Mean number of charged particles per unit of η measured in CMS at the world record collision energy of 2.36 TeV
While the LHC was being repaired, vivid activities in CMS took place in the field of calibrations as well as in preparations for physics analysis. The LHC shutdown allowed the CMS groups to do refinements and reparations to the CMS detector hardware as well as to tune the event reconstruction and analysis software. The LHC restarted the operations in November with 900 GeV collision energy. In December the LHC achieved the world record energy of 2.36 TeV. The total number of collision events recorded by CMS approached one million. A draft article on general features of pp collisions at 900 GeV was ready for the collaboration review by the end of the year.
Earlier in the year, prior to collisions, about 500 million cosmic events were recorded in the CRAFT09 runs (CRAFT=Cosmic Run At Four Tesla). These events proved very useful for tracker alignment in which the HIP group actively participated. A "Prompt Alignment Exercise" was organized in September in order to test CMS procedures for repetitive tracker alignment on a daily basis. In 2009 a total of 24 collaboration articles were prepared on cosmic events and submitted to scientific journals.
Another large scale CMS collaboration activity was the "October Exercise" in which a total of 900 TB of simulated data was processed and exchanged between the T1 and T2 computing centres. Helsinki T2 centre participated successfully in the exercise. The purpose of the exercise was to rehearse the CMS T2 analysis model and to test the readiness of the participating institutes to produce high quality results.
The two main fields of the CMS physics analysis in HIP were B-physics and the search for the charged Higgs boson. P.Eerola was the co-convener of the CMS B-physics analysis group and the HIP team continued to take the responsibility on the analysis of the H± in the τ lepton decay channel. The responsibility on the user support of the CMS software package CMSSW was another important HIP activity issue. The activities during the year 2009 are described in more detail in the following.
Physics analysis and simulation
Work on Higgs boson searches. The preparative study on the searches of the light charged MSSM Higgs bosons in a new discovery channel, gg → tt → bWbH± → bqqbτν with fully hadronic final state, and studies on the background measurements from data for charged Higgs boson searches were the main responsibilities of the HIP Higgs physics group (R.Kinnunen, M.Kortelainen, S.Lehti and L.Wendland). Other major responsibilities were the coordination of the τ trigger efficiency measurement from data, participation in the CMS October exercise for data analysis, participation in the CMS test runs and development of software and computing facilities with the local grid cluster.
An event display showing a simulated event with charged Higgs boson production and decay process gg → tbH+, t → Wb → scb, H+ → τν → hadrons with a radiative gluon jet simulated inside the CMS detector and visualized in the projection normal to the beam. The curved lines correspond to the reconstructed tracks of electrically charged particles and the rectangular bars visualize the amount of deposited energy in the electromagnetic and hadronic calorimeters.
The gg → tt → bWbH± → bqqbτν channel with hadronic final state is a potential discovery channel for the light charged Higgs bosons already during the first years of LHC operation. With this channel, the exclusion limits by the Fermilab experiments for charged Higgs bosons can be superseded with an integrated luminosity of 200 pb-1. A complete search strategy for the region mH+ < mtop was developed during 2009. The major challenge of this channel is the suppression of the QCD multi-jet background, reducible mainly with τ-jet identification and missing ET cut. A detailed study was performed on the identification of τ jets, reconstructed with the particle flow (PF) method, with luminosity-independent optimization of the cuts (L.Wendland). To select fully hadronic final states and thus a visible H± signal, methods to identify the electrons, muons and hadronically decaying τ's from the associated top quark were developed for vetoing the leptonic events. Methods for the measurement of the associated W and top masses were introduced. The algorithms to measure the missing ET with the calorimetric, track-corrected, and PF methods were tested for further suppression of the QCD multi-jet background.
Important progress was achieved in the development and tests of the methods to measure the residual ttbar, W+jet, and QCD multi-jet backgrounds from data. To measure the ttbar and W+jet backgrounds, a data driven method, based on replacing the identified isolated muon with a simulated τ in the muonic multi-jet events, was developed and tested with simulated data and implemented in the CMSSW software. Furthermore, the development of an energy correction algorithm for τ's based on combining the tracker and calorimeter information was continued. Measurement of the QCD multi-jet background from the early data with events from hadronic multi-jet triggers was investigated, assuming the probability for the hadronic jet to pass the τ-jet identification to be also measured from data. It was shown that this probability can be measured with the multi-jet events with a statistical uncertainty better than 2% for ET > 60 GeV with an integrated luminosity of 200 pb-1.
