/*=========================================================================
 *
 *  Copyright NumFOCUS
 *
 *  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
 *
 *         https://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 itkAmoebaOptimizer_h
#define itkAmoebaOptimizer_h

#include "itkSingleValuedNonLinearVnlOptimizer.h"
#include "vnl/algo/vnl_amoeba.h"
#include "ITKOptimizersExport.h"
#include <memory> // For unique_ptr.

namespace itk
{
/** \class AmoebaOptimizer
 * \brief Wrap of the vnl_amoeba algorithm
 *
 * AmoebaOptimizer is a wrapper around the vnl_amoeba algorithm which
 * is an implementation of the Nelder-Meade downhill simplex
 * problem. For most problems, it is a few times slower than a
 * Levenberg-Marquardt algorithm but does not require derivatives of
 * its cost function. It works by creating a simplex (n+1 points in
 * ND space). The cost function is evaluated at each corner of the
 * simplex.  The simplex is then modified (by reflecting a corner
 * about the opposite edge, by shrinking the entire simplex, by
 * contracting one edge of the simplex, or by expanding the simplex)
 * in searching for the minimum of the cost function.
 *
 * The methods AutomaticInitialSimplex() and SetInitialSimplexDelta()
 * control whether the optimizer defines the initial simplex
 * automatically (by constructing a very small simplex around the
 * initial position) or uses a user supplied simplex size.
 *
 * The method SetOptimizeWithRestarts() indicates that the amoeba algorithm
 * should be rerun after if converges. This heuristic increases the chances
 * of escaping from a local optimum. Each time the simplex is initialized with
 * the best solution obtained by the previous runs. The edge length is half of
 * that from the previous iteration. The heuristic is terminated if the total
 * number of iterations is greater-equal than the maximal number of iterations
 * (SetMaximumNumberOfIterations) or the difference between the current function
 * value and the best function value is less than a threshold
 * (SetFunctionConvergenceTolerance) and
 * max(|best_parameters_i - current_parameters_i|) is less than a threshold
 * (SetParametersConvergenceTolerance).
 *
 *
 * \ingroup Numerics Optimizers
 * \ingroup ITKOptimizers
 *
 * \sphinx
 * \sphinxexample{Numerics/Optimizers/AmoebaOptimizer, Amoeba Optimization}
 * \endsphinx
 */
class ITKOptimizers_EXPORT AmoebaOptimizer : public SingleValuedNonLinearVnlOptimizer
{
public:
  ITK_DISALLOW_COPY_AND_MOVE(AmoebaOptimizer);

  /** Standard "Self" type alias. */
  using Self = AmoebaOptimizer;
  using Superclass = SingleValuedNonLinearVnlOptimizer;
  using Pointer = SmartPointer<Self>;
  using ConstPointer = SmartPointer<const Self>;
  using NumberOfIterationsType = unsigned int;

  /** Method for creation through the object factory. */
  itkNewMacro(Self);

  /** \see LightObject::GetNameOfClass() */
  itkOverrideGetNameOfClassMacro(AmoebaOptimizer);

  /**  Parameters type.
   *  It defines a position in the optimization search space. */
  using ParametersType = Superclass::ParametersType;

  /** InternalParameters type alias. */
  using InternalParametersType = vnl_vector<double>;

  /** Start optimization with an initial value. */
  void
  StartOptimization() override;

  /** Plug in a Cost Function into the optimizer  */
  void
  SetCostFunction(SingleValuedCostFunction * costFunction) override;

  /** Set/Get the maximum number of iterations. The optimization algorithm will
   * terminate after the maximum number of iterations has been reached.
   * The default value is defined as DEFAULT_MAXIMAL_NUMBER_OF_ITERATIONS. */
  itkSetMacro(MaximumNumberOfIterations, NumberOfIterationsType);
  itkGetConstMacro(MaximumNumberOfIterations, NumberOfIterationsType);

  /** Set/Get the mode which determines how the amoeba algorithm
   * defines the initial simplex.  Default is
   * AutomaticInitialSimplexOn. If AutomaticInitialSimplex is on, the
   * initial simplex is created with a default size. If
   * AutomaticInitialSimplex is off, then InitialSimplexDelta will be
   * used to define the initial simplex, setting the ith corner of the
   * simplex as [x0[0], x0[1], ..., x0[i]+InitialSimplexDelta[i], ...,
   * x0[d-1]]. */
  itkSetMacro(AutomaticInitialSimplex, bool);
  itkBooleanMacro(AutomaticInitialSimplex);
  itkGetConstMacro(AutomaticInitialSimplex, bool);

  /** Set/Get the mode that determines if we want to use multiple runs of the
   * Amoeba optimizer. If true, then the optimizer is rerun after it converges.
   * The additional runs are performed using a simplex initialized with the
   * best solution obtained by the previous runs. The edge length is half of
   * that from the previous iteration.
   */
  itkSetMacro(OptimizeWithRestarts, bool);
  itkBooleanMacro(OptimizeWithRestarts);
  itkGetConstMacro(OptimizeWithRestarts, bool);

  /** Set/Get the deltas that are used to define the initial simplex
   * when AutomaticInitialSimplex is off. */
  void
  SetInitialSimplexDelta(ParametersType initialSimplexDelta, bool automaticInitialSimplex = false);
  itkGetConstMacro(InitialSimplexDelta, ParametersType);

  /** The optimization algorithm will terminate when the simplex
   * diameter and the difference in cost function values at the corners of
   * the simplex falls below user specified thresholds. The simplex
   * diameter threshold is set via SetParametersConvergenceTolerance().*/
  itkSetMacro(ParametersConvergenceTolerance, double);
  itkGetConstMacro(ParametersConvergenceTolerance, double);

  /** The optimization algorithm will terminate when the simplex
   * diameter and the difference in cost function values at the corners of
   * the simplex falls below user specified thresholds. The cost function
   * convergence threshold is set via SetFunctionConvergenceTolerance().*/
  itkSetMacro(FunctionConvergenceTolerance, double);
  itkGetConstMacro(FunctionConvergenceTolerance, double);

  /** Report the reason for stopping. */
  const std::string
  GetStopConditionDescription() const override;

  /** Return Current Value */
  MeasureType
  GetValue() const;

  /** Method for getting access to the internal optimizer. */
  vnl_amoeba *
  GetOptimizer() const;

protected:
  AmoebaOptimizer();
  ~AmoebaOptimizer() override;
  void
  PrintSelf(std::ostream & os, Indent indent) const override;

  using CostFunctionAdaptorType = Superclass::CostFunctionAdaptorType;

private:
  /**Check that the settings are valid. If not throw an exception.*/
  void
  ValidateSettings();

  NumberOfIterationsType        m_MaximumNumberOfIterations{};
  ParametersType::ValueType     m_ParametersConvergenceTolerance{};
  CostFunctionType::MeasureType m_FunctionConvergenceTolerance{};
  bool                          m_AutomaticInitialSimplex{};
  ParametersType                m_InitialSimplexDelta{};
  bool                          m_OptimizeWithRestarts{};
  std::unique_ptr<vnl_amoeba>   m_VnlOptimizer;

  std::ostringstream m_StopConditionDescription{};
};
} // end namespace itk

#endif