⚛️ high-energy experiments

High Multiplicity Trigger for Long-Lived Particles in CMS detector

This paper describes the development, implementation, and performance evaluation of a dedicated High Multiplicity Trigger (HMT) in the CMS experiment during LHC Run 3, which significantly enhances sensitivity to long-lived particles by efficiently selecting events with unusually large numbers of hits in the muon detectors while maintaining stable operation under high-luminosity conditions.

Original authors: M. Abbott, D. Acosta, D. Aebi, M. Ahmad, A. Albert, A. Apresyan, C. Arbour, C. Aruta, K. Banicz, V. Barashko, E. Barberis, L. Benato, A. Bolshov, J. Bonilla, R. Breedon, K. Bunkowski, H. Cai, C. Campa

Published 2026-06-26

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: M. Abbott, D. Acosta, D. Aebi, M. Ahmad, A. Albert, A. Apresyan, C. Arbour, C. Aruta, K. Banicz, V. Barashko, E. Barberis, L. Benato, A. Bolshov, J. Bonilla, R. Breedon, K. Bunkowski, H. Cai, C. Campagnari, J. Carlson, V. Cherepanov, M. Chen, R. Clare, R. Clark, P. T. Cox, R. De Los Santos, S. Dildick, M. Dittrich, X. Dong, L. S. Durkin, R. Erbacher, P. Fernández Manteca, P. Flanagan, E. Fontanesi, C. Galloni, Y. Gao, J. Gilmore, A. Greshilov, Y. Haddad, Y. Han, Z. Hao, J. Hauser, E. Hazen, H. He, J. Heikkilae, M. Herndon, C. Hill, T. Huang, M. Ignatenko, M. A. Iqbal, I. Israr, M. Jeitler, S. Jindariani, E. Juska, T. Kamon, V. Karjavine, H. Kim, J. Kim, A. Korytov, O. Kukral, K. H. M. Kwok, E. Kuznetsova, J. Liu, W. Liu, C. Lo, A. Madorsky, N. Manganelli, M. Matveev, H. Mei, N. Menendez, G. Mitselmakher, G. Mocellin, S. Mondal, D. M. Morse, M. Mulhearn, B. P. Padley, V. Palichik, A. Peck, C. Pena, V. Perelygin, D. Rabady, N. Rawal, J. Richman, A. Safonov, P. Siddireddy, M. Spiropulu, I. Suarez, N. Terentyev, M. Tripathi, N. Turner, V. Valuev, C. Wang, J. Wang, Y. Wang, Z. Wang, D. Wenzl, D. Wood, S. Xie, X. Yang, E. Yigitbasi, H. Zhang, Y. Zhang, J. Zheng, X. Zuo

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Large Hadron Collider (LHC) as the world's most powerful particle smasher. It fires protons at each other billions of times a second, creating a chaotic explosion of debris. The CMS detector is a giant, high-tech camera surrounding this collision point, trying to take pictures of the results.

However, the camera is overwhelmed. It can't save every single picture because there are too many, and the data storage would fill up instantly. So, the detector has a "smart filter" (called a trigger) that acts like a bouncer at a club. It quickly decides which events are interesting enough to keep and which boring ones to throw away.

The Problem: The "Ghost" Particles
Physicists are looking for "Long-Lived Particles" (LLPs). Think of these as ghostly particles that don't immediately disappear after being created. Instead, they travel a bit further—sometimes deep inside the detector's walls—before they finally decay (break apart).

Standard filters are designed to catch particles that behave normally: they zip straight through the detector and leave a clean, predictable trail, like a bullet hitting a target. But when an LLP decays deep inside the detector's "back room" (the muon system), it doesn't leave a clean trail. Instead, it crashes into the steel walls and creates a massive, messy explosion of secondary particles. It's like a ghost walking through a wall and causing a huge pile of furniture to crash down on the other side.

