august 12, 2003elizabethgallas / triggerexposurerules 1 rules for triggers in exposure groups...
TRANSCRIPT
August 12, 2003 ElizabethGallas / TriggerExposureRules
1
Rules for Triggers in Exposure Groups
Elizabeth GallasFermilab Computing Division
Michael BegelUniversity of Rochester
Special MeetingAugust 12, 2003
August 12, 2003 ElizabethGallas / TriggerExposureRules
2
This talk is based on a DO note in preparation (note number 4192) by Michael Begel and Elizabeth Gallas
• Current Resource Limitations• Systems involved in Exposure Group Allocation• Definitions:
• Triggers• Exposure Related NEOTERMS
• Combining triggers into Exposure Groups• Rules for Luminosity Normalization and Accounting• Other Issues• Alternatives
Outline
August 12, 2003 ElizabethGallas / TriggerExposureRules
3
Resource Limitations
• Online resources are allocated by the COOR program based on requests in trigger configuration files produced by the Trigger Database
• ‘Current’ Limitations:– 8 exposure groups (EG)– 128 L1 trigger bits and 128 L2 trigger bits
• where every L2 trigger must be paired with a L1 trigger in a one-to-one relationship,
– 4094 L3 trigger bits (‘after shutdown’)• sometimes called simply “trigger bits”• where multiple L3 triggers may depend from a single L2 trigger.
• Exceeding these limitations is outside the scope of this meeting
August 12, 2003 ElizabethGallas / TriggerExposureRules
4
Applications involved …• Luminosity system:
– Measure, record, provide normalization and/or accounting• for DO physics analyses and to DO leadership and the Fermilab Directorate
– provide feedback on beam conditions to DO Operations, Beams Division– Perform cross checks on system integrity, some of which depend upon
following Trigger Database Application Rules
• Trigger Database Application:– Store physics Trigger Lists used online, benchmark Lists for simulation– Report trigger configuration settings – Generate precise programming (trigger configuration xml) minimizing
resource utilization (e.g. sharing L1/L2 bits, EG where possible)Must feature: consistency, backward compatibility, reduce possibility for mis-
configuration (no hand written xml…)
• COOR– Reads configuration ‘xml’ files– Allocates resources for online operations
August 12, 2003 ElizabethGallas / TriggerExposureRules
5
A Trigger is a Logical Condition• identified by a trigger name
• w/a set of criteria called a Script at Level 1, Level 2, and Level 3> each of which is satisfied if all of its logical conditions or TERMS is satisfied
• Trigger bit = TRUE for an event if L1,L2 AND L3 Scripts=TRUE for the event
• A subset of the TERMs utilized at Level 1determine that trigger’s ‘exposure’ classification
2EM_HI_Z
CEM(2,10)TTK(2,pt3)AfastzALiveBX *ASkip0 *
2EM(.9,10.,.trk)
ele(2,.95,glob,..)
InvMass(ele1,ele2,75,100….)
Trigger Name Script
Terms
L1
L2
L3
August 12, 2003 ElizabethGallas / TriggerExposureRules
6
A Trigger List• identified by Triggerlist Name/Version
• contains one or more triggers
• like a tree with Triggers as branches– if any trigger is satisfied, the event is recorded and the trigger bit for that
trigger name is set to TRUE in the event record
global_CalMuon-?.??
2EM_HI
EM_MAX
3JET_HT
MU_JET_HI
Example:
…logicalTERMS(yes/no)...
