February 2009
26 February 2009
- worked with Avi and Claudio on muons for met correction , notes can be found here
- brown bag by Warren and fkw on mSugra study
25 February 2009
- morning meeting with Puneeth and Avi to clarify workflow for met muon corrections
- meeting with Johannes and Kevin to discuss definitions of vertex (reference) point and d0 - no resolutions yet
- started studying refinements to tcMET implementation
24 February 2009
- validated new tcMet made by Ronny, instrucitons for use on public twiki
- V00-04-02-16 RecoMET/Configuration
- emails were sent to Sasha, Chris Hill alerting them to the availability of tcMET tags
- emails sent to Elizabeth and Jim alerting them to the availability of the tcMET tags
- morning meeting with Filip, Claudio, Avi, Slava, Ronny, Puneeth and myself discussing muon associator for MET
- the following plan of action was drafted
hi all,
Today's discussion led to the following plan:
Each MET correction (muon and track) will store a bool and two floats for each muon in the event.
The muon correction will store, for each muon, (Ex, Ey) where Ei is the ith component of the energy removed from the calorimeter due to the MIP.
The tcMet correction wiill store, for each muon, (Ex, Ey) where Ei is the ith component of the energy removed from the calorimeter assuming the muon were corrected as a pion using the response function.
Attached below is a tex file containing some equations trying to clarify the words above.
From here, we agree to take the implementation of the corrections as is and freeze them until testing is done.
In the next week, we (myself and Puneeth) will create a first attempt to implement and store the above values and validate them against the current implementation. If things check out, we will commit and tag.
The muon experts from UCSB will then decide upon and submit improved baseline muon selections.
While this is going on: Filip, Claudio, and Avi will take care of politics and everything will be great.
-------------------------------------------------------------------------------------
Before ending, Avi and I talked a bit more on this morning's meeting and want to put forward a slight change to the proposed plan of work.
Rather than just storing a bool, we think it would be worthwhile to store an int. This would serve the purpose not only of telling the user whether or not a muon was corrected but also which momentum (global, tracker, standalone) was used in the correction. Thus, we suggest replacing the bool with:
0 - muon was not corrected
1 - muon was corrected using global momentum
2 - muon was corrected using track momentum
3 - muon was corrected using standalone momentum
Responses to this proposal would be greatly appreciated.
I think that summarizes this morning's meeting and our work for the next week. Anything missed or errantly reported?
Cheers,
Frank
- wrote draft of talk for presentation to MET subgroup on friday
23 February 2009
- validated new tcMet tags made by Ronny, instructions for use on public twiki
- V02-05-00-20 RecoMET/METAlgorithms
- V02-08-02-16! RecoMET /METProducers
- ntupled new samples
- /AstarJets-madgraph/Fall08_IDEAL_V9_v2/GEN-SIM-RECO
- /Wgamma/Summer08_IDEAL_V9_v1/GEN-SIM-RECO
- /Zgamma/Summer08_IDEAL_V9_v1/GEN-SIM-RECO
- added phi of corrected d0 to TrackMaker.cc and added function to tools.cc for on-fly correction
- it seems that we use vertex phi for d0 phi (in uncorrected case) but they are not the same?
21 February 2009
- cleaned up /data/tmp on uaf-6 to remove unnecessary remnants of old ntupling
- list of ntuples that still need to be made
- Zee, Zmm low mass DY samples
- Astar, Wgamma, Zgamma
- Zjets, ttbar, ww samples for tcmet study
- ttbar with single lepton filter
20 February 2009
- morning meeting with kevin, puneeth, yanjun, and avi to discuss starting an r&d project studying the utility of the hit pattern
- potentially useful for track selection (met) and electron identification (conversions)
- was decided to add hit pattern variables to track module and make a small ntuple from single pion sample to begin
- after, will run on large gamma and electron samples (get oli to make) to study ability to differentiate between electrons and conversions
- finalized tcmet code to be submitted for 22X
- unable to commit to cvs due to insufficient permissions, emailed files to ronny and asked him to commit, tag and publish
- derived method to calculate impact parameter and associated angle corrected for displaced beam spot (for use in jet probability b-tagging)
use linear approximation to model track as y cos(phi) - x sin(phi) + d0 = 0, where phi is the azimuthal angle of the momentum vector at the impact point and d0 is the uncorrected impact parameter (measured from origin)
NOTE: the sign convention used for d0 (and hence the sign on d0 above) was determined by trial and error
now, the displaced beamspot is at (bx, by) and we can define a unit vector perpendicular to the track {-sin(phi), cos(phi)}
using this the corrected impact parameter is then d0corr = d0 - bx sin(phi) + by cos(phi)
then, using the beamspot and the normal vector the corrected impact point is {bx + d0corr sin(phi), by - d0corr cos(phi)} from which the corrected impact parameter angle is
d0corrPhi = atan( -cot(phi) )
19 February 2009
- morning meeting with filip and teruki
- interest in us making MET comparision between tcMET, pfMET, etc.
