MnMachinePrecision.h

Go to the documentation of this file.

// @(#)root/minuit2:$Id$
// Authors: M. Winkler, F. James, L. Moneta, A. Zsenei   2003-2005
 
/**********************************************************************
 *                                                                    *
 * Copyright (c) 2005 LCG ROOT Math team,  CERN/PH-SFT                *
 *                                                                    *
 **********************************************************************/
 
#ifndef ROOT_Minuit2_MnMachinePrecision
#define ROOT_Minuit2_MnMachinePrecision
 
#include <math.h>
 
namespace ROOT {
 
   namespace Minuit2 {
 
 
/**
    Sets the relative floating point (double) arithmetic precision.
    By default the precision values are obtained from the standard functions
    std::numeric_limits<double>::epsilon.
    The values can optionally be computed directly using the ComputePrecision()
    member function. For a IEEE standard floating point arithmetic the computed values and
    the one from std::numeric_limits<double>::epsilon  are the same.
 
    SetPrecision() method can instead be used to override Minuit's own determination,
    when the user knows that the {FCN} function Value is not calculated to
    the nominal machine accuracy.
 */
 
class MnMachinePrecision {
 
public:
 
  MnMachinePrecision();
 
 
  /// eps returns the smallest possible number so that 1.+eps > 1.
  double Eps() const {return fEpsMac;}
 
  /// eps2 returns 2*sqrt(eps)
  double Eps2() const {return fEpsMa2;}
 
  /// override Minuit's own determination
  void SetPrecision(double prec) {
    fEpsMac = prec;
    fEpsMa2 = 2.*sqrt(fEpsMac);
  }
 
   /// compute Machine precision directly instead
   /// of using default values from std::numeric_limits
   void ComputePrecision(); 
 
private:
 
  double fEpsMac;
  double fEpsMa2;
};
 
  }  // namespace Minuit2
 
}  // namespace ROOT
 
#endif  // ROOT_Minuit2_MnMachinePrecision