Tau Trigger Vertical Slice

1
Tau Trigger Vertical Slice

• M. Pilar Casado (IFAE UAB)

• People involved
• Trigger levels for tau reconstruction
• Strategy
• Conclusions

8/2/05
2
The Tau/Etmiss/Jet Group

• A summary of future concentration is as follows
• ETmiss -- Caron, Moore, Pinfold
• Tau -- Soluk, Pinfold, Casado, Osuna
• Jets Steve Armstrong, De Jong (validating
  ATLFAST studies), McGill Group (3 faculty?)
• We will be asking for two new RAs one to work
  on Etmiss aspects and the other to join the
  effort on jets
• In July 2003 Roger Moore joined the Alberta
  effort
• The McGill group have expressed their intention
  to join the PESA Jet/Etmiss/Tau effort. They are
  all involved in other projects at the moment but
  they will move their main effort over to ATLAS
  relatively quickly.
• Of course if anybody else want to join the
  effort, dont hesitate!

3
LVL1
4
(No Transcript)
5
Processing steps of T2CALO at each step via HLT
Steering mechanism, data request is made and
accept/reject decision is possible
Total Electromagnetic Energy
E3x7/E7X7 in EM Sampling 2
(E1-E2)/(E1E2) in EM Sampling 1
Hadronic (Tile EM HEC) Energy

• Photon Selection Strategy
• Shower shape analysis to reject dominant
  background from jets with a leading p0
• Possibility to use track veto - identify
  conversions first

6
Tau/Hadron separation
7
Tau/Hadron separation
8
Event Filter

• The Event Filter is the 3rd stage of the online
  selection chain.
• It can access the full event information.
• It executes filtering algorithms based on the
  offline reconstruction (tauRec for taus).
• Seeding the algorithm is however a basic
  difference between offline and LVL2 and EF
  algorithm implementation.

9
Seeding

• Seeding consists of using the previous trigger
  levels to guide the reconstruction to those parts
  of the event relevant for preparing the trigger
  decision.
• The (h,f) coordinate from the previous level is
  used to define a Region of Interest. The user can
  set a size for this region according to physics
  and performance criteria.
• Only the information within the region is accesed

10
Progress Reported at the Frascati Meeting

• Defined a program of collaboration with BNL for
  upgrades to T2CALO
• We managed to get all the changes we required in
  the LVL2 T2CALO trigger into SW version 6.0.4
• We were well on the way to getting code working
  with version 8.8.1
• We have slighly improved LVL2 by using another
  ratio RL1 we did not find any isolation windows
  that improved matters significantly at LVL2
• Our EF TAUREC PT ran successfully
• Online
• EF tau trigger code (wrapped tauRec) compiled
  under version 8.5.x It was added as a PT and
  executed
• Next step was to run it under 8.7.x in order to
  run it on testbeam data.
• In the process of adding monitoring to check
  (used latest versions of Dataflow and online)

11
Work Since Frascati on T2Calo (1)

• L1 Postdiction
• Code by Steve Armstrong, Christian Gross
• Breaks T2Calo region into 0.1x0.1 cells and
  simulates L1 energy sums
• T2Calo region etaWidth phiWidth around L1
  center from ROI Descriptor
• Add index to L2 CBNT to provide a definitive link
  between the L1 CBNT instead of trying to deduce
  which RoI the L2 result used
• Problem Used ROIWord from RecEmTauRoI which
  means it is not compatible with test beam
  FakeLvl1Conversion
• Need to add RoIWord to TrigRoIDescriptor instead
• Aim here is to validate the LVL1 trigger at LVL2.

12
Work Since Frascati on T2Calo (2)

• Add new user adjustable regions (in order to
  facilitate optimization) in which to form energy
  sums
• In particular use these in EM sampling 1 to add a
  new tau cut based on the energy ratio in a
  24x2/64x2 area
• Utilize the HAD calorimeter samplings to sum
  energy over a region better suited to tau (rather
  then just a 0.1x0.1 veto region)
• Change existing hard coded eta-phi and strip
  selection ranges into ones controllable from
  job-options with defaults set to current hard
  coded values.

13
Work Since Frascati on T2Calo (3)

• Use number of strips in each sampling instead of
  numeric eta-phi ranges to give more consistent
  result and limit possible eta dependence.
• Problem when spanning a region where granularity
  changes or is not the same as the central region
  an integer number of cells cannot yield the same
  eta-phi area.
• Try to preserve same eta-phi range as much as
  possible. Optimum way to do this is not clear.
  Should existing sums be adjusted also?

14
Tau-hadron Trigger Milestones (1)

• Short term (finished by the summer of 2005)
• Validation of LVL1 trigger - in progress
• Get the HLT (LVL2 EF) code working with current
  release on both DC1 and DC2 data - in progress
• Study improvements to EF part of the trigger
  (update tauRec) in progress
• Eg. Tau hadronic ET
• Eg. The tau likelihood variable (uses fits to a
  number of variables)
• Any useful new tau variables from the offline tau
  reconstruction group
• Run with wrapped Etmiss (standard offline code)
  EF trigger start late spring
• Perform timing checks of code - soon to be
  started
• Get online code going with wrapped tauRec and
  use on canned testbeam data - starts in spring.

15
Tau-hadron Trigger Milestones (2)

• Longer term milestones (finished by the end of
  2005)
• Track version changes (in calorimetry,
  tracking.)
• Replace wrapped EF code (eq tauRec or tauRec
  prime) with first version of optimized true EF
  code.
• Get online version going
• Run in conjuction with high level Etmiss trigger
  for
• Optimized Etmiss ( based on standard offline)
  EF code.
• Wrapped Etmiss (based on Etmiss calculation
  using reconstructed objects) EF code
• Using optimized EF code (probably using
  reconstructed objects)
• Repeat timing/performance tests with DC2 and test
  beam data.

16
Conclusions

• At IFAE were trying to form a group working in
  tau physics across the different levels.
• Well start with the reconstruction at LVL2
• Studying different sizes of the core used to
  calculate the energy.
• Fraction of energy deposited in the core region.
• Effect of the isolation in the e.m. and hadronic
  regions.
• Were also interested in the EF e.g. effect of
  TopoCluster algorithm in the reconstruction.
• Finally, well see the effect of the trigger in
  one or various trigger channels, as for example
  in bbA production with A-gt t t.