⚛️ high-energy experiments

A study to suppress a sneaking cosmic muon background in the COMET experiment

This paper presents a methodology using track-fitting quality to discriminate the direction of cosmic muons sneaking into the COMET experiment's solenoid, successfully demonstrating a reduction in this critical background by an order of magnitude through Monte Carlo simulations.

Original authors: Manabu Moritsu, Hai-Bo Li, Tianyu Xing, Ye Yuan, Yao Zhang

Published 2026-07-01

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Manabu Moritsu, Hai-Bo Li, Tianyu Xing, Ye Yuan, Yao Zhang

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 COMET experiment as a high-stakes game of "spot the difference" happening inside a giant, super-sensitive magnetic tunnel. Scientists are looking for a very specific, rare event: a muon (a heavy cousin of an electron) turning into an electron. This "signal" electron has a very specific speed and moves in a specific direction, like a runner sprinting from the starting line (the target) toward the finish line (the detectors).

However, there's a sneaky problem. Sometimes, cosmic rays—particles raining down from space—slip through tiny cracks in the tunnel's roof and walls. These cosmic rays are mostly positive muons. If one of these sneaks in, hits a support beam, bounces around, and ends up moving in the opposite direction, it can look exactly like the signal electron the scientists are hunting for. It's like a runner on the wrong side of the track running backward; from a distance, they look identical to the real runner, but they are actually imposters.

The Problem: The "Backward Runner"

The paper explains that these "sneaking" cosmic muons are dangerous because they mimic the signal perfectly in terms of speed and path, but they are traveling in reverse.

The Signal: Starts at the target, goes out, and hits the outer detectors.
The Imposter: Starts at the outer detectors, goes in, hits the target, and looks like it came from the target.

If the scientists can't tell the difference, they might count a fake event as a real discovery, ruining their search for new physics.

The Solution: The "Fit Check"

To catch these imposters, the researchers developed a clever trick based on how well a path "fits" the data, similar to trying to force a square peg into a round hole.

They used a computer program to reconstruct the path of every particle. They tested two theories for every track:

Theory A: "This particle started at the target and went out." (The Normal Direction)
Theory B: "This particle started at the outer wall and came in." (The Reverse Direction)

The Analogy:
Imagine you are trying to trace a winding road on a map.

If you trace it in the correct direction, the road flows smoothly, and your pencil stays right on the lines. The "fit" is perfect.
If you try to trace the same road backward, you might find that the curves don't match up as well, or you have to make awkward corrections to stay on the line. The "fit" becomes messy and clumsy.

In the COMET experiment, the "messiness" is measured by a number called $\chi^2$ (chi-squared). A low number means the path fits perfectly; a high number means it's a bad fit.

How They Caught the Sneakers

The researchers simulated millions of these events on a computer. They found that:

Real Signal Electrons: When assumed to be moving "outward," they had a perfect fit. When assumed to be moving "inward," the fit was terrible.
Sneaking Cosmic Muons: When assumed to be moving "inward" (their true path), they had a perfect fit. When assumed to be moving "outward," the fit was terrible.

By comparing the quality of the fit for both directions, they could tell which way the particle was actually going. If the "inward" fit was much better than the "outward" fit, they knew it was a sneaking cosmic muon and could throw it out.

The Results

Using this method, the team showed they could:

Keep 87% of the real signal electrons (they didn't accidentally throw away the good stuff).
Remove 89% of the sneaking cosmic muon background (they successfully caught the imposters).

This means they reduced the background noise by about 10 times (an order of magnitude).

Why Precision Matters

The paper also notes that this trick works best if the detector is very sharp-eyed. The "spatial resolution" (how precisely the detector can see where a particle hit) is like the sharpness of a camera lens.

With a sharp lens (100 micrometers resolution), they could catch almost all the imposters.
With a blurry lens (200 micrometers resolution), the trick became less effective, and more imposters slipped through.

Summary

In short, the COMET experiment has a "sneaky" background problem where cosmic rays look like the signal they are hunting. The authors solved this by creating a mathematical test that checks if a particle's path makes sense in the forward direction or the backward direction. By rejecting the tracks that only make sense in the "wrong" direction, they successfully filtered out nearly 90% of the fake signals, making their experiment much more sensitive to the real physics they want to discover.

Technical Summary: Suppressing Sneaking Cosmic Muon Background in the COMET Experiment

Problem Statement
The COMET experiment, located at J-PARC, aims to search for neutrinoless muon-to-electron conversion with a target single-event sensitivity of $3 \times 10^{-15}$ . A critical challenge in Phase-I is the background arising from cosmic-ray muons. While the detector solenoid (DS) is surrounded by a Cosmic-Ray Veto (CRV) system, practical loopholes exist due to beam ducts and cabling. Cosmic muons can enter these loopholes, scatter off detector support structures (specifically the Cylindrical Trigger Hodoscopes, CTH), and traverse the Cylindrical Drift Chamber (CDC) to reach the stopping target.

