#include <FcnBeamSpotFitPV.h>

Public Member Functions
	FcnBeamSpotFitPV (const std::vector< BeamSpotFitPVData > &data)
unsigned int	nrOfVerticesUsed () const
double	operator() (const std::vector< double > &) const
void	setLimits (float xmin, float xmax, float ymin, float ymax, float zmin, float zmax)
double	Up () const
	~FcnBeamSpotFitPV ()
Private Attributes
const std::vector < BeamSpotFitPVData > &	data_
double	errorDef_
float	lowerLimits_ [3]
float	upperLimits_ [3]

Detailed Description

LogLH function for 3D beam spot fit from primary vertex data. Assumes a Gaussian resolution function for the vertices provided in the form of a vector of BeamSpotFitPVData and a Gaussian beam profile in 3D: based on a covariance without xz and yz correlations + linearized rotations in the xz and yz planes. Parameters: x, y, z, ex, corrxy, ey, dxdz, dydz, ez, scale x, y, z ...... beamspot position ex, ey, ez ... beamspot width corrxy ....... xy correlation in the system of the beam line dxdz, dydz ... rotations in xz and yz scale ........ scaling factor for vertex errors

Author:: WA, 9/3/2010

Definition at line 22 of file FcnBeamSpotFitPV.h.

Constructor & Destructor Documentation

FcnBeamSpotFitPV::FcnBeamSpotFitPV ( const std::vector< BeamSpotFitPVData > & data )

Definition at line 12 of file FcnBeamSpotFitPV.cc.

References lowerLimits_, and upperLimits_.

                                                                        : 
  data_(data), errorDef_(1.) { 
  //
  // default: no additional cuts
  //
  lowerLimits_[0] = lowerLimits_[1] = lowerLimits_[2] = -1.e30;
  upperLimits_[0] = upperLimits_[1] = upperLimits_[2] =  1.e30;
}

FcnBeamSpotFitPV::~FcnBeamSpotFitPV ( ) [inline]

Definition at line 26 of file FcnBeamSpotFitPV.h.

{}

Member Function Documentation

unsigned int FcnBeamSpotFitPV::nrOfVerticesUsed ( ) const

Definition at line 35 of file FcnBeamSpotFitPV.cc.

References data_, lowerLimits_, and upperLimits_.

{
  //
  // count vertices imposing the current limits
  //
  unsigned int nVtx = 0;
  double v1(0);
  double v2(0);
  double v3(0);
  for ( vector<BeamSpotFitPVData>::const_iterator ipv=data_.begin();
        ipv!=data_.end(); ++ipv ) {
    v1 = (*ipv).position[0];
    if ( v1<lowerLimits_[0] || v1>upperLimits_[0] )  continue;
    v2 = (*ipv).position[1];
    if ( v2<lowerLimits_[1] || v2>upperLimits_[1] )  continue;
    v3 = (*ipv).position[2];
    if ( v3<lowerLimits_[2] || v3>upperLimits_[2] )  continue;

    ++nVtx;
  }

  return nVtx;
}

double FcnBeamSpotFitPV::operator() ( const std::vector< double > & pars ) const

Definition at line 60 of file FcnBeamSpotFitPV.cc.

References gather_cfg::cout, data_, funct::log(), lowerLimits_, and upperLimits_.

{
  //
  // fit parameters
  //
  double vb1 = pars[0];
  double vb2 = pars[1];
  double vb3 = pars[2];
  double sigb1 = pars[3]; 
  double corrb12 = pars[4];
  double sigb2 = pars[5];
  double dxdz = pars[6];
  double dydz = pars[7];
  double sigb3 = pars[8];
  double escale = pars[9];
  //
  // covariance matrix of the beamspot distribution
  //
  typedef ROOT::Math::SVector<double,3> Vector3D;
  typedef ROOT::Math::SMatrix<double,3,3,ROOT::Math::MatRepSym<double,3> > Matrix3D;
  Matrix3D covb;
  double varb1 = sigb1*sigb1;
  double varb2 = sigb2*sigb2;
  double varb3 = sigb3*sigb3;
// parametrisation: rotation (dx/dz, dy/dz); covxy
  covb(0,0) = varb1;
  covb(1,0) = covb(0,1) = corrb12*sigb1*sigb2;
  covb(1,1) = varb2;
  covb(2,0) = covb(0,2) = dxdz*(varb3-varb1)-dydz*covb(1,0);
  covb(2,1) = covb(1,2) = dydz*(varb3-varb2)-dxdz*covb(1,0);
  covb(2,2) = varb3;

