280 likes | 386 Views
Update of D S + + . Liming Zhang & Sheldon Stone (Syracuse University). Outline. Results based on 314 pb -1 (data38-41) + 288 pb -1 (data47-48)=602 pb -1. D s - ( D s * - ) D s * + ( D s + ) TAGGED SIDE: (9 modes).
E N D
Update of DS+ + Liming Zhang & Sheldon Stone (Syracuse University)
Outline • Results based on 314 pb-1 (data38-41) + 288 pb-1 (data47-48)=602 pb-1
Ds-(Ds*-) Ds*+ (Ds+) TAGGED SIDE: (9 modes) SIGNAL SIDE: K+K-p- KsK- p- '(p+p-)p- h'(rg) p- K+K-p-p0 instead offr p+p-p- K*-K*0 (KsK-p+p-) r-h + or t+(p+n)n + g Find and look at MM2 Look at MM*2 Analysis Techniques e+e- (4170 MeV)
Tag side mbc in [2.015, 2.067] GeV to select DS* DS events 12 MeV mass window for Ks 150 MeV mass window for r+ and r0 100 MeV mass window for K* 3 s pullmass for p and h 10 MeV mass window for h' hpp (new from previous analysis) 20 MeV mass window and helicity angle of r decay |cosa|<0.8 for h' gr [it’s sin2adistribution] P(p/p0)>100 MeV in KKp, KKpp0, ppp modes to remove D*+ ~2010 MeV MDs window: 17.5 MeV MM*2 window: [3.782, 4.00] GeV2 Shower from Ds* Not hot No track match Pass E9/E25 Energy > 30 MeV if goodBarrel or >50 MeV if goodEndcap Selection Criteria of Tags
2D Fit to Obtain # of tags • 2D Binned Extended ML fit by RooFit • Fitting region: |MDs-mDs|<50 MeV, MM*2[3.5, 4.25] GeV2 • Three components of PDFs • Signal • BG1: real Ds combining with fake g • BG2: fake Ds • PDFs: production of PDFs in MDs and MM*2 • Signal: 2G(MDs) x CB(MM*2) • BG1: 2G(MDs) x P5(MM*2) • BG2: P1(MDs) x P5(MM*2) • Total 20(shape)+3(yields)-2(fixed for CB)=21 free parameters • Generic ddmix MC Data41 used (~9 times current dataset) to test this method
Double Tag Study (Data) All 9 modes combined a and n obtained and to be fixed in ST fitting MM*2 (GeV2)
MM*2 Distributions in mass 17.5 MeV KKp hp KS0K BG1 BG2 h'p KKpp0 ppp hr K*-K*0 h'(rg)p
Systematic Error on # of tags • Default fit is P5 for BG1 and BG2, all are allowed to float • Systematic error estimation • BG2: increase/reduce 1 order for MM*2 polynomial • BG1: fix its shape to MC, where its normalization is floated • We record the difference with the default fit, and imply to each of modes • We add all differences in quadrature as the systematic error
Data Fit Results (All DATA) # in MDs17.5 MeV and MM*2 [3.782, 4.00] GeV2 Total stat. error = 936 (2.1%) Total syst. error = 894 (2.0%) Differences of modes added linearly to get the total systematic error
Mbc Distribution • If DS tag comes from prompt DS , it peaks at mbc • If DS tag comes from prompt DS*, it’s flat at mbc • [2.015, 2.067] GeV window used to select Tags • In a wider window [2.0, 2.08], the amounts of DS (peaking curve) and DS*(flat curve) are the same, and the efficiency ratio is 1 (test by MC 1.007 0.009) • i.e. N(DS) = N(DS*) • Now we concentrate on DS* DS 2.067 2.015 DS*
C1 A1 B A2 C2 Idea • We look at DmS = m(Dsg) - m(Ds) • Tag from Ds* (flat) peaks in DmS, tag from Ds (peaking in mbc) is flat • Divide in three regions [Using A & B because A is clean, B have large BG • N(Ds*) = A + B + C • Total number of tags we used is N(Ds) + N(Ds*) – C = 2 N(Ds*) – C = 2(A+B) + C A=A1+A2 C=C1+C2 A, B, C are the # in the flat curve
