Available on the CERN CDS information server CMS PAS QCD-07-002 CMS Physics Analysis Summary Contact: cms-pag-conveners-qcd@cern.ch 2012/02/10 Zero bias and HF-based minimum bias triggering for pp collisions at 14 TeV in CMS The CMS Collaboration Abstract The analysis of the underlying event structure in pp collisions begins with an efficient and minimally biased data sample. This document describes the feasibility of obtaining zero bias data and presents an approach to triggering minimum bias collisions with the CMS detector. Triggering such collisions that often have just a handful of particles is difficult and can lead to severe biases on the data, thus the possibility to obtain zero bias data is evaluated for various beam bunch patterns and luminosity regions. For regions where zero bias triggering is not feasible, the merits of utilising the forward hadronic calorimeter (HF) as a minimum bias trigger is discussed and fully simulated. We find that using HF in a single-side configuration, a hard core (non-diffractive) efficiency of ∼80% can be attained, whilst retaining some sensitivity to diffractive type collisions. Estimated rates and suggested prescale factors are presented for each beam-bunch pattern configuration. The merits of zero-bias and HF-triggered minimum bias are also discussed. 1 1 Introduction Measurements of the underlying event structure, such as the pseudorapidity distributions and transverse-momentum spectra of charged particles, rely on a physics data sample which is 100% efficient at collecting the data and is free from bias. It is thus of interest to evaluate if it is possible to obtain such “zero bias” data during the various luminosity and beam bunch configurations expected during the LHC startup. For the cases where this is not possible, it is also equally prudent to be prepared with additional detector-based minimum bias triggering possibilities and schemes. The advantage of a detector based trigger is that it will be effective for all luminosities and beam bunch configurations. The disadvantage, of course, is that it is very difficult to obtain a detector triggered data sample that is 100% efficient; as with any triggered data sample some events are not triggerable, whilst other events are sacrificed to decrease background contributions. The aim of the study presented here is to determine regions where zero bias triggering is feasible and for all other regions to explore in detail a detector based minimum bias trigger that maximises the efficiency while still maintaining a low background level. It is proposed that an effective trigger for underlying event data utilises the CMS forward hadronic calorimeters (HF) [1] which are located at 3<|η |<5. The HF position is similar to what was done in other experiments that have measured underlying events in lower energy pp(p) collisions (for example CDF [2], PHOBOS [3]). For triggering, energy deposits from predefined regions of HF are summed to form “trigger towers” allowing a fast read out at Level-1. Two energy thresholds are applied to this trigger to remove effects of detector noise. First, a threshold at the calorimeter tower level of 0.8 GeV (1-σ) is applied to the fibres, before summing towers into the trigger towers. A second energy threshold is applied on each trigger tower with ET > 0.5 GeV (see Tables 2, 3 and 4). This minimum threshold corresponds to 2.3σ above the noise for η = 3.25 (outermost HF ring) and 10.4σ above the noise for η = 4.75 (innermost HF ring). In the case of ET > 1.0 GeV the thresholds are 4.6σ for η = 3.25 and 20.8σ for η = 4.75. To trigger the read out of the detector, the trigger towers (with energy above threshold) are counted (independently for the +η HF and -η HF). If the number of trigger towers struck is above a second (number, not energy) threshold, then the event will be considered for minimum bias triggering. Background triggering may comprise a considerable amount of data for the first run. Effects such as beam gas/halo and triggering noise pose problems which need to be overcome before the trigger will be effective. The balance of rates from background sources to real collision signals must be tuned carefully to minimise bias on the final data sample and maximise the number of true pp collisions written to tape for use in physics analysis. 