Towards Physics in Tevatron Run II

1
Towards Physics in Tevatron Run II
P. Grannis Manchester
Oct. 24, 2001
(some items for discussion)
Once the DØ detector, trigger commissioning
offline reconstruction of physics objects are
complete there will still be much to do to
optimize the physics analyses. This talk is
intended to stimulate discussion and draws on
some lessons from Run I. It is aimed at
identifying some of the areas where work is
needed for Electroweak studies and Searches
probably unimpeded by knowledge on my part of
details of recent work in the Physics Groups.

• Top quark
• Vector Bosons
• Higgs
• Supersymmetry

• Some issues to be considered for optimizing Run
  II analyses flagged throughout the talk

2
Top quark measurements
tt production 90 qq annihilation and 10 gg
fusion.
q
t t
t t
g
g
g
q
g

• Top BRs (SM) t ? W b (100)
• with W ? qq (67)
• ? en , mn
  , tn (11 each)
• So final states of tt ( lepton e or m)
• Dilepton bbl l nn 2 b jets, 2
  leptons, (low bknd, small signal) (BR
  5)
• Leptonjets bbqql n 2b, 2q jets,
  lepton, (moderate bknd signal) (BR
  30)
• All jets bbqqqq 2 b, 4 q jets
  (huge QCD bknd, 106 x signal) (BR
  45)
• Tausjets tnbbqq/tnl nbb
  (large bknd, t ID issues)
  (BR 20)
• In all cases, there is an additional hadronic
  system X formed from gluon radiation, and the
  spectator partons usually recoiling at small
  angles from the tt.

2
3
Top quark mass

• In Run I, top mass was extracted from l
  jets,
• l l 2 b-jets (and internally, all jets).
  
• In most cases, the method is to
• Choose a kinematic distribution that varies
  as mt changes
• Monte Carlo set of templates with various top
  masses, and the backgrounds included
• Likelihood fit for best Mt.
• (Many choices of kinematic distribution.)
• Illustration for l jets Use distribution of
  fitted top mass. Have 6 final state objects
  thus in zero mass approximation, 18 kinematic
  quantities needed. Measure 3-momentum of 4 jets
  and one lepton 2 components of ET. Constraints
  from equal t and t masses, and (qq), (l n) must
  have W mass, so have 20 known quantities ? 2C
  fit.
• BUT Did not in general identify the b-jets
  (only for b?mX in Run I), so have 12 ways to
  assign the jets to partons (6 if there is a
  b-tag).

Initial state radiation can give extra jets
this should be absorbed into the state X. Final
state radiation from top or jets these should
be combined with the parent jet.
true mass
METHOD Perform a 2C fit for all associations
of the 4 leading jets to the tt hypothesis
choose the smallest c2 solution. Call the best
fit mass the fitted mass. Do the same for
(Herwig) MC events with a family of known true
top mass and get template distributions of fitted
mass for each true mass.
fitted mass
3
4
Lepton jets analysis
Select l, n, jets with high ET (gt 20 GeV)

• Form topological variables for signal/bknd
  discrimination
• missing transverse energy
• HT SET(jets 2 to N) scalar transverse
  energy for non-leading jets. Large for top
  events due to large parent top mass, small for
  dijets.
• A (aplanarity) smallest eigenvalue of
  momentum tensor large for top events as the top
  quarks tend to give isotropic decays QCD jet
  backgrounds have parent 2-jet topology
• Dr minimum jet angular separation