- more info on comparison to come shortly
- teruki interested in getting 22X tcMET plots approved - timeline is end of month
- send them a link to a public twiki page with results for their viewing pleasure
- timeline for additions to 31X code is mid-march
- raised issue of new muon correction implementations, sounded interested, meeting with ucsb and met conveners sometime early next week to continue dicussion
- interest in having tcMET in 22X susy patTuples
- i will talk to ronny about submitting our latest version and getting a tag
- submitted draft of tcMET note
- emailed Rick Cavanaugh about accessing PFMET in Fall08 samples
- PFMET is not available in samples reconstructed with releases prior to 223
- monica Acosta was kind enough to provide us with a correct version (redigi-rereco) of the sample we want
- /ZJets-madgraph/Summer08_IDEAL_V11_redigi_v1/GEN-SIM-RECO
18 February 2009
- snt meeting
- meeting with filip and teruki tomorrow morning at 9 a.m.
- discussion with avi, claudio, and puneeth about muon corrections for met
- agree that we will store a bool and two doubles (delta_x, delta_y) of the incremental changes to met for each global muon
- will discuss with filip and teruki at meeting tomorrow morning
- make new Zjets ntuple with qoverp for muons, siscone jpt, and PFMET
17 February 2009
16 February 2009
- wrote talk for snt meeting
15 February 2009
- submitted some ntupling jobs
- added more plots to validation page
14 February 2009
- i've been remiss in keeping up the diary the last few days
- have mostly been continuing validation of tcMET
11 February 2009
- morning meeting with teruki kamon, new met convener
- talked about pending deadline for submitting changes to cmssw, further discussion with teruki and filip to follow over the weekend or early next week
- end of month deadline approaching for approval of tcMET plots, plots must be from 22X
- summary of this morning's meeting below
Hi all:
Below are brief minutes of today's informal meeting.
Filip -- will you be available on Saturday or Sunday
to have another informal meeting with Avi and Frank?
For example .. Sunday at 10:30/central, 8:30/California, 17:30/CERN?
The discussion items are operational issues, namely how to deliver
the tcMET code by when. Please let me know if you are flexible..
Teruki
(1) Feb 28 ... deadline of Feb 28 for approving analysis plots with 2_2_x
I believe this is a soft deadline set by Peter and Daniel (JetMET conveners)
to proceed the appovale process (JetMET to CMS) by May or so.
"May or so" is a hard deadline.
I will check with Filip and Peter/Daniel to understand the procedure.
(2) BIG Picture: provide tcMET as a package for CMS for physics analysis
namely the code implementation.
(*) CMSSW package ? ... 3_0_0?
(*) When? ... May? or later??
(*) How?
Technical issues such as "configurable" calorimeter responce functions
<Frank, could you list big items?>
(3) Frank has been working on tcMET with 2_2_x and 3_0_0.
Note that those codes are based on hard-coded calorimeter response functions.
(There will be many technical issues that I should know. But
this is one of them that Frank told me today.)
So we have to address those issues (related to (2) above).
Other than that, Frank can have updated plots of tcMET using 2_2_x
by early next week.
(4) Golden Plot(s) ... see the attachment for the starting point of the discussion
We have to discuss about Golden Plots to demonstrate the performance
of tcMET visually well.
Filip and I picked MET and tcMET ditributions for Z(-->mumu) and W (-->mu nu) events
based on Avi's and Frank's suggestion.
Avi and Frank also suggested the same plots, but with Z(-->ee) and W (-->e nu).
But I picked muon channels, because it will be one of fast validation process at CMS.
J/psi --> mumu, Z --> mumu, Z --> tau tau --> mu + tau_h, etc..
So in my mind, if I had to choose, the muon-channels are my choice.
Please correct/add as necessary.