2 Expected luminosity at the LHC The LHC will provide a steady increase of luminosity with differing beam bunch patterns starting with 43 × 43 (out of a possible 3564 bunch buckets) and a luminosity of ≈ 3.8 × 1029 cm−2 s−1 (see Table 1). This initial configuration will, perhaps, provide a significant fraction of the minimum bias dataset. For this bunch pattern configuration, the number of collisions (per bunch crossing) is expected to be small, O(10−3 ), as shown in Figure 1. As such, triggering with the ZeroBias scheme would only yield a collision every 1000 read-out events. At this time, the detector-based trigger would be more effective (see Figure 2). A more detailed view of the triggering rate can be found in Figure 2. Here, the triggered rate 2 3 Triggering Scheme and expected fraction of “ideal data” (one and only one collision per bunch crossing) are presented. Two regions are identified which would provide optimal detector-based triggering (at the lowest luminosities) and ZeroBias triggering (when the fraction of ideal data exceeds 0.2, i.e. at least 1 in 5 events contains a usable collision). It is immediately apparent that triggering significant amounts of ZeroBias data in the low luminosity region would “waste” the valuable bandwidth for detector read-out 1 . A similar analysis of each bunch configuration can be found in the Appendix. It should be noted that the regions are defined with specific physics analyses in mind, such as charged hadron transverse momentum spectra. Some ZeroBias events are necessary for all luminosities for calibration purposes; for example, to calibrate jet energy backgrounds. Table 1: Table representing possible expectations for beam bunch patterns and associated luminosities during early LHC operations. In the LHC commissioning plan, various combinations of bunch patterns and associated luminosity expectations are broken into “stages”. For the purpose of this study, the bunch pattern and associated beam luminosity are the relevant quantities. Bunch Pattern Luminosity Range [cm−2 s−1 ] 1×1 1027 43 × 43 3.8 × 1029 − 6.1 × 1030 156 × 156 1.1 × 1031 − 1.1 × 1032 936 × 936 2.3 × 1031 − 5.0 × 1032 2808 × 2808 1.7 × 1032 − 1.0 × 1034 It is possible to turn these expected rates into prescale factors to limit the Level-1 read-out rate to 10 Hz, see Figure 3. Accounting for the empty and multiple collisions per read-out, the lower panel of Figure 3 shows the expected rate of “ideal data” which will be written to tape. 3 Triggering Scheme Three specific Level-1 minimum bias triggers are proposed in order to provide maximum flexibility in obtaining the highest quality of minimum bias data under a wide range of possible backgrounds, beam conditions, luminosities and bunch pattern configurations. • ZeroBias is a beam bunch crossing-time trigger used to obtain a “zero bias” data sample, with 100% efficiency (by definition). Only active beam bunch crossings are to be read out to maximise the probability of obtaining events with valid collisions. • MinbiasHFsingle is the HF-based minimum bias detector trigger which requires at least one HF trigger tower to fire above an energy threshold on either side, i.e. a single-side trigger. This is more efficient than the double-side trigger, below, and depending on the thresholds can accept a reasonable fraction of the diffractive type collisions. This trigger is favourable over the ZeroBias when the luminosity, and thus the rate of collisions to tape, is too low for ZeroBias to be effective. • MinbiasHFdouble is a less efficient HF-based minimum bias detector trigger which requires at least one HF trigger tower to fire above an energy threshold on both sides, 1 This statement assumes that every event should be used for physics analysis. A reasonable amount of empty read-out events could provide useful information for detector noise evaluation, thus not “wasting” bandwidth. Average Number of Collisions