New results from CMS for PDF studies Oleksandr Zenaiev (DESY) on behalf of the CMS Collaboration PDF4LHC 7.3.7 / O. Zenaiev New results from CMS for PDF studies
Outline Measurement of double-differential cross sections for top quark pair production in pp collisions at s = 8 ev and impact on parton distribution functions [arxiv:73.63, submitted to EPJ C] Measurement of riple-differential Dijet Cross Sections at s = 8 ev with the CMS Detector and Constraints on Parton Distribution Functions [CMS-PAS-SMP-6-] Measurement and QCD analysis of double-differential inclusive jet cross-sections in pp collisions at s = 8 ev and ratios to.76 and 7 ev [arxiv:69.533, submitted to JHEP] Determination of the strong coupling constant from the measurement of inclusive multijet event cross sections in pp collisions at s = 8 ev [CMS-PAS-SMP-6-8] Measurement of associated Z + charm production in pp collisions at s = 8 ev [CMS-PAS-SMP-5-9] / O. Zenaiev New results from CMS for PDF studies
D t t [arxiv:73.63] t Why measure t t production? - GeV dσ dp σ -3 8 7 Dilepton 6 5 4 3 - CMS, 9.7 fb at s = 8 ev MadGraph+Pythia6 MC@NLO+Herwig6 Powheg+Pythia6 Powheg+Herwig6 Approx. NNLO (Phys.Part.Nucl. 45 (4) 74) m t provides a hard scale ultimate probe of pqcd (NLO, annlo, NNLO,...) Produced mainly via gg constrain gluon PDF Inclusive and D t t already in PDF fits Production sensitive to α s and m t May provide insight into possible new physics heory.6.4. 5 Stat. 5 5 3 35 4 Stat. Syst. EPJ C75 (5) t p GeV 54 5 5 5 3 35 4 t p GeV Why measure D? Previous D measurements revealed some trends D measurement: especially better PDF sensitivity 3/ O. Zenaiev New results from CMS for PDF studies
D t t: analysis overview [arxiv:73.63] Measurement in eµ channel: Leptons: at least oppositely signed e, µ p > GeV η <.4 Jets: at least p > 3 GeV η <.4 at least b-tagged (similar to previous D measurement) Kinematic reconstructions, D unfolding cross sections measured at parton level 4/ O. Zenaiev New results from CMS for PDF studies
D t t: measured distributions [arxiv:73.63] t production: y(t)-p (t): most simple, annlo publicly available t t production: M(t t) M(t t)-y(t t): most sensitive to PDFs (at LO x, = e s ±y(t t) ) M(t t)-p (t t): sensitive to radiation (at LO p (t t) ) t, t t mixed: M(t t)-y(t): sensitive to PDFs (at LO y(t t) = (y(t) + y( t))/) M(t t)- φ(t, t): sensitive to radiation (at LO φ(t t) π) M(t t)- η(t, t): correlated with p (t) as well as sensitive to radiation 5/ O. Zenaiev New results from CMS for PDF studies
D t t: comparison to fixed-order predictions [arxiv:73.63] ] - [GeV (t)dy(t) d dp - -3 - CMS 9.7 fb (8 ev) < y(t) <.75.75 < y(t) < 5 5 < y(t) <.45.45 < y(t) <.5 3 O(α s ) [MNR] C4 µ C4 µ + PDF HERAPDF. Appr. O(α 4 s ) [Diffop] C4 NNLO Ratio to C4-5..9 4 4 annlo provide better data description Substantial PDF sensitivity Moderate scale uncertainties at NLO (normalised distribution)..9 4..9 4 p (t) [GeV] HERA C4 C4 NNLO χ 46 4 3 (dof = 5) 6/ O. Zenaiev New results from CMS for PDF studies
