KalmanStepper< Dimension > Class Template Reference

Class to bundle all algorithms needed for the kalman update procedure. More...

#include <KalmanStepper.h>

Public Member Functions
double	kalmanStep (genfit::MeasuredStateOnPlane &measuredStateOnPlane, const genfit::MeasurementOnPlane &measurementOnPlane) const
	Kalman update of the mSoP using the measurement. Is just a wrapper around the other kalmanStepper working with bare matrices.
double	calculateResidual (const genfit::MeasuredStateOnPlane &measuredStateOnPlane, const genfit::MeasurementOnPlane &measurementOnPlane) const
	Helper function to calculate a residual between the mSoP and the measurement.

Private Types
using	MeasurementState = Eigen::Matrix< double, Dimension, 1 >
	Matrix class Dimension x 1.
using	MeasurementCovariance = Eigen::Matrix< double, Dimension, Dimension >
	Matrix class Dimension x Dimension.
using	HMatrix = Eigen::Matrix< double, Dimension, 5 >
	Matrix class Dimension x 5.
using	TrackState = Eigen::Matrix< double, 5, 1 >
	Matrix class 5 x 1.
using	TrackCovariance = Eigen::Matrix< double, 5, 5 >
	Matrix class 5 x 5.

Private Member Functions
double	kalmanStep (TrackState &x_k, TrackCovariance &C_k, const MeasurementState &m_k, const MeasurementCovariance &V_k, const HMatrix &H_k) const
	This now is the real "update" step, where we update the X_k and the C_k.

Detailed Description

template<unsigned int Dimension>
class Belle2::KalmanStepper< Dimension >

Class to bundle all algorithms needed for the kalman update procedure.

Its main functionality is to update a measured state on plane with a measurement on plane from a hit using the Kalman procedure described in https://doi.org/10.1016/0168-9002(87)90887-4.

Template Parameters

Dimension

The dimension of the hit - e.g. how many Kalman state parameters are defined by the hits. This defines the size of the matrices.

Definition at line 34 of file KalmanStepper.h.

Member Typedef Documentation

HMatrix

using HMatrix = Eigen::Matrix<double, Dimension, 5>

private

Matrix class Dimension x 5.

Definition at line 40 of file KalmanStepper.h.

MeasurementCovariance

using MeasurementCovariance = Eigen::Matrix<double, Dimension, Dimension>

private

Matrix class Dimension x Dimension.

Definition at line 38 of file KalmanStepper.h.

MeasurementState

using MeasurementState = Eigen::Matrix<double, Dimension, 1>

private

Matrix class Dimension x 1.

Definition at line 36 of file KalmanStepper.h.

TrackCovariance

using TrackCovariance = Eigen::Matrix<double, 5, 5>

private

Matrix class 5 x 5.

Definition at line 44 of file KalmanStepper.h.

TrackState

using TrackState = Eigen::Matrix<double, 5, 1>

private

Matrix class 5 x 1.

Definition at line 42 of file KalmanStepper.h.

Member Function Documentation

calculateResidual()

double calculateResidual ( const genfit::MeasuredStateOnPlane &  measuredStateOnPlane, const genfit::MeasurementOnPlane &  measurementOnPlane  ) const

inline

Helper function to calculate a residual between the mSoP and the measurement.

Definition at line 70 of file KalmanStepper.h.

    {
      // Extract matrices from the state
      const TrackState& x_k = convertToEigen<5>(measuredStateOnPlane.getState());
 
      // extract matrices from the measurement
      const MeasurementState& m_k = convertToEigen<Dimension>(measurementOnPlane.getState());
      const HMatrix& H_k = convertToEigen<Dimension, 5>(measurementOnPlane.getHMatrix()->getMatrix());
 
      const MeasurementState& residual = m_k - H_k * x_k;
      return residual.norm();
    }

kalmanStep() [1/2]

double kalmanStep ( genfit::MeasuredStateOnPlane &  measuredStateOnPlane, const genfit::MeasurementOnPlane &  measurementOnPlane  ) const

inline

Kalman update of the mSoP using the measurement. Is just a wrapper around the other kalmanStepper working with bare matrices.

Definition at line 48 of file KalmanStepper.h.

    {
      // Extract matrices from the state
      TrackState x_k = convertToEigen<5>(measuredStateOnPlane.getState());
      TrackCovariance C_k = convertToEigen<5, 5>(measuredStateOnPlane.getCov());
 
      // extract matrices from the measurement
      const MeasurementState& m_k = convertToEigen<Dimension>(measurementOnPlane.getState());
      const HMatrix& H_k = convertToEigen<Dimension, 5>(measurementOnPlane.getHMatrix()->getMatrix());
      const MeasurementCovariance& V_k = convertToEigen<Dimension, Dimension>(measurementOnPlane.getCov());
 
      const double chi2 = kalmanStep(x_k, C_k, m_k, V_k, H_k);
 
      // set the state back
      measuredStateOnPlane.setState(TVectorD(5, x_k.data()));
      measuredStateOnPlane.setCov(TMatrixDSym(5, C_k.data()));
 
      return chi2;
    }

kalmanStep() [2/2]

double kalmanStep ( TrackState &  x_k, TrackCovariance &  C_k, const MeasurementState &  m_k, const MeasurementCovariance &  V_k, const HMatrix &  H_k  ) const

inlineprivate

This now is the real "update" step, where we update the X_k and the C_k.

Definition at line 86 of file KalmanStepper.h.

    {
      const Eigen::Matrix<double, 5, Dimension>& K_k = C_k * H_k.transpose() * (V_k + H_k * C_k * H_k.transpose()).inverse();
      C_k -= K_k * H_k * C_k;
      x_k += K_k * (m_k - H_k * x_k);
 
      const MeasurementState& residual = m_k - H_k * x_k;
      const double chi2 = (residual.transpose() * (V_k - H_k * C_k * H_k.transpose()).inverse() * residual).value();
      return chi2;
    }

---

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

• tracking/ckf/general/utilities/include/KalmanStepper.h