- details of validation of tcMET in 22X can be found here
10 February 2009
- ntupled some more samples
- /VQQ-madgraph/Fall08_IDEAL_V9_v1/GEN-SIM-RECO
- /ZJets-madgraph/Fall08_IDEAL_V9_reco-v2/GEN-SIM-RECO
- /TTJets-madgraph/Fall08_IDEAL_V9_v2/GEN-SIM-RECO
- started looking at hybrid implementations of tcMET in 16X, more details here
9 February 2009
- worked on finished second round of ntupling Summer08/Fall08 MC
- WZ_incl sample published, dieted, and merged
- submitted jobs for reRECOed Zjets sample
- Wjets sample dieted and merged; unable to publish Wjets sample due to error
- update: was able to get wjets ntuples to publish by removing corrupt xml files
fgolf@uaf-6:crab$ crab -publish -c WJets-madgraph_Fall08_IDEAL_V9_v1
crab. crab (version 2.4.3) running on Mon Feb 9 12:12:51 2009
crab. Working options:
scheduler glite
job type CMSSW
working directory /home/users/fgolf/snt_nm2/CMSSW_2_2_3/crab/WJets-madgraph_Fall08_IDEAL_V9_v1/
crab. <dbs_url_for_publication> = http://ming.ucsd.edu:8080/DBS1/servlet/DBSServlet
Traceback (most recent call last):
File "/code/osgcode/ucsdt2/Crab/CRAB_2_4_3/python/crab.py", line 839, in ?
crab.run()
File "/code/osgcode/ucsdt2/Crab/CRAB_2_4_3/python/crab.py", line 796, in run
if act in self.actions.keys(): self.actions[act].run()
File "/code/osgcode/ucsdt2/Crab/CRAB_2_4_3/python/Publisher.py", line 234, in run
if reports[0].status == "Success":
IndexError: list index out of range
- compare ZSP and non-ZSP response functions
- working on making projections of response functions in eta, pt
pt |
eta |
zspRF |
non-ZSP RF |
2.5 |
0.2 |
0.37 |
0.54 |
2.5 |
2.1 |
0.31 |
0.66 |
7.5 |
0.2 |
0.35 |
0.72 |
7.5 |
2.1 |
0.73 |
0.81 |
14 |
0.2 |
0.79 |
0.74 |
14 |
2.1 |
0.83 |
0.83 |
26 |
0.2 |
0.77 |
0.84 |
26 |
2.1 |
0.87 |
0.83 |
36 |
0.2 |
0.77 |
0.85 |
36 |
2.1 |
0.93 |
0.83 |
46 |
0.2 |
0.77 |
0.86 |
46 |
2.1 |
0.95 |
0.83 |
- plots comparing tcMET implemented with ZSP and non-ZSP RFs can be found here
- have submitted jobs to ntuple qcd samples with four different pt ranges to include JPT collection, will use this to calculate hybrid tcMET as outlined here
December 2008
18 December 2008
- updated test_cfg.py for CMS2 so that it produces tcMET
- updated draft of contribution to jet met note
- added section on qcd
- updated figures
- added mention of samples
- updated plots
- took another look at JPT, looks like we can't just store a single float, will need to store four vector
- checked out jetmet note and added our stuff
- figures currently don't show, reason unknown
17 December 2008
- continuing working on JPT for CMS2
- sanjay explained how to set up a CMSSW release area in /store/disk01/data on cms-tas01
- . /afs/cern.ch/cms/LCG/LCG-2/UI/cms_ui_env.sh
- . /afs/cern.ch/cms/sw/cmsset_default.sh
- export CASTOR_HOME=/castor/cern.ch/user
- export STAGE_HOST=castorcms
- export RFIO_USE_CASTOR_V2=YES
- export CMS_PATH=/afs/cern.ch/cm
16 December 2008
- checked out and ran 22X tags for tcMET
- at first glance, things look OK
- will do a more detailed validation tomorrow
- wrote module for CMS2, commited to cvs
- list of 22X tags for tcMET
- cvs co -r V02-08-02-14 RecoMET/METProducers/
- cvs co -r V02-05-00-16 RecoMET/METAlgorithms/
- cvs co -r V00-06-02-09 DataFormats/METReco/
- cvs co -r V00-04-02-15 RecoMET/Configuration/
15 December 2008
- rewrote draft of contribution to jet met note
- started looking into JPT
- initial observations indicate that only an overall scale factor is involved so we should only have to store a float to connect each jet with the corresponding jpt
11 December 2008
- jobs (mostly) done running
- ran on four qcd samples
- /QCD_Pt_20_30/CMSSW_1_5_2-CSA07-2162/GEN-SIM-DIGI-RECO
- /QCD_Pt_50_80/CMSSW_1_5_2-CSA07-2049/GEN-SIM-DIGI-RECO
- /QCD_Pt_120_170/CMSSW_1_5_2-CSA07-2171/GEN-SIM-DIGI-RECO
- /QCD_Pt_230_300/CMSSW_1_5_2-CSA07-2050/GEN-SIM-DIGI-RECO
- output of jobs can be found in appropriately named folders in /data/tmp/fgolf on uaf-4
- results of qcd met study, will add comments to this linked page later
10 December 2008
- fkw showed me where old qcd samples are
- validated ntuple makers and submitted crab jobs
- took a look at signed impact parameter study while jobs were running
9 December 2008
- set up code, etc for qcd met study
- spent afternoon trying to find samples
- snt weekly meeting
- issues submitting tcMET producer to crab because of file handling, otherwise producer works
- need to create JPT producer
8 December 2008
- spent weekend helping barbara prepare thesis talk
- basically took day off to celebrate with barbara - she passed her thesis defense!
