Stato dell'arte The Higgs boson (H) is an essential particle of the ...

overratedbeltΤεχνίτη Νοημοσύνη και Ρομποτική

25 Νοε 2013 (πριν από 3 χρόνια και 4 μήνες)

64 εμφανίσεις

Stato dell'arte

The Higgs boson (H) is an essential particle of the Standard
Model (SM)

which explains
the origins of mass and electro
-
weak symmetry breaking. Within this model the Higgs is
expected to
de
cay predominantly to b bar
-
b jets if its mass is low
er than 135 GeV/c^2
.

The search for H
-
>bb
-
bar is extremely difficult at hadron colliders where the background
coming for QCD is several orders of magnitude higher than the signal.

For the moment, LHC experiments are not planning to use this decay channel
for Higgs
searches and recent studies [1] done with data collected at the Tevatron show that
sensitivity for H
-
>bb
-
bar is lower than what can be obtained when the Higgs is produced
in association with a vector boson, W or Z.

Figure 1
$IMM_PER_0001

shows th
e
expected

limit on the production cross section
with

respect to the SM expectation as function of the Higgs mass for the process pp
-
bar
-
>H
-
>bb
-
bar at the Tevatron collider for 1.1
i
nverse femtobarn of data. The background and
the signal have been evalua
ted using data collected with the CDF detector and Monte
Carlo simulation respectively. The expected number of background events in each mass
bin has been fluctuated following a Pois
s
on distribution and a systematic error of 10%
was assigned to the backgro
und shape given the poor
knowledge

of bb
-
bar spectrum. The
cyan band represents the 1
-
sigma error on the expected limit.

This study indicates that for the time being only WH and ZH can be used to set limit or
possibly

discover the Higgs at the Tevatron in
the low mass region.

Currently VH events are recorded by triggering on leptons (muons, electrons and
neutrinos) coming from W and Z. In order to keep the trigger rate acceptable as the
luminosity increases, the lepton momenta is raised loosing a considera
ble fraction of
signal events.

The other key element in the analysis is the di
-
jet invariant mass resolution. The jet
energy is corrected using the standard algorithm and it does not tak
e

into account the fact
that these are b
-
jets.

In order to enlarge th
e Higgs acceptance it is necessary to start triggering on H
-
>bb
-
bar
and to improve the sensitivity specific b
-
jet energy corrections must be applied to jets.

H
-
>bb
-
bar can be recorded by a specific trigger that allows to select online a data sample
enrich
ed of b
-
jets. The trigger algorithm can act on several quantities and in particular on
the jet energy that has to be treated with particular attention since this cut can bias the
invariant mass distribution.

Figure
1

The jet energy determination translates directl
y into di
-
jet invariant mass resolution and if
it is too poor the signal is washed out from the background. One of the best way to
improve it goes through the usage of the well known resonances, Z
-
>bb
-
bar.

The Padova unit has already studied this signal e
xtraction. The production cross section is
several order of magnitudes bigger than the VH for a mass of 120 GeV/c^2 and therefore
a large sample is expected to be collected at CDF. Extracting the Z
-
>

b bar
-
b resonance
allows to improve the b
-
jet energy s
cale determination and, most notably, opens the
doors to a direct test of algorithms that attempt to increase the resolution of the b
-
jets
energy.

The analyzed data are collected with a trigger that requires two central jets identified
with a very simple
algorithm implemented for the Run I data taking. In addition to the
jets the event must have at least two tracks reconstructed by Silicon Vertex Tracker
(SVT) (see Modello A for a description) .

Given the coarse
jet

identification
,

the only way to keep the

trigger rate under control is
to cut on the tracks impact parameter at 160 microns loosing the major part of the signal.
As

the Tevatron luminosity has increased by almost factor 3 since the start of data taking
,

the trigger requirements had to change se
veral times and in the last period
,

when the
luminosity reached 3.2x
10^
32

cm^
-
2s^
-
1
,

the trigger was dynamically prescaled, only one
event out of a given number which depends on the actual luminosity is accepted. The
variable conditions have made almost im
possible to analyze the full dataset due to the
difficulty of simulating the different trigger conditions to extract the trigger efficiency.
The analysis proceeds
getting rid of the non b background by requiring the two leading
jets to be identified as b
-
jets by the presence of secondary decay vertices.

