1. Alex Barbieri, Christof Roland, Ivan Cali, Bolek Wyslouch, Gunther Roland (MIT) Krisztián Krajczár, Yen-Jie Lee (CERN) Matthew Nguyen (LLR) Wei Li (Rice) CMS Upgrade Project Office 9 th August 2013 Stage-1 L1 calorimeter trigger upgrade for Heavy-Ion physics August 9, 2013 Upgrade Project Office Meeting1

2. August 9, 2013 Upgrade Project Office Meeting2 The case… The readout rate of the CMS detector in HI collisions is limited by the Pixels and the Tracker to about 3kHz No zero suppression in strips and pixel buffer overruns for large events 2011: Max. Interaction Rate ~4.5kHz  2.7kHz L1A rate The current L1 Calo trigger system is (barely) selective enough for HI run in 2011 For jets and photons about 50% of the hadronic interactions need to be accepted due to the large underlying PbPb event 2015: expect at least a factor of 4-6 lumi increase – There is strong overlap between the triggers reducing the L1A rate by 2 will require prescaling individual paths by factors of 10-20 – Need a factor of ~20 rejection factor for 2015 The Stage-1 L1 Calorimeter Trigger upgrade (2015) can achieve the desired rejection factor

3. August 9, 2013 Upgrade Project Office Meeting3 Trigger rate L1 AlgoL1 Accept Fraction HLT Accept Fraction 4.5 kHz (2011) 40 kHz (2015) 100kHz After LS2 SingleJet3635%0.5%(Jet65)1600Hz14 kHz36kHz SingleJet5228%0.08%(Jet80)1300Hz12 kHz29kHz SingleJet6824%1100Hz10kHz24kHz SingleJet9220%900Hz8.0kHz20kHz SingleJet12814%600Hz5.3kHz13kHz SingleEG522%0.3%(Photon20)1000Hz8.9kHz22kHz SingleEG88%350Hz3.1kHz7.8kHz SingleEG121%0.02%(Photon40)50Hz0.4kHz1.1kHz SingleEG150.6%26Hz0.2kHz0.6kHz DoubleMuOpenHQ1%1%40 Hz0.4kHz0.9kHz SingleMu31%50 Hz0.4kHz1.1kHz ETT10037%0.4%(Track14)1700Hz15kHz38kHz ETT14034%1550Hz15kHz34kHz ETT22031%1400Hz12kHz31kHz ETT80018%840Hz7.5kHz19kHz ETT200014%630Hz5.6kHz14kHz Total L1A60%5%2700Hz24kHz60kHz Rate can be controlled by the threshold. To be optimized Rate does not respond much to the threshold. Need change in algorithm Energy sum seeds for the single track trigger should be replaced by updated jet triggers Note: There is strong overlap between the triggers, reducing the L1A rate by 2 will require prescaling individual paths by factors of 10-20

4. August 9, 2013 Upgrade Project Office Meeting4 Background Subtraction Algorithm HLT/Offline background subtraction: - Process phi rings at const. eta - Calculate average and subtract - Jet finder runs after BG subtraction Current L1 Jet Finder: - Processes eta strips at const. phi - 2 x 11 sectors - 1 sector = 4x4 calo towers - Sliding window jet finder Region Energy Distributions PbPb central event Current strategy: Jet Finder at L1 and Jet Background subtraction at HLT The high non-uniformity in η of HI events does not permit a useful BG subtraction within a single 2x11 sector Access to the full eta phi map at L1 allows for a efficient underlying event subtraction (phi-rings) Peripheral and central events respond consistently to the thresholds applied to L1 jets

5. HI Stage-1 calo upgrade strategy Make decision in a central place !!! Trigger “primitives” are available in 18 separate 9U VME crates with custom high- speed backplanes corresponding to φ slices Insert a processing/communication board: optical Regional Summary Card (oRSC) into existing slot in each RCT crate Send regional calorimeter trigger products to L2 calorimeter trigger processor (MP7) boards Program FPGA to do background subtraction in full φ rings using 4x4 tower “regions” to estimate background Find jets at L1 speeds August 9, 2013 Upgrade Project Office Meeting5

6. August 9, 2013 Upgrade Project Office Meeting6 Stage-1 HI algorithm performance Accept rate can be controlled by L1 threshold using Stage-1 and Stage-2 system Stage-1 System: Reasonable trigger turn-on curve for both central and peripheral collisions L1 accept rate reduced by a factor of 10- 20 Stage-1 System Current System

