TopGroupStartUp
Top Group: project for start-up activities
The group
| Name | Position | Availability [mm/yy-mm/yy] | % time | Expertise |
|---|---|---|---|---|
| Jorgen D'Hondt | Professor | |||
| Catherine Vander Velde | Professor | |||
| Volker Adler | Post-doc | 08/09- ... | ~20% | SiStrip DQM, Analysis Software, from ZEUS: FCNC/l+MET, SiStrip (vertex) calib + digi |
| Stéphanie Beauceron | Post-doc | I believe close to whole period | ~60% | Calibration, Ecal, Trigger, Electron ID, Dead Channels, From my thesis on D0: calorimeter (Ecal+Hcal), noise studies. missing Et, W+jets with first data (with b-tagging) |
| Eric Chabert | Post-doc | 08/09- ... | ~50% | Tracker (noise & dead channels), Muon-ID, jet algo, Ana: Mttbar reconstruction |
| Grégory Hammad | senior PhD | |||
| Joris Maes | senior PhD | |||
| Petra Van Mulders | senior PhD | |||
| Ilaria Villella | senior PhD | |||
| Stijn Blyweert | PhD | |||
| Thierry Caebergs | PhD? | |||
| Alexis Kalopogeros | PhD | |||
| Michael Maes | PhD |
CRAFT09 Activities
2 Main axis:
- Muon Id: Michael (+Stephanie and Eric) - Jet Id/Soft Muon B-tag: Stijn (+Eric and Stephanie) - Studies of Recovery of Ecal Dead Channels in Cosmic: Stephanie
2 reprocessing is done:
- Cosmics: Use different timing for Muon chamber and non pointing track reconstruction, Ecal Amplitude reconstruction use Analytic Fit method (Fit the whole pulse [Aske Stephanie if more explanation are needed) - Collisions: Use collision timing for Muons (so upper muons should not be seen anymore), ask for the track to come from collisions point (tighter than SuperPointing Skim), Ecal Amplitude reconstruction is using the weight method.
Dataset of first reprocessing is [Cosmic sequence]: /Cosmics/CRAFT09-CRAFT09_R_V4_CosmicsSeq_v1/RECO Skim should arrive soon.
In more details (over SuperPointing Skim):
Muon Id studies
Following the note CFT-09-014 http://cms.cern.ch/iCMS/analysisadmin/cadi?ancode=CFT-09-014
Studies in parallel the different muons collections: StandAlone Muons, Tracker Muons, Global Muons (Tracker Muons should be the same than Global Muons except that we are considering ONLY the information coming from the track).
Perform correlations plots within this collection for given variables (basic ones are pt, eta, phi, normalized chi2, PCA, resolution of pt, eta, phi...).
Resolution on the variables should be obtained via the 2 legs method.
Others distributions:
Number of Hits for each muons collections, Number of tracks around the muons tracks and then looking at the different MuonId variables as defined in this note http://cms.cern.ch/iCMS/jsp/openfile.jsp?tp=draft&files=AN2008_098_v1.pdf
Study also the isolation variables (already done in CRAFT08).
In all this it is important to make sure that the distributions are similar to the one in the paper of CRAFT08.
Second point will be to make sure that distribution make sense.
All these distributions can also be compared to Cosmic MC.
Look at the trigger efficiencies for Muons (--> Stephanie).
Jet Id and Soft Muon b-tagging
Following the note AN2009_087 http://cms.cern.ch/iCMS/jsp/openfile.jsp?tp=draft&files=AN2009_087_v4.pdf
3 Jets collections should be looked at: Iterative Cone, SisCone and Anti-Kt (new default for CMS)
Check the usual distributions: Nbjets, pt, eta, phi (within |eta|<2.5)
Checks the correlation of this variable within the different algorithm used.
Look at the variables defined in the note (Fem, N90, Sigma_phiphi etc) and mainly at the ones design to reject HPD noise.
For the soft lepton b-tag:
Look at the DR between Muon and jet axis, look at pt_rel (should be pt_muon/pt_jet), look at the IP of the muon track within the jet vicinity, checks the number of tracks in the jet and compare the sum of their pt to jet pt...
Look at the different b-tagging discriminator which are run in cosmic sequence (not sure that they are really runs).
Make sure that distribution make sense.
Not sure that it makes sense to compare distributions to MC, except to see which kind of noise we are currently simulating in MC and compare it to real data.
From CRAFT to collisions
Once previous studies are performed, the same code can be rerun on collision like reconstructed CRAFT. Objects will be different (so we do not really expect any more muons), jets can remain and give us an idea of what is the rate of fake jets that we can expect from noise.
