Short Biased Minutes of

W meeting 01/09/1999 in Theory Amphi from 10 am to 12.30 am

1) Mass and width in the 4q channel (Roger)

Roger gave the detailed status of the mass and width analysis in the 4q channel at 189 GeV. He is using fixed width propagator for reweighting, abcfit for the kinematic fitting, NN14 for the selection, together with Roger binning (>120 evt bin). He has done in parallel a control mass analysis, a fixed mass width analysis and a 2D mass and width measurement. The window for the fits has been varied to check stability. For the cross section both the cases of nominal cross section (16.93 pb) and measured x-sec (14.98 pb) have been treated. In the second case the sm branching ratio has been assumed. John has done a similar analysis using kinfit instead of abcfit, a different number of evt/bin and a different interpolation grid. See the transparencies for the results in various cases. The bottom line is : the width result is very stable and typically one sigma higher than the SM value of 2.094 GeV, with a typical error of 270-300 MeV. Calibration plots have been checked and are ok. Results on MC samples are very sensible. The result goes down by a third of a sigma when the measured x-sec is used.This is thought to be an effect related to the lower purity, which affects the width (and of course doesn't affect the mass....). Systematic effects are being worked out. The correlation between the mass and width in the 2D is very small. The aim is to include the 2D mass/width analysis as an extra measurement in the forthcoming 189 GeV mass paper.

2) 4q selection and colour reconnection (Evelyn)

The comparison of data/mc variables sensitive to CR requires a robust selection based on variables with good data/mc agreement. This of course is true for other applications. Evelyn studied a simple hadronic selection inspired by opal using a small number of variable with a better purity at preselection level. In the after class 16 comparison she observed an excess of events at low multiplicity in data, most likely caused by untagged gamma-gamma interactions. More worrying is the effect of a known "feature" of the energy flow: when a charged tracks is measured as having momentum gt beam momentum the track is given by definition the beam momentum (what else to do ?). The fact is that we see more of these tracks in data than in MC. What to do about such badly measured tracks that is not clear, the effect on the analyses has to be evaluated (about 1% of events on data have such tracks). Evelyn does not use them in the computation of the variables for the selection. The same is true for sical/lcal clusters. She got a preselection 91.6% efficient and 59.9% pure (see transparencies for the variables...) and a weight based selection with performance similar to opal. The cross section, not surprisingly is low ....... Application to CR studies in progress, stay tuned.

3) Systematic error from BE correlations (Jorgen)

Together with Frederic he tried to identify variables sensitive to BE correlation in order to reject events for which the BE induced mass shift is more sizeable. The mass distortion going from events with BE in same W only to BE between W's has been clearly shown by Jorgen at generator level. The effect is seen with both 183 and 189 MC. The study has been repeated after reconstruction. It has been seen that reducing the mass window used to fit the mass reduces the absolute value of the BE shift without reducing significantly the statistical error. The other interesting idea is to cut out events with pions close in phase space. The observation is that these events have the largest BE shift, but are not located in the region of highest statistics, so in principle they could be rejected. An Aleph Note is in preparation.

4+5) Mass and Cross Section semileptonic selections (Barbara and Anne)

The two selection programmes have been compared in detail following the difference in selected events observed last time. One difference was due to the locking of bad tracks (mostly nuclear interactions) done for the lepton candidate search in the cross section but not in the mass analysis. This is now done also for the mass, which means n-tples are currently being redone. The preliminary studies indicates no mass shifts due to this difference (and to a small bug found in the makefile in the checking process). More news next time, no problem foreseen in readily complete the sl mass analysis.

Other small differerences in the selections have been identified (see the table in Anne transparencies) but have little impact on the performance. It would be desirable anyway to settle these differences to allow an easier comparison in the future. The most important point to sort out concerns the bremsstrahlung correction for muon, due to FRS. A photon is searched for in the vicinity of the lepton (muon in this case) and its energy added to the lepton candidate. The correction is applied in the mass analysis, but not in the cross section. A decision has to be taken wheater we have to apply this correction or not, based on its size and on the possible background contamination (which could not be the same in data and mc). After the meeting Evelyn provided useful information on this point ( click here).

Again on the semileptonic x-sec selection Anne has found that in the online programme 189 GeV efficiencies were used for 19x GeV data. This has been corrected and energy dependent cuts are now used. As a result of this the online sl cross section at 196 increased by more than half a sigma.

6) 4q and total WW cross section (Patrice)

The ultra-preliminary online plots can now be seen on the x-section web page ( online WW cross section ) .

The trend of having a lower WW cross section continues .... and this is always more true for the 4q channel. Among the many checks that have been done on the 4q channel Franco has developed a selection based on Linear Discriminant Analysis using calorimeter only information. He has spent quite some time in studying variables with reasonable data/montecarlo agreement. For calorimeters this means using global jet variables and in particular global angular variables . He has now a discriminant with shows decent stability. The preliminary results seem to confirm the NN14/NN19(energy flow) and NN6(charged tracks) numbers.

In the standard analysis there is some concern, already expressed by Eric last time, about jet energy variables after kinematic fitting. At least at 196 GeV the data - mc agreement seems to deteriorate after the fitting. In general it is felt that we should review the variables used in the standard NN selection following the various checks performed in the past few months and that we have to reach some definitive conclusion at least at 189 GeV that we wish to publish soon.

7) 2D reweighting in the semileptonic channel (Oliver)

Oliver has taken over from Eric Wannemacher the 2D reweighting analysis in the semileptonic channel. The two masses which are used are the 2C fit mass and the 1C mass on the hadronic side. The correlation between the two is about 40%. The analysis programme has been made very similar to the standard 1D analysis (same MC ref, same FRS corr, same selection, reco, and kinfit). Calibration curves are very similar to the standard analysis. The nice outcome is that the statistical error is reduced by 6-7% with respect to the 1D case, with very similar systematic errors. The central value is in agreement with the 1D case, even if the probability for electrons is sort of low. The idea is to include this method in the forthcoming 189 GeV mass paper.