Example HT distribution
ET
data
background
signal
Construct a measure D from variables that
indicates the relative probability for event to
be top vs. background two variants weighting
technique for low mass bias (DLB) and Neural
Network (DNN). Use this D to make a cut, or
as a weight for final event selection.
(b)
(a)
DNN
Obs. mass vs. DNN for (a) signal,
(b) background (c) data
(c)
fitted mass
4
5
Top mass in lepton jets channel
( lepton e or m )
Backgrounds
2/3 of bknd from W4 jets. Use parton
generator VECBOS with HERWIG/ISAJET fragmentation
to calculate. Check with W1, W2, W3 jet data
and Behrends scaling W(n1)jet/Wn
jet a 1/3 of bknd is from multijets, with a
jet faking lepton, and mismeasurements giving ET
. Determine from data with bad leptons.
Bknd (low D) sample
Best fits 171.3 6.0 GeV (neural net
analysis) 174.0 5.6 GeV (weight
analysis) (statistical errors) Combined
final DØ result for lepton plus jets Mt
173.3 5.6 5.5 GeV
Mfit
Mtrue
Signal (high D) sample
Mfit
5
6
Dilepton channel analysis
For the dilepton channel, there are two missing
neutrinos, so the fit is underconstrained by one.
DØ employed two methods Neutrino weighting
(nWT) and matrix element weighting (MWT). For
nWT, we assign a weight based on how much of the
nn phase space for signal is consistent with the
event kinematics, computed for a set of Mt
values. For MWT, compute a weight based on the
probability to have the lepton energies as
observed, and the product of PDFs required, for a
set of Mt values. After smearing many times with
the known resolution functions, and summing over
combinations, each event has a weight
distribution as a function of Mt
For the MWT analysis, weight distributions for
the 6 observed events
The weight distributions for all events are
averaged, and compared to MC templates for a
known input top mass. A best likelihood fit
gives the top mass.
Distribution of weight averages for 160 and 180
GeV MC samples, each with 6 events
6
7
Dilepton analysis results
nWT Analysis
MWT Analysis
Averaged dilepton mass Mt 168.4 12.3
3.6 GeV (lower systematic error than l
jets)
DØ Lepton jets and dilepton top mass
combined Mt 172.1 7.1 GeV
Tevatron average Mt 174.3 5.1 GeV
Other Run I methods
1. For l jet evts, assume set of Mts and do
3C fit at each. Form pseudolikelihood
L e-c2(fit)/2 Fit L for maximum. Get
Mt 176.0 7.9 4.8 GeV
2. Fit pT(b), pT(lepton), HT ( sum of object
ETs) etc. These have lower statistical
precision due to less dependence on Mt.

3. Use tt production matrix
elements in forming weights (looks promising)
7
8
Top mass in Run 2

• It will certainly be a different optimization
• With x20 data (Run 2a), or x200 (Run 2b),
  statistical errors get small thus systematics
  become more important.
• Much improved b-tagging (SVX) mean fewer
  combinatorics (though b vs. b is hard!), and
  much improved Signal/Bknd (e.g. double tagging)
• Many more overlaid minimum bias events at high
  L .
• Large improvement in MC capability (GRID farms
  around the world)

Look at the Run I systematic errors breakdown
Systematic errors dileptons Jet
energy scale 2.4 GeV Signal generator 1.8
Bknd model 1.1 Multiple Intns 1.3
MC statistics 0.3 Likelihood
fit 1.1 (Tot. syst. 3.6 )
(Stat. 12.3 )
Systematic errors l jets jet energy scale 4.0
GeV signal generator 1.9 bknd generator 2.5 multip
le intns 1.3 MC stats 0.9 likelihood
fit 1.0 method diff. 0.8 (Tot. syst.
5.5 ) (Stat. 5.6 )
How do we improve these?
8
9
Jet scale error
Jet energy scale (dominant error for l jets).
Error 2.5 0.5 GeV obtained from ET balance
in g jets events, taken from difference in
balance for data and MC. Check by ET balance in
Z(ee)jets events.

