marc goulette atlas cat physics meeting, cern, 5 september 2008 thanks a lot to: m. boonekamp, j....

16
Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D. Froidevaux, L. Gilbert, I. Hinchliffe, J. Huston, B. Kersevan, E. Richter Was, M. Seymour, T. Sjostrand, G. Stavropoulos and T. Weidberg. Recent paper: Evaluation of the theoretical uncertainties in the Wl Cross Sections at the LHC (N. E. Adam, V. Halyo, S. A. Yost, W. Theoretical uncertainties on W/Z cross-sections

Upload: cassandra-norton

Post on 13-Dec-2015

219 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

Marc Goulette

ATLAS CAT physics Meeting, CERN,5 September 2008

Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D. Froidevaux, L. Gilbert, I. Hinchliffe, J. Huston, B. Kersevan, E. RichterWas, M. Seymour, T. Sjostrand, G. Stavropoulos and T. Weidberg.

Recent paper: Evaluation of the theoretical uncertainties in the Wl CrossSections at the LHC (N. E. Adam, V. Halyo, S. A. Yost, W. Zhu), Aug. 08

Theoretical uncertainties on W/Z cross-sections

Page 2: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

2/165 September 2008 Marc Goulette

Outline

1. Introduction2. W and Z cross sections, pdfs and selection cuts3. Electron acceptances for W and Z4. Study of acceptance differences

1. Pythia documentation

2. Main sources of difference (ISR, kT) (except pdfs)

3. Comment about the intrinsic kT of the incoming partons4. Problem in interface between Herwig and Photos5. Impact of QED corrections and recombination6. pdfs

5. Overall systematic uncertainties6. NNLO corrections7. Ratio of W to Z cross sections8. Conclusion and outlook

Page 3: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

3/165 September 2008 Marc Goulette

We and Zee events at the generator level. Aim: Determine the systematic uncertainty on W/Z cross sections. Systematic uncertainty on cross-section is dominated by uncertainties

determined on acceptances. More details: ATLAS note submitted (ATL-COM-2008-010) Last talk: 13 December 07, ATLAS SM group

http://indico.cern.ch/conferenceDisplay.py?confId=12059 Update with release 12.0.6, standard DC3-jobOptions Monte Carlo Generators:

Pythia 6.4, Herwig 6.510+Jimmy 4.0, MC@NLO 3.2 Between 100 K and 500 K events for the 3 generators Some comments about the generation:

Same Mass and same Width of the bosons for all generators Herwig & Pythia cross sections: agreement better than 1% (see next slide) Same parton density function (pdfs) for all (CTEQ6L and CTEQ6M) Standard cuts: pT

leptons > 25 GeV and || < 2.5

Introduction

Page 4: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

4/165 September 2008 Marc Goulette

Cross sections, pdfs and filter cuts

Cross sections Pythia Herwig MC@NLO NNLO*

We (pb) 17210 (132)**

17315 (132) 19885 (138) 22097 (692)

Zee (pb) 1731 (42) 1676 (16) 1985 (155) 2018 (64)

* Ref: hep-ph/0307219 (S. Alekhin), 2003** only statistical errors

LEPTON FILTER: pT_cut = 0.0 _cut = 5.0

STANDARD CUTS: pT

e > 25 GeV for W pT

e > 20 GeV for Z pT

> 25 GeV || < 2.5 excluding the crack (1.371.52) Electron and photon merged if R = (2+2) < 0.04

PYTHIA: LHAPDF version 4.0 CTEQ6L1-LO with LO alpha_s

HERWIG: HWLHAPDF modpdf 10042 LHAPDF version 4.0 CTEQ6L1-LO with LO alpha_s

MCATNLO: HWLHAPDF modpdf 10000 LHAPDF version 4.0 CTEQ6m (NLO version of CTEQ6L1-LO)

pdfs

Page 5: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

5/165 September 2008 Marc Goulette

Acceptances as a function of max with pTleptons > 25 GeV (W) and

pTe > 20 GeV (Z) and |e| < 2.5 (crack excluded)

Acc. for We and Zee (STD)

