710 likes | 880 Views
First Combination of H1 & ZEUS Diffractive Cross-Sections. P. Laycock , M. Ruspa , E. Sauvan , L. Schoeffel , V. Sola , E. Tassi. Goal. Combine H1-ZEUS diffractive cross sections with Sasha Glasov’s combination code . Available samples H1 ‘97 LRG (published, DESY 06-049)
E N D
First Combination ofH1 & ZEUSDiffractive Cross-Sections P. Laycock, M. Ruspa, E. Sauvan, L. Schoeffel, V. Sola, E. Tassi
Goal Combine H1-ZEUS diffractive cross sections with Sasha Glasov’s combination code • Available samples • H1 ‘97 LRG (published, DESY 06-049) • H1 ‘99 LRG (preliminary) • H1 ’99 LRG minimum bias (preliminary) • H1 HERA II LRG (preliminary) • H1 HERA II FPS (published, DESY 10-095) • ZEUS 2000 LRG (published, DESY 08-175) • ZEUS 2000 LPS (published, DESY 08-175) H1 average LRG available (H1prelim-10-011) SFEX Session @ ZAW - 24/03/2011
Instructions for Use Combine H1-ZEUS diffractive cross sections with Sasha Glasov’s combination code • In the following slides χ2 / ndof comes from SG’s code in two cases • correlations among systematic uncertainties neglected, labelled ‘WA’ • (as equivalent to weighted average, where all the uncertainties are treated as • uncorrelated and added in quadrature) • full correlation among systematic uncertanties (treated as multiplicative), • labelled ‘SGfit’ • The output of the fit to the correlated systematic uncertainties will look like e.g. where e.g. dlar_h1 0.1723 0.3938 means that H1 LArhadronic energy scale changes by 0.17 original sigma and the related uncertainty is reduced of a factor 0.3938 Fitted systematics: 1 dlar_h1 0.1723 0.3938 2 dele_h1 3.3292 0.3340 SFEX Session @ ZAW - 24/03/2011
Cross Section Averaging Procedure Averaging procedure based on Least squares method Data model (assume first only additive systematics): Likelihood: with Pulls are defined as SFEX Session @ ZAW - 24/03/2011
Fit Ingredients • Binning is the same for all H1 LRG measurements; ZEUS measurements available • in this binning (only small shifts in Q2 of H1 points needed) • Correlated systematics from DESY 09-191 and DESY 06-049 • Normalization? Decided to • set normalization according to proton spectrometer results • (i.e. scale ZEUS and H1 according to LPS/FPS) • leave the normalization uncertainties out of the fit • (will be treated later as procedural uncertainty) SFEX Session @ ZAW - 24/03/2011
Digression: LPS/FPS Normalizations Reduced cross measured from Proton Spectrometer data are collected in a t range -> An integration in t is needed to measure the 3-diff cross section ( Q2, β, xIP, t ) -> ( Q2, β, xIP ) b-slope parameter is used in for the integration H1 FPS -> b ~ 5 ÷ 6 GeV-2 depending on xIP ZEUS LPS -> b = 7 GeV-2 “In DESY 10-095 we present the average H1 FPS /ZEUS LPS ratio: 0.85+/-0.01(stat)+/-0.03(unc.syst)+0.09/-0.12(norm.) which takes into account the normalisation uncertainty of 6% for H1 FPS and +11/-7% for ZEUS LPS. I recalculated the H1 FPS / ZEUS LPS ratio assuming the t-slope parameter B=7 GeV-2 as in the ZEUS LPS publication. The mean value for the H1 FPS / ZEUS LPS ratio is shifted from 0.85 to 0.95” [ Mikhail Kapishin, result confirmed by Sasha Proskuriakov ] SFEX Session @ ZAW - 24/03/2011
