1 / 43

New STAR tracker(Stv)

New STAR tracker(Stv). V. Perevoztchikov Brookhaven National Laboratory,USA. Why new tracker?. STAR has used ITTF (Sti) as its main tracker for over 6 years, enabling a successful physics program. However, there are important limitations which constrain us moving forward:

gail
Télécharger la présentation

New STAR tracker(Stv)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. New STAR tracker(Stv) V. Perevoztchikov Brookhaven National Laboratory,USA

  2. Why new tracker? • STAR has used ITTF (Sti) as its main tracker for over 6 years, • enabling a successful physics program. However, there are • important limitations which constrain us moving forward: • Geometry Description is too simple • Geometry ordering does not fit well, even for TPC, and will be nearly impossible for new detectors • ITTF requires sensitive detectors oriented along the z-axis, which precludes forward tracking detectors such as the FGT • Complicated task to integrate a new detector • Magnetic field must be constant and oriented along z • More than 6 years in STAR was used ITTF(or Sti) tracker. It works rather good but there are some important limitations. • Geometry is too limited to describe real material distribution • Geometry ordering , essential for Sti, does not fit well even for TPC • and practically impossible for the new detectors; • FORWARD tracking is FORBIDDEN. Only TPC like detectors are allowed. Sensitive planes must be oriented along Z axis. • Introducing of new detectors demands rather complicated job with a lot of additional coding; • Magnetic field must be constant and along Z. 2 Victor Perevoztchikov, BNL STAR Upgrade Review

  3. Stv tracker. Main features. • Standard ROOT geometry is used, produced by AgML and • ident • Geant track propagator is used in tracking. • Detector orientation is arbitrary; • Kalman fit is modified to be more precise for forward detectors; • Arbitrary magnetic field orientation (not yet); • Track extensions to other, not tracking detectors (not yet); • Multiple Seed Finders allowed. • Implemented • ROOT/TGeo geometry package is used. Ensures identical • geometry in simulation and reconstruction. • GEANT is used to propagate tracks through material • Detector orientation is arbitrary • Kalman fit is modified to be more precise for forward detectors • Support for running multiple seed finders • Planned • Track matching to other detectors, e.g. BEMC • Arbitrary magnetic field : 3 STAR Upgrade Review Victor Perevoztchikov, BNL

  4. Stv tracker, additional features. • Automatic recognition of detector orientation; • Automatic hits <-> detector elements relationship. No hand made lookup tables; • For energy loss and multiple scattering Geant machinery is used. • In a result, flexibility to add and modify new detectors. Practically only Geometry description is needed; STAR Upgrade Review

  5. ROOT TGeo and extensions • ROOT TGeo is chosen as the geometry description. • TGeo could be created from many sources. • TGeo allows detailed description of the detector. Thus energy loss and multiple scattering would be accurate; • The same geometry could be used for simulation and reconstruction; • TGeo Stv extension • Each TGeo volume has a proxy object which contains additional information. Most of information is automatically created during initialization stage; • Description of sensitive detector element: orientation, type (plane, cylindrical, …), hit error functors, activity status etc… • Hit container with hits associated with this detector element; • Etc… : STAR Upgrade Review

  6. Geant tracking engine • Geant VMC was selected as a tracking engine. • The standard Geant machinery is used to do tracking thru the complicated geometry, accounting the magnetic field, energy loss and multiple scattering; • When track is entering into a sensitive volume, then the according hit error functor, hit container, activity flag etc… are accessible. • The hits ownership with detector elements also provided via Geant. GEANT3 VMC was selected as the transport engine. Using the standard G3 machinery, tracking is done through the geometry accounting for magnetic field, energy loss, and multiple scattering. Stv associates sensitive volumes in the geometry with hit errors, hit containers, and other tracker parameters. This releases the developer from having to implement a custom interface for each new detector. : 6 STAR Upgrade Review Victor Perevoztchikov, BNL

  7. Integrating a new Detector with Stv Stv was designed to streamline the process of integrating a new detector into the tracker. In order to integrate a new detector, developers need to 1) Define the detector geometry in AgML 2) Create the detector's hit class and hit containers in StEvent. a) If hits are space points, done. b) Otherwise, space points should be provided or a custom seed finder will need to be implemented. 3) Implement an iterator over the hit container in StEventUtilities 4) Provide hit errors, either through the hit class or through StvHitErrorCalculator 5) Stv supports multiple seed finders running during reconstruction, as different detectors may well have different criteria for what makes a “good seed”.

