Trigger Meister Report

1
Trigger Meister Report

• Kyle Stevenson
• Elizabeth Gallas
• Terry Toole
• Trigger Board Meeting
• May 13, 2003

2
global_CMT-11.04 Week 1

• MarkAndPass monitor stream is enabled
• All triggers satisfying a L3FMarkAndPass filter
  are written to this stream (inclusively)
• 176869 - no store - 1.05 hours using 13E30 --
  final test run
• 176874 - store 2540 - 3.21 hours using 40E30
• 176875 - store 2540 - 3.98 hours using 30E30
• No surprise initial lum is now consistently in
  40E30 range
• NOTE this is NOT high luminosity by Run 2 specs.
• Run startup at beginning of Store have been
  smooth.
• Fewer trigger/DAQ problems than last week.

3
global_CMT this week

• Peak rates during stable Runs during Stores this
  week
• (New peak lum 43 in store 2538)
• Store 2540 2538 2536 2529 2523
  2521 2511
• Init L (E30) 42 43 30 40
  40 37 40
• L1 Accept 850 850 1170 1100 1150 1200
  900
• L2 Accept 510 370 440 490 500
  520 580
• L3 Accept 58 54 53 50
  55 gt60 gt65

4
Rate Caps and Prescales

• Rate caps remain (1100 / 800 / 50)
• L1 and L2 rate caps are based on DAQ limitations
• keep FEBlt5
• keep the system running
• L3 rate cap IS NOT A DAQ (online) limitation
• (though it occasionally affects DAQ stability)
• it is a limitation based on offline processing
  and reprocessing
• Prescales
• 99 of time are dominated by rate to tape (true
  for over a year)
• if your new trigger's rate to tape is too high
  relative to your priority due to
• Rate to tape is too high
• your MarkAndPass rate is too high (MP rate L2
  accept / pass_1_of_n)
• 'temporary' L3 rate to tape limitation (collector
  router ...)
• high luminosity (accelerator performance is on
  upward climb)
• decrease in rate cap by the powers that be (goal
  is to average 25Hz now)
• your trigger will be prescaled at L1.

5
TM Activity in last week

• Special Runs in various stages of preparation
• Cosmic Ray, Alignment, CPS calib, Jet_Gap, new L1
  muon and CTT A/O Terms
• Database Changes
• needed now to enter special run trigger list and
  for v12
• keep track of 2 releases
• need to be in before the end of the week
• Limited manpower is addressing the following
• commissioning triggers to be added
• special run requests (commissioning and special
  runs),
• prescale adjustments
• rate monitoring
• resolving requests for v12 (asking for some help
  from requesters),
• These things take considerable time and have
  higher priority.
• When requests can be resolved, we enter v12
  triggers (lower priority).
• A reasonable resolution of all trigger parameters
  must be arrived at before triggers can be entered
  because TM's do not have time to define 150
  triggers multiple times.
• Kyle starting today to make some v12 entries

6
Specific Progress on v12.00

• General comments
• Many triggers are being added.
• Have some triggers to delete (Nikos has standing
  request for all groups to report ?)
• Ultimate list may exceed 150 triggers - note
  this is a huge number (thousands of parameters).
• Changes are extensive, probably NO existing
  trigger unchanged due to tool and filter changes
  in L3 or change in low lying tools and
  L3FPrescale being replaced by PassFraction
  filter.
• An example
• 29 new tool definitions for EM request alone
• About 7 minutes per tool (based on 3.5 hours
  entering 29 tools withVolker with some
  interruptions, then about half the time spent
  resolving parameters/values with available
  experts)
• typical rates will be higher due to expertise of
  requester

7
Strategy for v12.00

• Define groups of triggers that have a common
  representative. For Example
• EM triggers (Volker) now has one set of EM
  triggers with a full suite of L3 scripts (this
  took over a day of Volker and TM time to
  resolve/enter into database due to large number
  of elements to be entered into the database).
• In the process of making the above definitions,
  many questions have arisen that must be
• resolved - what are the (real) effects, consult
  with experts
• agreed upon - by the representative known to be
  associated with requests
• announced - to the TB
• These groups of triggers must ultimately be
  combined into v12 trigger list with a minimum of
  changes/conflicts.

8
Examples of Issues for v12

• Calorimeter unpacking (TB too cavalier about
  multiple unpackers)
• L3 expert advise (friday) don't run more than 1
• many groups have informally agreed on non-linear
  corrections making this public
• 'option of NADA' should be enabled - making this
  public
• Calorimeter clustering (must run multiple
  clustering tools
• clustering cone size
• vertex or no vertex.
• if vertex based on tracks with what criteria ?
• L3 TOOL changes
• 3 new and 3 changing parameters for electron
  tool
• A reasonable resolution must be arrived at before
  further triggers can be entered because TM's do
  not have time to define 150 triggers multiple
  times.
• We encourage requesters to play a proactive role
  in resolving these requests with all concerned
  groups and circulate decisions widely -- this
  will speed up the process.

9
Outlook (1)

• Lots of progress on pieces of v12
• Some groups just becoming aware of issues
• TM's remain optimistic
• We have some good triggers in 11.04, but some are
  being prescaled due to fat cats (gorillas or
  guerrillas).
• We have some good triggers going into 12.00 (some
  triggers moving in right direction)
• Proposed triggers must include a rate estimate
  but ultimately rates are measured online to
  determine your real rates and these rates are
  used to calculate prescales.
• TM's experience
• prediction based on all previous major trigger
  list changes V12 will not relieve us of the
  existing rate to tape problems
• requests for unprescaled triggers will not
  withstand the test of reality

10
Outlook

• Advise
• Please include some ultra low rate triggers in
  your triggering strategy which minimize rate to
  tape (less than 1 Hz at highest luminosity).
• QCD group set a good example
• QCD triggers implemented 9 months ago can run up
  to 60E30 unprescaled because of excellent
  rejection.
• An additional advantage of their strategy is that
  their triggers changed very little over the
  course of the last year.
• Discussion
• Andrew Brandt email on recent prescale changes,
  makes some good points. QCD is not low priority.
  Highest ET jet triggers have highest priority
  but all QCD triggers have low rates. Specific
  request
• reinstate gap triggers at 30E30 and above (now
  prescaled away). During Run 176607 previous
  40E30 file at 40E30 Gap triggers had total of
  0.6Hz to tape.
• We need more rate reduction to tape. Any
  reduction in prescale must come with an increase
  in some other trigger.