DS* signal DS signal Fake gamma BG Fake DS BG DmS Distributions from MC DS mass and MM*2 signal windows applied to get DmS distribution • Fitted by CB function, a and n obtained from double tag • Sideband subtraction to remove this background
Double Tagging Signal DmS distribution (DS subtracted), to obtain a and n DS subtraction: Events in A - 0.09 B; 0.09 is ratio of # in A to B for DS signal (peak in mbc) MC DATA Single Tag in red Events / 2 MeV DmS (GeV) DmS (GeV)
DATA Fit • Fit to mass sideband subtracted distribution of all modes • Two backgrounds: DS signal (in green) and fake g (in blue) • Their shapes are fixed to MC, normalizations allowed to float A = 17125 434 B = 3513 230 C = 2077 136 Ntag = 2(A+B) + C = 43353 991 1% difference Ntag = 43859 936 from the MM*2 fit
Signal Dsm v Reconstruction • Only one extra track with opposite sign of charge to the tag [using “TrkmanApproved”] • The track has cos (opening angle) <0.90 [previous 0.81] • Eff. increase 11% • Resolution not change comparing with 0.81 • No any neutral energy cluster detected (Emax) > 300 MeV • Kaon veto using RICH [if ng(k)>2, and L(m)-L(K)>10] • Calculate MM2 and apply kinematical constrains fit with two hypotheses that g from signal side or tag side. The fits gives two fitted MM2, we choose the one with the lower fit c2 • Best photon candidate selection: if there is more than one photon candidates with the same Ds in a event we choose only the lowest c2 choice
160 54 63 Data MM2 Distribution We use [-0.1,0.2] GeV2 as the fit region to eliminatehigher sideK0p+(~0.25 GeV2) and hp+ (~0.3 GeV2) background Comparison of Candidates Lum. Ratio is 1.42 Case 1 [-0.05, 0.05] GeV2 5% lower
Signal m+n MC Efficiency Eff’s and shapes are weighted according to # of tags in the data
Tag bias from MC It’s easy to find tags in Dsmunu events than in the typical Ds decays
Ds+ t+n; t+ p+n MC Eff = 40.9% in [-0.1,0.2] GeV2
Backgrounds from Ds We study the other background sources as considered in the paper Expected # for data in [-0.1,0.2] GeV2 Case 2 Case 1 From MC
Fit Technique • Fit detail: a 2D fit • All Shapes in MM2 except fake Ds are fixed to MC, but normalizations are allowed to float • The signal shape in MDs (double Gaussian) obtained from Data when determining Ntag, and fixed • Fake Ds is described by a second order in MM2 and a first order polynomial in MDs, and the three shape parameters are allowed to float Generic MC fit (sum of two cases) Ntag = 336771+-1934 Nmunu = 1838.7+-36.9Br(Ds->munu)=(0.607+-0.013)% Input 0.61%
Data Fit (Sum two cases) • & dependent fit • Constrain N/N = Rbr Reff =1.059 0.455 = 0.482 • Rbr=R B(t+p+n); R = 9.72 B = (10.900.07)% • Breff (Ds ) = (0.603 0.037)% # in 17.5 MeV MDs & [-0.1, 0.2] MM2 region
Data Fit (case 1) • & independent fit • Br (Ds ) = (0.568 0.045)% # in 17.5 MeV MDs & [-0.1, 0.2] MM2 region
Simultaneous Fit to case 1&2 • & independent fit • Fix area in case 1 / 2 to 98.8%/1.2% • Fix area in case 1 / 2 to 55%/45% • Br (Ds t) = (6.7 0.8)% Plots need to be provide
Cross check: DsK0K+ Signal MC Resolution: s1*f+(1-f)*s2 (GeV2) Munu MC 0.0346+-0.0002 K0K MC 0.0344+-0.0003 K0K Data 0.0353+-0.0015 Data Fit • Yield 1066 42 • Eff: (77.0 0.6)% • Br(DsK0K+) = (3.15 0.14)% • Consistent with Peter’s • Br(DsKSK+) = (1.49 0.07 0.05)%