• Jet scale error statistically limited in Run I,
  but there are systematic issues that will arise
• jet widths
• min bias contributions to jet scale
• MC correction closure
• b-jet corrections ( light quarks)
• out of cone partons

• Are there alternatives that minimize the jet
  scale error?
• Is the dilepton channel better than l jets?
  (only 2 jets, so reduced jet scale effect)
• Ratio of W/top mass in tt events, then scale
  with accurate MW (partly cancel jet scale
  error?)
• Use variables less sensitive to jet scale (QCD
  radiation-proof variables) but with larger
  statistical error? (e.g. pT(lepton) for both l
  jet and l l )
• Can one do better with kT (JADE-like) jet
  definition? (force right jets?)
• Energy flow algorithm? (CFT for charged
  particles, CAL for neutral)
• Neural net/H matrix to correct n loss from
  b-jets?

9
10
MC generators
In Run I, HERWIG was the primary signal
generator. Error is set by study of top mass
variation as fn. of (x non b/W jets in 1st
4), (y jets 4), (z extra FSR jets).
Also used FSR-suppressed HERWIG. PDF variation
is negligible. VECBOS used for Wjets background.
Vary Q2 scale from jet ltpTgt2 to MW2. Vary
VECBOS to alter hW . But HERWIG, VECBOS are
not full NLO, and could have other features

• We need improved generators
• Implement new NLO top generators including
  interference of ISR/FSR (Orr, Stirling). These
  generators need interface to full fragmentation.
• Extend Wjets NLO to W4 jets? (DYRAD does W1
  jet). Add flavor accountability to VECBOS.
  Seek non-MC methods for Wjets (is Behrends
  scaling more accurate reliable? Verify
  theoretically at NLO?)

Likelihood fits
Likelihood curves are not parabolic error
depends on pts used, and on parametrization of
L fn.

• Get more rigorous definition of fitting
  procedure.

10
11
Other issues for top studies

• Min bias event overlaps Run I added 0.7 MC
  min. bias events on top, adding to Cal. energy.
  For Run 2, this up to 4-5. Good minbias MC
  modelling becomes crucial. Go to use of real
  minbias events taken at same luminosities for MC
  overlay (is now available, but needs work).

• Top mass definition Fitted mass (pole mass) is
  uncertain at level of 0.5 GeV due to
  non-perturbative renormalon effects. Can DØ
  help/test these effects ??

• Cross section Run 1 made 30 measurement,
  limited by jet scale background generator
  (20). If this improves, can use s(tt) as test
  of NNLO gluon resummed QCD! What precision for
  stt? Can this become a useful and meaningful
  test of resummed QCD? (Note that accurate s(tt)
  is needed for some Higgs backgrounds.)

• Mass measurements Can use of tt matrix
  elements improve the constrained fit? Explore
  the pseudolikelihood mass measurement (lower
  syst. error in Run 1)

11
12
Other issues for top studies

• Spin correlations Top decays before
  hadronization Run 1 made crude measurement of tt
  spin correlation l l direction correlations.
  How to make this QCD test most incisive? Get
  spin correlations into generators.

• Top EW measurements Single top production via
  W exchange are sensitive to Vtb , GTOT . Can one
  discover the single top production and then
  refine the selection of single top to enhance the
  precision of these EW parameters?

• FCNC and anomalous form factors what
  sensitivity for t ? Zc / g c to limit anomalous
  top couplings?

• Tau decays t BRs are sensitive to admixture
  of t ? H b with H ? t n . How much
  improvement on H mass limits?

12
13
W boson mass
Three (correlated) variables are sensitive to the
W mass Electron pT , Missing ET (neutrino),
Transverse mass mT mT 2pT(e) pT(n)
1 cos(f(e) - f(n)) mT is insensitive to the W
production dynamics (corrections O(pTW/MW)2 )
but requires the inferred neutrino pT, hence is
sensitive to detector response. pT(e) and
pT(n) Jacobian edges depend on pT(W). pT(e)
depends only on well measured electron
kinematics pT(n) also depends on hadronic
response
mT and pT(e) distributions as generated and pT(W)
0 (solid line) correct pT(W) (red points) and
after detector resolution effects (yellow
shading)
Method Fast MC to include W production, decay
and full detector modelling. Make templates of
mT , pT(e), pT(n), for set of MW values and find
best fit.
MW (DØ) 80.483 0.084 GeV 10 Run 1
measurements (electron channel)
13
14
Detector response modelling
The detector response functions for electrons,
recoil energy, radiative photons, etc. are mostly
determined from data distributions.