per Bunch Crossing 3 Regions of ideal data for different bunch patterns p+p - √s = 14TeV (σ = 79mb) 10 1 Region not useful for ideal data Region for ZeroBias and triggered minimum bias ideal data -1 10 10-2 h 56 1 6x 15 nc bu rn tte pa ern att hp 10-3 nc bu 36 x9 6 3 c un b ern att p h bu nc 8 0 28 ern att hp 8x 0 28 9 CMS preliminary 3 x4 43 Region for triggered minimum bias ideal data 1028 1029 1030 1031 1032 1033 1034 Luminosity (cm-2 s-1) Figure 1: The average number of collisions per beam bunch crossing versus beam luminosity for the four beam bunch schemes that are expected to be used during the LHC commissioning and early running. As shown on the figure, calculations assume a pp collision cross-section of σ = 79 mb. One can identify three general regions of relevance when desiring to trigger on “ideal data” (i.e. data with no pile-up). Regions were determined by requiring that ≥ 20% of accepted events must be ideal data. i.e. a double-side trigger. This trigger is most sensitive to hard-core collisions and has the possibility of rejecting most beam gas/halo collisions that occur within the CMS detector. Each of these triggers will have unique efficiencies, sensitivity to various backgrounds, and imposed biases on the resulting data sets, which must be understood to enable the reconstruction of, for example, the inelastic charged particle multiplicity. For this reason, a mixture of the above triggers will comprise the minimum bias data sets. This mixture will enable studies of bias for use in various physics analyses. 4 Estimated Efficiency The efficiencies of the single and √ double HF-based minimum bias triggers were evaluated using simulations of pp collisions at s = 14 TeV from a sample of PYTHIA [4] events with the DWT tune. The results for different HF trigger tower number requirements and energy thresholds are given in Tables 2 (single-side) and 3 (double-side). The corresponding sensitivity of each combination to the current expected level of noise in the HF is given in Table 4. 4 4 Estimated Efficiency Triggered data rates and fraction of ideal data Triggered Rate (Hz) 43x43 beam-bunch pattern 107 p+p - √s = 14TeV (σ = 79mb) 106 ZeroBias trigger 105 CMS preliminary 104 3 10 Fraction of ideal data AllBucket trigger 1.0 m Mini ias um b er trigg 28 10 29 10 30 10 31 10 Minimum bias trigger 0.8 0.6 Region for triggered minimum bias ideal data 0.4 0.2 Region for ZeroBias and triggered minimum bias ideal data ZeroBias trigger 0.0 1028 1029 1030 1031 Luminosity (cm-2 s-1) Figure 2: Triggering rate and fraction of ideal data for the 43 × 43 beam bunch pattern. Top panel is the average triggered rate using a beam crossing-time trigger (dashed line), which would yield “zero bias” data, and using a detector-based minimum bias trigger (solid line). Bottom panel is the fraction of triggered data that would be usable as ideal (i.e. no pile-up) minimum bias data in physics analyses. The detector triggered lines (solid) assume a 100% efficient detector trigger that samples the full 79 mb cross-section assumed for 14 TeV pp collisions. The lowest possible thresholds in both the single and double-side cases yield good efficiencies to all “types” of pp collisions (minimum bias, hard-core, single diffractive, and double diffractive). Unfortunately, for the currently expected noise levels in HF, the minimal triggering energy threshold of ET ≥ 0.5 GeV (compressed scale EC T ≥ 1) would allow a high rate of noise triggers and significantly dilute the effectiveness of the minimum bias trigger in the low luminosity regions where zero bias triggering is not feasible. A detailed study of efficiency, the expected noise levels, and the loss of ideal minimum bias data to same-bunch pile-up (also known as in-time pile-up) indicates that requiring one HF trigger tower to fire (on either or both sides) above a default triggering transverse energy threshold of 1.0 GeV (compressed scale EC T ≥ 2) is likely the most reasonable compromise. This transverse energy threshold setting is equivalent to a total energy threshold of 10 GeV at η = 3 and 80 GeV at η = 5. Using these default