5 December 2008
- validated tcMET EDProducer for cms2
4 December 2008
- wrote and debugged tcMET EDProducer for cms2
3 December 2008
2 December 2008
- worked on draft of cms note
1 December 2008
- worked on draft of cms note
November 2008
20 November 2008
- s&t meeting at fnal
- dave gives presentation on updating met cuts using tcMET
- for ttbar with at least 2 jets, proposes moving from MET > 45 to tcMET > 35
- for WW, have so little DY background that proposal is to accept a bit more and move from MET > 45 to tcMET > 35
- pythia and alpgen samples are in reasonable agreement - check the integral of the weighted pythia above 45 and remake overlay on linear scale
- integral of MET > 45 tail of alpgen is 7469, for pythia with weight is 9124
- slight improvement using tcMET for balance (angle) cut
- slightly worse using tcMET for pMET
- for WW, decision was to keep cuts the same but use tcMET instead of MET
- slava showed that with a tcMET > 50 GeV cut, one can include the Z peak in the ttbar analysis
- dave gives presentation on DY estimate method for WW analysis
- good first pass, need to keep working on it
- claudio gives presentation on ntuple wish list * there are things to do - i need to talk with sasha about JEC for JPT
- ingo and oli give presentation on DSG
- avi gives presentation on DSG big picture strategy
19 November 2008
18 November 2008
- Brief chat with Avi where we decided to
- write CMS note based on work done in 1_6_12
- test out using the non-ZSP from Sasha
- Weekly SnT meeting
- dave and dima will present status of revised MET cuts
- dave and slava will present status of DY estimates
- fkw and claudio will present on plans and procedures for future ntuple making
- status of dilepton search grid (DSG) by oli/ingo/avi
- dinner at lothar's
- looper tutorial by ingo/oli/johannes
- Make validation of tcMET EDProducer
- ttbar results are here - everything looks OK
- zmm results are here - NOT OK
- tcMET implementation is NOT validated
17 November 2008
- Added documentation to and cleaned up code for tcMET EDProducer, submitted to CVS
- Phone conversation with Ronny Remington
- tcMET EDProducer looks good - intial validation run on ttbar sample in CMSSW_2_1_9 successful validation plots
- we will not store a vector of additive corrections as these are not part of the default MET object and are easily obtainable from tcMET and CaloMET
- only potential issue is current hard-coding of response function in TCMETAlgo - we'll discuss with RecoJet conveners and probably need to transition to a text-based RF in future once Ronny has machinery in place
- Emailed Sasha to inform him that tcMET EDProducer is usable and facilitate exchange of our response functions
- List of additional features to add to tcMET EDProducer
- read quality and kinematic cuts from a .cfi rather than hard coding them
- store list of muons, electrons that have received "special treatment" from correction
- allow for multiple response functions (this needs to be thought about more)
- List of outstanding issues to work on for tcMET
- derive response function in 2_1_X
- study low pt tracks (can we lower pt threshold? if so, how do we properly implement this?)
- take another look at muon handling
- take another look at electron handling
- study type-I JES corrections, type-I muon corrections
- study use of nonZSP RF in core of jets
- I updated the WWMetStudies page with some additional comments and observations. Of particular note:
- the current table only contains pythia in the dymm column and this cannot be trusted as there is contamination from final states with taus. I will implement a filter that will produce correct numbers for the various DY processes shortly.
- IM conversation with Dima to sort out discrepancies between loopers when producing current table with alpgen samples, nTrks > 2, and MET -> tcMET
- was able to duplicate numbers from the WWstudy with the new looper
- Dima was able to track down the problem to an interference with ROOT in the met correction code for the old looper and has fixed the problem
- dima says, "basically static variable worked in some weird way and RF was always zero"
--
FrankGolf - 17 Nov 2008