 |
ATLAS Offline Software
|
|
ATLAS Offline Software
|
This package contains a new fitter application, which we called JetFitter, which permits to do a multi-vertex fit, where all the vertices are constrained to lie on a single
flight axis and the exact position of the various vertices along the flight length is found during the fit.
In practice this fit is very usefull to be applied to fit the B-D hadron cascade inside a b-quark jet. The fit is in this case initialized with the primary vertex position, where the flight axis is constrained to pass through, and with a first estimate of the flight direction of the B-D hadron cascade which is tipically the jet direction as reconstructed in the calorimeter. Then all the tracks are fit, constraining them to lie on this flight axis, which is exact for B-hadrons and a good approximation for tracks coming from the D-hadron. A clustering procedure is following. For more information please give a look at the documentation of the JetFitter finder in the InDetSecVxFinderTool package.
Trk::JetFitterRoutines This class contains the core functions for the "JetFitter" algorithm.
This is a new algorithm which uses a "Kalman filter" based approach to implement the "ghost track method" used for the first time by SLD (reference to be inserted here).
All the tracks inside a jet can be fitted with the constraint of lying on the same 'flight axis' from the primary vertex towards the jet. This class provides all the methods which are necessary to perform this fit and to evaluate the probability of two or more tracks to build one cluster on the jet axis.
The real finding algorithm is then implemented in the "SV" finder algorithm. In this class only the methods for 'fitting' are implemented.
There is no clear input and clear output in a similar fit, so this works more like a continous interaction between the finder and the fitter, which is provided in the form in a set of routines to use.
Trk::JetFitterHelper This class provides utilities for the JetFitter algorithm, mainly to manipulate the JetVxCandidate object during the fit, but also to perform linearization and similar tasks...
One very interesting feature is the application of exact constraints through the Kalman Filter formalism, a possibility illustrated for the first time in Babar with TreeFitter by Wouter Hulsbergen. This helps to have a fast estimation of the probability of forming a common cluster between two vertices. (reference will be put here)
Trk::JetFitterInitializationHelper This class is used to initialize a VxJetCandidate, which is the class to be provided to the 'JetFitter' algorithm in order to perform the fit of all the vertices on the JetAxis. For using it, please look at the doxygen documentation of its methods.
Trk::KalmanVertexOnJetAxisUpdator A concrete implementation of the Vertex Updator using the Kalman filter algorithm. This is a special implementation (part of the JetFitter algorithm) which is able to fit a vertex correcting the jet axis parameters to find the best overall agreement. --> Method updating the fit along the flight axis with a track which is added to one of the vertices in the fit.
Trk::KalmanVertexOnJetAxisSmoother A concrete implementation of the Vertex Smoother for smoothing the vertex along the Jet Axis. --> Update the vertex on the jet axis (chi2 and ndf) with the knowledge coming from the fitted cascade, updating also all the related track parameters of the tracks at the vertex.
For additional information or in case of problems please contact the authors...
1.8.18