D t t: comparison to fixed-order predictions [arxiv:73.63] ] - [GeV d dy(tt)dm(tt) - -3-4 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV 3 O(α s ) [MNR] C4 µ C4 µ + PDF HERAPDF. -5 Ratio to C4... PDF sensitivity exceeds scale uncertainties can use these data for PDF fits. y(tt) HERA C4 χ 9 6 (dof = 5) 7/ O. Zenaiev New results from CMS for PDF studies
D t t: PDF fit [arxiv:73.63] f f QCD factorisation: σ X = a,b dxdx f a(x, µ f )f b (x, µ f )ˆσ ab X (x, x, µ f,...) µ f m t x, = M(t t) e ±y(t t) s. x,.5 ) ref ) / xg(x, µ δxg(x, µ.4. -. -.4-3 CMS xg(x) µ = 3 GeV NLO f ± HERA + CMS W 8 ev + [p (t), y(t)] 8 ev + [y(t), M(tt)] 8 ev + [y(tt), M(tt)] 8 ev - - x Using xfitter (former HERAFitter): open-source QCD fit framework [EPJ C75 (5) 394, xfitter.org] Input data: HERA inclusive DIS data [EPJ C75 (5) 58] CMS W asymmetry [EPJ C76 (6) 469] these D t t NLO calculations for t t [NPB373 (99) 95] using MCFM ApplGrid Significant improvement of g at high x Best improvement comes from M(t t)-y(t t) 8/ O. Zenaiev New results from CMS for PDF studies
D t t: PDF fit: D vs D [arxiv:73.63] f f Repeated analysis for D distribution p (t), y(t), M(t t), y(t t) Checked their impact on PDFs ) ref ) / xg(x, µ δxg(x, µ.4. -. CMS xg(x) µ f = 3 GeV ± HERA + CMS W 8 ev + y(tt) 8 ev + [y(tt), M(tt)] 8 ev NLO D impact exceeds D First study of such kind Strongly suggests to use these data in global PDF fits -.4-3 - - x 9/ O. Zenaiev New results from CMS for PDF studies
riple-differential dijets [CMS-PAS-SMP-6-] Measurement at 8 ev, L = 9.7 fb, anti-k R =.7 3D cross sections p,avg = (p, + p,)/, rapidity separation y = y y dijet boost y b = y + y.6 Subprocess fraction..4 gg gq gq jets jets (xg < xq) jets (xg > xq) qiqi jets qiqj jets qiqi jets qiqj jets y = y y. yb < y * < 3. 3 5 p, avg / GeV.6 Subprocess fraction. gg gq gq jets jets (xg < xq) jets (xg > xq) qiqi qiqj jets jets qiqi qiqj jets jets.4 3 y b = y + y.. yb < 3 y * < 3 p, avg / GeV > 8% with at least one g at large y b / O. Zenaiev New results from CMS for PDF studies
riple-differential dijets [CMS-PAS-SMP-6-] d 3 dp, avgdybdy * / pbgev 8 7 6 5 4 3 3 CMS Preliminary 9.7fb (8 ev) yb < y * < ( ) yb < y * < ( ) yb < y * < 3 ( ) yb < y * < ( ) yb < y * < ( ) yb < 3 y * < ( ) NLOJE++ (NLO EW NP) NNPDF 3. = p, maxe.3y* anti kt R =.7 4 3 5 p, avg / GeV compatible with theory over a wide range of phase space / O. Zenaiev New results from CMS for PDF studies
