Muon Level-2 Outline

1
A. Maciel (N.I.U.) Muon-L2
The Level-2 Muon Trigger
ITEP -- LMU(Munich) -- Nevis Labs -- NIU
Arthur Maciel (N.I.U.) For The DØ b-ID Vertical
Review FNAL 04-04-01

• Muon Level-2 Outline
• Level-1 / Level-2 Muon-ID
• Example Algorithm
• Performance

2
Muon Level-2 Outline
A. Maciel (N.I.U.) Muon-L2
Data processing happens in three stages
(1) SLICs (track segment finding)
(C-code on T.I. DSPs)
INPUT PROCESSING OUTPUT

Muon Front End Data L1 Data
Pattern Recognition Re-Format
A - Layer Track Stubs BC-Layer Track
Stubs L1-objects (Aphi)
(2)ALPHA preprocessor (track assembly)
(C on Linux)
?Compare all SLIC reported objects, search for
matches ?Assemble matches (or singles) into final
L2 muon candidates ?Report track params. and
ID-quality to L2(global) and L3(seeds)
(3) GLOBAL Level-2 (trigger decisions)
(C on Linux)

• Relate muon L2 candidates to other L2 objects,
  report triggers

3
Overview of Muon L2 Algorithms
A. Maciel (N.I.U.) Muon-L2

• Pre-Global muon ID is based on muon system only,
  no cross referencing among different subdetectors
  (L1 CFT Pt-thresh).
• Track segment finding for A and BC layers
  separately. Run1 Pt.
• Segments built from both wire and scintillator
  elements.
• Calib central wires have drift and f(?T), Scints
  are T0 subtracted.
• L2 reports single-cell-wide 3D tracks in (Pt,?,f)
  coordinates.
• Central PDT (A)segments must satisfy track
  residual requirements.
• No vertexing, but hit patterns must be compatible
  with int. region.
• Added resolution (over L1) in track counting,
  pointing and timing.
• L2 candidate searches are not biased by or
  related to L1 results.

4
L2-Global Muon Candidates
A. Maciel (N.I.U.) Muon-L2

• Global L2 has two main functions object
  selection and object relations (objects
  Jet,EM,MEt,MU,TRK,IP)
• Muon selection (Number, Pt, h-range, Quality)
• Muon relations
• Angular separation, Dimuon mass CFT resolution
• Isolation, muon inside jet, track match
  (Pt,?,f) , IP
• Global can request object matches, but cannot do
  track re-fits, or ID building beyond CFT(Pt,?,f)
• Muon candidates sent to Global are also sent to
  L3 as seeds for data (detector module) unpacking

ALPHA
5
Overview of the L2mu Software Project
C Asm
C

• SLIC Driver
• Init / Config / Test Suite
• Soft Hard Resets, SCL init.
• Run control monitoring

• ALPHAS
• Worker Gen.Strategy,
• finalize formats
• AdministratorGlobal
• (expecting a free ride)

1
5

• DSP-OS
• Boot Kernel (Init /Config /Test)
• Data I/O driver and interrupts
• Data I/O memory management

2

• Simulator
• Framework
• Raw ? L2 data
• DSPs and Alpha

6

• DSP-Alg
• Interface w/ mem. management
• (A,BC)x(Central,Forward) , L1
• Error handlers, Monitoring

3

• MC Studies
• Backgrounds
• A-layer punch
• Thin region punch
• Halo and Multip.Int.
• (unmatched stubs)
• Eff./Rej./Rates
• L2 (Pt,?,f) resolution

7

• DSP-5
• Event sort/build/format
• Full readout events, MP
• Monitoring interface w/ VME

4
A. Maciel (N.I.U.) Muon-L2
6
A. Maciel (N.I.U.) Muon-L2
The Level-2 Muon Trigger
ITEP -- LMU(Munich) -- Nevis Labs -- NIU
Arthur Maciel (N.I.U.) For The DØ b-ID Vertical
Review FNAL 04-04-01

• Muon Level-2 Outline
• Level-1 / Level-2 Muon-ID
• Example Algorithm
• Performance

7
A. Maciel (NIU) L2 muon
L2 Muon ID
L1
MTC05
MTC10
L2 SLICALPHA

• 3D track resolution (counting)
• A BC segments separately
• A (3/4 hits) BC (3/6 hits)
• A BC crossing octants
• High PT A-layer w/out BC
• ? dependence of hit patterns
• (pointing to interaction region)
• simultaneous look at 2 or 3
• scintillator times on track

octant wide scint match (? wedge)
region wide scint match (f wedge)
CFT f wedge
PDT centroid
SLICs receive calibed data from Front Ends PDT
drift distances and phi (1st T-pair only) SCT
T0-subtracted times (w.r.to tile center) MDT hit
addresses only, no drifts. and From L1 MTC05,
MTC10, MTM, MCEN...
No cross referencing among different subdetectors
before Global-L2 (except L1 CFT-confirm
thres.) Run-1 muon PT (from toroid only) No
(muon) Z-vertex point back
8
L1
L2
L3
5 MHz
10 KHz
1 KHz
20 Hz
Expected L1 Muon Rates (KHz)
L1 muon bandwidth few KHz
Pt ? 4 GeV (2,2)GeV (2,4)GeV
? lt 1 1 0.2
? lt 1.5 2 2 0.2
? lt 2 4 10

• We can expect a rate reduction 2 from
  tracking
• Extra rejection can come from
• Tighter time gates
• Event-wide inspection