L1ScriptL2ScriptL3Script
August 12, 2003 ElizabethGallas / TriggerExposureRules
7
Simplified Trigger Database Diagram
Trigger Listtl_name/tl_version
Trigger Nametn_name/tn_version
1
N
L1 Scriptl1s_name/l1s_version
L3 Scriptl3s_name/l3s_version
L2 Scriptl2s_name/l2s_version
NEOTERMAnd/Or Term
1
N1
11 1
11
1
L2 Filtert_name/t_version
1
N2
L3 Filtert_name/t_version
1
N3
NAME/VERSION scheme is repeated throughout the design.The name is intended to reflect the conditions in that definition
August 12, 2003 ElizabethGallas / TriggerExposureRules
8
Trigger Nomenclature – L1• L1 Script decision = TRUE or FALSE
– Logical AND of one/more NEOTERMS
• NEOTERM == the “And/Or terms”– For any event: result is TRUE or FALSE
– Map into distinct bits of the L1 And/Or Framework
– Combine one/more to form a Level 1 Script decision• Example: TTK(1,10.)_CEM(1,9)_ncu / 1
– Have a flag called “Exposure Related” with allowed values ‘yes’ or ‘no’
• NEOTYPE == an L1 detector class– Classify NEOTERMS which share a common download protocol
• L1 trigger systems with a Trigger Manager have one NEOTYPE– Examples: ctt, fpd, fps, muo
• multiple or varied neotypes needed for Calorimeter, Framework, Pseudoterms– Examples: emcount, jetcount … specterm, pterm
August 12, 2003 ElizabethGallas / TriggerExposureRules
9
C -- Calorimeter -- based on Calorimeter “trigger towers”• emcount / CEM(n,Et[,Hv]) – Cal EM TTower• jetcount / CJT(n,Et) – Cal Jet (tot) TTower
M -- MUON – use Muon system scintillator, PDT,MDT and CFT• muo / MUO(n,Pt,eta,scint,wire,option)
R – * Forward Proton Detector• fpd / RDIFF% - Diffractive triggers• fpd / RA% -- Halo or Elastic triggers
T -- * CFT/CPS• ctt / TTK(n,p) – CFT track• ctt / TIS(n,p) -- Isolated track• ctt / TIQ(n,p,q) -- Isolated tracks in a quadrant• ctt / TIL - Isolated track(s) with low home-sector occupancy.
A – * Special (L1 Framework terms)– constructed from signals from: the Accelerator, Luminosity Monitor, Trigger
Timing and Control• specterm / Afastz, ALiveBX, ASkip0 …
Level 1 Trigger Systems:
* L1 Detector NEOTYPEs: fpd, ctt, spectermHave at least one NEOTERM ‘Exposure Related’ = ‘yes’
August 12, 2003 ElizabethGallas / TriggerExposureRules
10
Criteria for Exposure Related=‘yes’• Require/Veto on condition that biases one trigger over another
– e.g. Acaltc00 (Cal_Unsuppressed), AOut0
• Require/veto on luminosity counter signals other than require Afastz (min_bias)– e.g. ALMNorth or ALMNorth[v] …
• z vertex cut at level 1 other than the one used by the Afastz condition– e.g. tracking based L1 vertex requirement
• Beam structure dependent:– w/o violating poison distribution (# interactions per crossing at a given luminosity)
• beam crossing (e.g. ALiveBX, AfirstBXinSB…)
• tick selection (e.g. ATicSel0,1,2,3)
• unbiased disables (e.g. ASkip0, 1)
– violating poison statistics (# interactions per crossing at a given luminosity)• track occupancy < certain value (e.g. TOC) – prefers single interactions (low lum BX’s)
• fraction of hits > certain value (e.g. THT) – multiple interaction trigger (high lum BX’s)
• single interaction condition (from Luminosity system)
August 12, 2003 ElizabethGallas / TriggerExposureRules
11
Defining Exposure Groups• An Exposure Group is a group of triggers with a common dead
time (exposure to the beam while the detector is live)– defined by a crate list and a set of NEOTERMS
• Trigger Database Implementation is precisely that:– Groups consecutive triggers with common
• Exposure = ‘yes’ NEOTERMS (And/Or Terms) and• A distinct Device Group (crate list)
• Exposure Groups in global_CMT-12.20– All use same device group (allcrates/1)– 7 distinct sets of Exposure related = ‘yes’ NEOTERMS:
ALiveBX & NOT( ASkip0 ) & NOT( Acaltc00 )ALiveBX & NOT( ALMNorth ) & NOT( ASkip0 ) & NOT( Acaltc00 )ALiveBX & NOT( ALMSouth ) & NOT( ASkip0 ) & NOT( Acaltc00 )ALiveBX & NOT( ALMNorth ) & NOT( ALMSouth ) & NOT( ASkip0 ) &
NOT( Acaltc00 ) ALiveBX & NOT( ASkip0 ) & NOT( Acaltc00 ) & NOT( THT(5) ) ALiveBX & NOT( ASkip0 ) & NOT( Acaltc00 ) & THT(10)ALiveBX & Acaltc00 & NOT( ASkip0 )