The development of methods to measure the τ trigger efficiencies with the early LHC data first in the multi-jet events and later with genuine τ jets from W and Z decays was continued with simulated Monte-Carlo data. Work on software and computing, and the development work for new tools and methods for data analysis were continued. The activities for the multivariate analysis tools (TMVA) were also continued. The method was applied on the τ identification in H± → τν channel in the presence of the QCD multi-jet background.
Shift work and work on data quality monitoring. The group participated in the CMS test runs through shifts in the data quality monitoring group at CERN. Dr. Virginia Azzolini became member of the DQM (Data Quality Monitoring) group, has the task of data certification management, and acts as a DQM shift supervisor at P5. The results were presented regularly in the working group meetings at CERN and in group meetings and seminars at HIP.
Work on B-physics. B-physics analyses have been started as a new physics activity in the HIP CMS programme. The activities at HIP are led by Paula Eerola, who is also the CMS B-physics analysis group co-convener for 2009-2010. The B-physics analysis group is responsible for all aspects of B physics in CMS.
LHC provides extremely high rates of B-hadrons due to the high b-production cross-section, favourable signal-to-background ratio, and high luminosity. B-physics is a timely field in particular during the first years of data taking at LHC, when the luminosity will be relatively low, providing a clean event environment and low trigger thresholds.
During 2009, the plans for the first year's B physics deliverables (i.e. the B physics analyses which should be performed and published) have been specified. Assuming 200-300 pb-1 of collected data, our analysis plans include at least the following channels: J/Ψ production cross-section, including a measurement of the fraction of J/Ψ coming from B-decays; Υ meson production cross-sections; J/Ψ and Υ polarization; production cross-section and lifetimes of exclusive B decays B → J/Ψ KS0, B+ → J/Ψ K+, Bs → J/Ψ Φ, Λb → J/ψ Λ; B-quark production cross-section; and bbbar-correlations, will be among the first physics analysis topics. Apart from physics measurements, these channels also serve as crucial standard candles for the calibration of the mass and momentum scale of the Tracker and the muon system. Furthermore, these standard decays will serve later on as control channels for new and/or rare decays.
In 2009, the HIP group was particularly involved in the analysis preparations for the decay channel Bs → J/Ψ Φ. This Bs decay is interesting to test possible contributions from new physics beyond the Standard Model. The mixing induced CP-violation weak phase in the Bs decays Bs → J/Ψ Φ is expected to be very small in the SM, while effects from new physics beyond the SM could easily alter the phase by adding new contributions. A relatively large statistics, corresponding to several years of LHC running, is, however, required to measure the weak phase. During the start-up phase of LHC in 2010, the first goal is therefore to measure the differential production cross-section, and to cross-check the well-known properties of the Bs meson.
During
the data taking in 2009, the B physics group analyzed promptly all
the data, looking for events with two muons at both the 900 GeV and 2.36 TeV
data. In parallel, MC samples produced with the same center-of-mass
energies were analyzed for comparison. The statistics was too low to
see a J/Ψ signal peak, but one candidate event was found. These
results were approved by the CMS collaboration at the Plenary meeting
on Dec 16, 2009, and shown by the spokesperson in his report to the
CERN Council on Dec 18, 2009.
CMS computing and off-line user support
The HIP team continues the coordination (K.Lassila-Perini) of the user support to
CMS physicists. In 2009, several improvements were made to the CMS Software
Documentation Suite based on the results of the usability tests of the
documentation. The "Offline Software Guide" containing full details of the CMS
Software went through a comprehensive review. All major stake-holders providing
software or software tools for physics analysis were involved in the review and
the main emphasis was given to the accessibility and availability of the
information. The "Offline Workbook" containing essential instructions and basic
and advanced tutorials is being fully reviewed and the structure has been adapted
to the new phase of CMS operation.