The old filters didn't know what to do with this mess. They were looking for clean bullets, not furniture avalanches. As a result, these interesting "ghost" events were being thrown away by the bouncer before physicists could even see them.

The Solution: The "High Multiplicity Trigger" (HMT)
To fix this, the team built a new, specialized bouncer called the High Multiplicity Trigger (HMT).

Instead of looking for a single clean line, this new bouncer is programmed to look for clutter.

The Analogy: Imagine you are looking for a specific type of noise in a library. The old rule was: "If you hear a single loud shout, let them in." But the ghost particles are like someone knocking over an entire bookshelf. The new rule is: "If you hear a huge amount of noise coming from one specific spot, let them in immediately."

The HMT counts the number of "hits" (signals) in the detector's endcap chambers (the back rooms). If a single chamber sees an unusually high number of hits all at once, it flags the event as a potential "furniture avalanche" (an LLP decay) and saves the data.

How It Works (The Mechanics)
The detector is divided into different zones. The new system is very smart about where it looks:

The Inner Ring: This area is closer to the collision point and gets hit by a lot of "background noise" (regular particles). So, the bouncer here is very strict. It only lets in events with a massive amount of hits to avoid false alarms.
The Outer Ring: This area is quieter. Here, the bouncer is more relaxed, allowing events with fewer hits to pass through, ensuring no potential ghosts are missed.

The system works in two stages:

Level 1 (The Fast Bouncer): This happens in electronics in a fraction of a microsecond. It just counts the hits and makes a quick "Yes/No" decision.
High-Level Trigger (The Detective): If the Fast Bouncer says "Yes," a more sophisticated computer system (the Detective) takes a closer look. It groups the hits into "clusters" (like seeing the shape of the fallen furniture) and checks if they arrived at the right time. This filters out the few remaining false alarms.

The Results
The paper reports that this new system is a game-changer:

Efficiency: It is more than 30 times better at catching these specific "ghost" events than the old methods. Previously, the detector was missing almost all of them.
Stability: The system works reliably even when the collider is running at maximum speed (high "pileup," meaning many collisions happening at once).
Flexibility: It doesn't need to know exactly what kind of ghost particle is causing the mess. It just looks for the mess itself, making it useful for searching for many different types of new physics theories.

What's Next?
The paper mentions that in the future, they hope to install similar "clutter detectors" in the front part of the detector (the barrel) and use even smarter computer chips (machine learning) to recognize these patterns even faster. But for now, this new "High Multiplicity Trigger" has successfully opened the door to a whole new world of searching for long-lived particles that were previously invisible.

Technical Summary: High Multiplicity Trigger for Long-Lived Particles in CMS

Problem Statement
Searches for Long-Lived Particles (LLPs) at the CMS experiment often involve unconventional event topologies that are difficult to select using standard trigger strategies. Specifically, when LLPs decay within the CMS muon system, their decay products can initiate electromagnetic or hadronic showers as they traverse the steel absorbers interweaving the muon chambers. These "Muon Detector Showers" (MDS) produce a large number of charged particle hits, a signature distinct from the few hits deposited by standard muons. Prior to this work, published CMS LLP searches utilizing the MDS signature were severely limited by the lack of a dedicated trigger. Previous analyses relied on large missing transverse momentum (which often yields low signal efficiency) or associated electrons/muons, restricting the search channels.

Methodology and Algorithm Design
To address this, the authors developed and deployed a dedicated High Multiplicity Trigger (HMT) for LHC Run 3, implemented in both the Level-1 (L1) and High-Level Trigger (HLT) systems. The algorithm focuses on the multiplicity of hits in the Cathode Strip Chambers (CSCs), located in the endcap muon system ( $0.9 < |\eta| < 2.4$ ).

