  StopOptimization() override;
 
  itkSetMacro(AutomaticParameterEstimation, bool);
  itkGetConstMacro(AutomaticParameterEstimation, bool);
 
  itkSetMacro(AutomaticLBFGSStepsizeEstimation, bool);
  itkGetConstMacro(AutomaticLBFGSStepsizeEstimation, bool);
 
  itkSetMacro(MaximumStepLength, double);
  itkGetConstMacro(MaximumStepLength, double);
 
  itkSetMacro(MaximumNumberOfSamplingAttempts, SizeValueType);
 
  itkGetConstReferenceMacro(MaximumNumberOfSamplingAttempts, SizeValueType);
 
  using ThreadIdType = unsigned int;
 
  void
  SetNumberOfWorkUnits(ThreadIdType numberOfThreads)
  {
    this->m_Threader->SetNumberOfWorkUnits(numberOfThreads);
  }
 
protected:
  using FixedImageType = typename RegistrationType::FixedImageType;
  using MovingImageType = typename RegistrationType::MovingImageType;
 
  using FixedImageRegionType = typename FixedImageType::RegionType;
  using FixedImageIndexType = typename FixedImageType::IndexType;
  using FixedImagePointType = typename FixedImageType::PointType;
  using itkRegistrationType = typename RegistrationType::ITKBaseType;
  using TransformType = typename itkRegistrationType::TransformType;
  using JacobianType = typename TransformType::JacobianType;
  using ComputeJacobianTermsType = itk::ComputeJacobianTerms<FixedImageType, TransformType>;
  using JacobianValueType = typename JacobianType::ValueType;
  using typename Superclass2::SettingsType;
  using typename Superclass2::SettingsVectorType;
  using ComputeDisplacementDistributionType = itk::ComputeDisplacementDistribution<FixedImageType, TransformType>;
 
  using ImageSamplerBaseType = itk::ImageSamplerBase<FixedImageType>;
  using ImageSamplerBasePointer = typename ImageSamplerBaseType::Pointer;
  using ImageRandomSamplerBaseType = itk::ImageRandomSamplerBase<FixedImageType>;
  using ImageRandomSamplerBasePointer = typename ImageRandomSamplerBaseType::Pointer;
  using ImageRandomCoordinateSamplerType = itk::ImageRandomCoordinateSampler<FixedImageType>;
  using ImageRandomCoordinateSamplerPointer = typename ImageRandomCoordinateSamplerType::Pointer;
  using ImageSampleType = typename ImageSamplerBaseType::ImageSampleType;
 
  using ImageRandomSamplerType = itk::ImageRandomSampler<FixedImageType>;
  using ImageRandomSamplerPointer = typename ImageRandomSamplerType::Pointer;
  using ImageRadomSampleContainerType = typename ImageRandomSamplerType::ImageSampleContainerType;
  using ImageRadomSampleContainerPointer = typename ImageRadomSampleContainerType::Pointer;
 
  using ImageGridSamplerType = itk::ImageGridSampler<FixedImageType>;
  using ImageGridSamplerPointer = typename ImageGridSamplerType::Pointer;
  using ImageSampleContainerType = typename ImageGridSamplerType::ImageSampleContainerType;
  using ImageSampleContainerPointer = typename ImageSampleContainerType::Pointer;
 
  using RandomGeneratorType = itk::Statistics::MersenneTwisterRandomVariateGenerator;
 
  using TransformJacobianType = JacobianType;
  itkStaticConstMacro(FixedImageDimension, unsigned int, FixedImageType::ImageDimension);
  itkStaticConstMacro(MovingImageDimension, unsigned int, MovingImageType::ImageDimension);
  using CoordinateRepresentationType = typename TransformType::ScalarType;
  using AdvancedTransformType =
    itk::AdvancedTransform<CoordinateRepresentationType, Self::FixedImageDimension, Self::MovingImageDimension>;
  using NonZeroJacobianIndicesType = typename AdvancedTransformType::NonZeroJacobianIndicesType;
 
  using RhoType = itk::Array<double>;
  using SType = std::vector<ParametersType>;
  using YType = std::vector<DerivativeType>;
  using DiagonalMatrixType = itk::Array<double>;
 
