Validation of DC3 Validation of DC3 fully simulated fully simulated W W → → e e ν ν samples samples (NLO, (NLO, reconstructed in reconstructed in 11.0.42) 11.0.42) Laura Gilbert 01/08/06
Dec 19, 2015
Validation of DC3 fully Validation of DC3 fully simulated Wsimulated W→→eeνν samples samples (NLO, reconstructed in 11.0.42)(NLO, reconstructed in 11.0.42)
Laura Gilbert 01/08/06
WW→→eeνν from MC@NLO, full simfrom MC@NLO, full sim Original sample submitted for generation in Original sample submitted for generation in
December, some of the one million December, some of the one million WW→→ee--νν now fully now fully simulated on the grid, in batches of 10k events.simulated on the grid, in batches of 10k events.
47800 events available (using CBNTs)47800 events available (using CBNTs) Lepton generator level filter efficiency (10Gev, |Lepton generator level filter efficiency (10Gev, |ηη|<2.7) was |<2.7) was
~63.3%~63.3% Generated cross section is 8.4nb (cf. 30nb for Generated cross section is 8.4nb (cf. 30nb for WW→→ll--νν: TDR). : TDR).
Luminosity for 10k events is 1.9pbLuminosity for 10k events is 1.9pb-1-1.. Cuts applied at any stage are listed on top right of slide.Cuts applied at any stage are listed on top right of slide. ALL truth plots are normalised to weighting of simulated data ALL truth plots are normalised to weighting of simulated data
EXCEPT multiplicity plots.EXCEPT multiplicity plots.
ELECTRONSELECTRONS
- truth- full
sim
Sanity plots: multiplicitySanity plots: multiplicity
Electron Selection CutsElectron Selection Cuts
At least one electron with transverse energy > At least one electron with transverse energy > 25GeV. (Use ET rather than pT since brem 25GeV. (Use ET rather than pT since brem recovery not properly implemented yet0recovery not properly implemented yet0
Electron candidate has at least one associated Electron candidate has at least one associated track matched (exclude photons)track matched (exclude photons)
ηη < 2.4. This cuts out dubious candidates at < 2.4. This cuts out dubious candidates at the end of the electromagnetic calorimeter.the end of the electromagnetic calorimeter.
IsEM flag = 0, electron isolation cut. This IsEM flag = 0, electron isolation cut. This should now be tuned correctly but I need to should now be tuned correctly but I need to look into the bitwise cuts in more detail to look into the bitwise cuts in more detail to check.check.
ELECTRONSELECTRONS
- truth- full
sim
Sanity plots: multiplicity of electrons that pass Sanity plots: multiplicity of electrons that pass selection cutsselection cuts
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
Almost always only one electron
candidate per event passes cuts.
ELECTRONSELECTRONSCuts:
ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4Checking cuts with truth info:Checking cuts with truth info:
- before cuts- after cuts
Initially tracks of candiadates entered into egamma container include kaons, pions, muons. After cuts only electrons remain.
ELECTRONSELECTRONSSanity plots: phiSanity plots: phi
- truth- full
sim
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
ELECTRONSELECTRONSSanity plots: etaSanity plots: eta
- truth- full
sim
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
ELECTRONSELECTRONSIdentifying electrons with their partner in Identifying electrons with their partner in truth containers, matching (truth containers, matching (ηη,,φφ) space) space::
22 )()( TTR
If (ΔR<0.1) match is found
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
ELECTRONSELECTRONSSanity plots: ETSanity plots: ET
- truth- full
sim
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
ELECTRONSELECTRONSDifference between simulated and truth ET Difference between simulated and truth ET distributions:distributions:
No overall difference between simulated and truth ET shapes!
PROBLEM?
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
ELECTRONSELECTRONSElectron ET resolution: (sim-truth)/truth for Electron ET resolution: (sim-truth)/truth for matched sim:truth electron pairs.matched sim:truth electron pairs.
Cuts: ET(ele) > 25GeV At least one track
matched IsEM=0 η(ele) < 2.4 ΔR<0.1
(matching sim to truth)
Resolution looks roughly ok though…
ELECTRONS ELECTRONS "Wrong sign" electrons found from associated "Wrong sign" electrons found from associated track q/p, AFTER cuts:track q/p, AFTER cuts:
Wrong sign
Correct sign
*PROBLEM*
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
ELECTRONSELECTRONSET distribution of "wrong sign" electron ET distribution of "wrong sign" electron candidates:candidates:
- +ve charge electrons (from track q/p)
- all electron candidates (for shape comparison)
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
ELECTRONSELECTRONSConsider truth information: just cut on ET > 25GeVConsider truth information: just cut on ET > 25GeV
- before cuts- after cuts
There are positrons present in egamma truth, but none pass ET cut.
