KalmanFilterTracking Class Reference

Forward Kalman filter tracking algorithm, without smoothing step. More...

#include <KalmanFilterTracking.h>

List of all members.

Public Member Functions
const CLHEP::HepMatrix &	Jacobian_Track (const CLHEP::HepVector &p0) const
	Jacobian matrix, derivatives of measurement equation to track parameters.
Track	MakeRecoTrack (std::vector< Hit * > &hitvector)
	reconstruct a track with KF from given hits
Protected Member Functions
Track	CreateSeedTrack ()
	Create a seed track (from the given hits)
Track	CreateSeedTrack (const std::vector< Hit * > &hitvector)
	A track is seeded from the first and last hit.
Protected Attributes
CLHEP::HepMatrix	P
	projection matrix from track parameters to local measurement

---

Detailed Description

Forward Kalman filter tracking algorithm, without smoothing step.

This class allows straight track Kalman Filter tracking. This is done after pattern recognition, i.e. the hits that constitute the track have to be given to the algorithm.

---

Member Function Documentation

Track KalmanFilterTracking::CreateSeedTrack

(

)

[protected]

Create a seed track (from the given hits)

A track seed is created in the origin going parallel to Z with huge errors.

Assume track originating in center and going along Z with very large error

Track KalmanFilterTracking::CreateSeedTrack

(

const std::vector< Hit * > &

hitvector

)

[protected]

A track is seeded from the first and last hit.

The seed is created from the first and from the last hit in the hitvector. This already gives a good estimate of the track parameters. However, the track is created with large initial covariance matrix in order to keep the bias from the seed small.

When less than two hits are found on the track, CreateSeedTrack() is called without argument.

assume errors for initial covariance matrix

const HepMatrix & KalmanFilterTracking::Jacobian_Track

(

const CLHEP::HepVector &

p0

)

const

Jacobian matrix, derivatives of measurement equation to track parameters.

jacobian matrix, derivatives of measurement equation to track parameters evaluated at initial track parameters p0

The matrix is evaluated at the position p0, where p0 are the track parameters. p = (tx, ty, x, y)

Track KalmanFilterTracking::MakeRecoTrack

(

std::vector< Hit * > &

hitvector

)

reconstruct a track with KF from given hits

Process given hits and return a track built with Kalman Filter.

First, a track seed is built from the hit.

Then, the track is updated by the Kalman Filter with each hit. The method returns the final track after all hits are processed, with updated track parameters and updated covariance matrix.

In this method, no alignment is performed. The track is created taking into account only the nominal position of the detectors, thus the alignment is assumed not to be existing.

Now process hits and estimate track parameters

previous track state

actual measurement (it is only 2-dimensional in x and y)

predicted measurement given previous track state

---

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

• /home/mweber/Dokumente/CMS/Alignment/kalmanalignment/trunk/simulation/KalmanFilterTracking.h
• /home/mweber/Dokumente/CMS/Alignment/kalmanalignment/trunk/simulation/KalmanFilterTracking.cc