For the user training, the User Support team setup a model based on a new concept emphasizing the importance of integrating the acquired skills to the practical work. A successful pilot course on "Using Physics Analysis Toolkit in your analysis" was organized in the beginning of 2009. The course was well appreciated and it is now regularly given before major collaboration meetings. Approximately hundred physicists have now followed the course and many of them are actively contributing to the common software in CMS. Similar concept was followed for a course on statistical tools organized as a common effort between CMS and ATLAS collaborations.
Grid computing activities
To take advantage of the LHC repair time CMS organized two large computing
exercises in 2009 to stress test the computing system. HIP participated
both in the "Scale Testing for the Experimental Program in 2009" (STEP09)
and in the "October Exercise" (T.Lindén). Both exercises gave important lessons on
how to develop further the performance and procedures of the Finnish CMS
Tier-2 resource. The Finnish CMS Tier-2 resource accepts CMS grid
analysis jobs in a fully transparent way and it is commissioned for
analysis work. The close collaboration between HIP CMS and Technology
programmes, CSC (IT Center for Science Ltd) and the Nordic DataGrid
Facility (NDGF) resulted in good progress on many aspects of the CMS
software that are summarized here.
Hardware. The main CPU resource for CMS and ALICE was the 512 core Sepeli Linux cluster situated in the CSC premises, which has been very stable. In addition to that the 260 core Linux cluster Ametisti in Kumpula was used for CMS tasks. The 400 core Linux cluster Korundi also in Kumpula was connected to M-grid and configured for the CMS software environment. The 170 TB dCache disk system was upgraded with 100 TB of disk for CMS usage. The new Hitachi AMS 2500 disk space showed improved performance compared to the old Hitachi AMS 1000 racks, but still the disk I/O performance limited the CPU efficiency of CMS analysis jobs. A new cluster called Jade was purchased, which will replace Sepeli in 2010. The acquisition was made together with CSC and several other Finnish universities. The Jade cluster consists of 64 HP BL465c blades with dual 2.6 GHz AMD Istanbul six core CPUs totaling 768 cores connected with both 4x DDR Infiniband and gigabit ethernet networks. Each blade has a 300 GB 10 000 rpm SAS disk for scratch usage and 32 GB of RAM. In addition to the HP DL385 cluster fronted, CMS and ALICE will have separate grid front-ends to the cluster consisting of HP DL380 servers. Lustre and BeStMan will be studied as an replacement for dCache, so two Sun X4540 disk servers with 48 TB of raw disk each and a HP DL 360 server to be used as a Lustre metadata server have also been purchased.
Software. During 2009 many software services at the Finnish CMS Tier-2 became operational. Production Monte Carlo jobs started running at HIP in February. In March the CMS JobRobot started running at HIP. This is a site monitoring tool that is very useful for debugging and performance optimization. The stability of and usability of the dCache system improved when the access protocol was changed from xrootd to dcap, which was also seen in the JobRobot statistics. Studies of the dCache system performance was carried out to measure the available bandwidth and to improve the CPU efficiency of the CMS grid jobs. CMSSW grid installation jobs could also be run successfully. The CRAB ARC plugin was taken into use. During the October exercise the glidinWMS was reconfigured to work as during CSA08. The CMS Site Availability Monitoring (SAM) Compute Element (CE) jobs started running at HIP in November. Work on implementing a Local Scope DBS service and a CRABserver was started.
Operations. The Finnish CMS Tier-2 resources are operated, maintained and monitored jointly by HIP, CSC and NDGF. The beginning of the October exercise showed operations problems at CSC in connection with the Sepeli upgrade and dCache manpower allocation, but these problems have been addressed. According to the statistics collected with the CMS JobRobot, the Finnish Tier-2 resources performed very well and it was at the top of the CMS Tier-2 Readiness ranking list in December 2009.
The dCache system at CSC was very stable in 2009, but the available disk bandwidth was not enough to ensure good CPU efficiency for the CMSSW jobs.