The most dangerous background events involve positive muons ( $\mu^+$ ) entering from the downstream side. Due to the magnetic field and the detector's geometry, a "sneaking" $\mu^+$ traveling in the reverse direction (from CTH to the target) can mimic the trajectory of a signal electron ( $e^-$ ) from muon-to-electron conversion, which travels in the "normal" direction (from the target to CTH). Both particles have a momentum of approximately 105 MeV/c. Standard CRV systems cannot veto these events if the muon does not trigger the veto layers, and distinguishing them based on energy loss is difficult because the energy deposition of 105 MeV/c muons and electrons in the helium-based CDC gas is nearly identical.

Methodology
To address this, the authors propose a discrimination method based on track-fitting quality rather than energy deposition or timing alone. The core hypothesis relies on the mismatch between the assumed particle direction and the actual time origin of the drift time measurement.

Time Origin Mismatch: The CDC reconstructs particle tracks using the drift time of ionized electrons. The time origin is typically provided by the CTH hit time.
- Signal ( $e^-$ ): Generated at the target, passes through CDC, then hits CTH. The time origin (CTH hit) is correct for the drift time calculation.
- Background ( $\mu^+$ ): Enters from the downstream loophole, hits CTH first, then passes through CDC. If reconstructed assuming a "normal" direction (target $\to$ CTH), the time origin is misaligned by the time-of-flight between CTH and the CDC hits (up to 10 ns).
Drift Distance Error: A 10 ns timing error corresponds to a drift distance error of approximately 250 $\mu$ m (assuming a drift velocity of 25 $\mu$ m/ns). This error is comparable to the CDC's expected spatial resolution of 150 $\mu$ m.
Track Fitting Discrimination: The authors developed a track reconstruction program using the GENFIT2 framework. For each event, the track is fitted twice:
- Once assuming the normal direction (signal hypothesis).
- Once assuming the reverse direction (background hypothesis).
- The quality of the fit is evaluated using the reduced chi-squared ( $\chi^2/\text{ndf}$ ). The correct direction assumption yields a better fit, while the incorrect assumption introduces significant residuals and worsens the $\chi^2$ .
Discrimination Variable: The difference in fit quality is defined as $\Delta(\chi^2/\text{ndf}) = (\chi^2/\text{ndf})_{\text{Normal}} - (\chi^2/\text{ndf})_{\text{Reverse}}$ .

Key Contributions and Simulation Results
The authors performed a Monte Carlo simulation to validate this principle, generating signal electrons and reversed positive muons (simulated via a backward Monte Carlo method) with momenta of 100–110 MeV/c. The CDC hit positions were smeared with spatial resolutions of 100, 150, and 200 $\mu$ m.

Residual Distributions: When the wrong direction hypothesis is applied, the residual distribution (difference between drift distance and closest track-to-wire distance) shifts significantly, and the $\chi^2/\text{ndf}$ increases.
Separation Power: Using a threshold of $\Delta(\chi^2/\text{ndf}) < 0$ $Δ (χ^{2} / ndf) < 0$ :
- At the expected spatial resolution of 150 $\mu$ m, the method retains 87% of the signal electrons while suppressing 89% of the sneaking cosmic $\mu^+$ background (reducing background to 11%).
- The performance is sensitive to spatial resolution: at 100 $\mu$ m, suppression reaches 97% (3% background retention); at 200 $\mu$ m, suppression drops to 80% (20% background retention).
Magnitude of Suppression: The study demonstrates that this track-direction discrimination method can reduce the positive muon background by approximately one order of magnitude.

Significance and Claims
The paper claims that this methodology provides a viable, data-analysis-based solution to a severe background problem in the COMET Phase-I experiment without requiring hardware modifications to the detector. By leveraging the specific kinematic and timing differences between signal electrons and sneaking cosmic muons, the authors show that track-fitting quality is a robust discriminator.

The authors note that while the current study focuses on single-turn tracks and simple $\chi^2$ comparison, future improvements could involve:

Utilizing energy deposit differences in the CTH plastic scintillators (where muons lose slightly more energy than electrons).
Applying machine learning techniques that treat individual hit information and their order along the track as inputs, rather than relying solely on the global $\chi^2$ difference.

The work establishes a proof of principle that the "sneaking" background can be effectively mitigated, contributing to the overall sensitivity goals of the COMET experiment.