7. Heavy-Ion Contribution August 9, 2013 Upgrade Project Office Meeting7 Manpower: MIT provides 1 postdoc and 1 student Rice University provides 1 postdoc and 1 student (starting in September 2013) The HI group responsibility (in collaboration with WU and IC): Establishment of the detailed triggering specifications based on the requirements of heavy ion physics Testing of the boards and firmware at the trigger demonstrator at CERN 904 Development of FPGA firmware for MP7 boards with specific heavy-ion triggering algorithms Installation and commissioning of the new trigger in CMS experiment. Purchase and testing of the 3 prototype oRSC boards Purchase and testing of the final 22 oRSC boards Purchase of optical fibers and patch panel All the existing L1 calorimeter trigger electronics has been designed and produced by the University of Wisconsin and Imperial College (oRSC, CTP6/7, MP7)

8. Status and schedule Contributing to the setup of the test system/demonstrator in CERN 904. The system includes 2 RCT crates and 1 MCC crate XDAQ application to allow easy control and pattern test of trigger electronics (JCC, JSC, oRSC) Gained hand on experience with the RCT system and with oRSC and MP7 Specific HI algorithm implementation schedule: September 2013: Defined requirements for HI running October 2013: Algorithms performance studies offline completed (it will include also the test of the existing stage-1 pp algorithm) February 2014: Algorithm implemented in MP7 June 2014: Basic performance tests completed Fall 2014: System ready for data-taking MIT and Rice groups will collaborate with the Wisconsin and Imperial College groups for the installation and commissioning of the system in the CERN/P5 CMS experimental hall. August 9, 2013 Upgrade Project Office Meeting8

9. August 9, 2013 Upgrade Project Office Meeting9 Summary The upgraded Stage-1 system Significantly improve the online jet trigger Sufficient for data taking with heavy-ion collisions in 2015 Achieve similar performance for calo-jets as the HL-LHC system Active collaboration with Wisconsin and Imperial college group already started XDAQ application for 904 tests developed Hand on experience with the electronics acquired (RCT, oRSC, MP7) Definition of HI algorithm for the 2015 data-taking is ongoing Schedule/milestones for algorithm implementation is defined A first L1 jet UE background subtraction algorithm for HI already tested offline

10. August 9, 2013 Upgrade Project Office Meeting10 Backup slides

11. August 9, 2013 Upgrade Project Office Meeting11 Track trigger: trigger turn-on curves It is feasible to seed high p T track with L1 jet trigger Provides a factor of 5-10 reduction of the L1 accept rate “Peripheral event” 60-100% “Central event” 0-20% For Approval

12. August 9, 2013 Upgrade Project Office Meeting12 The high non-uniformity in η does not permit a useful BG subtraction within a single 2x11 sector Current L1: Sector wise subtraction BeforeAfter

13. August 9, 2013 Upgrade Project Office Meeting13 Test on 3 data samples Min. Bias, Jets and central events Can’t apply a threshold that keeps full efficiency for jets while rejecting a sufficient fraction of min bias events Current L1: Sector wise subtraction BeforeAfter

14. August 9, 2013 Upgrade Project Office Meeting14 Trigger Turn On: 2015 L1 system Jet + Central UE Jet + Peripheral UE Access to the full eta phi map allows for a efficient underlying event subtraction (phi-rings) – Peripheral and central events respond consistently to the thresholds applied to L1 Jets

15. August 9, 2013 Upgrade Project Office Meeting15 Trigger Turn On: Current L1 system Jet + Central UE Jet + Peripheral UE Lack of UE subtraction does not allow for a consistent threshold for central and peripheral events Poor control over L1 accept rates High Threshold Low Threshold

16. August 9, 2013 Upgrade Project Office Meeting16 Physics performance plots b jet quenching performance – Assuming the same amount of quenching as light jet 3-jet event performance – R 32 (3-jet / 2-jet ratio), access to gluon jet High p T track R AA and v2 performance – Increased the p T reach from 100 to 160 GeV/c

17. August 9, 2013 Upgrade Project Office Meeting17 b-jet physics performance Goal: di-b-jet asymmetry as done for inclusive jets in HIN-10-004 and HIN-11-013 Proposed observable: Dijet asymmetry (A J ) & R B (fraction of balanced di-b-jet) Expect similar systematics as light jets + (b tagging uncertainty & light jet contamination) Use 2011 kinematic cuts: p T,1 > 100 GeV/c and p T,2 > 30 GeV/c PAS

18. August 9, 2013 Upgrade Project Office Meeting18 Physics performance of 3-jet events Access to gluon jets: three jet events R 32 may be modified due to jet quenching Similar study as QCD-10-012 All jet p T threshold > 100 GeV/c No existing experimental measurements in heavy ion collision Simulated with PYTHIA at 5.5 TeV PAS

19. August 9, 2013 Upgrade Project Office Meeting19 High p T reach of tracks and jets High p T track High p T Jet (Anti k T R =0.3) Entries Track Jet p T (GeV/c) PAS HIG-12-054 Approved

20. Stage-1 Calo Upgrade system August 9, 2013 Upgrade Project Office Meeting20