Presentation on theme: "Computing EVO meeting, January 15 th 2013 Status of the Tracking Code Gianluigi Boca, Pavia University."— Presentation transcript:
Computing EVO meeting, January 15 th 2013 Status of the Tracking Code Gianluigi Boca, Pavia University
Progress on the online Pattern Recognition Yutie Liang, December meeting 2) adaptive Hough transform method ; 3) inclusion of the Mvd hits; Determination of the Radius ( = P of the track) and center of Helix : Results on multitrack event, no Mixing for the moment; P resolution = 3.1 % at 1 GeV/c at the PR level 1)conformal transformation for the Stt drift circles; use of the tangents to the drift circles to fill the Hough plot ;
3 Adaptive Hough transformation Initially, 16 X 16 Hough space, keep the best 16 bins for the next step. For a peak(r, θ), locate the area of r-2 ~ r+1, θ-2 ~ θ+1 Divide the area into 16 X 16 Hough space. (equivalent to 64 X 64 ) Iteration 0 1 2 3 4 5 Hough space 16X16 64X64 256X256 1024X1024 4096X4096 16384X16384
4 Inclusion of MVD Pt resolution @ 1GeV/c Iteration: 4 5 6 STT: 3.2% 3.2% 3.1% STT+MVD : 3.3% 3.2% 3.1% MVD Points Belong to MVD points No improvement on the pt resolution! X(cm) Y(cm)
Progress on offline Pattern Recognition (present status, Gianluigi) At the last collaboration meeting it was shown that the Cpu times on an Intel Xeon 2.13 GHz 64 bit Lenny computer was ~ 1 sec/evt with the event mixing. That was achieved by substituting the slow iterative fitting methods (GLPK) with much faster ‘analytical’ methods ( minimum 2 formula) or Hough transform methods.
The CpuTime still need and can be improved by changing the way the track candidates are found with the ‘road’ methods : a) use as seed only hits in the axial Stt at the boundary of the detector; b) confirm track with a SciTil hit at the earliest possible stage ; c) reject ghost tracks with ‘holes’ at the earlier possible stage (cleanup); d) parallelize the code. 6 10 Dec 2012 DONE ! Roadmap : Present result next week at the Pattern Recognition EVO.
7 Change the way the cluster search in XY plane is performed Improvement : only Stt axial hits on the geometrical boundary of the axial Stt region are considered as possible cluster seed. Mvd Pixel Mvd Strip Stt Parallel SciTil MC truth reco track 13 times
Progress on Forward Straw Pattern Recognition (December meeting, Martin) a) Parabola approximation in ZX plane, inside dipole plus straight lines outside; b) parameters of parabola found with Hough Transform method; c) so far implemented the non-skew Straw part of the algorithm, taking the drift circles into account;
Triplet Finding in Axial Straws – Once two triplets are found, calculate circle through origin – Associate nearby hits with track candidate Track Verification: – Currently mix of curvature and associated hits – Lots of room for improvement Further possibilites: – Distance sorting, Concurrent associations, Clutter veto, … – Missing straw compensation: Pivot Straw triggered by neighbor straws – Weight hits according to timestamp – Improved Triplet merging Darmstadt, 12.12. 2012Marius C. Mertens9 Interaction Point New ideas for the online STT pattern recognition (December meeting, Marius)
Progress on the Secondary Track Finder and related issues (present status, Lia) a) application of the Legendre transform in the search of the secondary vertices (work in progress); b) structure to integrate the primary and secondary track finder; c)new method to identify the neighbouring straw of a given Stt axial tube ; this tool is useful both in the cluster finding and in the cleanup code
New functions for STT 8/1/201211L. Lavezzi Parametrization of the geometry limiting tubes… … in order to assign to each tube a layerID and a sectorID LAYERS SECTORS III. Tool for STT