The CMS PhEDEx and Frontier services were run on silo3 in Kumpula by the HIP Technology programme. More than 213 TB (Monte Carlo and test) of data was transferred with PhEDEx to Finland. A total of 433 thousand CMS grid jobs using 541 thousand SI2k CPU hours were run in 2009. In addition to this a significant amount of local batch jobs were also run.
Tracker alignment
One of the most demanding calibration activities for the CMS Tracker is the
geometrical alignment of its 15148 modules with respect to each other. The
work aims at a determination of some 100000 calibration parameters which
need to be recomputed in regular time intervals. Most parameters are
determined by analyzing computed particle trajectories in order to optimize
the reconstruction precision. The detector alignment is a highly important
issue in view of physics discoveries.
As in the previous year, a large number of cosmic events were recorded in 2009 during several periods and used for alignment. Alignment studies were also immediately started with the first collision events in December 2009. The alignment algorithm mainly used in these alignment studies, the HIP algorithm, was originally introduced by our team (V.Karimäki and T.Lampén). In July 2009, the CRAFT09 exercise was carried out. As a continuation, a 5-day Prompt Alignment Exercise took place in August 2009. During this exercise new alignment constants for the CMS Tracker were calculated on a daily basis with a cumulative sample of cosmic muons. In October-November, yet another sample of cosmic muons were recorded, and alignment with these data revealed a significant deformation of the pixel detector caused by a cooling problem of the TIB detector. Therefore new alignment calibration was performed with the latest data as a preparation for the data from the first collisions.
As part of the CMS Tracker alignment group, the HIP team participated actively in all these exercises, this time devoting mainly to validation and comparison of different alignment constants.
Geant4 development
The Geant4 development effort was focused on the INCL intra-nuclear cascade and
ABLA de-excitation model development in collaboration with a group at
Commissariat à l'Énergie Atomique, Saclay (A.Heikkinen and P.Kaitaniemi).
These models can be used for incident p, n, d, t, He-3, alpha and pions on
nuclei ranging from carbon to uranium with projectile energies from 200 MeV up
to 3 GeV. In 2009 we developed a new physics list that makes INCL/ABLA models
easier to use in Geant4 based applications. This new physics list has been
included in the latest official Geant4 9.3 release. We have also started the
implementation of the entirely new redesigned version of the INCL cascade code.
The light ion projectile support of the INCL intra-nuclear cascade model was extended up to carbon. This allows us to calculate ion-ion reactions, such as carbon on oxygen target, that are relevant to the Geant4 medical user community. We also participated the development of the Geant4 Hadrontherapy application in an IAEA coordinated research project 'Heavy charged-particle interaction data for radiotherapy'. This application allows us to test the new ion-ion features of the INCL model and directly benchmark it against other Geant4 models, such as the Binary cascade.
Test beam off-line data analysis
The development of the CMSSW-based off-line analysis code for the Silicon Beam
Telescope collaboration (SiBT) continued together with the CMS Tracker
Operations project. Our project provided software and analysis support during
the beam tests of summer 2009 (M.Kortelainen and T.Lampén). The project
participated actively in the subsequent data analysis, for which the CMS
Tracker Operations project carried the main responsibility.
Outreach activities
The group members were active in presenting CERN and the LHC experiments
to visitor groups from Finland. The groups included high school students
and a group of high school teachers. The members also approached the
general public by writing popular articles and giving popular talks, by
giving interviews and by maintaining a blog describing the status and
goals of the LHC experiments. Special outreach sessions were arranged to
recruit graduate level summer students.
V. Karimäki, Docent, Proj. Leader
V. Azzolini, Senior Scientist (at CERN)
R. Kinnunen, Senior Scientist
K. Lassila-Perini, Senior Scientist (at CERN)
S. Lehti, Senior Scientist
T. Lindén, Senior Scientist, Grid Coordinator
T. Lampén, Senior Scientist
L. Wendland, PostDoc Scientist
A. Heikkinen, Grad. Student (- June 30)
P. Kaitaniemi, Grad. Student (in Saclay)
M. Kortelainen, Grad. Student
J. Välimaa, Grad. Student
G. Danielsen, Summer Trainee (at CERN)
E. Ruokokoski, Summer Trainee (at CERN)