No attempt has been done to tag b
-
jets by requiring a soft lepton from b.

The two b
-
tagged jets must be almost back
-
to
-
back to reject
part of the

QCD b bar
-
b
background. Figure 2
$IMM_PER_0002

shows the i
nvariant mass distribution of the b
-
jets (blue points) with the data
-
driven background (green area) and the expected signal
(red area) superimposed. The inset in the upper right shows the data minus the
background. In total 5674

448 events are selected in
584 pb
-
1

of data. Fitting the
invariant mass distribution and comparing it with Monte Carlo predictions a b
-
jet energy
scale of 0.974

0.011 has been determined[2].












The
difficulty of
this
measuremen
t comes mainly from two factors: the poor jet

energy resolution that widen the resonance
and the background peaking almost under the signal due to the trigger requirements.

Figure 2

The b
-
jet energy resolution can be improved correcting for the missing neutrino that
couples the lepton in the semileptonic deca
y. This is completely missing in the current
jet energy corrections
and

its development is part of this project.

The shift of the di
-
jet invariant mass toward high values is partially solved by the new
calorimetric trigger at Level
-
2 (details in Modello A
)
. The new hardware allows to
reconstruct jets with a sophisticated algorithm

and gives the possibility to keep the jet
energy threshold low enough not to have signal and background peaking at the same
value.

The Padova unit has actively participated to th
e calorimeter upgrade by making the
system monitor. The online L2CAL online monitoring controls system functionalities
and eases and speed
-
up problem finding and maintenance. It is composed of two parts:
the first part checks hardware behavior by comparin
g data readout from boards to a
system emulation. The second part monitors clustering algorithms run on the
Level
-
2 PC
.
This structure allow
s

to control both the hardware, firmware and software components of
the system. The monitoring system has been devel
oped in compliance of the CDF
standards, and is constantly run during normal data taking operations.


Another

sector where the Padova unit is already collaborating is the production of the
GigaFitter

mezzanine boards needed to host the FPGA for the trac
ks fit. The project
developed by the Siena unit has been given to the Padova e
-
shop and the prototype board
mounted.
The GigaFitter will allow to
extend the lepton coverage in the forward region
of the detector and to improve the b
-
jets identification, in
creasing the acceptance on the
searched signals.


Once the data are collected,

the capability to select signal events with high efficiency
and to extract them from the huge background heavily relies on the analysis strategy.

Up to now, the usual appro
ach was to select only the best reconstructed objects and to
apply selection techniques based on sequential cuts on the variable distributions. This
approach is being surpassed by multivariate analysis techniques which can exploit the
subtle correlations p
resent between the variables. Neural networks are beginning to be
widely used in HEP analysis, as well as support vector classifiers and nearest neighbor
methods. But new methods for combining classifiers have been developed in the last
decade and are la
rgely unexplored in HEP. The Padova unit intends to apply these new
techniques to the search of the Higgs boson, where we need to exploit at the maximum
level all the information carried by the variables of the event, as we have to face with
really small
ratios between signal and background events.






[1] Search for direct production of H
-
>bb
-
bar

Authors: S. Amerio, S. Camarda, M. d'Errico, D. Lucchesi, S. Pagan Griso

CDF Note 9677

[2]
W and Z Physics at Tevatron

Author(s): J. Donini, the CDF and D0 C
ollaborations


Proceedings 42nd Rencontres De Moriond: Electroweak Interactions and


Unified Theories, La Thuile, Italy, March 10
-
17, 2007. FERMILAB
-
CONF
-
07
-
155
-
E