  8. Stv Seed Finders By design Stv uses several seed finders. The first one is the default seed finder and others specialized ones. • Stv allows list of seed finders called one after another; • Each seed finder could be called several times; • If special seed finder does not exist, default one is used; • At the end default seed finder is called repetitively up to no more tracks is founded. Three seed finder right now are implemented : STAR Upgrade Review

  9. Modified Kalman fit Stv used modified Kalman fit. More precise, not a Kalman fit was modified but the system of coordinates is different than usual. In standard Kalman fit detector plane is used for the fit. But when track is crossing this plane with angle far from orthogonal, result is unstable and inaccurate. In Sti this angle < 1 radian; In Stv we use DCA plane. This plane is orthogonal to track and crossing the hit. Hit errors projected into this plane; Due to orthoganality accuracy is better for forward tracks. STAR Upgrade Review

  10. Simple example 1. : StarSoft local

  11. Simple example 2. : StarSoft local

  12. Now Sti and Stv comparison for TPC Captions : • Red is negative charge track, Black is positive one; • +pt> 0.1 • pt> 0.1 • +pt> 0.5 • -pt> 0.5 • +pt> 1.0 • -pt> 1.0 • +pt> 2.0 • -pt> 2.0 STAR Upgrade Review

  13. Xi2(nHits) comparison • Sti13 Stv13 <Xi2> must be 1. Stv Erros 1.4 times less thane needed STAR Upgrade Review

  14. Track angle Psi(nHits) deviation comparison • Sti13 Stv13 Stv and Sti comparable for reconstructing azimuthal angle STAR Upgrade Review

  15. Track angle Psi(nBadHits) deviation comparison • Sti13 Stv13 Stv appears more stable with respect to bad points STAR Upgrade Review

  16. .q/pt deviation(nHits) • Sti13 Stv13 Stv has comparable charge-sign discrimination to Sti STAR Upgrade Review

  17. q/pt deviation(Eta) • Sti13 Stv13 Comparable STAR Upgrade Review

  18. q/pt deviation(pT) • Sti13 Stv13 Stv has a problem in high pT STAR Upgrade Review

  19. Efficiency(Phi) • Sti13 Stv13- • +pt> 0.1 • pt> 0.1 • +pt> 0.5 • -pt> 0.5 • +pt> 1.0 • -pt> 1.0 • +pt> 2.0 • -pt> 2.0 Sti has outsiders, Stv not. But Sti~80 Stv~75. Sti better STAR Upgrade Review

  20. Efficiency(Eta) • Sti13 Stv13- Sti in average better, but high pT and big eta Stv better. STAR Upgrade Review

  21. Something wrong • You noticed that in previous slides was typed Stv-. Why (-)? • Three weeks ago some modifications was made, mostly result become better, but efficiency… • List of modifications: • Kalman Track initialization; • Hit error refitted; • … STAR Upgrade Review

  22. Efficiency Stv 3 weeks ago and current one • Stv13- Stv13 Something happened. Current Stv efficiency mach worse. STAR Upgrade Review

  23. Summary for Sti/Stv comparison. • It is clear that hit errors are too small. Factor is about 1.4 . Right now we know the reason. FitHitError application will be modified, and on next week we will see result. • Efficiency in average is still better in Sti. It could be explained again, by too small estimations of hit errors. • In other parameters, Sti/Stv are comparable. • Tuning is still needed, hopefully 3-4 week will be enough. STAR Upgrade Review

  24. 20 GeV Muon Proof of Principle FGT Tracking with Stv • Stv is capable of tracking with the FGT in the forward • direction, reconstructing tracks with hits in both TPC and FGT. • MC sample: 1 mu- with pT > 15 GeV in FGT acceptance on each event • FGT clusters are combined into space points • Space points passed to customized seed finder • Seeds are extended and fit as tracks using Stv • Default seed finder was used

  25. Estimated Efficiency Efficiency and Charge Sign The good news... The bad news... Ideal TPC alignment Charge is pulled to the wrong value in the region where tracks meet the FGT acceptance. Reconstruction efficiency for global tracks is reasonable out to eta = 1.6. Clone tracks in red. The take away: more work needs to be done to understand why tracking falls off at eta of 1.6, and why charge sign gets pulled in wrong direction.

  26. Forward Tracking in the ETTR At the STAR upgrades workshop Last year, Stv was shown to be capable of forward tracking in the proposed ETTR upgrade detector. ETTR provides 3 x 30 hit points on the track. ETTR

  27. Integrating a new Detector with Stv Stv was designed to streamline the process of integrating a new detector into the tracker. In order to integrate a new detector, developers need to 1) Define the detector geometry in AgML 2) Create the detector's hit class and hit containers in StEvent. a) If hits are space points, done. b) Otherwise, space points should be provided or a custom seed finder will need to be implemented. 3) Implement an iterator over the hit container in StEventUtilities 4) Provide hit errors, either through the hit class or through StvHitErrorCalculator 5) Stv supports multiple seed finders running during reconstruction, as different detectors may well have different criteria for what makes a “good seed”.

