Download presentation
Presentation is loading. Please wait.
Published byMaurice Craig Modified over 9 years ago
1
Parton-level study of Z l + l - for luminosity measurement Motivation PDF uncertainties Parton-level study & rate estimation Relaxed cuts & Conclusions Ricardo Gonçalo
2
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 052 Z l + l - for luminosity measurement Proposed in: Dittmar, Pauss & Zurcher Phys.Rev.D56 7284 (97) Basically just count nr of Z ee and Z Shown in Rome meeting by J.Pinfold and Bryan L.Caron (see also next slide): lumi = 8.6 – 28 hours First Year Operations L = 2 x10 33 cm -2 s -1 W : (12.0 – 2.4) Hz Z : (2.0 – 0.4) Hz Significant problem seems to be PDF uncertainties (several % uncertainty)
3
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 053 NNLO Calculations & PDFs (J.Pinfold’s talk) Bands indicate the uncertainty from varying the renormalization ( R ) and factorization ( F ) scales in the range: M Z /2 < ( R = F ) < 2M Z At LO: ~ 25 - 30 % x-s error At NLO: ~ 6 % x-s error At NNLO: < 1 % x-s error Anastasiou et al., Phys.Rev. D69:094008, 2004 Similar improvement in calculation for W at NLO and NNLO VRAP code at http://www.slac.stanford.edu/~lance/Vrap/http://www.slac.stanford.edu/~lance/Vrap/ PDFs - See talk from A. Tricoli from the Standard Model session for some estimated PDF uncertainties upon measured W→ l distributions. Currently PDFs contribute several percent to the errors.
4
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 054 What do the PDF uncertainties look like at Q 2 ~M Z ? MRST2001 - 30 PDF “eigenvector” sets
5
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 055 Can we improve the situation? The idea is to try to select Z bosons produced almost at rest in the lab frame to try to reduce the range of Bjorken x in the hard scatter. To do this, select events with Z decaying to two leptons back-to-back in the R- Z plane using pseudorapidity ( 1 = - 2 0.1) Should get for free lepton energies close to 45GeV and back-to-back leptons in phi => can also cut on these variables Hopefully this would limit x to a region where PDF uncertainties are small. At LO we should have log 10 x~-2.2: P1P1 Z0Z0 P2P2 X2P2X2P2 X1P1X1P1 (x 1 +x 2 ).P=q
6
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 056 Naïve case Pythia 6.321; CTEQ5L PDF set; 200k events: No initial-state parton showers No multiple interactions Only pp Z (ISUB=1) Simple cuts only: Preselection: 2 electrons or muons of opposite charge from a Z decay (truth only, no mis-ID) Selection: | l |<1 to have a clean sample 2 = l 0.1 to select Z “at rest” in beam direction
7
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 057 Naïve case
8
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 058 Naïve case Lepton p T above 25GeV for selected sample good trigger efficiency Hadronic p T low (note this also includes very forward particles which are missed by detector) Expect hadronic p T to increase in more realistic case
9
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 059 Naïve case m Z calculated from stable lepton momentum only, no correction for FSR (and plot slightly distorted by generator-level cut) All seems to work for naïve case but resolutions will be smeared when initial PS included Most critical is the effect on x
10
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0510 With initial-state parton showers Initial-state parton showers Approximates higher-order processes Range of x now very wide but still limited to x>6.5x10 -3 This may still be good to minimise PDF errors due to DGLAP evolution integrals P1P1 Z0Z0 P2P2 X1P1X1P1 X2P2X2P2
11
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0511 With initial-state parton showers
12
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0512 With initial-state parton showers Lepton p T > ~20 GeV trigger efficiency should still be ok
13
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0513 With initial-state parton showers Z 0 ’s momentum still pretty much ok
14
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0514 Rates and efficiencies Apart from the low statistics, some (naïve?) estimates can be made: Assume preselection is all that can be seen in detector with 100% acceptance in |eta|<2.5 Production rate is: L~10 33 cm -2 s -1 ~ 1nb -1 s -1 ; x-section ~ 43nb (Pythia) => R=43Hz Other factors (ex.BR to e and mu ~2x3.36% =>2.89Hz; etc…) Preselection (|eta| 9) gives ~2.9%-3.0% (~6000sel./200k gen.) Reduction factor from the eta cuts is: 1.9%+-0.2% (no initial PS) and 2.1%+-0.2% (with PS) This means: Preselection: ~1.3Hz Additional “back-to-back” cuts: ~0.026Hz And so (at low luminosity): 3.8x10 3 s ~ 1 hour to get 10% statistical accuracy (~100 ev) 15x10 3 s ~ 4.2 hours to get 5% 384x10 3 s ~106 hours to get 1% Other acceptance factors on top of these would be trigger and reconstruction/tagging efficiencies for e/mu which could be ~0.8-0.5 or less… Must relax cuts to get higher rates!
15
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0515 Relaxed angular cuts Same sample as the previous one (parton showers ON) Relaxed cuts on back-to-back- ness of leptons: | l |<2.5 instead of | l |<1 2 = l 0.3 instead of 2 = l 0.1 Sharp edge on Bjorken x still in the same place But minimum lepton p T now very low (next 2 slides) Also wide lepton range means that crack regions become relevant etc This means other sources of systematic uncertainty become important
16
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0516 Relaxed angular cuts
17
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0517 Relaxed angular cuts Minimum lepton p T is now very low even if energy still ~45 GeV
18
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0518 Relaxed angular cuts With these cuts: Preselection: 6080 sel./200k gen. After angular cuts: 534 ev. (factor of ~4 over previous cuts) But events clearly not as clean as before
19
R.Goncalo ATLAS-UK Higgs meeting - Glasgow, 24 June 0519 Relaxed angular cuts: rates and conclusion Production rate is: L~10 33 cm -2 s -1 ~ 1nb -1 s -1 ; x-section~43nb (Pythia)=> R=43Hz Preselection (|eta| 9) gives ~2.9%-3.0% (~6000sel./200k gen.) Selection (eta cuts): 8.8%+-0.4% So: Preselection: ~1.3 Hz Selection: ~0.11 Hz 10% stat. accuracy in ~15mins. 5% stat accuracy in ~1 hour 1% stat accuracy in ~24 hours Conclusions: The numbers in these page start to look encouraging But: This is a parton level study Detector and trigger acceptances need to be taken into account Systematic uncertainties may make us loose what we may gain from limiting the region in x that we're sensitive to.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.