riple-differential dijets [CMS-PAS-SMP-6-] uncertainties: jet energy scale (.5 %) statistical uncertainties %, up to % at high p,avg heory uncertainties larger than experimental y = y y 3 3 y b = y + y Ratio to NNPDF 3. NLO EW NP Ratio to NNPDF 3. NLO EW NP.6.4...6.4.6.4...6.4 CMS Preliminary Exp. Unc. heo. Unc. C4 NLO EW NP MMH 4 NLO EW NP ABM NLO EW NP 3 5 CMS Preliminary Exp. Unc. heo. Unc. C4 NLO EW NP MMH 4 NLO EW NP ABM NLO EW NP 3 9.7fb (8 ev) yb < y * < p, avg / GeV 9.7fb (8 ev) yb < 3 y * < p, avg / GeV are well described in most of the phase space, but some differences at high p,avg, y b data can constrain PDFs / O. Zenaiev New results from CMS for PDF studies
riple-differential dijets: PDF fit [CMS-PAS-SMP-6-] xg(x, Q ) Rel. Uncert. 8 6 4 CMS Preliminary HERAPDF Method (Hessian) HERA I+II DIS HERA I+II DIS + CMS Dijets Q = 4 GeV.4....4 4 3 x 3D cross sections ultimately probe Q, x, x in PDFs PDF fit using HERA DIS + these 3D dijet data strong improvement of gluon distribution, precise α s extraction: 3/ O. Zenaiev New results from CMS for PDF studies
Inclusive jets [arxiv:69.533] Measurement at 8 ev, anti-k R =.7: 74 < p < 5 GeV (L = 9.7 fb ) < p < 74 GeV (L = 5.6 pb ) NEW! D cross sections as function of p and y pb GeV σ dy d dp 7 4 CMS 8 ev - Open: L = 5.6 pb int - Filled: L int = 9.7 fb C NLO NP EWK C NLO NP 8 5 4 7 6 ) y <.5 ( 5.5 < y <. ( ) 4. < y <.5 ( ) 3.5 < y <. ( ). < y <.5 ( ).5 < y < 3. ( ) 3. < y < 4.7 ( ) 3 4 3 Jet p [GeV] 4/ O. Zenaiev New results from CMS for PDF studies
Inclusive jets: comparison to theoretical predictions [arxiv:69.533] Ratio to C.8.6.4..6 CMS /(C NLO NP EWK) /(C NLO NP) ABM HERAPDF.5 MSW8 NNPDF3. otal exp. sys. unc. 8 ev.4 - y <.5 Open: L = 5.6 pb int. - anti-k t (R =.7) Filled: L int = 9.7 fb 3 4 3 Jet p (GeV) uncertainties: jet energy scale ( 45% depending on y), lumi (.6%) NLO uncertainties: scales (5 4%), PDFs ( %) well distinguish between PDFs Ratios.76/8, 7/8 available: partial reduction of uncertainties (8 ev) σ dy d dp (.76 ev) / σ dy d dp CMS.6 / heory.4...8.6.4. y <.5 /NP C heo. prediction C PDF uncertainty NNPDF3. ABM HERAPDF.5 MMH4 8 3 4 5 Jet p (GeV) CMS 3.5 3.5.5.5 y <.5 /NP C heo. prediction C PDF uncertainty NNPDF3. ABM HERAPDF.5 MMH4 8 3 4 5 Jet p (GeV) 5/ O. Zenaiev New results from CMS for PDF studies
Inclusive jets: PDF fit [arxiv:69.533] x g (x, Q ) CMS NLO HERAPDF Method (Hessian) 5 HERA I+II DIS Q = GeV HERA I+II DIS + CMS jets 8 ev 5 Fract. uncert..4. -. -.4-4 -3 - - : combined HERA DIS + these jet data heory (for all jet results): NLOJE++ interfaced to fastnlo + EWK, if available Strong improvement of gluon distribution, precise α s extraction: x 6/ O. Zenaiev New results from CMS for PDF studies
D t t vs D inclusive jets vs 3D dijets f f ) ) / xg(x, µ ref xg(x, µ.4..6 arxiv:73.63 CMS -3 xg(x) µ = 3 GeV NLO f ± HERA + CMS W 8 ev HERA + CMS jets 8 ev ± HERA + CMS W 8 ev + [y(tt), M(tt)] 8 ev - - x Rel. Uncert. xg(x, Q ) 9 8 7 6 5 4 3.4 CMS-PAS-SMP-6- CMS Preliminary HERAPDF Method (Hessian) HERA I+II DIS + CMS Inc. Jets HERA I+II DIS + CMS Dijets Q = 4 GeV....4 4 3 x Consistent and competitive PDF constraints using different CMS data! Complementary sensitivity to α s and m t one should use all the data sets to better test/constrain different QCD aspects 7/ O. Zenaiev New results from CMS for PDF studies