Systematics • Correlated systematics treated in the fit • H1 LRG: • Larhadronic energy scale • SPACAL electromagnetic energy scale • scattered electron angle measurement • calorimeter noise treatment • non diffractive background subtraction • plug energy scale • SPACAL hadronic energy scale • xIP reweighting • β reweighting • Q2 reweighting • Swimming factors applied to H1 points to reach ZEUS Q2s • will be treated as a procedural uncertainty (not done yet) • Normalization uncertainties not included in the fits • (only in the checks described on backup slides) • ZEUS 2000 LRG: • energy scale • xIP reweighting • ZUFO cut SFEX Session @ ZAW - 24/03/2011
Fit Results SFEX Session @ ZAW - 24/03/2011
Fit Results Study the consistency H1 vs ZEUS across the phase space Full kinematics ZEUS LRG + H1 average LRG: χ2 / ndof = 562 / 153 = 3.4 (wa); χ2 / ndof = 1045 / 153 = 6.8 (SGfit) β < 0.6 & xIP > 0.002 ZEUS LRG + H1 average LRG: χ2 / ndof = 211 / 100 = 2.1 (wa) -> χ2 / ndof = 262 / 100 = 2.6 (SGfit) MX > 4 GeV ZEUS LRG + H1 average LRG: χ2 / ndof = 234 / 113 = 2.1 (wa) -> χ2 / ndof = 360 / 113 = 3.2 (SGfit) [reminder: β~ Q2 / (Q2 +MX2) ] We concentrate our combination in a safe region! Don’t PANIC !!! Still large room for improvement (see next) SFEX Session @ ZAW - 24/03/2011
Full kinematics ZEUS LRG + H1 average LRG β = 0.17 (i = 3) β = 0.05 (i = 6) β = 0.27 (i = 2) β = 0.08 (i = 5) β = 0.43 (i = 1) β = 0.13 (i = 4) β = 0.67 (i = 0) β = 0.2 (i = 3) β = 0.32 (i = 2) β = 0.5 (i = 1) β = 0.8 (i = 0) MX ≤ 4 GeV H1+ZEUS Plenary - 18/03/2011
β = 0.017 β = 0.005 β = 0.027 β = 0.008 β = 0.013 β = 0.043 β = 0.02 β = 0.067 β = 0.032 β = 0.11 β = 0.05 β = 0.17 β = 0.08 β = 0.27 β = 0.13 β = 0.2 β = 0.43 β = 0.67 β = 0.32 β = 0.5 β = 08 • We take separate H1 datasets for next studies H1+ZEUS Plenary - 18/03/2011
H1 Datasets -> Experimental errors are treated as uncorrelated between all datasets * ’97 - HERA II (wa) χ2 / ndof = 121 / 92 = 1.3 -> (SGfit) χ2 / ndof = 148 / 92 = 1.6 * ’97 - ’99+2000 (wa) χ2 / ndof = 120 / 93 = 1.3 -> (SGfit) χ2 / ndof = 144 / 93 = 1.5 * ’97 - ’99 mb (wa) χ2 / ndof = 200 / 131 = 1.5 -> (SGfit) χ2 / ndof = 216 / 131 = 1.6 * ’99+2000 - HERA II (wa) χ2 / ndof = 85 / 92 = 0.92 -> (SGfit) χ2 / ndof = 131 / 92 = 1.4 * ’99 mb - HERA II (wa) χ2 / ndof = 45 / 57 = 0.8 -> (SGfit) χ2 / ndof = 53 / 57 = 0.9 SFEX Session @ ZAW - 24/03/2011
H1 Datasets -> Experimental errors are treated as uncorrelated between all datasets * ’97 - ’99+2000 - ’99 mb (wa) χ2 / ndof = 315 / 224 = 1.4 -> (SGfit) χ2 / ndof = 373 / 224 = 1.7 * ’99 +2000- ’99 mb (wa) χ2 / ndof = 48 / 57 = 0.8 -> (SGfit) χ2 / ndof = 52 / 57 = 0.9 * ’99+2000 - ’99 mb - HERA II (wa) χ2 / ndof = 130 / 149 = 0.9 -> (SGfit) χ2 / ndof = 185 / 149 = 1.2 * ‘97 - ’99+2000 - ’99 mb - HERA II (wa) χ2 / ndof = 400 / 316 = 1.3 -> (SGfit) χ2 / ndof = 509 / 316 = 1.6 SFEX Session @ ZAW - 24/03/2011
