KalmanCalculator Class Reference

does the calculation of the gain matrix, updates the cov and fitpars More...

#include <KalmanCalculator.h>

Public Member Functions
	KalmanCalculator (int sizeRes, int sizeState)
	constructor
void	updateState (FitParams &fitparams)
	update statevector
void	updateState (FitParams &fitparams, FitParams &oldState)
	update statevector
void	updateCovariance (FitParams &fitparams)
	update the statevectors covariance
double	getChiSquare () const
	get chi2 of this iteration
ErrCode	calculateGainMatrix (const Eigen::Matrix< double, -1, 1, 0, 7, 1 > &residuals, const Eigen::Matrix< double, -1, -1, 0, 7, MAX_MATRIX_SIZE > &G, const FitParams &fitparams, const Eigen::Matrix< double, -1, -1, 0, 7, 7 > *V=0, double weight=1)
	init the kalman machienery
double	getConstraintDim () const
	get dimension of the constraint

Private Attributes
int	m_constrDim
	dimension of the constraint
double	m_chisq
	chi2
Eigen::Matrix< double, -1, 1, 0, 7, 1 >	m_res
	we know the max sizes of the matrices we assume the tree is smaller than MAX_MATRIX_SIZE parameters and the largest constraint is the track constraint with 7
Eigen::Matrix< double, -1, -1, 0, 7, MAX_MATRIX_SIZE >	m_G
	G former H, transforms covraince of {residuals}<->{x,p,E}.
Eigen::Matrix< double, -1, -1, 0, 7, 7 >	m_R
	R residual covariance.
Eigen::Matrix< double, -1, -1, 0, 7, 7 >	m_Rinverse
	R inverse.
Eigen::Matrix< double, -1, -1, 0, MAX_MATRIX_SIZE, 7 >	m_K
	K kalman gain matrix.
Eigen::Matrix< double, -1, -1, 0, MAX_MATRIX_SIZE, 7 >	m_CGt
	C times G^t

Detailed Description

does the calculation of the gain matrix, updates the cov and fitpars

Definition at line 23 of file KalmanCalculator.h.

Constructor & Destructor Documentation

KalmanCalculator()

KalmanCalculator	(	int	sizeRes,
		int	sizeState
	)

constructor

Definition at line 14 of file KalmanCalculator.cc.

    :
    m_constrDim(sizeRes),
    m_chisq(1e10),
    m_res(sizeRes),
    m_G(sizeRes, sizeState),
    m_R(sizeRes, sizeRes),
    m_Rinverse(sizeRes, sizeRes),
    m_K(sizeState, sizeRes),
    m_CGt(sizeState, sizeRes)
  {
    m_R = Eigen::Matrix < double, -1, -1, 0, 7, 7 >::Zero(m_constrDim, m_constrDim);
  }

Member Function Documentation

calculateGainMatrix()

ErrCode calculateGainMatrix	(	const Eigen::Matrix< double, -1, 1, 0, 7, 1 > &	residuals,
		const Eigen::Matrix< double, -1, -1, 0, 7, MAX_MATRIX_SIZE > &	G,
		const FitParams &	fitparams,
		const Eigen::Matrix< double, -1, -1, 0, 7, 7 > *	V = 0,
		double	weight = 1
	)

init the kalman machienery

Definition at line 31 of file KalmanCalculator.cc.

  {
    m_res = residuals;
    m_G = G;
 
    Eigen::Matrix < double, -1, -1, 0, MAX_MATRIX_SIZE, MAX_MATRIX_SIZE > C = fitparams.getCovariance().triangularView<Eigen::Lower>();
 
    m_CGt = C.selfadjointView<Eigen::Lower>() * G.transpose();
    Eigen::Matrix < double, -1, -1, 0, 7, 7 > Rtemp = G * m_CGt;
    if (V && (weight) && ((*V).diagonal().array() != 0).all()) {
 
      const Eigen::Matrix < double, -1, -1, 0, 7, 7 > weightedV  =
        weight * (*V).selfadjointView<Eigen::Lower>();
 
      m_R = Rtemp + weightedV;
 
    } else {
      m_R = Rtemp.triangularView<Eigen::Lower>();
    }
 
    Eigen::Matrix < double, -1, -1, 0, 7, 7 > RInvtemp;
    RInvtemp = m_R.selfadjointView<Eigen::Lower>();
    m_Rinverse = RInvtemp.inverse();
    if (!m_Rinverse.allFinite()) { return ErrCode(ErrCode::Status::inversionerror); }
 
    m_K = m_CGt * m_Rinverse.selfadjointView<Eigen::Lower>();
    return ErrCode(ErrCode::Status::success);
  }

getChiSquare()

double getChiSquare ( ) const

inline

get chi2 of this iteration

Definition at line 42 of file KalmanCalculator.h.

{ return m_chisq;}

getConstraintDim()

double getConstraintDim ( ) const

inline

get dimension of the constraint

Definition at line 53 of file KalmanCalculator.h.