Multijets [CMS-PAS-SMP-6-8] Measurement at 8 ev, L = 9.7 fb, anti-k R =.7 -jet and 3-jet event cross sections as function of average transverse momentum H,/ = (p, + p,) of two leading jets /) (pb/gev) dσ/d(h 6 5 4 y <.5 3 3 4 5 6 7 3, - CMS Preliminary 9.7 fb (8 ev) R =.7 anti-k t n j C NP EWK n j 3 C NP 4 5 6 H, / (GeV) are described by theory predictions 8/ O. Zenaiev New results from CMS for PDF studies
Multijets [CMS-PAS-SMP-6-8] 3-jet to -jet cross section ratio R 3: many uncertainties cancel sensitive α s α s extracted, also in ranges of H,/ R 3.35.3.5..5 - CMS Preliminary 9.7 fb (8 ev) R =.7 anti-k t y <.5 /NP C α s (M ) =. - Min. Value z C α s (M ) = z C α s (M ) =.7 - Max. Value z α S (Q) +.55.8 CMS R 3 preliminary, s = 8eV, α S (M ) =.5 z -.3 CMS R 3 preliminary, s = 8eV.6 CMS Incl.Jet, s = 8eV CMS R 3, s = 7eV.4 CMS Incl.Jet, s = 7eV CMS tt, s = 7eV...6.4 CMS Preliminary CMS 3-Jet Mass, ALAS EEC, D Incl.Jet s s = 7eV = 7 ev D Angular Correlation H ZEUS World Avg α S (M ) = ± z.....5 3 4 5 6 68 H, / (GeV).8 5 6 7 8 3 4 3 Q (GeV) Precise α s extraction, consistent with other CMS results: 9/ O. Zenaiev New results from CMS for PDF studies
Measurement of associated Z + charm production [CMS-PAS-SMP-5-9] Measurement at 8 ev, L = 9.7 fb Cross section of Z + c and ratio Z + c/z + b as function of p Important for searches beyond SM, sensitive to possible intrinsic charm ] ]/[dσ(z+b)/dp [dσ(z+c)/dp Z Z CMS Preliminary 4 3-9.7 fb ( s = 8 ev) MADGRAPH MG5_aMC MCFM (MSW8) MCFM (C) MCFM (NNPDF3IC) Stat. uncertainty otal uncertainty 5 5 Z p [GeV] ] jet ]/[dσ(z+b)/dp jet [dσ(z+c)/dp CMS Preliminary 4 3-9.7 fb ( s = 8 ev) MADGRAPH MG5_aMC MCFM (MSW8) MCFM (C) MCFM (NNPDF3IC) Stat. uncertainty otal uncertainty 5 5 [GeV] jet p are compared to NLO calculations with different PDFs (including those with IC) / O. Zenaiev New results from CMS for PDF studies
Summary Many new results from CMS for PDF studies: D t t [arxiv:73.63]: first PDF fit of D t t: demonstrated better impact on PDFs w.r.t D results competitive to those from jets 3D dijets [CMS-PAS-SMP-6-]: ultimate probe PDFs improved gluon distribtuion, accurate α s determination Inclusive jets [arxiv:69.533]: improved gluon distribtuion, accurate α s determination ratios 7/8,.76/8 available: partial cancelation of theory and exp. unc. Multijets, R 3 [CMS-PAS-SMP-6-8]: many uncertainties cancel in ratio, α s sensitivity remains precise α s determination Z + charm [CMS-PAS-SMP-5-9]: compared to PDFs with intrinsic charm For many of these measurements data are typically more precise than theory: need higher order calculations to fully reveal data precision / O. Zenaiev New results from CMS for PDF studies