H1 + ZEUS Combination -> Experimental errors are treated as uncorrelated between all datasets MY < 1.6 GeV, LPS/FPS relative normalization MX > 4 GeV * H1 Preliminary - ZEUS 2000 (wa) χ2 / ndof = 234 / 113 = 2 -> (SGfit) χ2 / ndof = 360 / 113 = 3.2 * H1 ’97 - ZEUS 2000 (wa) χ2 / ndof = 137 / 113 = 1.2 -> (SGfit) χ2 / ndof = 149 / 113 = 1.3 * H1 HERA II - ZEUS 2000 (wa) χ2 / ndof = 154 / 63 = 2.4 -> (SGfit) χ2 / ndof = 237 / 63 = 3.8 * H1 ’99+2000 - ZEUS 2000 (wa) χ2 / ndof = 109 / 64 = 1.7 -> (SGfit) χ2 / ndof = 142 / 64 = 2.2 * H1 ‘99 mb - ZEUS 2000 (wa) χ2 / ndof = 161 / 79 = 2 -> (SGfit) χ2 / ndof = 239 / 79 = 3 SFEX Session @ ZAW - 24/03/2011
H1 + ZEUS Combination -> Experimental errors are treated as uncorrelated between all datasets MY < 1.6 GeV, LPS/FPS relative normalization MX > 4 GeV * H1 ’97 - H1 ‘99+2000 - ZEUS 2000 (wa) χ2 / ndof = 252 / 191 = 1.3 -> (SGfit) χ2 / ndof = 308 / 191 = 1.6 * H1 ‘99+2000 - H1 ‘99 mb - ZEUS 2000 (wa) χ2 / ndof = 267 / 152 = 1.8 -> (SGfit) χ2 / ndof = 382 / 152 = 2.5 * H1 ‘97 - H1 ’99+2000 - H1 ’99 mb- ZEUS 2000 (wa) χ2 / ndof = 449 / 287 = 1.6 -> (SGfit) χ2 / ndof = 587 / 287 = 2.0 * H1 ‘97 - H1 ’99+2000 – H1 HERA II - ZEUS 2000 (wa) χ2 / ndof = 375 / 268 = 1.4 -> (SGfit) χ2 / ndof = 510 / 268 = 1.9 * H1 ‘99+2000 - H1 ‘99 mb - H1 HERA II - ZEUS 2000 (wa) χ2 / ndof = 386 / 229 = 1.7 -> (SGfit) χ2 / ndof = 590 / 229 = 2.6 SFEX Session @ ZAW - 24/03/2011
H1 + ZEUS Combination -> Experimental errors are treated as uncorrelated between all datasets MY < 1.6 GeV, LPS/FPS relative normalization MX > 4 GeV * H1 ’97 - H1 ‘99+2000 - H1 ‘99 mb - H1 HERA II - ZEUS 2000 (all samples) (wa) χ2 / ndof = 563 / 364 = 1.55 -> (SGfit) χ2 / ndof = 787 / 364 = 2.16 SFEX Session @ ZAW - 24/03/2011
MX > 4 GeV ZEUS LRG + H1 ‘97 Fitted systematics: 1 dlar_h1 0.7214 0.8094 2 dele_h1 2.0889 0.4931 3 dtheta_h1 -0.2327 0.7147 4 dnoise_h1 -0.6382 0.5362 5 dxpom_h1 -0.5184 0.9125 6 dbeta_h1 -0.0285 0.8725 7 dbg_h1 -0.0284 0.5974 8 dplug_h1 -0.1401 0.9193 9 dq2_h1 0.4204 0.9523 10 dspa_h1 0.2422 0.7603 11 scale_zeus 1.6913 0.6814 12 xpomW_zeus -0.1568 0.4810 13 zufoc_zeus 0.1784 0.9914 β = 0.05 (i = 6) β = 0.08 (i = 5) β = 0.13 (i = 4) β = 0.2 (i = 3) β = 0.32 (i = 2) β = 0.5 (i = 1) SFEX Session @ ZAW - 24/03/2011
β = 0.017 β = 0.005 β = 0.027 β = 0.008 β = 0.013 β = 0.043 β = 0.02 β = 0.067 β = 0.032 β = 0.11 β = 0.05 β = 0.17 β = 0.08 β = 0.27 β = 0.13 β = 0.2 β = 0.43 β = 0.67 β = 0.32 β = 0.5 β = 08 SFEX Session @ ZAW - 24/03/2011
MX > 4 GeV ZEUS LRG + H1 HERA II β = 0.05 (i = 6) Fitted systematics: 1 dlar_h1 2.5667 0.4267 2 dele_h1 -0.6915 0.4924 3 dtheta_h1 -0.7374 0.4003 4 dnoise_h1 -2.6210 0.5017 5 dxpom_h1 -0.3954 0.9219 6 dbeta_h1 0.9755 0.7614 7 dbg_h1 -0.4995 0.9972 8 dplug_h1 0.0000 1.0000 9 dq2_h1 0.3755 0.9746 10 dspa_h1 0.0000 1.0000 11 scale_zeus 0.9122 0.4441 12 xpomW_zeus -0.0122 0.2764 13 zufoc_zeus -0.7171 0.9688 β = 0.08 (i = 5) β = 0.13 (i = 4) β = 0.2 (i = 3) β = 0.32 (i = 2) β = 0.5 (i = 1) SFEX Session @ ZAW - 24/03/2011