  AdaptiveStochasticLBFGS();
  ~AdaptiveStochasticLBFGS() override = default;
 
  SettingsVectorType m_SettingsVector;
 
  SizeValueType m_NumberOfGradientMeasurements;
  SizeValueType m_NumberOfJacobianMeasurements;
  SizeValueType m_NumberOfSamplesForExactGradient;
 
  typename AdvancedTransformType::Pointer m_AdvancedTransform;
 
  typename RandomGeneratorType::Pointer m_RandomGenerator;
 
  double m_SigmoidScaleFactor;
 
  virtual void
  AutomaticParameterEstimation();
 
  virtual void
  AutomaticParameterEstimationOriginal();
 
  virtual void
  AutomaticParameterEstimationUsingDisplacementDistribution();
 
  virtual void
  AutomaticLBFGSStepsizeEstimation();
 
  virtual void
  SampleGradients(const ParametersType & mu0, double perturbationSigma, double & gg, double & ee);
 
  virtual void
  GetScaledDerivativeWithExceptionHandling(const ParametersType & parameters, DerivativeType & derivative);
 
  virtual void
  AddRandomPerturbation(ParametersType & parameters, double sigma);
 
  virtual void
  StoreCurrentPoint(const ParametersType & step, const DerivativeType & grad_dif);
  virtual void
  ComputeDiagonalMatrix(DiagonalMatrixType & diag_H0);
 
  virtual void
  ComputeSearchDirection(const DerivativeType & gradient, DerivativeType & searchDir);
 
  DerivativeType m_PreviousCurvatureGradient;
 
  double       m_NoiseFactor;
  unsigned int m_CurrentT;
  unsigned int m_PreviousT;
  unsigned int m_Bound;
 
  RhoType m_Rho;
  SType   m_S;
  YType   m_Y;
  RhoType m_HessianFillValue;
  double  m_WindowScale;
 
private:
  elxOverrideGetSelfMacro;
 
  // multi-threaded AdvanceOneStep:
  struct MultiThreaderParameterType
  {
    ParametersType * t_NewPosition;
    Self *           t_Optimizer;
  };
 
  static itk::ITK_THREAD_RETURN_TYPE
  AdvanceOneStepThreaderCallback(void * arg);
 
  inline void
  ThreadedAdvanceOneStep(ThreadIdType threadId, ParametersType & newPosition);
 
  bool   m_AutomaticParameterEstimation;
  bool   m_AutomaticLBFGSStepsizeEstimation;
  double m_MaximumStepLength;
 
  SizeValueType m_MaximumNumberOfSamplingAttempts;
  SizeValueType m_CurrentNumberOfSamplingAttempts;
  SizeValueType m_PreviousErrorAtIteration;
  bool          m_AutomaticParameterEstimationDone;
 
  SizeValueType m_OutsideIterations;
 
  SizeValueType m_MaxBandCovSize;
  SizeValueType m_NumberOfBandStructureSamples;
  SizeValueType m_NumberOfInnerLoopSamples;
  SizeValueType m_NumberOfSpatialSamples;
 
  bool m_UseNoiseCompensation;
  bool m_OriginalButSigmoidToDefault;
  bool m_UseNoiseFactor;
 
  bool m_UseAdaptiveLBFGSStepSizes;
 
}; // end class AdaptiveStochasticLBFGS
 
 
} // end namespace elastix
 
#ifndef ITK_MANUAL_INSTANTIATION
#  include "elxAdaptiveStochasticLBFGS.hxx"
#endif
 
#endif // end #ifndef elxAdaptiveStochasticLBFGS_h