Trigger for Run 8

Trigger for Run 8

Rates, Yields, Backgrounds…

Debasish DasPibero Djawotho

Manuel Calderon de la Barca

Analysis MeetingBNL

October 16, 2007

Prospects for in Run 8• p+p

– 30 pb-1 sampled luminosity• Run 6: 9.2 pb-1 sampled, S/B~ 1 : 2.31

• Physics goals: – Improved yield (stat. error can go down by ~2)– pT spectrum?

• Needs Seff~80, or ~450 counts (Seff~40 in |y|<0.5).

• dAu – minbias = 2.2 b.– rare processes:

• = pp x (2·197) – peak: L = 30x1028 cm-2 s-1,

• rate = 660 kHz (pileup!!)– L Delivered: 120 nb-1

– L Sampled:• Fast Detectors : 60 nb-1

• Slow Detectors : 30 nb-1

– Physics goals:• Yield, RdAu pT integrated• Needs 100-200 counts in |y|<0.5

(Final?) Numbers from 2006

• Trig ID 117602 + 137603• S = 175 • B = 405 • Seff = 31.1 Seff = 5.58 • εϒ = 0.094 • dy=2.0 Ldt = 9.23387 pb-1 • BR × dy/dσ = 100.808 pb

– Note: signal above in dy=2 – in dy=1; S=87.5, Seff = 16

Upsilon Estimates for Run 8• What can we get?• BR x d/dy = 91 pb @ 200 GeV.• Efficiency :

• Geometrical Acceptance: 26%• L0 Efficiency : 93%• L2 Efficiency : 86%• Offline electron pair efficiency :

47% – Total =9.4%, use ~10%

• Yields:– pp:

• 30 pb-1 x 91 pb x 10% = 273 ’s

– dAu: • 60 nb-1 x 91 pb x (2·197)0.95

x 10% = 160 ’s

• With how much bandwidth?• pp Rejection

– Run 6: 10 Hz / 550kHz• rejection factor : 55000

– Trigger rate Run 8 pp : 1.5 MHz/55000 = 27 Hz peak

• Au Au Run 7 : 10 Hz / 25kHz (<Nbin>=235)

– rejection factor : 2500• Assumption: rejection decreases linearly

with increasing <Nbin>• dAu <Nbin> = 17

– estimated rejection factor : 51000

– Trigger rate Run 8 dAu : 660 kHz/51000 = 13 Hz peak

Options to decrease rate

• Focus on p+p case (27 Hz is the most bandwidth requested!)– Reductions in p+p can be carried over to d+Au case.

• Raising Cluster Energy of High-energy electron– Advantage:

• Rate goes down quickly with higher E.– Disadvantage:

• So does the efficiency…• Hits low-pt hardest.

• Cutting tighter on opening angle– Advantage: efficiency does not drop dramatically– Disadvantage:

• Rate does not decrease dramatically either• Hits high-pt hardest

• Vertex cuts?– Almost the same as a prescale.– Some Differences:

• For previous runs, tight vertex could help with Brems. of electrons• With low material in 2008, Brems. not so much of an issue.• It can still help with analysis offline, tracks will be close to center, many hits.

Upsilon L2 parameters

Algorithm Parameter Set I Set II Set III

i0 L0 ADC threshold ? ? ?

i1 L2 clustering seed threshold 150 150 200

i2 CTB macthing 0 0 0

i3 Use ZDC vertex z information 0 0 0

i4 Tower per EMC cluster 3 3 3

f0 Higher energy electron threshold 4 4.5 5

f1 Lower energy electron threshold 2.5 3 3.5

f2 Minimum invariant mass 6 6.5 7

f3 Maximum invariant mass 15 15 15

f4 Maximum cos(theta) 0.5 0.25 0

Raising “f0” (E of leading-e)

• For 4<Et<6, Rejection increases decreases exponentially• Slope independent of multiplicity• Should also hold true for E of high energy cluster at L2

Estimate Signal Reduction

• Plot is for Et of tower at L0• Cluster energy at L2 should

show similar dependence.• Take value at 4.5 GeV to

be baseline (~ threshold used in Run VI and Run VII)

• Calculate reduction in signal relative to 4.5 GeV.

• Calculate Seff = S/(2*(B/S)+1) to compare.

Estimate Reduction in Rate

• Rejection depends exponentially with Et– See blue lines

• Slope is independent of multiplicity– Minbias & Central Au+Au have same

slope• Use fit to rejection to estimate rate

reduction• Caveat:

– Rate reduction does not equal offline background.

– Rate: photons and electron combinations

– Offline background: electron combinations

– Assume offline background also decreases, but with half the slope.

E threshold, Rate Reduction

E Eff. Rej. Rate Red.

Rate S red. B red. Seff Seff

4.5 .78 23 100% 27 Hz 100% 100% 48.5 6.96

5.0 .63 51 45% 12 Hz 80% 71% 42.7 6.5

5.5 .45 103 22% 6 Hz 57% 50% 30.7 5.5

6.0 .31 151 15% 4 Hz 39% 35% 20.6 4.5

pt vs Cos()

• With good cos() determination, at cos(q)<0.4 we only lose pt>6 GeV.

• Issue is resolution.

From 2006 Offline data

• Nominal cut has been at cos()=0

• Seff drops linearly with cos()

– How much tighter can we live with?

cos(), Rate Reduction

cos() S B Rate Red.

Rate Seff

0 37 31 100% 27 Hz 6.96

-0.2 35 29 94% 25.4 Hz 6.5

-0.3 31 27 85% 23 Hz 6.3

-0.4 33 (??) 22 81% 21.9 Hz 6.2 (??)

-0.5 24 21 66% 17.8 Hz 5.5

Other possibilities?

• Vertex cut?– VPD needed, available in run 8?– Run 8 is low material, so no issue with

Bremsstrahlung– Slightly preferable to prescale,

• EMC design has projective geometry to z=0– Better E resolution for single tower.

– Caveat: normalization needs a mb trigger (prescaled) with identical vertex cut.

Conclusions

• Raising E threshold on high energy cluster.– f0=5.5 GeV,

• rate goes to 12 Hz Seff~6.5, Seff~42.7 in |y|<0.5

• Can get enough counts for pt spectrum?

– More reduction is not desirable because impacts low pT.

• cos()~-0.3– Additional reduction possible (~10-20% rate reduction) – Impacts high-pT part.

• Run 8 trigger can work with slightly tighter cuts.