β = 0.017 β = 0.005 β = 0.027 β = 0.008 β = 0.013 β = 0.043 β = 0.02 β = 0.067 β = 0.032 β = 0.11 β = 0.05 β = 0.17 β = 0.08 β = 0.27 β = 0.13 β = 0.2 β = 0.43 β = 0.67 β = 0.32 β = 0.5 β = 08 SFEX Session @ ZAW - 24/03/2011
MX > 4 GeV ZEUS LRG + H1 all datasets separately ( ‘97 + ’99/2000 + ‘99 mb + 2000 ) β = 0.05 (i = 6) β = 0.08 (i = 5) β = 0.13 (i = 4) β = 0.2 (i = 3) β = 0.32 (i = 2) β = 0.5 (i = 1) SFEX Session @ ZAW - 24/03/2011
β = 0.017 β = 0.005 β = 0.027 β = 0.008 β = 0.013 β = 0.043 β = 0.02 β = 0.067 β = 0.032 β = 0.11 β = 0.05 β = 0.17 β = 0.08 β = 0.27 β = 0.13 β = 0.2 β = 0.43 β = 0.67 β = 0.32 β = 0.5 β = 08 SFEX Session @ ZAW - 24/03/2011
Fitted systematics: 1 dlar_h197 0.8010 0.7806 2 dele_h197 1.9614 0.4472 3 dtheta_h197 -0.3554 0.6541 4 dnoise_h197 -0.9624 0.4360 5 dbg_h197 0.1299 0.5491 6 dplug_h197 -0.1369 0.9062 7 dspa_h197 -0.0478 0.6481 8 dlar_h199 1.2746 0.5296 9 dele_h199 -0.8711 0.5018 10 dtheta_h199 -0.3798 0.7718 11 dnoise_h199 -1.1697 0.5794 12 dbg_h199 0.2959 0.9954 13 dplug_h199 0.2936 0.8690 14 dspa_h199 0.0000 1.0000 15 dlar_h1mb99 -1.3700 0.5512 16 dele_h1mb99 1.0456 0.6328 17 dtheta_h1mb99 -2.2038 0.6136 18 dnoise_h1mb99 -0.0670 0.4789 19 dbg_h1mb99 -0.6154 0.7403 20 dplug_h1mb99 0.7485 0.8007 21 dspa_h1mb99 0.0000 1.0000 22 dlar_h1hera2 2.1364 0.3781 23 dele_h1hera2 -0.8466 0.4160 24 dtheta_h1hera2 -0.5377 0.3265 25 dnoise_h1hera2 -2.3700 0.4589 26 dbg_h1hera2 -0.6695 0.9953 27 dplug_h1hera2 0.0000 1.0000 28 dspa_h1hera2 0.0000 1.0000 29 dxpom_h1th -0.4754 0.7258 30 dbeta_h1th 0.9496 0.5408 31 dq2_h1th -0.0331 0.8510 32 scale_zeus 0.8825 0.3694 33 xpomW_zeus 0.1275 0.2155 34 zufoc_zeus -0.7470 0.9587 SFEX Session @ ZAW - 24/03/2011
Fractional Differences SFEX Session @ ZAW - 24/03/2011
H1 ‘97 + H1 HERA II ( 97 - HERA2 ) / 97 ( fit - 97 ) / fit ( fit - HERA2 ) / fit SFEX Session @ ZAW - 24/03/2011
H1 ’97 + ZEUS 2000 ( 97 - ZEUS ) / 97 ( fit - 97 ) / fit ( fit - ZEUS ) / fit SFEX Session @ ZAW - 24/03/2011
H1 HERA II + ZEUS 2000 ( HERA2 - ZEUS ) / 99mb ( fit - HERA2 ) / fit ( fit - ZEUS ) / fit SFEX Session @ ZAW - 24/03/2011
Pull Distributions SFEX Session @ ZAW - 24/03/2011
* H1 ‘97 - ’99+2000 - ’99 mb - HERA II -> (fit) χ2 / ndof = 509 / 316 = 1.6 H1 ’97 H1 ‘99/2000 H1 99 mb H1 HERA II SFEX Session @ ZAW - 24/03/2011
* H1 ’97 - H1 ‘99+2000 - H1 ‘99 mb - H1 HERA II - ZEUS 2000 -> (fit) χ2 / ndof = 787 / 364 = 2.16 H1 ’97 H1 ‘99/2000 H1 99 mb H1 HERA II ZEUS 2000 SFEX Session @ ZAW - 24/03/2011