Level-1 Implementation: Due to bandwidth constraints, the L1 algorithm operates on a per-chamber basis. It counts anode wire hits and cathode strip hits within a single bunch crossing (BX).
- Thresholds: A chamber is identified as containing a shower if hit counts exceed configurable thresholds. These thresholds are categorized into "Loose," "Nominal," and "Tight" and are optimized per station-ring (e.g., ME1/1, ME2/1) to balance signal efficiency against background rates, which vary significantly with pseudorapidity ( $\eta$ ).
- Logic: The system encodes shower presence into 2-bit categories. In outer-ring chambers (legacy TMBs), only anode hits are used. In inner-ring chambers (OTMBs), both anode and cathode hits are utilized, with cathode hits evaluated over a three-BX window due to timing resolution differences.
- Global Trigger (GT): The Global Muon Trigger aggregates data from 12 Endcap Muon Track Finder (EMTF) sectors. It generates three L1 seeds: "One-Nominal" (at least one Nominal/Tight shower in any sector), "One-Tight," and "Two-Loose" (at least two Loose showers from different sectors).
High-Level Trigger (HLT) Implementation: With looser latency constraints, the HLT reconstructs "cluster objects" by combining hits from multiple CSC chambers and stations.
- Reconstruction: Hits are reconstructed into 2-D positions ("rechits") and clustered using the Cambridge-Aachen algorithm in $\eta-\phi$ space with a distance parameter of 0.4.
- Selections: The HLT applies stricter criteria to background clusters, which tend to be concentrated in a single station and at higher $\eta$ . The "One-Nominal" path requires specific hit multiplicity thresholds ( $N_{hits}$ ) ranging from 100 to 500 depending on $\eta$ and the number of stations involved. It also vetoes clusters with hits in the ME1/1 and ME1/2 stations (where background rates are highest) and requires timing consistency ($-5.0$ to $12.5$ ns). The "Two-Loose" path requires two distinct clusters with a minimum of 75 hits each.

Key Contributions

First Hardware Implementation: This paper reports the first implementation of an LLP-specific hardware trigger for the CMS experiment, moving beyond reliance on generic triggers like missing transverse momentum.
Optimized Thresholds: The authors provide optimized hit multiplicity thresholds for both anode and cathode strips across all CSC stations, derived from zero-bias data and Monte Carlo simulations based on a Twin Higgs benchmark model ( $H \to 2S \to 4b$ ).
Dual-Stage Architecture: The work details a two-stage trigger strategy where L1 performs rapid hit counting to reduce rates, and HLT performs sophisticated clustering and timing analysis to further suppress backgrounds while maintaining high efficiency.

Results

Efficiency: The HMT demonstrates a dramatic improvement in signal efficiency compared to Run-2 analyses. For benchmark models, the efficiency improved by more than a factor of 30 (from $\sim1\%$ to $>40\%$ for single LLP decays). For events with two LLPs decaying in the CSC acceptance, the "Two-Loose" seed provides an additional $\sim10\%$ efficiency gain over the "One-Nominal" seed alone.
Rate and Stability: The trigger rates show a linear dependence on pileup, demonstrating stability across different data-taking conditions in 2022 and 2023. The HMT occupies approximately 1-2% of the total CMS L1 trigger bandwidth and less than 0.5% of the total HLT output rate under typical Run-3 conditions.
Background Suppression: The requirement for multiple clusters or specific station vetoes effectively suppresses background clusters, which are typically concentrated in single stations and lack the spatial spread of signal showers.

Significance
The paper claims that the HMT substantially extends the CMS sensitivity to non-standard signatures associated with LLP decays. By targeting the generic MDS signature, the trigger provides a flexible platform for searching for a broad range of Beyond the Standard Model (BSM) scenarios, including exotic Higgs decays, Heavy Neutral Leptons, axion-like particles, and hidden valley models. The authors note that this implementation opens a new program of LLP searches using the muon detector shower signature, which was previously inaccessible due to trigger limitations. Future developments discussed include lowering thresholds in association with other objects (like charged leptons) and implementing similar triggers in the barrel muon system following Phase-2 hardware upgrades.

More like this