• Electron response Emeas aEtrue d taken
  from Z?ee and precision LEP Z mass
• Electron resolution s/E c s/ E n/E
  taken from fit to Z lineshape
• Electron directions chamber and calorimeter
  position calibration from muons and Z?ee
• Hadronic response require pT balance of ee-
  from Z and the hadronic recoil
• Hadronic resolution shape of pT(X)
  distributions along/perpendicular to Z?ee
• Trigger efficiencies measured with special data
  sets
• Energy corrections to pT(e) and UT for hadronic
  energy falling into electron window
• Correct selection bias for UT close to pe (loss
  of events due to energy isolation cut)
• Radiative decays (W? eng) taken from theory and
  modelled in MC
• Effect of extra minimum bias events underlying
  the W production taken from special inclusive
  triggers overlay these events at the same
  luminosity as for signal events
• Backgrounds (mainly from QCD jets misidentified
  as electrons at 10-4 level) taken from special
  data sets by selecting bad electrons. W?tn
  ? ennn is included in the decay MC

14
15
Sample MC fits
The MC, with parameters determined from data, can
be confronted with data to show the validity of
the model. Some examples
Z?ee distributions for central/end and end/end
es showing validity of electron response,
resolution and background determinations
ratio of data h distribution to that of MC (an
important constraint on the PDF)
recoil energy along electron direction
recoil transverse energy
15
16
Mass fits

• The W mass is obtained by comparing MC templates
  with various assumed MW to the data, and
  performing a likelihood fit. This example shows
  the mT distribution fit for end electrons.
• (similar fits for pT(e) and pT(n) distributions )
• A variety of cross checks are performed
• consistency of mT, pTe, pTn fits
• vary fit region in both mass and h
• bin results as a function of time, thus L
• vary the recoil pT cut
• fit for the Z mass using transverse mass
• compare result from two end calorimeters
• compare for different electron impact position
• compare for different EM energy fractions

mT distribution
c2 distribution
16
17
Combined DØ W mass fits
Systematic errors arising from uncertainty in
detector or theory parameters are computed the
parameter errors are themselves correlated in
some cases, and when the same data sets are
employed, are correlated because of the data set.
The simultaneous measurement of MW for central
and end electrons is important in reducing the
theoretical error due to uncertainty in the PDF.
New measurement uses electrons aimed near
edge of central calorimeter modules (14 gain in
statistics for W). This measurement helps
primarily by constraining response parameter for
non-edge electrons. The full correlation
matrix is determined, yielding these W mass
values
Run Ia (central e) mT 80.35 0.25
GeV Run Ib (central e) mT
80.44 0.12 Run Ib (central e)
pTe 80.48 0.14 Run Ib
(central e) pTn 80.37
0.18 Run Ib (end e) mT
80.76 0.23 Run Ib (end e)
pTe 80.55 0.24 Run Ib
(end e) pTn
80.74 0.35 Run Ib (central e-edge)
mT 80.60 0.44 Run Ib (central
e-edge) pTe 80.73
0.53 Run Ib (central e-edge) pTn
80.51 0.61 Overall DØ average 80.483
0.084 GeV
17
18
Combined World results
Combining with the CDF result, Tevatron MW
80.454 0.060 GeV . LEP experiments precision
(per experiment) about the same as Tevatron.
LEP MW has increased over the past two years, so
now good agreement between LEP and Tevatron LEP
MW average 80.450 0.039
World Avg MW 80.451 0.033 GeV
Mt, MW give strong constraint on the Higgs boson
mass in the framework of the SM (green ellipse)
Indirect prediction (red ellipse) from the
precision LEP/SLC/nN measurements in reasonable
agreement (lt than 2s), but with the new higher
MW, there is a weak hint of the effect of new
physics. Supersymmetry would provide new
particles whose virtual effects would predict
higher MW.
The indirect MW indication from Z, n, top
measurements is 80.373 0. 023 GeV, nearly 2s
from the measured value.
18
19
W boson mass errors
Dominant combined end central electron errors
Z stats special run stats
W stats