What is the Conclusion? ✓ Very well known and long debated normalization discrepancy H1-ZEUS Shape consistency tested for the first time ever and... can’t say it looks ok! And there is no much room for improvement... ✓ The consistency ZEUS - H1 ‘97 is comparable to that between H1 ‘97 and the other H1 samples ✓ Fit results do not change when fixing the normalization of each H1 sample to ZEUS ✓ H1 FPS and ZEUS LPS will be added to the fit (and also fitted separately) H1+ZEUS Plenary - 18/03/2011
Backup SFEX Session @ ZAW - 24/03/2011
Uncertainties H1 ’97: stat ~ 10% - unc~ 2->11% - tot ~ 13% H1 99/2000: stat ~ 5% - unc~ 2% - tot ~ 6% H1 99mb: stat ~ 7% - unc~ 3% - tot ~ 9% H1 HERA II: stat ~ 1.3% - unc~ 1.5 - tot ~ 4% dLAr: < 1% dEle: ~ 2 % dTheta: ~ 1.5% dNoise: ~ 2% dNonDiffBg: 0 -> 7% (increasing with xIP) dPlug: < 1% dSpa: ~ 1% dXpom: ~ 1% dBeta: ~ 1% dQ2: ~ 0.5% ZEUS: stat ~ 3% - unc~ 2% Escale~ 2% xIPweight~ 3.5% ZUFOcut~ 1%
Pulls Full Fit – H1 ‘97 SFEX Session @ ZAW - 24/03/2011
Pulls Full Fit – H1 ’99/2000 SFEX Session @ ZAW - 24/03/2011
Pulls Full Fit – H1 ’99 mb SFEX Session @ ZAW - 24/03/2011
Pulls Full Fit – H1 HERA II SFEX Session @ ZAW - 24/03/2011
Pulls Full Fit – ZEUS 2000 SFEX Session @ ZAW - 24/03/2011
Normalization? Reminder: H1 measurement for MY < 1.6 GeV (normalization uncertainty 7%) ZEUS measurement for MY = Mp (normalization uncertainty 5%) When ZEUS corrected to H1 MY region remaining discrepancy (~13%) What to use as input normalization in the fits and how to deal with the normalization uncertainties? See next 4 slides SFEX Session @ ZAW - 24/03/2011
Normalization uncertainty - tests • Tests done playing with the normalization uncertainty • Fit details: • ZEUS data forced to H1 normalization (in practice corrected for 13% discrepancy • mentioned in previous slide) • Normalization uncertainties treated as correlated errors in the fit • (7% for H1, 5% for ZEUS) • Highest β bins not included in the fit • Results: • - Weighted average (WA): χ2 / ndof = 131 / 127 • - Fit to all correlated systematics (including normalization): χ2 / ndof = 676 / 127 • - Normalization fitted only, all other sources treated as unc.: χ2 / ndof = 326 / 127 • - All systematics additive, only normalization multiplicative: χ2 / ndof = 560 / 127 • - Normalization uncertainties removed from input files: χ2 / ndof = 679 / 127 • Weighted average without normalization uncertainties: χ2 / ndof = 288 / 127 • Normalization uncertainties decoupled from fit quality SFEX Session @ ZAW - 24/03/2011
Input normalization - Tests • Tried different solutions for the input normalization of ZEUS and H1 • ZEUS forced to H1 normalization • ZEUS and H1 @ MY < 1.6 GeV • ZEUS and H1 @ MY = MP • Shape of the combined data not affected by input normalizations (see next two slides) • No reason to prefer any of A, B and C • Setting the normalization according to the proton spectrometer • measurements (ratio LPS/FPS) seems the most rigorous solution SFEX Session @ ZAW - 24/03/2011
β = 0.17 (i = 3) β = 0.05 (i = 6) β = 0.27 (i = 2) β = 0.08 (i = 5) β = 0.43 (i = 1) β = 0.13 (i = 4) β = 0.2 (i = 3) β = 0.32 (i = 2) β = 0.5 (i = 1) SFEX Session @ ZAW - 24/03/2011