OFFLINE

 

Introduction

A first round of improvements targeting the post-LS1 period has been the primary activity of the Offline group since the last CMS Week. These improvements have gone into the 6_1 release cycle, which we plan on finalising during CMS week. They span all areas, from simulation to reconstruction algorithms, in the context of an adoption of a Global Event Description as the basic paradigm for all physics objects measurement.

The Offline and Computing Week at the beginning of November was an opportunity to define the work plan for the first part of 2013. In particular the plan for the deployment of the core components of the new multi-threaded CMSSW framework was presented, and discussed with Trigger and PPD as it will have far reaching effects in those areas. Our next milestone is a release to be used for mass production of post-LS1 simulation samples in the Fall of 2013.

Fast and Full Simulation

Recent efforts in the Simulation group have been directed at improving the agreement of Fast and Full Simulations with collider data. The Fast Simulation has been modified to enable the use of the Full Simulation code that emulates the digitisation step. This will allow the use of more realistic noise models, calibration constants, and greater ease of maintenance, all with essentially no loss in CPU performance. On the Full Simulation side, work with Geant4 continues, with the latest work incorporating the mid-2012 9.5 patch01 release into the CMS development environment. Other modifications include the addition of a new bremsstrahlung model that has been shown to improve the simulation of electromagnetic showers. In both Full and Fast Simulation, extensive additions and modifications have been made to incorporate various upgrade geometries and detector configurations. Full Simulation models of the HCAL and pixel upgrade geometries were used for the TDR studies this summer, and simulations of the L1 Trigger upgrades are ongoing. Further work in all of these areas will take place over the next few months.

Reconstruction

Several improvements and new developments have been made in the reconstruction for CMSSW 6.1.0 in the last few months. The content of the AOD has been reviewed, and the removal of several, little used collections has allowed reducing the size of the AOD by approximately 30%. The most time-consuming parts of the reconstruction have been reviewed to reduce the CPU time. This effort has also shown further areas of the code that could be improved during the next development phase after CMSSW 6.1.0. Muon reconstruction has seen an improvement to the selection of the best refit for the high-PT (TeV) muons. Track reconstruction has been improved in several areas, such as the identification of duplicate tracks or the reconstruction of muons in the Tracker. The latter uses a dedicated outside-in seeding algorithm with information from the muon sub-detectors. In addition, the retuning of iterative tracking has allowed reducing the fake rate. In the next year, the track reconstruction will be substantially rewritten to improve CPU and memory performance and allow parallel execution.

Analysis Tools

The analysis tools packages are stable over several latest release series. Critical updates are backported to otherwise frozen 2012 analysis releases CMSSW_5_2_X/CMSSW_5_3_X. A major restructuring of PAT (unscheduled mode) is completed in its core part: it will become available in CMSSW_6_1_0. Several high-level PAT tools are being adopted for the unscheduled mode: tau, MET and PF2PAT. This work and the global event description development continue in CMSSW_6_1_X and beyond. The existing statistics tools are stable and the documentation is being expanded. The PAT tutorial is routinely offered to the CMS collaborators. A new statistics tools tutorial has been developed and successfully tested at CMSDAS. New tools are considered, with cmssh –– an integrated CMS development environment –– being the most recent addition. It will be demonstrated at CMS Week.

Core Software

The framework group is completing the re-design of the core of CMSSW to allow cmsRun to utilise many CPUs in the same job. Work will begin in earnest at the beginning of the year to implement the design. The plan is to have the re-design ready by fall 2013.


by L. Sexton-Kennedy, F. Cossutti, P. Elmer with contributions from A. Giammanco, M. Hildreth, V. Ivantchenko, V. Krutelyov, T. Speer, V. Adler, G. Kukartsev, R. Wolf and C. Jones. Edited by K. Aspola.