• Notes
• most errors are stats limited
• use mT, pT(e), pT(n) for consistency check (3
  error reduction over mT alone)
• pT(e) least affected by hadronic corrections
• overlaid min bias affects recoil resolution will
  become worse
• need both end and central es to control pdf
  uncertainty
• PDF error will be further constrained in Run 2 by
  W/Z asymmetries
• Measurement in m channel not as accurate, but
  good cross check (note CDF problem in Run 1)

W statistics 61 Z statistics 59 Calorimeter
linearity 25 electron resolution 19 electron
angle calib. 10 recoil response 25 recoil
resolution 25 E in electron window 12
Backgrounds 9 pT(W) modelling 15
GW 10 radiative corrections 12 PDF
uncertainty 7
Simple scaling of 84 MeV Run 1 error by 1/ N
would give dMW 20 MeV for Run 2 however, most
of the errors will have some irreducible
systematic component. The most problematic of
these will be the detailed electron response
function (non-linearity, offset), effect of min.
bias overlay, pT(W) modelling.
19
20
Questions for Run 2 W mass
The very first Run 2 W transverse mass plot !

• Minimum bias overlay Can one improve with
  algorithm that seeks to measure underlying energy
  from other vertices in the same/neighboring
  events for each event? (recall that the DØ
  calorimeter samples several crossings. Run 2b
  may have digital filter to tag energy per cell
  from adjacent crossings. Use modification of
  energy flow algorithm to tag collisions at other
  z in same event?)

• Which distribution? In presence of extra min.
  bias evnts, is pT(e) the distribution of choice?

• The W muon decay is less useful than electron
  (dp)m 2.5 GeV (dE)e 1 GeV at W peak, so
  expect (dMW)m 2 x (dMW)e . Nevertheless, the
  W? mn channel should be used

• Muons How will one best cross-calibrate the
  muon momentum and electron energy scales? What
  subsidiary measurements are needed to understand
  the effects of dead material, radiative effects,
  etc.

• Muonic Zs Can Z ? mm events be used for
  establishing hadronic recoil parameters (even if
  Z ? mm resolution not so good)? (Double Z
  statistics)

20
21
Questions for Run 2 W mass

• W pT Is there a (different) optimum cut on
  pT(W) that will minimize the combination of
  statistical and W production model error?

• Radiation With a solenoid, radiated photons no
  longer overlap with the electron so well what
  errors for the radiative corrections and how best
  to establish them?

• New methods Internal Run 1 analysis used ratio
  of W and Z transverse mass and scale to LEP Z
  mass (thus is limited by Z statistics). This
  cancels many effects due to electron, underlying
  event response/ resolution. Remaining systematic
  errors from small differences in W/Z production
  distributions. How far can this measurement be
  taken?

• Combined channels At present, detector
  parameters (e.g. had. response) are determined in
  subsidiary data sets and used to generate
  separate mT, pT(e), pT(n) distributions. Can
  the fact that the underlying MW is the same for a
  given event in each of the distributions, can one
  add a constraint in the parameter determinations?
  (Like the constraint that the unknown t and t
  masses are equal.)

21
22
Higgs search

• The canon
• (a) SM Higgs now constrained to lt 200 GeV
  
• (b) maybe LEP sighted it at 115 GeV
• (c) Tevatron constraints from top/W will tighten
  indirect indication by x2-3.
• (d) Tevatron can discover/see evidence up to 180
  GeV before LHC.
• At masses below 135 GeV, use W/Z (H)
  production. l n (bb), l l (bb),
  nn (bb), qq (bb) , and l
  n (tt) (??), qq (tt) (??) channels all can
  contribute.

expected run 2 precision

• Above 135 GeV, H ? WW dominates can use
  gluon fusion production.