BACKUP / O. Zenaiev New results from CMS for PDF studies
BACKUP. Inclusive jets: determination of α s χ 95 9 - CMS 9.7 fb (8eV) χ [α S (M )] Z Polynomial Fit +.9 α S (M ) =.64 Z -.3 (µ /µ, µ /µ) = (, ) R F χ /N min Bins = 86.5/85 α S (Q) CMS.4...6.4.. CMS Incl.Jet, CMS Incl.Jet, s = 8eV, s = 8eV CMS R 3, s = 7eV CMS Incl.Jet, s = 7eV CMS tt, s = 7eV CMS 3-Jet Mass, D Incl.Jet D Angular Correlation H ZEUS s = 7eV World Avg α S (M ) = 5 z +.6 α S (M ) =.64 z -.43 ±.6.8.3.6.9.3.7 α S (M ) Z 5 6 7 8 3 4 3 Q (GeV) Using CNLO PDFs Result consistent with other determinations Performed in separate p bins: determine running α = α(q) 3/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Kinematic reconstruction Measured input: leptons, jets, ME Unknowns: p ν, p ν (6) Constraints: m t, m t () m W +, m W () ( p ν + p ν) = ME () ake weighted average of reconstructions with inputs smeared by their resolution his helps to recover events with no solution because of detector effects [Phys. Rev. D73 (6) 545] 4/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Kinematic distributions op quark pairs - CMS Preliminary 9.7 fb (8 ev) 8 eµ 7 tt Signal 6 tt Other Single op 5 Diboson 4 Z / γ* ττ Z / γ* ee/µµ 3 W+Jets QCD Multijet tt+z/w/γ op quark pairs - CMS Preliminary 9.7 fb (8 ev) eµ 4 35 3 5 5 5 op quark pairs - CMS Preliminary 9.7 fb (8 ev) eµ 5 5 5 MC.4..6 3 4 5 6 p (t), GeV MC.4..6 -.5 - -.5 - -.5.5.5.5 y t MC.4..6 -.5 - -.5 - -.5.5.5.5 y tt op quark pairs MC 4 3.4. - CMS Preliminary 9.7 fb (8 ev).6 5 5 5 3 35 4 p (tt), GeV eµ op quark pairs MC 4 3.4. - CMS Preliminary 9.7 fb (8 ev).6 4 6 8 4 6 8 eµ M(tt), GeV MadGraph + Pythia6 provides overall good description, but: p (t) harder y(t) more central y(t t) less central 5/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Unfolding Unfold [JINS 7 () 3] χ minimisation with regularisation ( %) 6/ O. Zenaiev New results from CMS for PDF studies
BACKUP. D t t: comparison to MC [arxiv:73.63] ] - [GeV (t)dy(t) d dp - < y(t) <.75.75 < y(t) < 5 5 < y(t) <.45.45 < y(t) <.5-3 - CMS 9.7 fb (8 ev) MADGRAPH+PYHIA6 POWHEG+PYHIA6 POWHEG+HERWIG6 MC@NLO+HERWIG6 Ratio to MG+P -5. 4 Harder p (t) distribution in all MC his trend is in wide y(t) range. 4 4. 4 p (t) [GeV] MP PP PH MH χ 96 58 4 46 (dof = 5) 7/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Comparison to MC ] - [GeV d dy(t)dm(tt) -3 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV MADGRAPH+PYHIA6 POWHEG+PYHIA6 POWHEG+HERWIG6 MC@NLO+HERWIG6-4 Ratio to MG+P.6.4..6.4..6.4..6.4. y(t) 8/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Comparison to MC ] - [GeV d dy(tt)dm(tt) - -3-4 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV MADGRAPH+PYHIA6 POWHEG+PYHIA6 POWHEG+HERWIG6 MC@NLO+HERWIG6-5 Ratio to MG+P.... y(tt) 9/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Comparison to MC ] - [GeV d (tt)dm(tt) dp -4-5 -6 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV MADGRAPH+PYHIA6 POWHEG+PYHIA6 POWHEG+HERWIG6 MC@NLO+HERWIG6-7 Ratio to MG+P.. 4. 