We Zee

HW/PY: 0.7 % for W and 2 % for Z HW/NLO: 7 % for W and 11 % for Z

max2.5

MC@NLO 51.02 0.10

HERWIG 45.27 0.14

PYTHIA 46.20 0.07

HERWIG/PYTHIA 97.98 0.34

HERWIG/MC@NLO 88.72 0.33

PYTHIA/MC@NLO 90.55 0.23

max2.5

MC@NLO 36.84 0.10

HERWIG 34.27 0.15

PYTHIA 34.51 0.07

HERWIG/PYTHIA 99.29 0.48

HERWIG/MC@NLO 93.02 0.47

PYTHIA/MC@NLO 93.94 0.30

Acceptance with all effects switched OFF (no UE, ISR, kT nor ME) give:

HERWIG/PYTHIA 98.42 0.38 HERWIG/PYTHIA 98.98 0.32

Page 6: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

6/165 September 2008 Marc Goulette

Study of acceptance differences

- check main distributions (see note or previous talks…)- check printouts, variables, comparison with standalone mode- study individual sources of differences:

- ISR

- kT - ME- UE- impact of the electroweak (EWK) corrections (Photos)- impact of the electron/photon recombination- pdfs

- Impact of boson mass and width (difference between W and Z)- Impact of the individual cuts

- Re-weighting method on pTW and pT

Z - NNLO corrections

Slide 9

Slides 10

Slide 11

Slide 13

slide 8

Slide 10

slide 7

Page 7: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

7/165 September 2008 Marc Goulette

Problem in Pythia documentation

Problem here …

- We should have : Px_s~+Px_c = Px_W and the same for Py and Pz …- Torbjorn S. checked on 10 K events (Thanks !): nothing in standalone mode

Info from common blocs not directly accessible in Athena Should this be fixed (so that same problem is not rediscovered later ?)

(MeV)

(MeV)(MeV)

Px Py

Pz

Should be a perfect delta function

Page 8: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

8/165 September 2008 Marc Goulette

Impact of ISR on acceptances

ISR

xu xd

xc xs

_

_

Fluctuations for MC@NLO due to sligthly less statistics (100 k compared to 500 k) and most probably due to different pdfs model !

Page 9: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

9/165 September 2008 Marc Goulette

Intrinsic kT of the incoming partons

Tail for Pythia

(doc ? Heavy quarks ? Other ?)

Mean 1.753 GeV

Ptrms=1.753 GeV Mean 2.37 GeV

RMS=935 MeV(mstp 91)

RMS=1187 MeV

Pythia

Herwig

(needed to reproduce Tevatron data on pT

Z)

Slight difference in parameters between Pythia and Herwig:

kT

(Note: kT not switched on by default in Herwig)

Wmass

Page 10: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

10/165 September 2008 Marc Goulette

Impact of QED correctionsProblem in interface between Herwig and Photos for Z: 7% !!

So no numbers discussed for Herwig here

Note: - no electron/photon(s) recombination - 2% on acceptance- with recombination effect vanishes- to be checked for systematics with more realistic simulation

1.1 %effect

2.3 %

2.1 %

Page 11: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

11/165 September 2008 Marc Goulette

Pdf impact on acceptances

W- NLO 6.1

W+ NLO 6.15W+ MC@NLO 6.1

W- MC@NLO 6.1

Thanks toJoey Huston.

Calculations for the W acceptance using ResBos

W+ MC@NLO 3.2

W- MC@NLO 3.2

syst ~

+ 0.72 %- 0.69 %

+ 0.89 %- 0.90 %

+ 0.66 %- 0.69 %

+ 0.68 %- 0.84 %

+ 0.89 %- 0.90 %

syst = [(acc(i)-mean)2]) (stat is not significant)

Please note that MC experts said that this recipe is not good enough because moving to another pdfs model could introduce an uncertainty on the acceptances (shift of the central value of the set) that could be higher than the fluctuations …

Page 12: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

12/165 September 2008 Marc Goulette

Overall systematic uncertainty

From the Pythia / Herwig comparison (i.e at LO), we get the following uncertainties (using in the worst case 20 % of the impact on the acceptances from offon, and 10 % for QED):

For W: 0.11 % (ISR), 0.38 % (kT), 0.0 % (ME), 0.19 % (UE) and 0.11 % (Photos).