settings (Ntt =1, ET ≥ 1.0 GeV) results in a hard-core triggering efficiency 5 Ideal Rate (Hz) Prescale Factor (for 10Hz rate) Prescale Factor To Achieve 10 Hz Trigger Rate p+p - √s = 14TeV (σ = 79mb) 106 CMS preliminary 105 104 103 102 10 8 6 4 2 0 43×43 156×156 936×936 2808×2808 triggered minimum ZeroBias bias 1028 1029 1030 1031 1032 1033 1034 1028 1029 1030 1031 1032 1033 1034 Luminosity (cm-2 s-1) Figure 3: Upper panel: required prescale factor to achieve a data rate-to-tape of 10 Hz for each bunch pattern at the LHC, versus luminosity, shown for triggering by zero bias (dashed lines) and (100% efficient) minimum bias collision (solid lines). Lower panel: corresponding rate of ideal data expected to be recorded to tape. of ∼80.8% for the single-side trigger (Table 2). In this case, the diffractive type collisions are also recorded at an efficiency level of ∼15%, although this trigger would be more susceptible to beam gas/halo collisions. The double-side trigger (Table 3), with the default settings, has a hard core efficiency of ∼47.5%. Unfortunately, the efficiency for recording diffractive collisions is negligible (∼ 0.6%) in this case as the symmetric triggering coupled with the high energy thresholds largely removes the predominantly asymmetric diffractive events. Studies of the triggering efficiencies using the EPOS [5] and QGSII [6] event generators indicate that the efficiencies obtained using PYTHIA are a lower limit. Increasing the energy thresholds and/or requiring a higher number of trigger towers hit, results in an increased loss of events, predominantly diffractive and low multiplicity collisions. Decreasing the thresholds allows a larger fraction of diffractive collisions to be recorded, but at the expense of allowing more noise triggers (see Table 4). 6 5 Conclusion Table 2: Single-side HF trigger efficiencies. The number of trigger towers required to trigger is shown in the first column. For example, the first row requires at least one trigger tower on either the positive or negative HF. The compressed and the actual trigger tower energy thresholds are noted in the second and third columns, respectively. In the remaining columns the efficiencies are broken down into collision type. Number Required 1 HF Trigger Towers Energy Threshold Compressed Actual (GeV) (EC ≥ ) (ET >) T 1 0.5 Collision Type Efficiency (%) Minimum Hard Single Double bias core diffractive 91.9 98.9 74.9 76.8 1 2 1.0 61.1 80.8 15.0 15.3 1 2 2 2 3 1 2 3 1.5 0.5 1.0 1.5 41.2 78.7 40.8 23.6 57.4 93.9 57.6 33.7 3.1 41.6 1.5 0.1 3.4 44.8 1.3 0.0 Table 3: Double-side HF trigger efficiencies. The columns are as for Table 2, except that the meaning of the first column has changed to be the number of trigger towers required to fire on each of the positive and negative HF. Number Required 1 5 HF Trigger Towers Energy Threshold Compressed Actual (GeV) C (ET ≥) (ET >) 1 0.5 Collision Type Efficiency (%) Minimum Hard Single Double bias core diffractive 71.5 88.7 31.3 30.9 1 2 1.0 33.5 47.5 0.6 0.7 1 2 2 2 3 1 2 3 1.5 0.5 1.0 1.5 18.7 53.3 20.3 10.7 26.7 73.0 29.0 15.2 0.0 7.8 0.0 0.0 0.0 5.9 0.0 0.0 Conclusion Studies of minimum bias triggering for pp collisions at 14 TeV indicate that there are regions of luminosity and beam bunch patterns that will require a detector-based minimum bias trigger as well as regions where a zero bias trigger could additionally be employed. The CMS forward hadronic calorimeters (HF) are detectors that can be utilised to provide a reasonably efficient trigger for minimum bias collisions, while also providing flexibility to handle possible high backgrounds from beam bas/halo events. Recommended default HF-based minimum bias triggers are presented with corresponding hard core efficiencies of ∼80% for a single-side trigger and ∼ 47% for a double-side trigger, which provides additional rejection capabilities to beam gas/halo backgrounds. Sensitivity to diffractive collision processes with the default threshold settings exists for the single-side trigger with an efficiency of ∼15%. Higher efficiencies are possible, depending on the noise levels in HF. 7 Table 4: Single- and double-side HF noise-only trigger efficiencies (Ncoll = 0) compared to the single collision minimum bias efficiency (Ncoll = 1). Efficiencies for diffractive and nondiffractive collision types can be found in Tables 2 and 3. Number Required 1 HF Trigger Towers Energy Threshold Compressed Actual (GeV) C (ET ≥) (ET >) 1 0.5 Acceptance Efficiency (%) Single-side Double-side Ncoll = 0 Ncoll = 1 Ncoll = 0 Ncoll = 1 50.2 91.9 15.0 71.5 1 2 1.0 0.04 61.1 0.00 33.5 1 2 2 2 3 1 2 3 1.5 0.5 1.0 1.5 0.00 16.1 0.00 0.00 42.1 78.7 40.8 23.6 0.00 1.52 0.00 0.00 18.7 53.3 20.3 10.7 References [1] CMS Collaboration, “The Compact Muon Solenoid Technical Proposal”, CERN/LHCC 94-38 (1994). [2] F. Abe et al., “Pseudorapidity distributions of charged particles produced in pp √ interactions at s=630 and 1800 GeV.”, Phys. Rev. D41 (1990) 2330. [3] B. B. Back et al., “Charged antiparticle to particle ratios near midrapidity in p+p collisions √ at s NN =200 GeV”, Phys. Rev. C71 (2005) 021901. [4] T. Sjostrand, S. Mrenna, and P. Skands, “PYTHIA 6.4 physics and manual (*version 8 was used)”, JHEP 05 (2006) 026. [5] K. Werner et al., “Parton ladder splitting and the rapidity dependence of transverse momentum spectra in deuteron gold collisions at RHIC”, Phys. Rev. C74 (2006) 044902, arXiv:hep-ph/0506232. [6] N. N. Kalmykov, S. S. Ostapchenko, and A. I. Pavlov, “Quark-gluon string model and EAS simulation problems at ultra-high energies”, Nucl. Phys. Proc. Suppl. 52B (1997) 17–28. 8 A A Additional Supporting Figures and Tables Additional Supporting Figures and Tables LHC Luminosity Ideal data fractions Fraction of ideal data for ZeroBias and triggered minimum bias 1.01 0.8 1 (a) 0.8 bunch pattern bunch pattern 43×43 936×936 0.6 0.6 0.6 0.4 0.4 0.4 minimum bias ZeroBias 0.2 0.2 0.00 1.01 10 28 0.8 0.8 (c) 0.8 Region for ZeroBias and triggered minimum0.2 bias ideal data Region for triggered minimum bias ideal data Region not useful for ideal data 0 29 10 30 10 31 32 10 10 (b) 30 1 10 31 32 10 10 33 34 10 10 (d) 0.8 bunch pattern bunch pattern 156×156 2808×2808 0.6 0.6 0.6 0.4 0.4 0.4 CMS preliminary 0.2 0.2 0.2 0.00 0 28 10 28 10 29 10 29 10 30 10 30 10 31 10 31 10 32 30 10 32 1030 10 10 31 10 32 1031 10 1032 33 10 1033 34 10 1034 Luminosity (cm -2 s-1) Figure 4: The fraction of triggered data that would be usable as “ideal data” in physics analyses for two different triggers, a beam crossing-time trigger (that would yield “zero bias” data) and a 100% efficient detector-based trigger (that would yield “minimum bias” data). The shaded regions have the same meaning as in Figures 1 and 2. Table 5: Table of prescale factors to achieve 10 Hz of actual data-taking rate and the corresponding ideal data rates for the LHC Stage A commissioning. The ideal data rate refers to minimum bias data with one and only one collision per bunch crossing. Stage A Physics Run Plan∗ Average Results for 10 Hz Data-Taking Rate ZeroBias Trigger Minimum Bias Trigger∗∗ Bunch Luminosity Events per Prescale Ideal Data Prescale Ideal Data Pattern (cm−2 s−1 ) Crossing Factor Rate (Hz) Factor Rate (Hz) 1×1 1.6 × 1027 0.011 1.1 × 103 0.11 1.3 × 101 9.9 43 × 43 7.0 × 1028 0.011 4.8 × 104 0.11 5.5 × 102 9.9 43 × 43 1.1 × 1030 0.18 4.8 × 104 1.5 8.7 × 103 9.1 43 × 43 6.1 × 1030 1.0 4.8 × 104 3.7 4.8 × 104 5.8 156 × 156 2.2 × 1031 1.0 1.8 × 105 3.7 1.7 × 105 5.9 156 × 156 1.1 × 1032 5.0 1.8 × 105 0.35 1.8 × 105 0.35 ∗ CMS Week, 19sep07, Helmut Burkhardt, “LHC Backgrounds, Luminosity and more” ∗∗ These values assume a 100% efficient detector-based minimum bias trigger. 9 dNch dη from pp collisions at 1 dNch N event ddN η ch 1 N event d η Generated p+p - √s = 14TeV 66 (b) 6 minimum bias collisions hard core collisions 4 22 2 66 -10 0 10 (c) 0 -10 0 double diffractive collisions 4 22 2 -10 0 -10 10 (d) CMS preliminary 6 single diffractive collisions 44 00 s = 14 TeV (a) pythia simulation 44 0 √ 10 0 10 0 -10 0 -10 0 10 10 η η p+p - √s = 14TeV 8 1 dNch N event d η 1 dNch N event d η Figure 5: The number of charged particles per unit pseudo-rapidity (averaged over many events) for minimum bias collisions (a), hard core collisions (b) and single- (c) and doublediffractive collisions (d). The distributions were generated via the PYTHIA Monte-Carlo generator (DWT tune). Note in (c) open and closed symbols represent the possible fragmentation directions. Effective triggering of all collision types is the main aim of the minimum bias trigger, although this is challenging in the context of diffractive collisions. CMS preliminary EPOS simulation pythia simulation p+p - √s = 14TeV 8 6 6 4 4 2 2 0 -10 0 0 10 η CMS preliminary QGSII simulation pythia simulation -10 0 10 η Figure 6: Comparison of the minimum bias dNch /dη distributions from P YTHIA (open circles) to EPOS (left panel, closed circles) and QSGII (right panel, closed circles). Both EPOS and QSGII calculations predict more charged particles than P YTHIA. 10 A Additional Supporting Figures and Tables Table 6: HF trigger efficiencies for P YTHIA, EPOS and QGSII monte-carlo generators. The number of trigger towers required to trigger is shown in the first column. For example, the first row requires at least one trigger tower on either the positive or negative HF. The compressed (actual) trigger tower energy threshold is noted in the second (third) column. The remaining columns show the minimum bias efficiencies for single- and double-side triggers for each model used. 1 dNch N event d η HF Trigger Towers Number Energy Threshold Required Comp’d Actual (EC ≥ ) (E T >) (GeV) T 1 1 0.5 1 2 1.0 1 3 1.5 2 1 0.5 2 2 1.0 2 3 1.5 Trigger Efficiency (%) Single-side Triggers Double-side Triggers P YTHIA EPOS QGSII P YTHIA EPOS QGSII 91.9 61.1 41.2 78.7 40.8 23.6 95.6 68.7 50.2 83.2 51.5 33.5 p+p - √s = 14TeV 6 94.7 77.9 61.9 85.9 60.2 35.8 71.5 33.5 18.7 53.3 20.3 10.7 77.1 44.3 29.7 60.9 31.8 17.9 82.2 48.4 28.1 66.5 30.5 11.9 CMS preliminary pythia simulation minimum bias collisions Fixed Target (7TeV) hijing 1.383 simulation p+O p+p HF region 4 2 0 -10 p(7TeV) 0 O(0TeV) 10 η Figure 7: The number of charged particles per unit pseudo-rapidity for minimum bias collisions compared to beam gas collisions occurring at the centre of the detector. Closed circles represent the minimum bias collisions at 14 TeV, open circles represent the collision of protons on stationary protons (i.e. hydrogen nuclei). Closed grey squares represent the collision of protons on stationary oxygen nuclei, the largest nuclei of the expected beam gas/halo interactions. The beam gas collisions show a distinctly one-sided distribution relative to the collision point. P(Nch )×100 11 p+p - √s = 14TeV 1.5 CMS preliminary pythia simulation minimum bias collisions minimum bias hard core single diffractive double diffractive 1.0 0.5 0.0 0 100 200 300 Number of Charged Particles Figure 8: The frequency distribution of the total number of charged particles, as calculated over the whole pseudo-rapidity range for minimum bias (solid black circles), hard core (grey circles), single-diffractive (black squares) and double-diffractive (open circles) collisions. 100 p+p - √s = 14TeV 80 Number of HF Trigger Towers (η>0) Number of HF Trigger Towers Triggering minimum bias collisions with HF CMS preliminary pythia simulation minimum bias collisions 60 40 20 0 0 50 100 150 200 Number of Generated Particles (3<|η|<5) 50 p+p - √s = 14TeV CMS preliminary 40 pythia simulation minimum bias collisions 102 30 20 10 10 1 0 0 10 20 30 40 50 Number of HF Trigger Towers (η<0) Figure 9: Left panel: correlation between the total number of trigger towers in the forward hadronic calorimeters (HF) and the number of generated charged particles impinging the pseudo-rapidity region covered by HF. Right panel: Correlation between the number of hit trigger towers from the positive and negative forward hadronic calorimeters (HF). 12 ch P(N )×100 A p+p - √s = 14TeV 1.5 1.5 (a) (b) 1.5 pythia simulation minimum bias collisions hard core collisions 1.0 1.0 1.0 0.5 0.5 0.5 0.0 0.0 0 100 Additional Supporting Figures and Tables 0.0 3000 200 non-triggered No. of Trig. Towers Hit≥1 Trig. Tower TPG ET ≥1 (Comp.) OR triggered AND triggered 100 200 (c) 1.5 1.5 CMS preliminary 1.5 single diffractive collisions double diffractive collisions 1.0 1.0 1.0 0.5 0.5 0.5 0.0 0.0 0 0 100 100 300 (d) 200 200 0.0 3000 0 100 100 200 200 300 300 Number of Charged Particles Figure 10: Illustration of the triggering efficiency for minimum bias collisions (a), hard core collisions (b) and single- (c) and double-diffractive collisions (d). 13 fake rate due to pile-up p+p - √s = 14TeV 100 (a) pythia simulation minimum bias collisions 100 50 50 0 1 2 0 1 2 100 50 0 0 1 1 2 2 0 1 2 0 (e) 100 100 50 50 0 (g) 0 1 2 0 (h) 100 100 50 50 0 0 0 1 1 2 2 0 0 1 2 1 2 1 2 (f) 0 CMS preliminary(i) 〈N coll〉 = 1.00 50 0 50 0 50 0 50 (d) 50 0 (c) 100 OR Trigger AND Trigger 100 0 100 (b) 100 〈N coll〉 = 0.51 0 100 0 fake rate due to empty events and pile-up 〈N coll〉 = 0.25 Single-Collision Minimum Bias Efficiency (%) fake rate due to empty events 0 0 1 2 Relative Rate: Fake Ideal Figure 11: The single collision (no pile-up) minimum bias efficiency versus the rate of fake collisions (relative to the ideal data collision rate). The columns show the fake rate for empty events - random noise triggering - ((a),(d),(g)), the fake rate of pile-up only ((b),(e),(h)) and for both combined ((c),(f),(i)). The rows show three average number of collision cases, illustrating the relative interplay of noise and pile-up on the trigger. Each data point represents a different triggering setup (various energy thresholds and number of towers hit). Dark (light) symbols represent the single-side (double-side) triggering scenarios. 14 A 100 100 p+p - √s = 14TeV ideal data loss due to pile-up (a) pythia simulation minimum bias collisions 50 50 OR Trigger AND Trigger 50 100 100 (d) 50 100 50 50 0 0 0 0 50 50 100 (e) 0 0 50 50 (g) 0 0 50 100 (h) 0 0 100 100 50 50 0 0 100 0 50 50 100 0 0 100 0 50 100 50 100 50 100 (f) CMS preliminary (i) 〈N coll〉 = 1.00 100 100 〈N coll〉 = 0.51 0 0 50 100 50 100 0 0 100 50 0 (c) 100 50 0 0 (b) 100 50 0 100 ideal data loss due to empty events and pile-up 〈N coll〉 = 0.25 Single-Collision Minimum Bias Efficiency (%) ideal data loss due to empty events Additional Supporting Figures and Tables 50 100 Fraction of Ideal Data (%) Figure 12: The single collision (no pile-up) minimum bias efficiency versus the fraction of ideal data (expected to be recorded to tape). The columns show this fraction for empty events - random noise triggering - ((a),(d),(g)), the ideal fraction when only considering pile-up only ((b),(e),(h)) and for both combined ((c),(f),(i)). The rows show three average number of collision cases, illustrating the relative interplay of noise and pile-up on the trigger. Each data point represents a different triggering setup (various energy thresholds and number of towers hit). Dark (light) symbols represent the single-side (double-side) triggering scenarios. 15 Effect of HF triggering on physical measurements p+p - √s = 14TeV CMS preliminary ch pythia simulation P(N non-triggered No. of Trig. Towers Hit≥1 Trig. Tower TPG E ≥1 (Comp) T OR triggered AND triggered 10 Efficiency (%) |η|<2 )×100 15 100 5 50 00 50 100 Number of Charged Particles (|η|<2) 0 0 50 100 150 Number of Charged Particles (|η|<2) p+p - √s = 14TeV 1 dNch N event d η 1 dNch N event d η Figure 13: Multiplicity distribution of charged particles (|η | < 2). The black symbols represent the total inelastic cross-section. Dark (light) grey symbols show the distribution for single-side (double-side) HF triggers. The insert figure shows the efficiency versus the number of charged particles (|η | < 2) for both HF triggers. (a) CMS preliminary pythia simulation 6 4 p+p - √s = 14TeV 6 4 non-triggered OR triggered AND triggered No. of Trig. Towers Hit≥1 Trig. Tower TPG E ≥1 (Comp) 2 non-triggered OR triggered AND triggered No. of Trig. Towers Hit≥1 Trig. Tower TPG E ≥1 (Comp) 2 T T 0 (b) CMS preliminary pythia simulation -10 0 0 10 η -10 0 10 η ch Figure 14: Panel (a) shows the dN dη per event measured for the total (non-triggered) inelastic cross-section (black symbols) single- (dark grey) and double-side (light grey) triggered events. Panel (b) shows the same data as (a) but events are inverse-weighted by the efficiency found in Figure 13.