Cuts: ET (ele truth)
> 25GeV
ELECTRONSELECTRONS
What are the wrong sign electrons?What are the wrong sign electrons? Not generated positronsNot generated positrons Don't have high pT tracks, so probably Don't have high pT tracks, so probably
not electrons mis-reconstructed.not electrons mis-reconstructed. Possibly wrong track-cluster Possibly wrong track-cluster
combinations. Needs further combinations. Needs further investigation.investigation.
Cuts: ET(ele) > 25GeV At least one
track matched IsEM=0 η(ele) < 2.4
ELECTRON ISSUESELECTRON ISSUES
E over p always 0! → needs to be fixed in future.
Where do the "positrons" come from?
Need to look into IsEM flags in detail.
Further investigation needed:
ELECTRON JETSELECTRON JETS
W Transverse mass plots look fishy if I apply W Transverse mass plots look fishy if I apply exclusions on events containing jets with exclusions on events containing jets with ET>30GeV.ET>30GeV.
Therefore many jets come from electrons.Therefore many jets come from electrons.
Identify and exclude these electron jets. Identify and exclude these electron jets.
Can only be one on jets with Can only be one on jets with ηη<2.5 to make <2.5 to make electron matching possible.electron matching possible.
(Jets reconstructed using cone algorithm with R<0.7, see (Jets reconstructed using cone algorithm with R<0.7, see later)later)
ELECTRON JETSELECTRON JETSCuts:
ET(ele) > 25GeV At least one
track matched IsEM=0 Jet η < 2.5
22 )()( jetejeteR
Identify jets with electrons if ΔR<0.7
- truth- full
sim
Matching electrons and jets in (Matching electrons and jets in (ηη,,φφ) space:) space:
Log scale→
ELECTRON JETSELECTRON JETSCuts:
ET(ele) > 25GeV At least one
track matched IsEM=0 Jet η < 2.4
Jets are corrected to have more energy than electron candidates so -ve values more likely to be a match.
Trying to match energy as well as (η,φ) distribution (no cut made):
- truth- full
sim
Matching electrons and jets in energy:Matching electrons and jets in energy:
ELECTRON JETSELECTRON JETSCuts:
ET(ele) > 25GeV At least one
track matched IsEM=0 Jet η < 2.4A closer look at electron-jet A closer look at electron-jet ΔΔR and R and
energy distributions: simulationenergy distributions: simulation
Smattering of events
here
ELECTRON JETSELECTRON JETSCuts:
ET(ele) > 25GeV At least one
track matched IsEM=0 Jet η < 2.4A closer look at electron-jet A closer look at electron-jet ΔΔR and R and
energy distributions: truthenergy distributions: truth
no events here
ELECTRON JETSELECTRON JETSCuts:
ET(ele) > 25GeV At least one
track matched IsEM=0 Jet η < 2.4A closer look at electron-jet A closer look at electron-jet ΔΔR and R and
energy distributions: (truth - simulation)energy distributions: (truth - simulation)
This is the normalised difference of the last two plots!
ΔR
Energy difference (MeV)
Truth - simulated
JETSJETSCuts:
ΔR (e-jet) > 0.7 Jet η < 2.4
Looking at all jets which have not been matched Looking at all jets which have not been matched with electrons: ie. Jets which are isolated from with electrons: ie. Jets which are isolated from electron candidates by at least 0.7 in (electron candidates by at least 0.7 in (ηη,,φφ).).
Still looking at jets with Still looking at jets with ηη<2.4 for accurate <2.4 for accurate electron matching.electron matching.
There are three available clustering algorithms:There are three available clustering algorithms: Kt alorithm: apparently unsuitable for low pT jets.Kt alorithm: apparently unsuitable for low pT jets. Cone algorithm with R<0.4. Discounted as produces Cone algorithm with R<0.4. Discounted as produces
very low statistics (few jets reconstructed)very low statistics (few jets reconstructed) Cone algorithm with R<0.7. Use this.Cone algorithm with R<0.7. Use this.