4. 4 4 p (tt) [GeV] 3/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Comparison to MC ] - [GeV d d η(t,t)dm(tt) -3-4 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV MADGRAPH+PYHIA6 POWHEG+PYHIA6 POWHEG+HERWIG6 MC@NLO+HERWIG6-5 Ratio to MG+P. 4. 4. 4. 4 η(t,t) 3/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Comparison to MC ] - rad - [GeV d d φ(t,t)dm(tt) - -3-4 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV MADGRAPH+PYHIA6 POWHEG+PYHIA6 POWHEG+HERWIG6 MC@NLO+HERWIG6-5 Ratio to MG+P.4..4. 3.4. 3.4. 3 3 φ(t,t) [rad] 3/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Comparison to NLO ] - [GeV d dy(t)dm(tt) -3 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV 3 O(α s ) [MNR] C4 µ C4 µ + PDF HERAPDF. -4 Ratio to C4.... y(t) 33/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Comparison to NLO ] - [GeV d d η(t,t)dm(tt) -3-4 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV 3 O(α s ) [MNR] C4 µ C4 µ + PDF HERAPDF. -5 Ratio to C4. 4. 4. 4. 4 η(t,t) 34/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Comparison to NLO ] - rad - [GeV d d φ(t,t)dm(tt) - -3-4 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV 3 O(α s ) [MNR] C4 µ C4 µ + PDF HERAPDF. -5 Ratio to C4.5.5.5.5 3.5.5 3.5.5 3 3 φ(t,t) [rad] 35/ O. Zenaiev New results from CMS for PDF studies
Comparison to fixed-order predictions ] - [GeV d (tt)dm(tt) dp -4-5 -6 - CMS 9.7 fb (8 ev) 34 <M(tt)<4 GeV 4<M(tt)<5GeV 5<M(tt)<65GeV 65<M(tt)<5 GeV 3 O(α s ) [MNR] C4 µ C4 µ + PDF HERAPDF. -7 Ratio to C4.5.5.5.5 4.5.5 4.5.5 4 4 p (tt) [GeV] 36/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Systematics Following D cross section measurement (OP--8): pile-up lepton selection trigger efficiency jet energy scale and resolution ( %) b-tagging efficiency kinematic reconstruction efficiency background variation: DY varied separately by ±3% other backgrounds varied simultaneously by ±3% model uncertainties ( %): perturbative scale variation matching scale variation m t variation PDFs Hadronization (PowhegHerwig - PowhegPythia) Hard scattering (PowhegPythia - MadgraphPythia) 37/ O. Zenaiev New results from CMS for PDF studies