This steps will be in preparation of the validation and studies of the samples where cosmic event will be added to ttbar Monte Carlo ones (in view of Top Commissioning Activities)
Top commissioning
Tim's Proposal
hypernews https://hypernews.cern.ch/HyperNews/CMS/get/top/614.html
slides http://indico.cern.ch/materialDisplay.py?contribId=0&materialId=slides&confId=54470
Summary
a) Object commissioning:
- Performance of lepton-ID and isolation (instrumental impact on lepton isolation can already be looked at in CRAFT data) - Tag and Probe on Muon using the second leg ==> efficiencies - Jet energy scale - Fake rates - Impact of dead and/or noisy channels on top selection. - MET resolution measurement with Z (+njet) events. MET in CRAFT and at 900 GeV... - Beam-spot stability. - Object disambiguity. - B-tagging - ....
b) Trigger validation:
- tag & probe in Z events to measure the trigger performance "per lepton" (Z and top topologies are very different!), but to complement the picture by studying - multi-jet triggered events with >= 1 reco lepton to independently measure the lepton-trigger performance in l+jets topologies slides 19 of Marta's talk
http://indico.cern.ch/materialDisplay.py?contribId=10&sessionId=5&materialId=slides&confId=50147).
c) Top-specific commissioning: O(1/pb) of data. For each of the main top channels (e+jets, mu+jets, di-leptons)
- jet and lepton pT spectra - jets over threshold in eta/phi - jet multiplicities, "Berens scaling" in Nj=0,1,2..3, comparison to MC (so QCD, W/Z+jets at this stage) - the data-driven methods for QCD-background estimation can be tested to separate W/Z+jets in lower Njet bins, compare to expectations ... - ==> re-optimization of base thresholds/selection (iso, lepton/jet-pT, MET, |eta|, ...) - ...
d) "automatic" physics validation or DQM
- use of DQM @ T0 vs. CAF express-stream monitoring vs skimming/pat-tuple production (e.g. at T2) - definition and implementation of top-pag skims - definition of a set of high-level quantities that we can monitor on a run-to-run basis and that can influence run-quality labeling - ...
Additional thought from Stephen Winpenny
hypernews https://hypernews.cern.ch/HyperNews/CMS/get/top/614/1.html
Variables to study at start up (very easy and prompt), kind of DQM plots for top
1) Muons: - eta x phi maps (look for hot region etc)
- Pt spectrum and MC tuning
- Calorimeter/Tracker Isolation
2) Met: - phi maps (look for hot region)
- muon corrections on MET
3) Jets: - eta x phi maps (look for hot region etc)...
4) Primary Vertex: d0 comparison vs MC
5) Study of inclusive W-> lnu etc
CRAFT data
Workshop Turin (march 09) http://cms-torino-2009.to.infn.it/index.html
Summer Students Work (For CERN group ones, we will see in detail later on who is going to do what)
3 main activities:
1) Multiple Muons studies: - No MC develop so far - Check to make sure that Calibration events are not mixed up with physic events - Identify them, check the rate, trigger efficiencies etc - Muon chamber resolution (all multiple muons should come the same vertex, by constraining to the vertex, we can study segment resolution) - Test magnetic field, alignment etc
2) Showering Muons: - Will be identified as Multiple Muons but it is single muon that will start a shower in the detector (iron) - Can test the material budget (iron but also calorimeter wide)
3) Charged Ratio: Muon charge ratio should be around ~1.3 - Determining the charge of cosmics muon, understanding mischarge measurement - Determine efficiencies (mainly trigger as trigger efficiency is not identical with respect to charge) and acceptance - Study charge ratio at high energy (pt > 500 GeV) - wiki page of on going analysis https://twiki.cern.ch/twiki/bin/view/CMS/MuonChargeRatio
--> This study is a long term study [not only for summer student]
+ Possibility to study recovery of dead channels in Ecal (over cosmics)
Additional Possible Analysis from Muon POG
Wiki https://twiki.cern.ch/twiki/bin/view/CMS/MuonPOGandCRA
Additional Ideas for Post Docs / PHD [but can be done by summer students too]
4) Look at reco objects in CRAFT data using skimmed events (can be super pointing skim that ask a track going through the center of the detector --> should mimic quite closely collisions). Apply basic cuts on objects (pt, eta etc) and see what is remaining (quantify number of jets, electrons, muons, value of MET, perhaps we have already some W->l nu ?) ;-) . Study rate of such events (see if it is a negligible background or if cuts should be developed to remove them)
5) Run Particle Flow reconstruction over super pointing skim (should not have trouble as tracks are pointing towards vertex)
6) Use cosmics event to superimpose on MC events. Study the properties of the objects when doing so.
Proposals
Constraints
Ideas
Stephanie: For me, working on 3) with summer students or on 6) is very interesting. I like also a lot 4). 4) and 6) are in any case highly correlated, so we can split a bit the post doc in 2 over summer time (to put in place infrastructure in both cases), and join together some time in September where we exchange findings (removing hot cells from cosmics, applying it on MC, are we removing good cells etc). 3) is quite decorrelated from top activities. 4) and 6) and full fill the requirements a) and c) of Tim email.