Tracking strategies used in LHCb


* TODO: *

  • * add link to pages describing fitting *
  • * add link to CloneKiller *
  • * check if there is a good description of track types somewhere (with sketch) *

Tracks are found by different tracking strategies. A tracking strategy is mainly concerned with how tracks are found in the different subdetectors in general, while tracking algorithms specify how tracks are found in detail. This page attempts to give an overview of the different tracking strategies, the algorithms that implement them and the types of tracks resulting from their application. It also lists the input required by algorithms implementing a strategy.

The order in which strategies are discussed respects dependencies among the strategies, so we will not discuss a strategy before we explain where its input comes from. This means that one has to start with the strategies and algorithms that reconstruct tracks from hits rather than those which already need tracks as input. Furthermore, all strategies have been placed in different categories, depending on the type of tracks they produce.

Velo Tracking

This category contains strategies to find tracks in the Velo (See also this page.).

Velo tracking in RZ

Method: Starting from hits in Velo R sensors, form 2D Velo tracks by assuming all tracks originate in the same point.

Input: Velo R hits

Output: RZ Velo tracks

Default Location in TES: Rec/Track/RZVelo


Velo space tracking

Method: Starting from 2D Velo tracks in RZ projection and hits in the Phi sensors of the Velo, this strategy will produce full 3D Velo tracks, again under the vertex assumption.

Input: RZ Velo tracks, Velo Phi hits

Output: 3D Velo tracks

Default Location in TES: Rec/Track/Velo


General Velo tracking

Method: This strategy aims to find 3D tracks in the Velo without assuming anything about the origin of the tracks. While being more general, algorithms in this category tend to be slower than the combination of Velo tracking in RZ and Velo space tracking.

Input: Velo hits in both R and Phi sensors

Output: 3D Velo tracks

Default Location in TES: Rec/Track/Velo


  • PatVeloOpenTracking: special algorithm for reconstructing tracks without assumptions about track parameters (e.g. to reconstruct with the Velo in open position), requires hits in consecutive sensors
  • PatVeloGeneric: general purpose reconstruction, no assumptions about track parameters, produces very clean track sample for alignment and test beam studies
  • PatVeloGeneralTracking: can run on unused tracks after PatVeloRTracking and PatVeloSpaceTracking to recover some K_s, halo and beam gas tracks

Standalone T station reconstruction

This category discusses how to find T station tracks without using information from other subdetectors.

Standalone T station reconstruction (combinatorial approach)

Method: Track reconstruction works be choosing strategically combinations of very few T station hits from which a track hypotesis is derived. Then, algorithms collect hits in a window around this hypothesis. The resulting track candidates have to satisfy a number cuts before a hypothesis is accepted as track.

Input: T station hits

Output: T station tracks

Default Location in TES: Rec/Track/Seed


  • TsaSeeding (used per default, has been in LHCb software for a long time, tunings for magnetic field off or HLT usage)
  • PatSeeding (newer, has special tunings for magnet off, HLT usage, finding cosmics or to run with misaligned detector)
  • TrackSeedFind (seems to be rarely used nowadays, can't say much)

Long track reconstruction

This category describes strategies to find long tracks.

Forward tracking

Method: Starting from seeds in the Velo, tracks are searched in the T stations by parametrizing the expected position in T as function of the Velo seed track parameters and the position of a single hit in T. Further T station hits in a window around the expected position in different stations and layers are picked up. If the combination of a Velo seed and some T station hits satisfies some quality cuts, it is promoted to a long track. Hits in TT are picked up if they are close enough to a track through Velo and T station hits.

Input: 3D Velo tracks as seeds, hits in T and TT

Output: Long tracks

Default Location in TES: Rec/Track/Forward


Track matching

Method: Starting from a set of tracks reconstructed in the Velo and a second set reconstructed in the T stations, track matching attempts to match the tracks in the two sets to one another. This is done by extrapolating both Velo and T station track segments to the bending plane of the magnet and evaluating various quantities (e.g. position in the bending plane, slope change, number of compatible hits in TT) to determine if a Velo and a T track segment do actually belong together. TT hits close to the resulting tracks are added afterwards.

Input: 3D Velo tracks, T station tracks, TT hits

Output: Long tracks

Default Location in TES: Rec/Track/Match


Downstream tracking

Method: Using tracks in the T stations, algorithms implementing this strategy search for matching TT hits.

Input: T station tracks, TT hits

Output: Downstream tracks

Default Location in TES: Rec/Track/Downstream


Upstream tracking

Method: Starting from seed tracks in the Velo, upstream tracks are constructed by adding matching TT hits

Input: 3D Velo tracks, TT hits

Output: Upstream tracks


Edit | Attach | Watch | Print version | History: r8 | r6 < r5 < r4 < r3 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r4 - 2008-09-15 - ManuelSchiller
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LHCb All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback