Update on Trigger simulation Spasimir Balev /CERN/ 13.07.2011

Introduction 13/7/2011 The rate with ideal CHOD_Q0 signal was 200 kHz

What we learned: The “default” rate with ideal CHOD_Q0: 200 kHz The “default” rate with realistic CHOD_Q0: 500 kHz  We can no longer play with simple kinematics signals Taking into account the interactions in the trigger simulation is important All but ~60 kHz of this (500 kHz) rate is junk  decays happening after Z = 165 m. Realistic and not so realistic solutions: put SAC and IRC or STRAW at L0  not feasible; use RICH reconstruction at L0  a drastic improvement which will allow fairly simple cuts in CHOD; play with CHOD multiplicity  complicated signals; interplay between signal and junk triggers losses; requires well established analysis procedure  getting in shape, but still key ingredients missing… We can not survive without CHOD: the output L0 rate becomes ~1 MHz, with significant halo contribution (mainly from p+m+ upstream GTK3) The purpose of this talk: to get an idea of optimal CHOD design in terms of trigger rates quantify the effect on L0 output rate with different signals 13/7/2011

CHOD acceptances p+ from pnn @CHOD after all cuts NA48 CHOD Not in |X|,|Y|<130 mm |X|,|Y|<1210 mm + octragon Analysis CHOD 150 < R < 1080 mm Trigger CHOD R(350 mm,0) < 1080 mm Not in |Y|<130 mm and -130 < X < 400 mmm The trigger CHOD cut costs only 0.8% of the signal 13/7/2011

CHOD pseudo-digitization At the moment, no digitization is available in NA62FW for CHOD In order to check the rates, we need to define signals Kinematic signal  count every charged particle from K+ decay, or halo muon Realistic signal  see below The energy deposits are available for each single hit in NA62MC. One charged particle can produce many “hits” (single energy deposit), which are stored separately with their coordinates, energy, time. This is useful to study any possible design (given that CHOD consists of two planes with 2 cm thick vinyltoluene) Recipe for realistic signal construction: add each energy deposit within a single channel; define a signal from this channel if the total energy deposit is Etot>2 MeV. 13/7/2011

Example for the NA48 design Edep, MeV Edep, MeV One pp0 event: p+ and one g reach LKr the other g converts at the end of the RICH See the backsplashes from LKr: y,mm thit – t0, ns Edep, MeV Edep, MeV x,mm 13/7/2011

The backsplashes Complicated time profile of CHOD hits thit – t0, ns Complicated time profile of CHOD hits Backsplashes from LKr Backsplashes from IRC CHOD H and V planes Ehit, MeV 13/7/2011 single charged track pair?

Rates calculation Two designs studied: Two types of rates studied: the NA48 CHOD with 2 planes of vertical and horizontal slabs (64 + 64) with varying size: 6.5 cm width in the central area and 9.9 cm in the peripheral; tiles design: 6.5x6.5 cm2 in the central area; 6.5x9.9 and 9.9x9.9 cm2 in the peripheral. Two types of rates studied: kinematical rate (for comparison purposes) real rate (taking into account the interactions) All the components simulated: K+ decays downstream GTK3 (6 main decay modes) p+ beam (with pm decays enabled) after GTK3 halo and non-halo muons from K± and p± decays upstream GTK3 Detailed rates by components given at MUV3/CHOD meeting; here – only the summary 13/7/2011

Rates - Summary Kinematics signals Real signals kHz kHz kHz kHz 13/7/2011

CHOD optimal design yCHOD, mm xCHOD, mm 13/7/2011

Signal acceptance The generated quantities used here (reconstructed in progress) Total number of generated pnn events: ~105 Total number in 105 < Z < 165 m range: 79 599 15 < P < 35 GeV 37.5% No pm decay 34.7% 15 < RCHOD < 108 cm 29.6% !LAV, RICH, LKR, !IRC, !SAC acc. 28.9% MM2 cut 16.6% 13/7/2011

p+ interactions Rint, mm Zint, m 8.1% of selected events are with p+ interacting before reaching CHOD 13/7/2011

p+ interaction example LKR display CHOD display y, mm Ecell, MeV Ehit, MeV x, mm The signal event mimics pp0 decay. What the reconstruction will see? 13/7/2011 16 crossing points

A glimpse to the trigger rates Default signals: CHOD * !MUV3 * RICH5 * !RICH40 * !LKR * !LAV12 CHOD signal Rate of signal [kHz] L0 rate Signal loss [%] Eout<500 MeV CHOD_ACC_Q0 9916 212 3.6% CHOD_ACC_TR_Q0 7928 166 4.5% CHOD_Q0 11580 489 4.6% 1.9% CHOD_Q1 9309 258 10.2% 4.9% CHOD_Q0 * !CHOD_QX 9909 314 6.5% 2.5% CHOD_TR_Q0 9493 402 5.7% 3.1% CHOD_TR_Q1 7692 214 10.0% 5.0% CHOD_TR_Q0 * !CHOD_TR_QX 8212 264 7.1% 3.5% CHOD_X14 10018 286 9.1% 2.9% CHOD_X16 10337 315 8.2% CHOD_NSLABS_10 11496 390 5.8% 1.2% CHOD_NSLABS_8 10620 324 8.7% 2.2% Ideal Real NA48 acc. Real trigger acc. Multiplicity cuts 0  at least 1 quadrant; 1  exactly one quadrant; X  two opposite quadrants X14(X16)  number of crossing points between 1 and 4(6) For signal loss: all cuts applied for the normalization Eout < 500 MeV  energy in LKr outside a circle with 20 cm radius around the p+ (cell threshold 100 MeV) 13/7/2011

Put CHOD time in the game Italo’s idea [to be presented at MUV3/CHOD meeting]: use two planes with horizontal tiles with the same geometry; in the trigger use the time coincidence of the single tile from the first plane and the corresponding tile in the second plane, or the neighbouring ones; suppress the rate from backsplashes or interacting photons. Try simple implementation of the idea with the existing tools: segment CHOD as in the 32 x 32 “tiles” design (much finer segmentation than what we will have  only to test the effect in extreme case); for each tile count the minimum time of hits above the threshold (1 MeV); no resolution smearing applied  times ideally computed by G4 construct a signal from a given pair of tiles from 1st and 2nd planes (including the neighbouring ones) if Dt < 5 ns ~1.5% of the selected pnn are without a signal (8.5% if the 2 neighbouring tiles in the second plane are ignored). Multiplicity signals defined: Nt  total number of tile pairs Nq  total number of quadrant with tile pairs giving a signal 13/7/2011

Trigger rates Default signals: CHOD * !MUV3 * RICH5 * !RICH40 * !LKR * !LAV12 CHOD signal Rate of signal [kHz] L0 rate Signal loss [%] Eout<500 MeV 0 < Nt < 3 7812 222 6.9% 2.1% 0 < Nt < 4 8133 241 6.0% 2.0% 0 < Nt < 5 8267 259 5.4% Nq = 1 7400 202 6.4% 2.4% Significantly better separation for signal and background, leading to acceptable rate. However, this is a much finer segmentation than what we will have  to be investigated. No effect from 5 ns  2 ns time difference between the two planes Strong dependency on the Energy threshold due to the way the process of energy loss is simulated. 13/7/2011

CHOD multiplicity rates [summary] 13/7/2011

Summary The design of the new CHOD could be optimized for efficient triggering purpouses  a suitable segmentation can allow to exclude areas, not populated by the signal at high intensities. With more complicated CHOD signals it is possible to keep the L0 output rate at 200-300 kHz for the main trigger, without RICH. It is still mainly junk, however… This is (almost) the maximum with the tools available at the moment. To progress further in trigger simulation it is needed: Overlay MC reasonable LKr reconstruction CHOD digitization and reconstruction multi-ring RICH reconstruction MUV1/2/3 reconstruction 13/7/2011

S P A R E S 13/7/2011

CHOD studies for pnn 13242 events in analysis cuts: 1.4% (182 events) with 0 crossings 92.2% (12208 events) with 1-4 crossings (80% with 1 crossing point) 6.4% (852 events) with more than 4 crossings 9% (1231 events) with Eout > 500 MeV 1083 (8.2%) with interactions before CHOD 0.5% (61 events) with 0 crossings 2.5% (329 events) with 1-4 crossings 5.2% (693 events) with more than 4 crossings: 0.5% (63 events) are with Eout < 500 MeV 4.7% (630 events) are with Eout > 500 MeV 6.3% (863 events) with Eout > 500 MeV Eout is the total LKr energy outside 20 cm around the impact point of p+ if Ecell > 100 MeV 12159 (91.8%) w/o interactions before CHOD 0.9% (121 events) with 0 crossings 89.5% (11879 events) with 1-4 crossings 1.2% (159 events) with more than 4 crossings 2.8% (368 events) with Eout > 500 MeV Using 1-4 crossing points or Eout > 500 MeV cut, the signal acceptance goes to 15% (from 16.6%) 18% of the good events are with more than 1 crossing point To be compared with 30% obtained by Gianluca on 2007 data (1 track + 1 cluster) 13/7/2011