Future dates:
Thurs. Jan. 13th - Next meeting (Final results from most channels.)
Mon. Jan 17th - Final inputs to BEHOLD (no meeting).
Jan. 26th?? - HTF meeting during ALEPH week. Final results.
Middle February - Results given to LEP Working Group.
End February - Draft 1 of paper ready for Moriond
end of March - Last day to show improvements (with proof).
early April - Final decisions for analyses in BEHOLD 2000.
1. - Summary of Final Analysis Decisions
--------------------------------
The following decisions were based upon previously shown improvements.
As some of the debates were long and the physics content was discussed in
previous meetings, I have only outlined the final decisions below.
hA 4 jet Improvements
----------------
The benefits of removing the unneccessary thrust cut, and including the
Dtheta3 variable had previously been shown and were accepted. The Dtheta3
variable is designed to remove qqg events with gluon splitting. See HTF on
Sep. 16 1999. Boris also showed comparisons of the hA limit using various
mass fit algorithms. Although the 5C fit was the most performant, it was
agreed to use the 4C fit which performed nearly as well and does not strongly
affect the background mass distribution.
hZ 4 jet NN Improvements
--------------------
The 4C fit was agreed, as in the hA analysis, to replace the Higgs mass
obtained from 4 jet rescaling. The decay angle variables used in last year's
cut analysis pair selection were included into the 4 jet neural net, giving
a total of 19 variables in the 4 jet neural net.
hZ 4 jet Cut Improvements
--------------------
To retain consistency with the other 4 jet analyses, it was decided to
also use the 4C fit for the Higgs mass instead of 4 jet rescaling. A bug
was found in the previous work concerning the best jet pair choice. John
Kennedy redid the work following the meeting, and also incorporated the
switch to the 4C fit. His conclusions were that the best pairing choice is
the choice given by the decay angle PDF, i.e. the same choice used last year.
John will show the results of his work at the next HTF meeting.
Other Analysis Improvements
--------------------
The other neutral Higgs analyses are using the same analyses as last year
with the exception of new Neural Net trainings and optimizations.
2. - Summary of Other Decisions
--------------------------------
Overlap between Analyses
--------------------
To account for multiple analyses selecting the same events, like the
4 jet selection selecting Hll events, the following scheme similar to the
method used at LEPC was adopted. Overlaps with the Hll analysis are removed
by using a new routine which encapsulates the entire Hll code. To remove
remaining overlaps, the event selections are ordered:
Htt --> Hll --> Hnn --> 4 Jets. Events selected by a previous analysis can
not be reselected. The only remaining overlap with the 4 jet analysis
selecting Htt events is small enough to be ignored in the Monte Carlo for
shape generation, etc. Precise signal and background numbers could be found
quickly, once analyses were finished. Data treatment will be determined by
the event ordering.
Overlap between hZ and hA
--------------------
The 4 jet NN analysis will use the 2b/4b overlap treatment instead of the
AND and exclusive OR's between the hZ and hA analyses used in last year's
limits. In the 4b overlap treatment, the hZ NN and hA F variable are
alternately used in the confidence level calculation depending upon the choice
which gives the best expected limit as a function of sin2(b-a).
The cuts analysis will continue to use the same method as last year.
LEP Higgs Working Group Limit
--------------------
The calculation of the ALEPH Higgs limit for the LEP working group
last year required two limit calculations per point (whether to treat
events like hZ or hA) and used the calculation which gives the best
expected limit. This year, the determination of whether to use the hZ
or hA treatment will be determined by parameterizing the
optimal transition curve in the expected Higgs limit plane.
3.- Minutes of Remaining Meeting
-----------
Interpolation of 4 jet Shapes - D. Smith and Jinwei Wu
--------------------
Work involving the interpolation of signal mass shapes to different
center-of-mass energies and to intermediate masses was shown. Some MC
was generated at a center-of-mass energy of 191.6 GeV to help the study
on energy interpolation. By comparing the limits set by using 4 different
sets of energy interpolated shapes, David showed that the energy interpolation
which measures masses as a function of distance from threshold agrees to
the level of 50 MeV. Jinwei showed that the confidence level could be affected
by as much as 500 MeV by comparing interpolated shapes to real Monte Carlo.
The reconciliation of these seemingly divergent results is that the
differences seen by Jinwei are due to mass interpolation near the threshold,
and not the energy interpolation. Pedro suggested that one way to increase
the number of MC samples near threshold would be to first interpolate the
MC by energy according to distance to threshold, and then to interpolate
between different masses. This method would provide many Monte Carlo events
near the threshold where the mass interpolation method fails.
Hnn Progress - Gavin, Jenn, and Marumi
--------------------
Everything is on schedule. The final optimizations are nearly done, and
combination of the NN channels will begin next week.
Jet Smearing - Marumi Kado
--------------------
Marumi showed the Jet energy and angle corrections using the same
method as last year, where the corrections were determined from the 1999
Z peak data. The results were consistent with last year's results, and
consistent for data both before and after the TPC short. As a consequence
the average corrections for the entire year (ignoring TPC short dependencies)
were to be applied to this years analyses. The precise numbers will be
sent in an e-mail by Marumi to the HTF.
B-tag smearing - Boris Tuchming
--------------------
Boris showed the results of data/MC comparisons for smearing the
d0 and z0 of the tracks. Good agreement was found after smearing with
QIPBTAG. The 4 and 6 Variable Neural Nets still showed some disagreement
after smearing. An oversmearing (~double the correction) was then applied,
and reasonable agreement between data and MC was achieved for the neural
networks. It was agreed to use this oversmearing and to take half of the
correction as a systematic. The card to use:
SMEA 98/ 1.06 1.12
Charged Higgs - 4 jets - Elizabeth Locci
--------------------
Elizabeth showed the latest Charged Higgs results using the new LDA
(shown previously). At all center-of-mass energies, the new LDA showed
significant improvement over the old LDA. In addition, Elizabeth showed
a performance plot for a Higgs mass of 80 GeV with curves for all of the
different energies. The performances wer very similar, indicating that
interpolation of performance to different energies would not be too
difficult. Agreement between data and MC showed general agreement with
fewer WW events than expected in the mass peak near 80 Gev. With all
data combined, and 1 sigma less background subtracted to account for
the systematic uncertainty, an expected limit of 75.8 GeV is attained. The
observed limit is about 81 GeV.
Charged Higgs - cs tau nu Channel - Phil Seager
--------------------
Phil showed two improvements to the tau nu channel. The tau minijet
code from the htt channel was included to identify the tau's in the event.
This change is expected to lead to cleaner taus and consequently a better
Higgs mass resolution. Secondly, these tau's were used to measure the
tau polarization. By taking into account the polarization, it is hoped that
the taus from the decay of W+ can be separated from the H+. The tau events
were categorized by decay. For each event, then a number of polarization
variables were obtained. (Please see the transparencies for the descriptions
of the variables.) These variables all showed good separation between the
W and H+ decays. Placing these polarization variables into a linear
discriminant led to good results at mH = 80 GeV. This will be reoptimized
at a smaller Higgs mass (~73 GeV) closer to the expected limit. Further studies
are under way.