| 1 |
#include "TrackingTools/PatternTools/interface/CollinearFitAtTM.h"
|
| 2 |
#include "FWCore/MessageLogger/interface/MessageLogger.h"
|
| 3 |
|
| 4 |
CollinearFitAtTM2::CollinearFitAtTM2 (const TrajectoryMeasurement& tm) :
|
| 5 |
valid_(false),chi2_(-1),ndof_(-1) {
|
| 6 |
//
|
| 7 |
// check input
|
| 8 |
//
|
| 9 |
if ( !tm.forwardPredictedState().isValid() ||
|
| 10 |
!tm.backwardPredictedState().isValid() ||
|
| 11 |
!tm.updatedState().isValid() ) {
|
| 12 |
edm::LogWarning("CollinearFitAtTM2") << "Invalid state in TrajectoryMeasurement";
|
| 13 |
return;
|
| 14 |
}
|
| 15 |
//
|
| 16 |
// prepare fit
|
| 17 |
//
|
| 18 |
initJacobian();
|
| 19 |
AlgebraicVector5 fwdPar = tm.forwardPredictedState().localParameters().vector();
|
| 20 |
AlgebraicSymMatrix55 fwdCov = tm.forwardPredictedState().localError().matrix();
|
| 21 |
AlgebraicVector5 bwdPar = tm.backwardPredictedState().localParameters().vector();
|
| 22 |
AlgebraicSymMatrix55 bwdCov = tm.backwardPredictedState().localError().matrix();
|
| 23 |
|
| 24 |
LocalPoint hitPos(0.,0.,0.);
|
| 25 |
LocalError hitErr(-1.,-1.,-1.);
|
| 26 |
if ( tm.recHit()->isValid() ) {
|
| 27 |
hitPos = tm.recHit()->localPosition();
|
| 28 |
hitErr = tm.recHit()->localPositionError();
|
| 29 |
}
|
| 30 |
|
| 31 |
fit(fwdPar,fwdCov,bwdPar,bwdCov,hitPos,hitErr);
|
| 32 |
}
|
| 33 |
|
| 34 |
CollinearFitAtTM2::CollinearFitAtTM2 (const AlgebraicVector5& fwdParameters,
|
| 35 |
const AlgebraicSymMatrix55& fwdCovariance,
|
| 36 |
const AlgebraicVector5& bwdParameters,
|
| 37 |
const AlgebraicSymMatrix55& bwdCovariance,
|
| 38 |
const LocalPoint& hitPosition,
|
| 39 |
const LocalError& hitErrors) :
|
| 40 |
valid_(false),chi2_(-1),ndof_(-1) {
|
| 41 |
//
|
| 42 |
// prepare fit
|
| 43 |
//
|
| 44 |
initJacobian();
|
| 45 |
|
| 46 |
fit(fwdParameters,fwdCovariance,bwdParameters,bwdCovariance,hitPosition,hitErrors);
|
| 47 |
}
|
| 48 |
|
| 49 |
void
|
| 50 |
CollinearFitAtTM2::initJacobian ()
|
| 51 |
{
|
| 52 |
//
|
| 53 |
// Jacobian
|
| 54 |
//
|
| 55 |
for ( int i=0; i<12; ++i ) {
|
| 56 |
for ( int j=0; j<6; ++j ) jacobian_(i,j) = 0;
|
| 57 |
}
|
| 58 |
for ( int i=1; i<5; ++i ) {
|
| 59 |
jacobian_(i,ParQpOut+i) = jacobian_(i+5,ParQpOut+i) = 1;
|
| 60 |
}
|
| 61 |
jacobian_(0,ParQpIn) = 1.;
|
| 62 |
jacobian_(5,ParQpOut) = 1.;
|
| 63 |
}
|
| 64 |
|
| 65 |
bool
|
| 66 |
CollinearFitAtTM2::fit (const AlgebraicVector5& fwdParameters,
|
| 67 |
const AlgebraicSymMatrix55& fwdCovariance,
|
| 68 |
const AlgebraicVector5& bwdParameters,
|
| 69 |
const AlgebraicSymMatrix55& bwdCovariance,
|
| 70 |
const LocalPoint& hitPos, const LocalError& hitErr)
|
| 71 |
{
|
| 72 |
|
| 73 |
if ( hitErr.xx()>0 )
|
| 74 |
jacobian_(10,ParX) = jacobian_(11,ParY) = 1;
|
| 75 |
else
|
| 76 |
