CMS Bulletin

OFFLINE

The Offline group is now finishing the first production release of 2013, CMSSW_6_2_0. It will contain the latest release of Geant4 as planned. We will use it to start the generation and simulation of 13 TeV datasets for a first round of evaluation of technical and physics performance of the software under the data-taking conditions of 2015. We have also participated in preparing the CMSSW_5_3_X series release to run on 2011 data, and regenerate matching Monte Carlo samples, after the decision this June to create legacy datasets for 2011 that matched 2012.

The Framework group continues its development of a multi-threaded framework. However the big news in this area is the large amount of work that has gone into transitioning the CMSSW source code repository from CVS to Git, and the decision to use a free and robust commercial service for hosting it, github. We recommend people using the UserCode repository also migrate their code there and we provide tools and assistance to do so.

The Fast Simulation team has nearly finished all of the new pile-up Mixing code, including all of the ingredients to use minbias events made with Full Simulation. The hadronic response model has been retuned to take into account the relative times of the energy depositions.  Modelling out-of-time pile-up in FastSim is now possible in the 6_2_X series. Other ongoing work is related to the Phase 2 Upgrades. The simulation of the future forward calorimeters is underway, the ageing models for ECAL and HCAL have now been integrated, and the generalisation of the Tracker geometry description is nearly complete. Future work will concentrate on upgrade integration.

For the Full Simulation, recent work has continued the focus on improving the performance of our implementation of the Geant4 simulation step.  Several rounds of testing of the “Russian Roulette” sampling technique have converged on a default parameter set that will be tried in future Monte Carlo productions for the 13 TeV running.  Several technical improvements in the performance of the digitisation algorithms at very high pile-up have been implemented. As in the Fast Simulation, detailed ageing models for ECAL and HCAL have been implemented in the Geant4 modelling of the detectors.  Ongoing work also includes the development of Geant validation suites and the development of “PreMixing” for modelling of multiple interactions. For this technique, the minbias events that we use to model pile-up are combined into a single “overlay event” before the Monte Carlo production jobs are run. Then, a single overlay event is used for each Monte Carlo hard-scatter event, instead of almost 1000 individual minbias events. This will dramatically simplify large scale MC production.

Major additions were made in the reconstruction algorithms aiming towards completion later in the year:

1. MVA discriminants are now saved for tracks, which can be used for an improved “high purity” selection;

2. the particle-flow-based ECAL clustering has improved both on the CPU performance side and on the physics algorithm side, now using moustache shapes;

3. the electron/photon reconstruction, aiming to give a consistent Global Event Description (GED), using, in particular, the clustering mentioned above, now has the unified eγ candidates and their photon and electron “projections” stored in data for further studies, nearing their use for physics.

Several developments in other areas were completed, including updates in b-taggers, cleanup of older jet collections and addition of collections with substructure information.  Full details will be available in the release notes once the release is made available. Many changes are still ahead of us: some are technical, required to be able to run in multi-threaded environment; some are aiming to improve physics performance.

The Analysis Tools software is very stable and routinely used in physics analyses throughout the collaboration. Latest developments include the final implementation the “unscheduled” processing mode for multi-threaded production with CMSSW. Also the Statistics Tools software is in a mature state, and tutorials are regularly offered in the context of the CMSDAS.

The software for the upgrade projects has been organised in a series of releases in two-week intervals. These releases have been based on the CMSSW_6_1_2 release, which is the most recent fully validated CMSSW release. The upgrade software itself is evolving on a more rapid timescale and has been facilitated by regular releases that move rapidly from new feature integration to production deployment. The Phase 1 Tracker and HCAL upgrades and potential Phase 2 Tracker geometries are currently included in these releases.

Significant progress towards high-pileup reconstruction has been made and large event samples are available with an average of 140 pile-up events per bunch crossing. Current samples include the Phase 1 Pixel detector upgrade and models of degradation due to detector ageing corresponding to integrated luminosities from 300 fb^–1 up to 3000 fb^–1. A new round of samples will soon be generated including the Phase 1 HCAL detector upgrade and an improved ageing model for the HE sub-detector starting from the CMSSW_6_1_2_SLHC5 release.

We anticipate that the upgrade software will soon migrate to the CMSSW_6_2_0 release cycle once it has been validated. This release includes important changes to low-level tracker code and to the fast simulation that could not be easily included in the current CMSSW_6_1_2 based SLHC releases. In addition, the software to simulate and reconstruct the post LS1 shutdown CMS detector is fully integrated into the CMSSW_6_2_0 release cycle and no special release will be needed for its use.