Introducing a Markov Chain-Based Time Calibration Procedure for Multi-Channel Particle Detectors: Application to the SuperFGD and ToF Detectors of the T2K Experiment

This paper presents and validates an iterative, Markov Chain-based time calibration method that utilizes only correlated hit pairs to determine precise per-channel offsets without external references, successfully improving the timing resolution of the SuperFGD and ToF detectors in the T2K experiment.

The Gist

The Big Picture: The "Orchestra" Problem

Imagine a massive orchestra with 56,000 musicians (the detector channels). They are all trying to play the same song (detecting particles). To sound good, they must all start at the exact same moment.

However, in the real world, some musicians are sitting further from the conductor, some have longer instrument cables, and some have slightly slower reaction times. Even if they try to start together, their notes arrive at the audience's ears at slightly different times. This creates a "muddy" sound where the music loses its crispness.

In particle physics, this "muddy sound" means the detector can't tell exactly when a particle arrived. If the timing is off, scientists can't measure the speed or energy of particles accurately.

The Problem: Usually, to fix this, you need a "Master Clock" (a perfect reference signal) to tell every musician exactly when to start. But in huge, complex detectors like the ones in the T2K experiment, getting a perfect Master Clock signal to every single wire is incredibly hard, expensive, or sometimes impossible.

The Solution: This paper introduces a clever new method that acts like a self-correcting choir. It doesn't need a Master Clock. Instead, the musicians listen to each other, figure out who is late, and adjust their own timing until everyone is perfectly in sync.

---

How It Works: The "Matching Pairs" Game

The core idea relies on correlated pairs.

Imagine two musicians, Alice and Bob, who are playing a duet. They know that whenever they play, their notes should be exactly 1 second apart because of the distance between them.

• The Reality: Alice plays her note, and Bob plays his. But because of their individual delays, the audience hears them 1.2 seconds apart.
• The Clue: The audience knows they should be 1.0 second apart. The extra 0.2 seconds is the "error."

The paper proposes a method where the system looks at millions of these "duets" (pairs of signals from different parts of the detector). It asks: "If Alice is late, is Bob early? Or are they both late?"

By comparing millions of these pairs, the system can mathematically deduce exactly how late or early each individual channel is, without ever needing to know the "true" start time of the universe.

The Secret Sauce: The "Markov Chain" (The Smoothing Blanket)

The paper uses a mathematical concept called a Markov Chain. Think of this as a game of "Telephone" or a "Smoothing Blanket."

1. The Iteration: The system starts with a guess. It calculates the average error for every channel based on its neighbors.
2. The Correction: It nudges the timing of Channel A based on what Channel B said, and Channel B based on Channel A.
3. The Loop: It does this over and over again (like a loop). With every pass, the "noise" gets smoothed out.
4. The Convergence: Eventually, the system settles down. Everyone agrees on the timing. The "late" channels are pulled forward, and the "early" channels are pushed back, until they all meet in the middle.

The authors proved mathematically that this process always works, provided the detector is connected enough (like a well-connected social network where everyone knows someone else). They also added a "damping factor" (a volume knob) to control how fast the system settles, preventing it from oscillating wildly before finding the right answer.

Real-World Applications: The T2K Experiment

The authors tested this on two very different detectors in the T2K neutrino experiment in Japan:

1. The SuperFGD (The Giant Cube):

   • What it is: A giant block made of nearly 2 million tiny plastic cubes, each with fibers running through them.
   • The Analogy: Imagine a 3D grid of light bulbs. When a particle hits a cube, light travels down three different fibers to three different sensors.
   • The Fix: The system looked at the light arriving at the three sensors from the same cube. Since it knows the distance to each sensor, it knew exactly how much time the light should have taken. It used these differences to calibrate all 56,000 sensors.
   • Result: The timing precision improved from 1.81 nanoseconds to 1.36 nanoseconds. This is like turning a blurry photo into a sharp one, allowing scientists to measure the speed of neutrons (which are hard to catch) much better.

