CMS reconstruction improvements for the tracking in large pileup events

The CMS tracking code is organized in several levels, known as iterative steps, each optimized to reconstruct a class of particle trajectories, as the ones of particles originating from the primary vertex or displaced tracks from particles resulting from secondary vertices. Each iterative step consists of seeding, pattern recognition and fitting by a kalman filter, and a final filtering and cleaning. Each subsequent step works on hits not yet associated to a reconstructed particle trajectory. The CMS tracking code is continuously evolving to make the reconstruction computing load compatible with the increasing instantaneous luminosity of LHC, resulting in a large number of primary vertices and tracks per bunch crossing. The major upgrade put in place during the present LHC Long Shutdown will allow the tracking code to comply with the conditions expected during RunII and the much larger pileup. In particular, new algorithms that are intrinsically more robust in high occupancy conditions were developed, iterations were re-designed (including new ones, dedicated to specific physics objects), code optimizations were deployed and new software techniques were used. The speed improvement has been achieved without significant reduction in term of physics performance. The methods and the results are presented and the prospects for future applications are discussed. 1. CMS Track Reconstruction The central component of the CMS experiment [1] is the world’s largest all-silicon detector, composed of an inner Pixel detector (arranged in three barrel layers and two forward disks, for a total of 66 millions of channels, with pixel size of 100×150 μm2) and an outer Strip detector (arranged in a barrel-forward complex geometry, for a total of 9.6 millions of channels, with pitch in the range 80-180 μmand length in the range 10-20 cm). A key component of the Strip detector is the usage of double-sided modules, which are composed of two different strip detectors glued together with a stereo angle of 100 mrad, to provide three-dimensional position measurements in global coordinates. The CMS tracking [2] is based on a Kalman filter and can be logically divided into four steps: the seeding, in which a proto-track is formed starting from two or three consecutive hits with either a beamspot or a vertex constrain; the pattern recognition, during which the proto-track is propagated into the CMS tracker and compatible hits are associated to the proto-track; the fitting, in which the best parameters’ estimate is computed for all hits along the trajectory; and the final selection, in which quality criteria are applied to the tracks to reject the badly 21st International Conference on Computing in High Energy and Nuclear Physics (CHEP2015) IOP Publishing Journal of Physics: Conference Series 664 (2015) 072040 doi:10.1088/1742-6596/664/7/072040 Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. Published under licence by IOP Publishing Ltd 1 reconstructed ones and to reduce the fake rate1. This procedure is run iteratively: at each iteration, the hits associated to high quality tracks are masked so that the next iterations face a much reduced combinatorics problem. The driving principle is, though, to reconstruct easy tracks first, allowing for more complex algorithm to run in the later iterations. During RunI, seven iterations were used, as illustrated in Table 1. The logic behind the definition of the iterations used during RunI could be evinced looking at Figure 1 and Figure 2. The very first iterations are aimed at reconstructing prompt tracks within the full pTspectrum, identified by the blue-ish colors in Figure 1; the later iterations, thanks to the much lower combinatorics they have to face, are tuned to reconstruct much more difficult tracks, namely displaced and very displaced ones, identified by the green-ish color in Figure 2. Table 1. Tracking iterations used during RunI data taking. Order Name Seed Target Track 0 Initial pixel triplets prompt, high pT 1 LowPtTriplet pixel triplets prompt, low pT 2 PixelPair pixel pairs recover high pT 3 DetachedTriplet pixel triplets displaced 4 MixedTriplet pixel+strip triplets displaced 5 PixelLess strip pairs more displaced 6 TobTec strip pairs very displaced Figure 1. Performance of iterative tracking in RunI: efficiency vs η Figure 2. Performance of the iterative tracking in RunI: efficiency vs the production radius. 1 A fake track is a reconstructed track that has no correspondence in reality. In Monte Carlo simulation, fake tracks are characterized as tracks not matching the trajectory of any generated charged particle. The fake rate is the ratio between fake tracks and all tracks. 21st International Conference on Computing in High Energy and Nuclear Physics (CHEP2015) IOP Publishing Journal of Physics: Conference Series 664 (2015) 072040 doi:10.1088/1742-6596/664/7/072040