August 12, 2003 ElizabethGallas / TriggerExposureRules
12
Report: global_CalMuon-12.20 (1)
August 12, 2003 ElizabethGallas / TriggerExposureRules
13
Report: global_CalMuon-12.20 (2)
August 12, 2003 ElizabethGallas / TriggerExposureRules
14
L1 specterm: Special And/Or Framework
August 12, 2003 ElizabethGallas / TriggerExposureRules
15
L1 fpd NEOTERMS
August 12, 2003 ElizabethGallas / TriggerExposureRules
16
L1 ctt NEOTERMS (1)
August 12, 2003 ElizabethGallas / TriggerExposureRules
17
L1 ctt NEOTERMS (2)
August 12, 2003 ElizabethGallas / TriggerExposureRules
18
Alternatives and Points for Discussion• Reminder : global trigger list is not for commissioning
– Get permission for Special Run during Global run• Advantage: write events to a separate stream outside of 50 Hz limit (if current
online DAQ problems are solved)
• Demonstrate your NEOTERM is Exposure Related = ‘no’• Alternative: Don’t use Exposure Related criteria at L1
– Level 2 and Level 3 triggers have access to the L1 And/Or Framework: write a L2 or L3 filter to impose the criteria
• Trigger DB Application change: – current system options:
• normalized by LUM ID group• normalized by someone else
– request seems to be to add an unnormalizable category• Add one/more additional allowed values of Exposure_Related• need algorithm that determines precisely
– grouping of triggers that contain those classes of TERMS– exposure group criteria
• to the collaboration: explain how you expect to use triggers in unnormalizable exposure groups which do not have the same live fraction.
August 12, 2003 ElizabethGallas / TriggerExposureRules
19
Other Issues
• We have current problems managing triggers in new EGs without luminosity normalization– Lack of cross section measurement in reference runs
– GM lmTrigger ratio makes no sense
– Prescale generator requires special coding by trigger
– Recent spikes seen in diffractive triggers: this behavior warrants the trigger to be disabled.
• Requirements for running these types of triggers in global: – Requestors must provide expected rate as a function of luminosity
– If those triggers exceed rate expectations or demonstrate erratic behavior, they should be disabled.
August 12, 2003 ElizabethGallas / TriggerExposureRules
20
Two Special Trigger Classes – related to luminosity/accounting
• zero_bias_%– in every EG for which D0 wants accounting (live fraction)– Level 1 only trigger
• Required NEOTERM == Exposure Group NEOTERMS– simplest example: ALiveBX (and ASkip0 – in every trigger)
» true on each of the 36 beam crossings of a single turn of the accelerator» About 1.7 M times per second
• Used to cross check the luminosity measurement and trigger system functionality• Really is unbiased
• min_bias_% (‘minimum biased’)– In every Exposure Group for which D0 wants luminosity from the LUM ID group– Makes no sense in an EG which includes a L1 condition violating Poisson Statistics– Level 1 only trigger
• requiring NEOTERM ‘Afastz’ – (and ALiveBX and ASkip0) – every trigger
• Based on Luminosity monitor:– North, South scintillator array on beamline
• Requires N and S pulse heights above threshold in timing coincidence• Gives a quick measure of the z vertex
– Necessary to measure luminosity– Is undoubtedly biased physics-wise
August 12, 2003 ElizabethGallas / TriggerExposureRules
21
Trigger List Rules ...Examples of rules for valid Triggers, Lists…
– all Trigger Names must • be unique (in that Trigger List)• len(TriggerName) 16 (thumbnail)• cannot contain special characters
– cannot use more than 4 Level1 Calorimeter EM or JET thresholds– cannot use more than 32 L1 muon terms from the set of 256 valid terms– cannot use more than 128 unique L1L2 bits– L3 filters and tools mustn’t use different versions of tools of the same
name– L3 filters and tools may call other tools, but tools may not call filters (not
true at L2)– L3 tool names must conform to SR parsing rules
–…Many rules checked upon db entry, but the ‘xml’ generator
checks many features as well ...