2. The ToF Detector (The Scintillating Bars):

   • What it is: A set of long plastic bars surrounding the main detector, used to catch particles entering or leaving.
   • The Analogy: Think of long batons. When a particle hits a baton, light travels to both ends.
   • The Fix: The system looked at particles that hit two different batons in the air. By knowing how fast the particle was moving (close to the speed of light) and the distance between the batons, it could calculate the expected time difference.
   • Result: The timing improved from 298 picoseconds to 175 picoseconds. This is a massive improvement, allowing the detector to tell the difference between particles moving forward and backward, which is crucial for identifying what kind of particle was found.

Why This Matters

• No Master Clock Needed: This method is "self-healing." It doesn't rely on a perfect external clock signal, which is a huge advantage for building massive, complex detectors in the future.
• Scalable: It works whether you have 100 channels or 10 million.
• Unbiased: Because it uses the internal relationships between the detectors, it doesn't accidentally "fake" the data to fit a theory.

The Bottom Line

This paper presents a brilliant piece of mathematical engineering. It turns a complex synchronization problem into a simple game of "listen to your neighbor and adjust." By using the natural correlations between particles hitting different parts of a detector, the system automatically fixes its own timing errors, making the T2K experiment (and future experiments) much sharper and more accurate.

It's like giving a massive, chaotic orchestra a way to tune itself by listening to the harmony of their own music, rather than waiting for a conductor to shout "Start!"

Technical Summary

1. Problem Statement

High-performance particle detectors with thousands of readout channels often suffer from inter-channel mis-synchronisation. Even when channels share a common primary clock, factors such as:

• Imperfect cable length and trace matching.
• Comparator switching delays.
• Secondary clock phase-locking errors.
• Variations in electronics crate synchronization.

...introduce fixed time offsets ($T^{(0)}$) between channels. As the intrinsic time resolution of individual channels improves (often reaching sub-nanosecond levels), these fixed offsets become the limiting factor for the overall detector timing resolution.

Traditional calibration methods require reconstructing particle tracks to estimate a "reference time" ($t_{ref}$) and calculating offsets based on the difference between measured and expected arrival times. This approach is computationally expensive and can introduce biases if track reconstruction is imperfect. The authors propose a method that eliminates the need for an external reference time or full track reconstruction.

2. Methodology: Markov Chain-Based Iterative Calibration

The core of the proposed method is an iterative algorithm that exploits the correlation between "matching hit pairs" (signals from the same particle detected in two different channels) to solve for channel offsets.

A. The Concept of Matching Hit Pairs

The method relies on identifying pairs of hits ($ch_1, ch_2$) generated by the same ionizing particle. For these pairs, the expected time difference ($\Delta t_{exp}$) can be calculated purely from geometry and known light propagation speeds, without knowing the absolute time of the event.

• SuperFGD: Light generated in a scintillating cube travels along three optical fibers to SiPMs. The distance from the cube to the sensors is known, allowing precise calculation of $\Delta t_{exp}$.
• ToF Detector: Particles crossing two scintillator bars in different panels generate hits. The distance between hit positions allows calculation of $\Delta t_{exp}$ based on the speed of light.

The discrepancy $\Delta$ is defined as:
$$\Delta = (t_1 - t_2) - (s_1 - s_2)$$
Where $t$ are measured times and $s$ are expected times.

B. The Iterative Algorithm

The algorithm seeks a vector of offsets $\mathbf{T}^{(0)}$ such that applying them minimizes $\Delta$ globally.

1. Initialization: Start with raw measured times.
2. Iteration:
   • Calculate the average discrepancy $\Delta_i$ for all hit pairs where channel $i$ is the first hit.
   • Update the time for channel $i$ by subtracting a fraction of this discrepancy (controlled by a damping factor $\alpha$).
   • The update rule is: $t_i^{(k+1)} = t_i^{(k)} - \alpha \Delta_i^{(k)}$.
3. Convergence: The process repeats until the corrections become negligible (below a threshold $T_{converge}$).