  28. Stv tracker status • What is implemented: • Track propagator based on Geant geometry and Geant propagator; • Arbitrary orientation of sensitive detectors; • Forward oriented Kalman fit and filter; • Multiple Seed Finders; • Automatic recognition of detector orientatation and types; • Automatic relationship between hits and sensitive detectors, without hand make lookup tables; : STAR Upgrade Review

  29. Stv status continue • What was tested and tuned: • Stv & Sti give mostly comparable results; • Two forward detectors was introduced and successfully tested. Some additional tuning is needed; • What improvement is needed : • Tpc Hit errors is still too small. Reason is probably understood. New results expected in a week; • Efficiency is about 5% less than in Sti. Probably related to underestimated hit errors. • Performance was not yet investigated; • Expected time scale for tuning is 2-3 weeks with the help of assigned workforce team; • Common work with FGT group iz expected. STAR Upgrade Review

  30. . STAR Upgrade Review

  31. Arbitrary magnetic field orientation • We use Geant as a tracking engine. Geant allows tracking in arbitrary magnetic field. But tracking only is not enough. To do fitting the error propagation is also needed. Right now error propagation is implemented for Z oriented magnetic field only. But Stv design allows such implementation with arbitrary magnetic field. : STAR Upgrade Review

  32. Stv Seed Finders continue Non standard seed finders are needed in two cases: • Increase performance for concrete detector; • Non standard detector, for which default seed finder does not work; The typical example of point 2. is Fgt. Fgt hits are elongated. • Some hits define only Z and Rxy, others Z and Phi. In addition, there are amplitudes which could be used as well. • Right now we have: • Default, :following nose”, seed finder, for any detector ; • KNN (K-Nearest Neighbor) seed finder, for any detector ; • CA seed finder oriented for TPC only; • 2d hit generic seed finder in development : STAR Upgrade Review

  33. 2D hit Seed finder. It is based on the following ideas: • All 2d hits are joined into 3D hits, with a many fake ones. Amplitudes could be used to decrease amount of fake hits. • 1st hit selected. Its coordinates considered as zero. • All hits in neighborhood projected onto 2d plane, using their φ,λ angles. Real track on this plane represented by spot. • Using well known “K neighbor” method we find the spot. • If there is no spot, the first hit is fake. Select the next. • When spot is found, the seed is created from all related hits This is not yet ready, only part of code is written. STAR Upgrade Review

  34. What is Kalman • The simple example. We have 4 measurements of the same parameter . • To estimate the value we search the minimum of: The solution is trivial, this is a “global fit”: STAR Upgrade Review

  35. Kalman continue • It was a global solution. The Kalman fit is: STAR Upgrade Review

  36. Kalman continue Kalman fit features: • The result of Kalman fit is exactly the same as for a global one. No mystics; • In global fit, size of matrix to invert, depends on number of measurements. In Kalman fit not. As a result, machine accuracy for Kalman fit are more critical; • In Kalman fit an estimation is updated with each new measurement. This allows to make a decision: is this new measurement belong to our object, or not? So we can use not only Kalman fit but Kalman filter. Not possible in global; • By the points above, Kalman fit and filter are so popular. STAR Upgrade Review

  37. Modified Kalman fit Stv used modified Kalman fit. More precise, not a Kalman fit was modified but the system of coordinates is different than usual. Standard Kalman fit: Let consider simplified, 2d case. Hitting plane along Y axis, track crossed Y axis with angle α wrt X axis. So : • global track parameter α projected into local Y and proportional to tan(α). • then δY ~ δ α *(1+ tan(α) *(δα) /2)/ • when α << 1, then cos(α ) =1, tan(α) = α, second term is very small and projection from global to local system is linear. • Projections of error matrix is also linear. In local frame fit is linear. Transformation to global of fitted parameters and errors is linear too. So life is good. But when α >1 , cos(α ) ~0, life is bad. Linearization is wrong, linear fit is wrong, Backward transformation into global is also wrong. All times, when I saw unstable fit in Sti, it was α >1 STAR Upgrade Review

  38. Detector Frame hit plane hit δα d D Fitted track α Y STAR Tracking Review

  39. Fit continue Could we do something with bad fit when α > 1? Yes, we can! Why we fit in local frame? There are only two reasons: • We know that track is crossing our hit plane; • We know the errors in this frame; Let invent another local frame, where linearization is always working . The evident candidate is Dca frame. Dca frame is a track coordinate system where origin is in Dca point to hit. In this frame plane perpendicular to the track and crossing the hit point is a Dca plane. Look the following picture.

  40. Dca Frame hit plane Dca plane hit d α D δα α δd δD Y Fitted track STAR Tracking Review

  41. Stv12 & Stv13 comparison. • Stv12 Stv13 Xi2 still high, but shape now is correct STAR Upgrade Review

  42. q/pt(nHits) • Stv12 Stv13 STAR Upgrade Review

  43. Dca(xy) • Stv12 Stv13 STAR Upgrade Review

More Related