Документ взят из кэша поисковой машины. Адрес оригинального документа : http://zebu.uoregon.edu/~uochep/talks/talks04/neut.pdf Дата изменения: Fri Oct 1 20:03:03 2004 Дата индексирования: Tue Oct 2 09:54:36 2012 Кодировка:

The Use of Neutrals for an Improvement of Higgs BR
A Review of and an Update on Higgs BR
Jan Strube, Jeffrey J. Early and James E. Brau
jstrube@darkwing.uoregon.edu

University of Oregon

ALCPG Victoria LCWS, 2004 ­ p.1/18


Overview
Review of previous analyses Battaglia (Tesla TDR) Potter, Brau, et al. (Oregon) T. Kuhl (DESY) Yu, et al. (JLC) Presentation of new approach

ALCPG Victoria LCWS, 2004 ­ p.2/18


Battaglia - 1999



s =350 GeV, mH =120 GeV

Two studies (cut-based): 1. CDR Ver texer 2. Improved impact parameter resolution Modes: 1. e+ e- Z H (a) Z ll (b) Z 2 jets (c) Z 2. e+ e- H
http://tesla.desy.de/new_pages/TDR_CD/PartIII/partIII.pdf http://arxiv.org/PS_cache/hep- ph/pdf/9910/9910271.pdf
ALCPG Victoria LCWS, 2004 ­ p.3/18


Yu et al. - 2003

s =250 GeV, mH =120 GeV

JLC (Yu, Kang, Miyamoto, Park) 1. no Z ll ­ too few events 2. Z 2 jets 3. Z

0.257 / 0.201 relative error on H cc Ї

http://arxiv.org/pdf/hep- ex/0309041

ALCPG Victoria LCWS, 2004 ­ p.4/18


Oregon - 2003

s =350 GeV, mH =120 GeV

flavor preselection cuts 16 parameter neural net only Z lЇ l cleanest mode Tagging the Z Use scaling to include Z q q Ї

ALCPG Victoria LCWS, 2004 ­ p.5/18


Kuhl - 2004



s =350 GeV, mH =120 GeV

preselection cuts neural net tagging e+ 1. 2. 3. e- Z Z Z ZH ll 2 jets

LCWSParis, 2004

ALCPG Victoria LCWS, 2004 ­ p.6/18


Comparison I
BR

s =350 GeV, mH =120 GeV

Battaglia
CDR imp. Vtx.

DESY

OR 0.016 0.19 0.10

Yu et al.
s = 250GeV

H bЇ 0.024 0.024 0.011 b H cc 0.135 0.083 0.121 Ї H g g 0.055 0.055 0.083 Ї

0.201 -

ALCPG Victoria LCWS, 2004 ­ p.7/18


Comparison I
BR

s =350 GeV, mH =120 GeV

Battaglia
CDR imp. Vtx.

DESY

OR 0.016 0.19 0.10

Yu et al.
s = 250GeV

H bЇ 0.024 0.024 0.011 b H cc 0.135 0.083 0.121 Ї H g g 0.055 0.055 0.083 Ї

0.201 -

Using also e+ e- Z ( q q + )H Ї Ї

ALCPG Victoria LCWS, 2004 ­ p.7/18


Comparison I
BR

s =350 GeV, mH =120 GeV

Battaglia
CDR imp. Vtx.

DESY

OR 0.016 0.19 0.10

Yu et al.
s = 250GeV

H bЇ 0.024 0.024 0.011 b H cc 0.135 0.083 0.121 Ї H g g 0.055 0.055 0.083 Ї Using also e+ e- h

0.201 -

Using also e+ e- Z ( q q + )H Ї Ї

ALCPG Victoria LCWS, 2004 ­ p.7/18


Comparison II



s =350 GeV, mH =120 GeV

Results on cc Ї BR DESY all modes H cc Ї 0.121 H bЇ b 0.011 H gg 0.083

DESY Oregon e + e - Z ( l Ї + q q )H l Ї 0.154 0.19 0.013 0.016 0.114 0.10

ALCPG Victoria LCWS, 2004 ­ p.8/18


And now for something completely different...

ALCPG Victoria LCWS, 2004 ­ p.9/18


The Use of Neutrals for an Improvement of Higgs BR
on behalf of Jeffrey J. Early

ALCPG Victoria LCWS, 2004 ­ p.10/18


Useful Neutrals I
raw ver tex mass using charged tracks only has contamination

ALCPG Victoria LCWS, 2004 ­ p.11/18


Useful Neutrals I
raw ver tex mass using charged tracks only has contamination What if ... ? Adding all the gammas increases the ver tex mass.

ALCPG Victoria LCWS, 2004 ­ p.11/18


Useful Neutrals II
Number of Neutrals versus Raw Mass for C's 10 9 8 7 6 5 4 3 2 1 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
numberNeutrals Entries 3902 Mean 1.004 0.4146 RMS

Number of Neutrals from B and D Mesons vs. Raw mass of Seed Ver tex

Number of Neutrals versus Raw Mass for B's 10

numberNeutrals Entries 1164 Mean 1.027 0.3838 RMS

8

6

4

2

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

ALCPG Victoria LCWS, 2004 ­ p.12/18


Useful Neutrals II
Number of Neutrals from B and D Mesons vs. Raw mass of Seed Ver tex High Potential of Separation in ambiguous region below 2 GeV

ALCPG Victoria LCWS, 2004 ­ p.12/18


Using Neutrals
1. simple counting to use as NN input parameter The more input parameters, the better (?) 2. try to assosciate with seed ver tex to increase flavor separation pT - corrected mass is already very powerful method not possible with calorimeter hits, because of low(er) spatial resolution 3. combine to pions
ALCPG Victoria LCWS, 2004 ­ p.13/18


Methods
E sin vs. E cos for neutrals yields regions of high purity using a cut to add those to ver tex breaks even at error on BR sum of neutrals must have same pT as
True Decay Vector

Neutrals

Vertex Vector

p

neutrals

=p

tracks

- 2p

T

ALCPG Victoria LCWS, 2004 ­ p.14/18


Neutral Finding
Find neutrals that come from a ver tex. Employ neural net for each neutral Two uses: 1. use number of neutrals from ver tex as input to neural net 2. include the 4-vector in calculation of m

p

T

ALCPG Victoria LCWS, 2004 ­ p.15/18


First Results
use of neutrals H decay mode bЇ b cc Ї gg Ї

a) without 0.027 0.19 0.15 b) counting (jet) 0.023 0.15 0.12 c) counting (ver tex) 0.022 0.16 0.13

ALCPG Victoria LCWS, 2004 ­ p.16/18


finding
distinguish from background gammas by combining to 0 Oxford groupa : relative reduction of b-jet background to c-tag by 10-25 % Use MC to match into ver tex We could hand them to our NeutralFinder NN
a

0

Hillert, LCWSParis2004

ALCPG Victoria LCWS, 2004 ­ p.17/18


Summary
Compared on equal footing, the reviewed studies are comparable The results can be improved even before adding new modes First results are encouraging Room for improvement on existing methods We have some more ideas on what could be done

ALCPG Victoria LCWS, 2004 ­ p.18/18