JETSJETSSanity plots: multiplicity before any selection:Sanity plots: multiplicity before any selection:
- truth- full
sim
JETSJETSCuts:
ΔR (e-jet) > 0.7 Jet η < 2.4
Sanity plots: multiplicity after electron/jet Sanity plots: multiplicity after electron/jet matches removed, jet matches removed, jet ηη<2.4<2.4
- truth- full
sim
Log scale→
Multiplicity greatly reduced
JETSJETSSanity plots: phiSanity plots: phi
Cuts: ΔR (e-jet) > 0.7 Jet η < 2.4
- truth- full
sim
JETSJETSSanity plots: etaSanity plots: eta
Shape before η
cut
Cuts: ΔR (e-jet) > 0.7 Jet η < 2.4
- truth- full
sim
JETSJETSCuts:
ΔR (e-jet) > 0.7 Jet η < 2.4 ΔR(jet-
truth)<0.5
Matching jets with truth in (Matching jets with truth in (ηη,,φφ) space:) space:
22 )()( TTR
Log scale→If (ΔR<0.5) match is found
JETSJETSSanity plots: ETSanity plots: ET
Cuts: ΔR (e-jet) > 0.7 Jet η < 2.4
- truth- full
sim
Jets not reconstru
cted < 10GeV
JETSJETSET resolution: (sim-truth)/truthET resolution: (sim-truth)/truth
Cuts: ΔR (e-jet) > 0.7 Jet η < 2.4 ΔR(jet-truth)<0.5
JETSJETSSanity plots: EnergySanity plots: Energy
Cuts: ΔR (e-jet) > 0.7 Jet η < 2.4
- truth- full
sim
JETSJETSEnergy resolution: (sim-truth)/truthEnergy resolution: (sim-truth)/truth
Cuts: ΔR (e-jet) > 0.7 Jet η < 2.4 ΔR(jet-truth)<0.5
JETSJETSSanity plots: ETSanity plots: ET
Cuts: ΔR (e-jet) > 0.7 Jet η < 2.4
- truth- full
sim
Strange: data is shifted down from truth.
22 pEmass
JET ISSUESJET ISSUES
Jets from electrons are easily identified and Jets from electrons are easily identified and removed by an exclusion cut in (removed by an exclusion cut in (ηη,,φφ) space.) space.
Not certain this is the best algorithm, but Not certain this is the best algorithm, but other available algorithms have more other available algorithms have more problems.problems.
Jet statistics low after removal of electrons, Jet statistics low after removal of electrons, therefore resolutions not well defined.therefore resolutions not well defined.
Meaning of jet “mass” not well understood, Meaning of jet “mass” not well understood, left-shift of simulation w.r.t. truth left-shift of simulation w.r.t. truth unexpected.unexpected.
MISSING ENERGYMISSING ENERGY
- truth: calculated from non-interacting particles
- full sim: corrected, inc. muons
Sanity plots: Missing ETSanity plots: Missing ET
Probably not properly
calibrated
Cuts: Ptele > 25GeV At least one
track matched IsEM=0 Electron η < 2.4 Jet η < 2.4 ΔR (e-jet) > 0.7 MET>25GeV
Cut on MET>25GeV
MISSING ENERGYMISSING ENERGYCuts:
Ptele > 25GeV At least one
track matched IsEM=0 Electron η < 2.4 Jet η < 2.4 ΔR (e-jet) > 0.7 MET>25GeV
Missing ET Resolution:Missing ET Resolution:
Asymmetry reflecting right-
shift of data w.r.t. truth
MISSING ENERGYMISSING ENERGY
- truth- full
sim
Cuts: Ptele > 25GeV At least one
track matched IsEM=0 Electron η < 2.4 Jet η < 2.4 ΔR (e-jet) > 0.7 MET>25GeV
Missing ET Missing ET φφ distribution distribution::
W RECONSTRUCTIONW RECONSTRUCTION
- truth- full
sim
Cuts: Ptele > 25GeV At least one
track matched IsEM=0 Electron η < 2.4 Jet η < 2.4 ΔR (e-jet) > 0.7 MET>25GeV
W transverse mass reconstructed from W transverse mass reconstructed from Missing ET and highest ET electron:Missing ET and highest ET electron:
)cos1(2 Tmiss
Te
TW ppm
Conclusions:Conclusions:
The W looks like TDR (right)! The W looks like TDR (right)! Electrons are well reconstructed.Electrons are well reconstructed. The e/p variable in the egamma container The e/p variable in the egamma container
should probably be fixed.should probably be fixed. Electrons with mis-identified charge not Electrons with mis-identified charge not
yet understood.yet understood. Missing energy doesn’t seem best Missing energy doesn’t seem best
corrected to match truth.corrected to match truth. Insufficient stats for decent jet resolution.Insufficient stats for decent jet resolution. Jet “mass” shift seems strange.Jet “mass” shift seems strange.