  //
  // calculation of the likelihood function
  //
  double sumLL(0);
  double v1(0);
  double v2(0);
  double v3(0);
  double ev1(0);
  double corr12(0);
  double ev2(0);
  double corr13(0);
  double corr23(0);
  double ev3(0);
  //
  // vertex - beamspot difference and total covariance matrix
  //
  Vector3D dv;
  Matrix3D cov;
  Matrix3D wgt;
  //
  // iteration over vertices
  //
  for ( vector<BeamSpotFitPVData>::const_iterator ipv=data_.begin();
        ipv!=data_.end(); ++ipv ) {
    //
    // additional selection
    //
    v1 = (*ipv).position[0];
    if ( v1<lowerLimits_[0] || v1>upperLimits_[0] )  continue;
    v2 = (*ipv).position[1];
    if ( v2<lowerLimits_[1] || v2>upperLimits_[1] )  continue;
    v3 = (*ipv).position[2];
    if ( v3<lowerLimits_[2] || v3>upperLimits_[2] )  continue;
    //
    // vertex errors (after scaling) and correlations
    //
    ev1 = (*ipv).posError[0];
    corr12 = (*ipv).posCorr[0];
    ev2 = (*ipv).posError[1];
    corr13 = (*ipv).posCorr[1];
    corr23 = (*ipv).posCorr[2];
    ev3 = (*ipv).posError[2];
    ev1 *= escale;
    ev2 *= escale;
    ev3 *= escale;
    //
    // vertex covariance matrix
    //
    cov(0,0) = ev1*ev1;
    cov(1,0) = cov(0,1) = ev1*ev2*corr12;
    cov(1,1) = ev2*ev2;
    cov(2,0) = cov(0,2) = ev1*ev3*corr13;
    cov(2,1) = cov(1,2) = ev2*ev3*corr23;
    cov(2,2) = ev3*ev3;
    //
    // total covariance and weight matrix
    //
    cov += covb;
    int ifail;
    wgt = cov.Inverse(ifail);
    if ( ifail ) {
      //edm::LogWarning("FcnBeamSpotFitPV") 
      cout << "Inversion failed" << endl;
      return -1.e30;
    }
    //
    // difference of vertex and beamspot positions
    //
    dv(0) = v1 - vb1;
    dv(1) = v2 - vb2;
    dv(2) = v3 - vb3;
    //
    // exponential term and normalization
    // (neglecting the additional cut)
    //
    sumLL += ROOT::Math::Similarity(dv,wgt);
    double det;
    cov.Det2(det);
    sumLL += log(det);
  }

  return sumLL;
}

void FcnBeamSpotFitPV::setLimits	(	float	xmin,
		float	xmax,
		float	ymin,
		float	ymax,
		float	zmin,
		float	zmax
	)

Definition at line 22 of file FcnBeamSpotFitPV.cc.

References lowerLimits_, and upperLimits_.

Referenced by PVFitter::runBXFitter(), and PVFitter::runFitter().

{
  lowerLimits_[0] = xmin;
  lowerLimits_[1] = ymin;
  lowerLimits_[2] = zmin;
  upperLimits_[0] = xmax;
  upperLimits_[1] = ymax;
  upperLimits_[2] = zmax;
}

double FcnBeamSpotFitPV::Up ( ) const [inline]

Definition at line 32 of file FcnBeamSpotFitPV.h.

References errorDef_.