FuncMinimum.h

Go to the documentation of this file.

#ifndef ALGTOOLS_FUNCMINIMUM_H
#define ALGTOOLS_FUNCMINIMUM_H 1
// Include files
// from CLHEP
#include "CLHEP/GenericFunctions/AbsFunction.hh"
#include "CLHEP/GenericFunctions/Argument.hh"
#include "CLHEP/GenericFunctions/GenericFunctions.hh"
#include "CLHEP/Matrix/SymMatrix.h"
// from Gaudi
#include "GaudiAlg/GaudiTool.h"
// from GSL
#include "gsl/gsl_multimin.h"

// local
#include "GaudiGSL/IFuncMinimum.h"

class FuncMinimum : public extends<GaudiTool, IFuncMinimum>
{
public:
  typedef std::vector<const GenFunc*> Gradient;

public:
  StatusCode minimum( const GenFunc& func, Arg& arg ) const override;

  StatusCode minimum( const GenFunc& func, Arg& arg, Covariance& covar ) const override;

  StatusCode initialize() override;

  class FuncMinimumMisc final
  {
  public:
    FuncMinimumMisc( const GenFunc& func, Arg& arg );

  public:
    inline const Arg& argument() const { return m_argum; }
    inline Arg& argument() { return m_argum; }
    inline const GenFunc* equation() const { return m_eq; }
    inline const Gradient& gradient() const { return m_grad; }

  private:
    // default constructor is disabled
    FuncMinimumMisc() = delete;
    // copy constructor is disabled
    FuncMinimumMisc( const FuncMinimumMisc& ) = delete;
    // assignment operator is disabled
    FuncMinimumMisc& operator=( const FuncMinimumMisc& ) = delete;

  private:
    Arg m_argum;
    const GenFunc* m_eq;
    Gradient m_grad;
  };

  using extends::extends;

private:
  FuncMinimum() = delete;
  FuncMinimum( const FuncMinimum& ) = delete;
  FuncMinimum& operator=( const FuncMinimum& ) = delete;

private:
  Gaudi::Property<std::string> m_algType{this, "Algorithm", "conjugate_fr",
                                         "type of the algorithm for multidimensional minimization"};
  Gaudi::Property<double> m_max_iter{this, "Iteration", 200, "maximum of iteration"};
  Gaudi::Property<double> m_norm_gradient{this, "Gradient", 1.0e-10,
                                          "absolute tolerance for the Euclidean norm of the gradient"};
  Gaudi::Property<double> m_step_size{this, "Step_size", 0.01, "size of the first trial step"};
  Gaudi::Property<double> m_tol{this, "Tol", 1e-10, "accuracy of the line minimization"};

  const gsl_multimin_fdfminimizer_type* m_type = nullptr;
};

// ============================================================================
// The END
// ============================================================================
#endif // ALGTOOLS_FUNCMINIMUM_H
// ============================================================================