22
23
General Higgs search questions

• For some channels (Z(nn)H(bb), Z(tt)H(bb),
  Z(qq)H(tt) and Z?bb), triggering is an issue,
  even with STT detached vertex trigger. The Run
  2a trigger should have t trigger capability using
  isolated high pT track triggers. The Run 2b
  trigger upgrade envisions correlating CFT and CAL
  information to further aid in t triggering.

The Z?nn H?bb dijet opening angle and missing
ET correlations differ from QCD dijet events.
Run 2b trigger improvements to CAL trigger to
allow topological analysis of energy towers, and
finer CFT/CAL correlation could improve t and ET
triggers at L1.
dijet opening angle (in trk
sectors)

• Triggering Study trigger rejections possible
  with CFT/CAL correlations, and topological
  triggers using CAL. Refine the triggers for Z?
  nn / tt H?bb, and Z? qq H? tt.

opening angle missing ET
correlation
23
24
General Higgs search questions

• The H?WW analysis simulation is at present
  based on series of simple cuts on kinematic
  quantities and is not optimized. There is no
  mass peak left after cuts, so the analysis is a
  counting experiment.

• H?WW Pursue a multivariate analysis for the
  H?WW channel to improve its sensitivity. Seek
  variables that survive after cuts that carry mass
  information.

• There may be a substantial diffractive
  production of Higgs, leading to two diffractive
  protons and a gap in rapidity to the di-b
  jets. Trigger on leading protons and rapidity
  gaps. Can one discover Higgs this way?

• Diffractive Higgs Sort out the theory of
  diffractive Higgs production simulations needed
  to see if this is a useful channel and to find
  trigger efficiency.

24
25
Issues for Higgs search
Higgs will not be a precision business in Run 2 !
The Run 2 Higgs Working Group study has
examined many of the critical issues for refining
the search sensitivity. The two main keys for the
Higgs search are dijet mass resolution, and
b-tag efficiency and purity.
NN algorithm
Eflow algorithm

• 1. Dijet mass resolution issues
• Jet energy resolution is key (angular error is
  small)
• Separate optimization of b and q/g jet
  resolutions
• Getting a good sample of Z?bb will be crucial
  for studying b-jet optimization.

• Energy flow (Use charged tracks neutrals in
  CAL). Optimize this algorithm (using data).
  Recent studies show perhaps 25 improvement over
  cone alg., (15 over NN alg.) largely due to ICR
  region, may be possible.

Effect of corrections to b jet on Higgs mass
resolution

• b-jet resolution Devise multivariate algorithm
  to correct b-jet resolution use the
  multiplicity, leading track momenta, jet mass,
  topology etc. in each event to optimize dE/E

• Z?bb Improve the trigger rejection at L1.
  STT is essential at L2 what else? (only 2 jets?
  equal ET jets?

25
26
Issues for mass resolution, contd

• Effect of initial state radiation (should
  ignore) and final state radiation (should combine)

FSR only
ISR only
ISR FSR

• Handling extra jets Can one devise a way to
  guess which are ISR and FSR jets? (pT, mass,
  angular weighting) Can JADE-like algorithms
  forcing event to right jets help?

• Choice of jet algorithm (Higgs WG found little
  help from kT algorithm)

• Jet algorithm What is the optimum jet
  algorithm. Can kT be better than cone?
  (Recent studies suggest it does not!)

• Event pileup Higgs WG 25 deterioration from
  overlaid min bias events

• Event pileup Can one learn to recognize energy
  due to pileup and subtract? Run 2b CAL trigger
  with digital filter could help tag events from
  other crossings.