For Z: 0.34 % (ISR), 0.10 % (kT), 0.02 % (ME), 0.04 % (UE) and 0.0 % (Photos).

In addition we have an estimate of the pdfs uncertainty from CTEQ (at NLO): 0.90 % (for W), same applied for Z (resbos data. not available).

Overall syst. excluding crack is: ~1.01 % for W and ~0.97 % for Z for effects studied here.

Is there anything missing ? Yes, impact of higher orders.

Page 13: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

13/165 September 2008 Marc Goulette

NNLO corrections

Ref: hep-ph/0603182(Melnikov & Petrielo), 2006W- production: pp W- X e- X

From 1.8 to 2.8 % if pTe min > 20, 30 GeV, so take ~ 2.3 % for pT

e min > 25 GeV

_1) Comparison with NNLO predictions:

2) NNLO corrections:

pTe min

(GeV)

A (NLO) A (NNLO) A (MC@NLO)

My data,

A (MC@NLO)A (my data)/

A (NNLO)

in %

20 0.488 0.492 0.485 0.4835 0.0007

1.8

30 0.378 0.376 - 0.3654 0.0007

2.8

- Remember that LO to NLO gave a 7 % difference for We- Comparison here between NLO and NNLO gives ~ 2.3 %

- Reasonable assumption ? Assume systematic of ~ 30 % of acceptance difference between NLO and NNLO (0.76 %) and add it (quadr.) to previous overall systematic: 1.23 % for W and 1.19 % for Z

Page 14: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

14/165 September 2008 Marc Goulette

Ratio of We to Zee

- Precision measurements of the ratio of We to Zee production cross sectionsare used to test the Standard Model.- The ratio of cross sections can be expressed in terms of measured quantities:

- where F is the correction factor for converting the measured */Zee cross sectioninto the cross section for pure Z boson exchange.

- Uncertainties from Background dominant until we have a lot of data (QCD corrections)

- Efficiencies depends on Trigger and Reconstruction.- Uncertainty on the ratio of the efficiencies should also be less big than the individualuncertainties.

. . . 1 Nw

sig - Nwbck AZ Z

F NZsig - NZ

bck AW W

R =

Page 15: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

15/165 September 2008 Marc Goulette

Which uncertainties cancel in the ratio ?

- The integrated luminosity terms cancel completely in the ratio calculation.- Uncertainties on the acceptances arising from the PDF model are significantlysmaller for the ratio than for either individual acceptance. See for instance a nicepaper produced in 2007 by CDF:“Measurements of inclusive W and Z cross sections in pp collisions at s = 1.96 TeV”.- Can’t be shown here because pdfs data not available for Z.

- In my study:

_

Ratio (AZ/AW) Pythia Herwig

ALL off 1.0653 0.01 1.0714 0.01

STD photos on 1.1330 0.01 n.a (*)

STD photos off 1.1321 0.01 n.a

ISR on 1.3375 0.02 1.3291 0.02

kT on 1.1756 0.01 1.1587 0.01

ME on 1.1469 0.01 1.1536 0.01

UE on 1.1466 0.01 1.1502 0.01

Don’t know yetwhat to deduce

from that…

Photos seems ok, but the other sources ?

(*) due to problemin interface

Page 16: Marc Goulette ATLAS CAT physics Meeting, CERN, 5 September 2008 Thanks a lot to: M. Boonekamp, J. Butterworth, A. Cooper-Sarkar, E.Dobson, N. Ellis, D

16/165 September 2008 Marc Goulette

Conclusion and outlook

I) Final numbers for acceptances with high statistics

- use MC@NLO as central value and extract acceptances- W: = 36.84 0.10 % (with final definition of crack)- Z: = 51.02 0.10 %

II) Systematic uncertainties

- use 1.23 % for W and 1.19 % for Z with current knowledge- Point out that dominant sources (ISR, kT, pdfs) will be further constrained with data:

• pTW, pT

Z spectrum• Zee found with one electron with |e|>>2.5

• Outlook

- Need to recompute acceptance with large statistics using MC@NLO and full simulation/reconstruction- Requires different evgen files beyond standard configuration (no QED, different pT

W, different pTZ, anything else ?)

- Probably do this only at 14 TeV ?