{ return m_constrDim; }

updateCovariance()

TREEFITTER_NO_STACK_WARNING void updateCovariance

(

FitParams &

fitparams

)

update the statevectors covariance

Definition at line 87 of file KalmanCalculator.cc.

  {
    Eigen::Matrix < double, -1, -1, 0, MAX_MATRIX_SIZE, MAX_MATRIX_SIZE > fitCov  =
      fitparams.getCovariance().triangularView<Eigen::Lower>();
 
    Eigen::Matrix < double, -1, -1, 0, MAX_MATRIX_SIZE, MAX_MATRIX_SIZE > GRinvGt =
      m_G.transpose() * m_Rinverse.selfadjointView<Eigen::Lower>() * m_G;
 
    //fitcov is sym so no transpose needed (not that it would have runtime cost)
    Eigen::Matrix < double, -1, -1, 0, MAX_MATRIX_SIZE, MAX_MATRIX_SIZE > deltaCov  =
      fitCov.selfadjointView<Eigen::Lower>() * GRinvGt * fitCov.selfadjointView<Eigen::Lower>();
 
    double eps = Eigen::NumTraits<double>::epsilon();
    Eigen::Matrix < double, -1, -1, 0, MAX_MATRIX_SIZE, MAX_MATRIX_SIZE > delta =
      (fitCov.array() - deltaCov.array()).matrix() + eps * Eigen::MatrixXd::Identity(fitCov.rows(), fitCov.cols());
 
    fitparams.getCovariance().triangularView<Eigen::Lower>() = delta.triangularView<Eigen::Lower>();
 
  }//end function

updateState() [1/2]

void updateState

(

FitParams &

fitparams

)

update statevector

Definition at line 65 of file KalmanCalculator.cc.

  {
    double eps = Eigen::NumTraits<double>::epsilon();
    fitparams.getStateVector() = (fitparams.getStateVector().array() - (m_K * m_res).array()).matrix()
                                 + eps * Eigen::MatrixXd::Identity(fitparams.getStateVector().rows(), fitparams.getStateVector().cols());
    m_chisq = m_res.transpose() * m_Rinverse.selfadjointView<Eigen::Lower>() * m_res;
  }

updateState() [2/2]

void updateState	(	FitParams &	fitparams,
		FitParams &	oldState
	)

update statevector

Definition at line 73 of file KalmanCalculator.cc.

  {
    double eps = Eigen::NumTraits<double>::epsilon();
    Eigen::Matrix < double, -1, 1, 0, 7, 1 > res_prime =
      m_res + m_G * ((oldState.getStateVector().array() - fitparams.getStateVector().array()).matrix() +
                     + eps * Eigen::MatrixXd::Identity(fitparams.getStateVector().rows(), fitparams.getStateVector().cols()));
    fitparams.getStateVector() = (oldState.getStateVector().array() - (m_K * res_prime).array()).matrix()
                                 + eps * Eigen::MatrixXd::Identity(oldState.getStateVector().rows(), oldState.getStateVector().cols());
 
    m_chisq = res_prime.transpose() * m_Rinverse.selfadjointView<Eigen::Lower>() * res_prime;
  }

Member Data Documentation

m_CGt

Eigen::Matrix< double, -1, -1, 0, MAX_MATRIX_SIZE, 7 > m_CGt

private

C times G^t

Definition at line 89 of file KalmanCalculator.h.

m_chisq

double m_chisq

private

chi2

Definition at line 60 of file KalmanCalculator.h.

m_constrDim

int m_constrDim

private

dimension of the constraint

Definition at line 57 of file KalmanCalculator.h.

m_G

Eigen::Matrix< double, -1, -1, 0, 7, MAX_MATRIX_SIZE > m_G

private

G former H, transforms covraince of {residuals}<->{x,p,E}.

Definition at line 77 of file KalmanCalculator.h.

m_K

Eigen::Matrix< double, -1, -1, 0, MAX_MATRIX_SIZE, 7 > m_K

private

K kalman gain matrix.

Definition at line 86 of file KalmanCalculator.h.

m_R

Eigen::Matrix< double, -1, -1, 0, 7, 7 > m_R

private

R residual covariance.

Definition at line 80 of file KalmanCalculator.h.

m_res

Eigen::Matrix< double, -1, 1, 0, 7, 1 > m_res

private

we know the max sizes of the matrices we assume the tree is smaller than MAX_MATRIX_SIZE parameters and the largest constraint is the track constraint with 7

-> Eigen puts this on the stack

Eigen::Matrix < double, col, row, ColMajor, maxCol, maxRow> -1 = Eigen::Dynamic 0 = Eigen::ColMajor (don't touch) vector holding the residuals

Definition at line 74 of file KalmanCalculator.h.

m_Rinverse

Eigen::Matrix< double, -1, -1, 0, 7, 7 > m_Rinverse

private

R inverse.

Definition at line 83 of file KalmanCalculator.h.

---

The documentation for this class was generated from the following files:

• analysis/VertexFitting/TreeFitter/include/KalmanCalculator.h
• analysis/VertexFitting/TreeFitter/src/KalmanCalculator.cc