L2
gt
single track
L2 is both an object-ID trigger and a physics
trigger
9

• L2 Impact on L1 Muon-ID
• Smallest granularity 3D tracking, improved
  wire-scint and L-L correl
• More accurate object counting removes
  multiple triggers (stubs) resolves closely spaced
  tracks
• Pt matching to L1 threshold
• Narrower rejection time gate
• Scint-TOF discrimination for Prompt - Slow - Out
  of Time
• Recover crossing octant tracks
• ? dependence of hit patterns

L1 Muon ID
MTC05
non-CFT fwd.
region wide scint match (f wedge)
CFT f wedge
dimuons
MTC10
punch
calib
octant wide scint match (? wedge)
MSP searches
(Pt,?,f)
PDT centroid
pointing
A. Maciel (N.I.U.) Muon-L2
10
Comments for L3
A. Maciel (N.I.U.) Muon-L2

• Muon-subdetector-only ID largely done _at_L2
• Exploit other subdetectors !
• Muon vertex - MTC vertex - min.H vertex is
  the muon vertex compatible with the rest of the
  event ? (multiple interactions)
• Kinematics refitting, dimuon masses, reln to
  other phys. objects
• Back to back tracks - correlate scint times
  (low Pt dimuons)
• Number of hits on track (L2 fixed _at_6) -
  combinatorics(fwd)

T
11
A. Maciel (N.I.U.) Muon-L2
The Level-2 Muon Trigger
ITEP -- LMU(Munich) -- Nevis Labs -- NIU
Arthur Maciel (N.I.U.) For The DØ b-ID Vertical
Review FNAL 04-04-01

• Muon Level-2 Outline
• Level-1 / Level-2 Muon-ID
• Example Algorithm
• Performance

12
A. Maciel (N.I.U.) Muon-L2
SLIC Algorithms
Central A Layer
Ave. time budget 25 µs

• Sweep octant in search for wire triplets
• check that matching z-row has been hit
• check triplet validity, calculate residual
• assign scint time and phi, wire eta
• report stub, continue octant sweep

• Track stub finding (wirescint)
• driven by lookup tables with
• valid wire hit patterns
• wire geometry for track fit
• valid wire-scint correlations

Scint. z-rows
PDT
A 3x4 Search Window
beam line
PDT hit triplet
obs wire-only or scint-only stubs are not being
considered
13
A. Maciel (N.I.U.) Muon-L2
L2 Muon 3D Track Pointing
Valid hit-triplet Patterns are listed on a lookup
table
Table has ? dependence. One A-layer octant holds
40 valid patterns. At each ?, only up to 18max.
are possibly valid.
Track receives a detector(?,f) Track ? is given
by innermost cell Track f is given by Af or
Pixel PDT ?T (triplet average) resolves doubles
sigma 0.09 LSB(?) 0.05
14
A Layer Track Segment
dout
Requiring 3 cells hit (out of max.4)
y
PDT Hit Triplet. Wire Positions are ( z ,
y ) for cells in , mid , out d Signed
Drift Distances Hit Position ( zd , y
)
dmid
din
z
The straight line segment constraint
det zi di zm dm zo do
0 yi ym yo
1 1 1
determines a track stub residual equation for the
drift times, c o d Offset
Res(dim L) where
A. Maciel (N.I.U.) Muon-L2
15
A. Maciel (N.I.U.) Muon-L2
Track Stub Residual Equation c o d
Offset Res(dim L)
Offsets are in 1 1 correspondence with
Triplet patterns
Offset 1 zo zm zi
yo yi yo ym yi
1 1 1
Cout sign x (ym yi) / (yo yi) Cin
sign x (yo ym) / (yo yi) Cmid sign
x - 1
Coefficients are in 1 1 correspondence with
missing deck
PDT triggering must test residuals over all
valid sign combinations
16
A. Maciel (N.I.U.) Muon-L2
The Level-2 Muon Trigger
ITEP -- LMU(Munich) -- Nevis Labs -- NIU
Arthur Maciel (N.I.U.) For The DØ b-ID Vertical
Review FNAL 04-04-01

• Muon Level-2 Outline
• Level-1 / Level-2 Muon-ID
• Example Algorithm
• Performance

17
A. Maciel (N.I.U.) Muon-L2
DSP Evaluation
DSP Timing

• Trigger code in assembly format
• Download trigger code and tables
• Each A layer DSP holds one octant
• Code and tables are octant specific
• Timing is for single track events
• Backgnd 5 random hits added
• DSP internal memory has 64kB
• Code and table have 20kB each
• DSP management code under study
• SLIC 4 worker DSPs, one manager
• Central-A 8 DSPs in parallel
• Central-BC 8 SLICs in parallel

search all
search one
window sweep
µsec
18
The L2 Muon Object Finding
3x4 Search Window
A-layer PDT
For each search window, find all valid hit
triplets, deck ordered inside out. Each triplet
is tested against trigger table for track segment
and drift residual (here no scintillator match)
A. Maciel (N.I.U.) Muon-L2
19
(No Transcript)
20
A. Maciel (N.I.U.) Muon-L2
Central A-Layer PDT
ROAD 009 - 011 - 013 015 Triplet
009 013 015 Pattern 0,0,1
y
. . .
column
95


15
6
7
14
15
5
4

12
13
28
Four decks
.
.

3
2
. . . 24 columns
10
11
26
1
8


1
89
9
24
Cell 0 Coln 0
Col_in 3
Barrel 1 , 2 , 3
z
ROAD 025 026 - 028 030 Triplet
017 018 022 Pattern -1,-1, 2
A triplet pattern is defined as two cell
projections, and the missing deck number
Nb patterns are triplet shapes, and have Z
translation invariance
21
(No Transcript)
22
(No Transcript)
23
(No Transcript)