C. Mathematical Proof and Markov Chain Analogy

The authors prove that this iterative process is mathematically equivalent to a Markov Chain process.

• They define a Crossing Matrix $\mathbf{X}$, where elements $x_{ij}$ represent the probability (or weight) that channel $i$ forms a matching pair with channel $j$.
• The update dynamics are governed by a transition matrix $\mathbf{M} = (1-\alpha)\mathbf{I} + \alpha\mathbf{X}$.
• Convergence Proof: Using the Perron-Frobenius theorem, the authors demonstrate that $\mathbf{M}$ is a row-stochastic matrix. Under general conditions (irreducibility and aperiodicity of the detector geometry), the system converges to a state where the residual time offsets are uniform across all channels.
• Since the method only uses time differences, the absolute common offset is arbitrary. The authors fix this by imposing that the average offset across all channels is zero.
• Rate of Convergence: The speed of convergence is determined by the spectral gap of the matrix $\mathbf{M}$, which is directly controlled by the damping parameter $\alpha$. This allows users to tune the convergence speed.

3. Key Contributions

1. Reference-Free Calibration: The method solves for time offsets using only intrinsic correlations between channels, removing the need for external reference clocks or complex track reconstruction.
2. Mathematical Rigor: Provides a formal proof of convergence using Markov Chain theory, ensuring the algorithm finds the unique solution for time offsets (up to a global shift).
3. Scalability and Tunability: The convergence rate is controllable via a single parameter ($\alpha$), making the method adaptable to detectors with vastly different geometries and channel counts.
4. General Applicability: The framework is detector-agnostic, provided a definition for "matching hit pairs" can be established based on geometry.

4. Results and Applications

The method was validated through numerical simulations and applied to two distinct detectors in the T2K Near Detector (ND280) upgrade.

A. SuperFGD (Super Fine-Grained Detector)

• Setup: ~56,000 SiPM channels reading scintillating cubes via optical fibers.
• Validation: Simulations with 600 channels confirmed the algorithm recovers input offsets with high precision.
• Real Data Application: Calibrated using 5 million matching hits from cosmic muons and neutrino beam data.
• Outcome:
  • Improved single-channel time resolution from 1.81 ns to 1.36 ns.
  • This enables the measurement of neutron time-of-flight within the detector, allowing for precise kinetic energy reconstruction of neutrons produced in neutrino interactions.

B. ToF (Time of Flight) Detector

• Setup: 118 scintillator bars arranged in panels surrounding the tracking volume.
• Validation: Simulations with 30 channels (5 bars per panel) confirmed offset recovery.
• Real Data Application: Calibrated using 180,000 cosmic muon tracks.
• Outcome:
  • Improved single-bar time resolution from 298.2 ps to 175.3 ps.
  • This enhancement is crucial for distinguishing forward-going vs. backward-going tracks and improving particle identification via time-of-flight.

5. Significance and Future Outlook

This paper presents a robust, scalable solution to a critical bottleneck in modern particle physics: synchronizing massive arrays of readout channels.

• Performance: It successfully mitigates hardware-induced timing errors (cables, traces, electronics) without requiring external timing references.
• Extensibility: The authors note the method can be extended to correct for time-walk effects (making offsets charge-dependent) and light propagation variations (making offsets distance-dependent) by expanding the dimensionality of the offset vector.
• Broad Impact: While demonstrated on scintillator detectors, the framework is applicable to any large-scale, time-synchronized system where channel correlations can be established, including energy response calibration.

In summary, the Markov Chain-based procedure offers a computationally efficient, unbiased, and mathematically proven method to achieve sub-nanosecond timing precision in complex, multi-channel detector systems.