1. Trigger  Detectors at 420m can be included in the HLT
 Would need to increase L1 latency for inclusion in L1, which might be possible in special runs
(use post-L1Accept event buffers as pre-L1Accept storage)
For standard CMS running triggering with forward detectors has been studied in great detail,
see trigger appendix in CMS PTDR-II
CMS trigger menus in PTDR-II foresee 1% of the bandwidth on L1 and HLT for a
dedicated forward detectors trigger stream
H(140GeV) -> WW:
Can trigger ~20% of events via the standard L1 muon trigger
H(120GeV) -> bb:
Can trigger 10% of the events via the standard L1 muon trigger
Can gain an additional 10% from the jet trigger when combining jet trigger and tag at 220m,
because can lower dijet ET thresholds substantially, to about 40GeV per jet, for bandwidth
limit of 1kHz
In order to retain this 10% gain from the jet trigger on L1 and stay within bandwidth limit of 1Hz,
need FP420 information on HLT
Monika Grothe, Prospects for diff and fwd physics at the LHC, March 2007

2. Pile-up background
TOTEM
xL=P’/Pbeam= 1-x
d(epeXp)/dxL [nb]
Number of PU events with protons
within acceptance of near-beam
detectors on either side:
~2 % with 420m
~6 % with 220m
Translates into a probability of obtaining a
fake DPE signature caused by protons from PU:
Eg at 2x 1033 cm-2s-1 10% probability for a fake DPE signature for any background source that
looks identical to the signal in the central CMS detector only.
This is independent of the type of signal.
Depends critically on the leading proton spectrum at the LHC which in turn depends on size of soft rescattering effects (rapidity gap survival factor) !
Monika Grothe, Prospects for diff and fwd physics at the LHC, March 2007

3. Pile-up background ; 1 2 s = M2 Can be reduced by:
Requiring correlation between
ξ, M measured in the central detector and
ξ, M measured by the near-beam detectors
Fast timing detectors that can determine
whether the protons seen in the near-beam
detector came from the same vertex as the
hard scatter within 3mm
Condition that no second vertex be
found within 3mm vertex window
left open by fast timing detectors
Additional cuts exploiting difference in
multiplicity between diff signal and non-diff
background
; 1 2 s = M2
CEP H(120) bb
incl QCD di-jets + PU
(jets)
(p tagger)
CEP of H(120 GeV) → b bbar and
H(140 GeV) → WW:
S/B of unity for a SM Higgs
M(2-jets)/M(p’s)
Monika Grothe, Prospects for diff and fwd physics at the LHC, March 2007

4. Reserve

5. Trigger 2 x 1033 cm-2 s-1 L1 jet trigger rates Efficiency pp  p jj X
no fwd
detectors
condition
pp  p jj X
2-jet trigger
Efficiency
Events per pb-1
single-arm
220m
L1 output
bandwidth:
100 kHz
single-arm
420m
L1 trigger threshold [GeV]
Attention: Gap survival probability not taken into
account; normalized to number of events with
0.001 <  < 0.2 and with jets with pT>10GeV
At 2x 1033 cm-1 s-1 without any additional condition on fwd detectors:
L1 1-jet trigger threshold O(150 GeV)
L1 2-jet trigger threshold O(100 GeV)

6. Trigger (II)
→ CMS trigger thresholds for nominal LHC running too high for diffractive events
→ Use information of forward detectors to lower in particular CMS jet trigger thresholds
→ The CMS trigger menus now foresee a dedicated diffractive trigger stream
with 1% of the total bandwidth on L1 and HLT (1 kHz and 1 Hz)
single-sided
220m condition
without and with
cut on 
Achievable total reduction: 10 (single-sided 220m) x 2 (jet iso) x 2 (2 jets same hemisphere as p) = 40
Adding L1 conditions on the near-beam detectors provides a rate reduction sufficient to lower the 2-jet threshold to 40 GeV per jet while still meeting the CMS L1 bandwidth limits for luminosities up to 2x 1033 cm-1 s-1
Much less of a problem is triggering with muons, where L1 threshold for 2-muons is 3 GeV
Monika Grothe, Prospects for diff and fwd physics at the LHC, March 2007

7. Trigger (III) Central exclusive production
pp pHp with H (120GeV)  bb
Trigger is a major limiting factor !
Efficiency
H(120 GeV) → b bbar
L1 trigger threshold [GeV]
Level-1:
2-jets (ET>40GeV, ||<3) & single-sided 220m condition results in efficiency ~12% (L1)
HLT:
Efficiency of above 2-jet condition ~7%
To stay within 1 Hz output rate, needs to either prescale or add 420 m detectors in trigger
 Can add another ~10% efficiency by introducing a 1 jet & 1 (40GeV, 3GeV) trigger condition
Monika Grothe, Prospects for diff and fwd physics at the LHC, March 2007

8. Pushing the limits even harder: Squeezing more latency out of L1 (II)
Tracker memory buffer partitioning
upon L1A
cell 1
Revolving part of 128 BX depth Storage part can hold
8 L1A à 3 buffer cells each
129
192
...
pre-L1A storage
post-L1A buffer
“Long latency” mode: Only option to include RPs >220m in CMS L1
Uses post-L1A event buffers for pre-L1A storage
For trigger rate < 10 kHz, L1 latency 4.8 ms (189 BX)
Resulting trigger rule: Each L1A separated by at least 8.75 ms because
events has to be read out from Tracker buffer with 8.75 ms latency
Total deadtime cost at 10 KHz L1A rate of the order 10%
Monika Grothe, Prospects for diff and fwd physics at the LHC, March 2007

9. Pile-up
Monika Grothe, Prospects for diff and fwd physics at the LHC, March 2007