jacobian_(10,ParX) = jacobian_(11,ParY) = 0;
|
| 77 |
|
| 78 |
for ( int i=0; i<12; ++i ) {
|
| 79 |
for ( int j=0; j<12; ++j ) weightMatrix_(i,j) = 0;
|
| 80 |
}
|
| 81 |
|
| 82 |
for ( int i=0; i<5; ++i ) measurements_(i) = fwdParameters(i);
|
| 83 |
weightMatrix_.Place_at(fwdCovariance,0,0);
|
| 84 |
for ( int i=0; i<5; ++i ) measurements_(i+5) = bwdParameters(i);
|
| 85 |
weightMatrix_.Place_at(bwdCovariance,5,5);
|
| 86 |
if ( hitErr.xx()>0 ) {
|
| 87 |
measurements_(10) = hitPos.x();
|
| 88 |
measurements_(11) = hitPos.y();
|
| 89 |
weightMatrix_(10,10) = hitErr.xx();
|
| 90 |
weightMatrix_(10,11) = weightMatrix_(11,10) = hitErr.xy();
|
| 91 |
weightMatrix_(11,11) = hitErr.yy();
|
| 92 |
}
|
| 93 |
else {
|
| 94 |
measurements_(10) = measurements_(11) = 0.;
|
| 95 |
weightMatrix_(10,10) = weightMatrix_(11,11) = 1.;
|
| 96 |
weightMatrix_(10,11) = weightMatrix_(11,10) = 0.;
|
| 97 |
}
|
| 98 |
//
|
| 99 |
// invert covariance matrix
|
| 100 |
//
|
| 101 |
if ( !weightMatrix_.Invert() ) {
|
| 102 |
edm::LogWarning("CollinearFitAtTM2") << "Inversion of input covariance matrix failed";
|
| 103 |
return false;
|
| 104 |
}
|
| 105 |
|
| 106 |
//
|
| 107 |
projectedMeasurements_ = ROOT::Math::Transpose(jacobian_)*(weightMatrix_*measurements_);
|
| 108 |
//
|
| 109 |
// Fitted parameters and covariance matrix
|
| 110 |
//
|
| 111 |
covariance_ = ROOT::Math::SimilarityT(jacobian_,weightMatrix_);
|
| 112 |
if ( !covariance_.Invert() ) {
|
| 113 |
edm::LogWarning("CollinearFitAtTM2") << "Inversion of resulting weight matrix failed";
|
| 114 |
return false;
|
| 115 |
}
|
| 116 |
|
| 117 |
parameters_ = covariance_*projectedMeasurements_;
|
| 118 |
|
| 119 |
//
|
| 120 |
// chi2
|
| 121 |
//
|
| 122 |
chi2_ = ROOT::Math::Similarity(measurements_,weightMatrix_) -
|
| 123 |
ROOT::Math::Similarity(projectedMeasurements_,covariance_);
|
| 124 |
ndof_ = hitErr.xx()>0 ? 6 : 4;
|
| 125 |
|
| 126 |
valid_ = true;
|
| 127 |
return true;
|
| 128 |
}
|
| 129 |
|
| 130 |
Measurement1D
|
| 131 |
CollinearFitAtTM2::deltaP () const {
|
| 132 |
//
|
| 133 |
// check validity
|
| 134 |
//
|
| 135 |
if ( !valid_ ) return Measurement1D();
|
| 136 |
//
|
| 137 |
// deltaP = 1/qpout - 1/qpin ; uncertainty from linear error propagation
|
| 138 |
//
|
| 139 |
double qpIn = parameters_(0);
|
| 140 |
double sig2In = covariance_(0,0);
|
| 141 |
double qpOut = parameters_(1);
|
| 142 |
double sig2Out = covariance_(1,1);
|
| 143 |
double corrInOut = covariance_(0,1);
|
| 144 |
double pIn = 1./fabs(qpIn);
|
| 145 |
double pOut = 1./fabs(qpOut);
|
| 146 |
double sig2DeltaP = pIn/qpIn*pIn/qpIn*sig2In - 2*pIn/qpIn*pOut/qpOut*corrInOut +
|
| 147 |
pOut/qpOut*pOut/qpOut*sig2Out;
|
| 148 |
|
| 149 |
return Measurement1D(pOut-pIn,sig2DeltaP?sqrt(sig2DeltaP):0.);
|
| 150 |
}
|
| 151 |
|
