Short Biased Minutes of
W meeting 07/07/1999 in 40/SS-C01 – 10.30 am to 1 pm
Ben analysed the preliminary energy files provided by the LEP energy group. The present uncertainty on the LEP energy is 50 MeV, expected to improve when more depolarizations measurements will be available. He weighted the energy points with the luminosity selection for Tampere (Benjamin used Tampere I). The beam energy was extracted for every CLAS 7 event. At about 192 GeV the beam average energy is 95.808 GeV with a RMS of 53 MeV and an energy spread of 180 MeV. At about 196 GeV the beam average energy is 97.802 GeV with a RMS of 56 MeV and an energy spread of 186 MeV. As a cross check with the calibration data taken at the Z peak the result is 45.621 GeV (see alnews 4181 in aleph.general folder)
The 27.287 pb-1 at 192 GeV and 26.234 pb-1 at 196 GeV have been analysed in the e,mu,tau channel. We have nice peaks in all cases . We will propose to give a shot of these for the pleasure of Tampere participants.
One other good news is that with increased Monte Carlo statistics at the 80.35 reference point the slope for the muon calib (189 GeV) is now consistent with 1. More news about this at next W meeting.
3. 19x GeV hadronic mass and progress on 4q mass analysis (Ann)
Also for the 4q’s we have a nice peak at 19x GeV. We will show it with the nominal 80.35 GeV MC superimposed.
There have been considerable progresses in debugging the 4q fitting code, also increasing to double precision when applicable (thanks Evelyn !). Now we do not have failures any longer and the Minos and Hesse errors are parabolic and consistent with each other. With this new version of the code the expected and actual error on 189 GeV data are the same.
Ann discussed the statistical error on W mass measurement from the 4 experiments, ("…..why Delphi has less luminosity but same stat. error as us?"). It was pointed out that stat.+syst. have to be compared, but also that we can learn by comparing all these different techniques.
The 4q analysis seems to give similar stat. error (what about syst. ?) enlarging the window from an upper limit of 86 to 90 GeV, which would be desirable not only for better presentation but also for the width analysis.
The online hadronic selection provides a preliminary x-section with two methods (cuts and fit) at 192 and 196 GeV (official lumi, see point 2)). The two gives very similar results. Selection based on charged tracks gives very similar numbers as well. The measured X-section is right on SM at 192 and a couple of sigmas lower at 196. The various input variables have been examined and no particularly worrying misbehaviour has been seen.
Leptonic analysis has been run at 192 and 196, E12 cut inefficiency has been computed with instantaneous lumi (it’s about 8%). Average efficiency for CC03 selection is 64.1% (similar to 189 GeV). Bkg have been recomputed. The X-section is in agreement with SM. Same comments apply to semileptonics.
Preliminary total x-section at the 2 energies :
192 GeV è 16.96 +/- 0.93 (stat.) pb (SM=17.01+/-0.34)
196 GeV è 16.13 +/- 0.87 (stat.) pb (SM=17.29+/-0.34)
We propose to give the updated preliminary plot to Tampere.
The cut analysis has been repeated at the two new energy points with results in agreement with expectations. The X-section plot with the two additional points is proposed for Tampere.
MLPfit allows to find the best functional description using PAW and neural network techniques. Philippe has used this tool to describe the shape of the w invariant mass in the 4q channel. Basically he has started from the invariant mass plot (1-d plot, 2 dijet summed to the same plot) from MC at 3 mass points. He has fed MLPfit with this and he got back the "best" function with two parameters : W mass and normalization. He has then used this to fit our data. On 200 MC subsamples he has show that this method and standard reweighting have similar statistical power. The mass difference, sample by sample, is centered at 0 with about 50 MeV width. The result on data is consistent with the standard analysis. He has then studied the effect of enlarging the window, obtaining a reduction of the stat. error of about 10%.