BACKUP. χ comparison w.r.t NLO Distribution MC NLO nominal (including PDF unc.) NDoF MP PP PH MH HERAPDF MMH4 C4 NNPDF3 ABMnf5 JR4 CJ y(t) p (t) 5 96 58 4 46 46 (4) 6 (4) 4 () 8 (5) 6 (5) 47 (47) 4 (3) M(t t) y(t) 5 53 3 5 (44) () 9 (8) 4 (4) 7 (55) 44 (44) () M(t t) y(t t) 5 9 5 9 (5) 5 (5) 6 (5) () 4 (3) 5 (5) 5 (5) M(t t) η(t t) 63 33 39 46 (43) 3 (3) 3 (3) 45 (4) 48 (44) 39 (39) 3 (3) M(t t) p (t t) 5 3 83 3 33 485 (49) 377 (3) 379 (64) 5 () 553 (46) 48 (45) 38 (378) M(t t) φ(t t) 7 354 (336) 93 (7) 96 (59) 48 (43) 386 (335) 39 (34) 97 (95) Monte Carlo MP: MadGraph(CEQ6) + Pythia6 PP: Powheg(C) + Pythia6 PH: Powheg(C) + Herwig6 MH: MC@NLO(CEQ6) + Herwig6 Overall description by MCs: best: Powheg + Herwig worst: MadGraph + Pythia NLO Some comments on PDFs: MMH4, C4 and NNPDF3: use LHC data other PDFs: no LHC data Overall description by different PDFs: best: MMH4, C4, NNPDF3, CJ worse: HERAPDF., JR4, ABMnf5 Particular distributions described by best PDFs: bad description of φ(t t) and p (t t) also bad description of η(t t) reasonable description of the rest 38/ O. Zenaiev New results from CMS for PDF studies
BACKUP. Framework of QCD analysis In short: similar to HERAPDF. fit with t t data included (any publicly reproducable PDF fit would serve the purpose) Platform: xfitter (former HERAFitter) [www.xfitter.org] Input data: HERA e ± p inclusive data [56.64], Q min > 3.5 GeV + CMS W asym. 8eV [63.83] + t t normalised D data R optimal variable-flavour-number scheme with n f = 5, (M Z) =, M c =.47 GeV, M b = 4.5 GeV α n f =5 s Predictions for t t: MNR calculations [NPB373 (99) 95] via MCFM ApplGrid xfitter pole top mass m t = 7.5 GeV scales µ r = µ f = m t + (p (t) + p ( t))/ PDF parametrisation: 8 free parameters HERAPDF style Uncertainties: Experimental: from χ = Model: from theoretical and model parameter variations Parametrisation: from µ f and parameterisation form variation 39/ O. Zenaiev New results from CMS for PDF studies
BACKUP. NLO vs NNLO Is it OK to use NLO calculations? Yes, for normalised distributions p (t), y(t), M(t t), y(t t) with current data precision, because: hese are the only publicly available calculations Missing higher order corrections (estimated by scale variation) for NLO are small w.r.t to data: % fot total x-section 3% for p (t), y(t), M(t t), y(t t) shapes Confirmed by exact NNLO calculations [Czakon et al. arxiv:5.549] (slightly larger effect on p (t)). dσ/dm tt - [pb/gev] NNLO/NLO /NNLO (/σ)dσ/dp,t [/GeV x -3 ] 7 6 5 4 3.3...9.5.75.5.5...9 (pictures taken from arxiv:5.549) Czakon, Heymes, Mitov (5) NNLO NLO LO CMS(l+j) PP tt - +X m t =73.3 GeV MSW8 µ F,R /m t {.5,,} p (t), % 5 5 5 3 35 4 (/σ)dσ/dy t /NNLO Czakon, Heymes, Mitov (5).35 NNLO NLO.3 LO CMS(l+j).5..5. PP tt - +X m t =73.3 GeV MSW8 µ F,R /m t {.5,,}.5 -.5 - -.5 - -.5.5.5.5.5 5 5 5 3 35 4 -.5 - -.5 - -.5.5.5.5 p,t [GeV] y t Czakon, Heymes, Mitov (5) NNLO NLO LO M(t t), 3% NNLO/NLO dσ/dy - tt [pb]..5.95 9 8 7 y(t), % Czakon, Heymes, Mitov (5) NNLO NLO LO y(t t), 3% 6 PP tt - +X m t =73.3 GeV 5 PP tt - +X MSW8 µ F,R /m t {.5,,} m t =73.3 GeV 4 MSW8 3 µ F,R /m t {.5,,} 4 5 6 7 8 9.5.5.5...9 NLO/LO.6.4. 4 5 6 7 8 9 4 5 6 7 8 9 m - tt [GeV] NLO/LO.5.5.5.6.4..5.5.5 y - tt 4/ O. Zenaiev New results from CMS for PDF studies