26
27
b-tagging for Higgs
Yes, the low mass Higgs search is all about
b-tagging! The full simulation MC b -tag
efficiency mistag rate are improved from the
Higgs WG (parametrized MC).
old Higgs Wkshp parameterized MC
27
28
b-tagging for Higgs
High tanb Susy Higgs can emphasize multi-b
events for example bbA production this may
indicate different b-tagging needs (looser?)

• b tag algorithms Using data, work to improve
  the b-tag efficiency and decrease the mistag
  rates. (A Profound Statement !!)

Run 2 reach in MA vs tanb

• bb event tag Is it possible to improve upon a
  simple tight/loose single b-tag, to make a
  bb-pair tag based on a single combined goodness
  variable? (and by extension a 4b variable for
  selecting Susy Higgs events.

28
29
Supersymmetry
Run 2 has discovery reach beyond LEP in 25
different channels. Leptons and ET are the key
ingredients. DØ low ET lepton triggering in Run
2 is important addition. Tevatron has good
potential for sighting SUSY in Run 2.


Trilepton final states probe chargino pair
production (Run III Run 2b 15 fb-1).
GMSB gg ET search (c10 ? G g )

• Trigger questions Develop more powerful lepton
  triggers. Take the lepton thresholds lower using
  CFT/CAL correlations. Design restrictive t
  triggers. There is room for new ideas and
  simulations.

• Pointing photons GMSB has g decays from NLSP
  to LSP gravitino. Develop algorithms for photons
  pointing to vertex.

29
30
Supersymmetry
Stop searches in several channels

• squark, gluino decays place a premium on
  missing ET resolution, control of calorimeter
  noise.

• Missing ET How well can missing ET be
  incorporated in the trigger. Include ICR
  detectors in L1 trigger. Modify thresholds for
  summing CAL tower energy into ET to optimize
  resolution at L1.

• Kinks Develop the algorithms needed to detect
  charged particle kinks associated with c1 ?
  c10 p decays (slow p ) characteristic of AMSB.

• And

• Framework Tevatron in Run 1 lacked a solid
  phenomenological framework (defined benchmarks)
  for interpreting SUSY limits (as LEP had).
  These need to be defined for the several types of
  SUSY breaking. Also should develop the
  methodology for combining DØ and CDF results for
  Tevatron averages.

30
31
Summary

• Its not all over when the detector is
  commissioned !
• New clever ideas are needed to gain more
  sensitivity for Run 2 physics studies and
  searches
• Much work remaining to develop generators and
  better control of QCD processes
• Great opportunities to develop triggers that
  increase sensitivity to t, ET, special
  topologies
• Develop better algorithms dijet mass,
  b-tagging, min bias event tags, etc.

32
abcdefghijklmnopqrstuvxyz Abcdefghijklmnopqrstuvwx
yz in System VT special ABCDEFGHIJKLMNOPQRSTUVW
XYZ caps in 1234567890-
1234567890- in System VT
special !()_ caps in
\,./ lt?_at_ \,./
ltgt?_at_ in System VT special



gt

ET
gt


33
Combined World results
Combining with the CDF result, Tevatron MW
80.454 0.060 GeV . LEP experiments precision
(per experiment) about the same as Tevatron.
LEP MW has increased over the past two years, so
now good agreement between LEP and Tevatron LEP
MW average 80.450 0.039
World Avg MW 80.451 0.033 GeV
The indirect MW indication from Z, n, top
measurements is 80.373 0. 023 GeV, nearly 2s
from the measured value.
Mt, MW give strong constraint on the Higgs boson
mass in the framework of the SM (green ellipse)
Indirect prediction (red ellipse) from the
precision LEP/SLC/nN measurements in reasonable
agreement (lt than 2s), but with the new higher
MW, there is a weak hint of the effect of new
physics. Supersymmetry would provide new
particles whose virtual effects would predict
higher MW.
18
34
(No Transcript)