Novel method to indirectly reconstruct neutrinos in collider experiments

This paper introduces a novel inclusive-tagging scheme based on an asymptotically recursive vector sequence that enables the first indirect reconstruction of multiple undetected particles, such as neutrinos, in collider experiments, thereby significantly enhancing the precision of Standard Model measurements and the search for new physics.

The Gist

Imagine you are a detective trying to solve a crime scene where a valuable object (a particle) has vanished without a trace. In the world of high-energy physics, this "object" is often a neutrino. Neutrinos are like ghosts: they zip through detectors without leaving a single footprint, making them impossible to see directly.

For decades, physicists have faced a frustrating rule: if a collision produces one ghost, they can figure out where it went by looking at what did get caught. But if two or more ghosts are present, the case becomes unsolvable with traditional tools. The clues are too jumbled, and the "ghosts" hide in the noise.

This paper, written by Hongrong Qi and Paoti Chang from National Taiwan University, introduces a brand-new detective technique to solve this exact problem. Here is how their method works, explained in everyday terms:

The "Magic Mirror" Analogy

Imagine a collision event as a sealed room where two people (let's call them Signal and Tag) are dancing.

• Signal is the person we are studying. They drop a visible item (A) and a ghost (B, the neutrino).
• Tag is the partner. They drop a visible item (C) and a messy pile of other stuff (D) that we can't fully sort out.

The rule of the universe is that the total momentum (the "push" of the dance) must balance out. If we know where the visible items went, we can calculate where the invisible ones should have gone. But because the "messy pile" (D) is so chaotic, we can't get a perfect answer immediately.

The "Infinite Zoom" Trick

The authors propose a clever mathematical trick called an "asymptotically recursive vector sequence." That's a fancy way of saying: "Keep guessing, but get smarter with every guess."

Think of it like trying to find the exact center of a dartboard while blindfolded, but you have a magical assistant who tells you, "You are off by this much," and then you adjust your guess.

1. The First Guess: You make a rough estimate of where the ghost went based on the visible items.
2. The Correction: You realize your estimate was slightly off because of the messy pile (D).
3. The Loop: You take your previous guess, add a tiny correction based on the messy pile, and make a new guess.
4. The Magic: The authors show that if you repeat this process over and over (mathematically, infinitely), the "messy pile" gets "eaten" or cancelled out. The error shrinks by half every time you loop.

After about 15 loops, the error becomes so tiny (less than 0.01%) that your guess is practically perfect. You have effectively "reconstructed" the ghost's path without ever seeing it.

The "Ghost Eating" Concept

The paper uses a vivid metaphor: the missing information (the messy pile D) is "eaten" by the infinite iterations. Just like a Pac-Man game where the character eats the dots, this mathematical process "eats" the uncertainty until only the true path of the neutrino remains.

What They Tested

The authors didn't just do this on paper; they simulated it using computer models (pseudo-experiments) that mimic real particle colliders like Belle II, BESIII, and LHCb. They tested scenarios involving:

• B-mesons decaying into muons and neutrinos.
• Tau particles decaying into pions and neutrinos.
• Lambda-c particles decaying into electrons and neutrinos.

In every test, their new method successfully pinpointed the neutrino's momentum with high precision, whereas traditional methods produced blurry, indistinguishable results.

Why This Matters

Currently, if physicists want to study neutrinos in complex collisions, they often have to throw away data or rely on rough estimates, which lowers the precision of their measurements.

This new method is like giving the detective a high-powered telescope. It allows them to:

• See the invisible: Reconstruct the four-momentum (speed and direction) of undetected particles like neutrinos or neutral Kaons.
• Solve harder cases: Handle events with multiple missing particles, which was previously impossible.
• Find new physics: By measuring Standard Model parameters more precisely, they can spot tiny deviations that might hint at "New Physics" (things we don't know about yet).

The authors also suggest that this "infinite guessing" math might be useful in other fields, such as machine learning, acting as a filter to clean up unknown or missing data.

In short: The paper claims to have solved a 50-year-old problem in particle physics by inventing a mathematical loop that "eats" uncertainty, allowing scientists to finally track the untrackable ghosts of the subatomic world.

Technical Summary

Technical Summary: Novel Method to Indirectly Reconstruct Neutrinos in Collider Experiments

Problem Statement
In collider experiments, neutrinos and other long-lived neutral particles (such as $K^0_L$) cannot be directly tracked. While the four-momentum of a single missing particle can be determined using recoil techniques, events containing two or more neutrinos present a longstanding challenge where traditional methods fail. In such scenarios, experimental groups are typically limited to extracting signal yields from momentum distributions. However, because both signal and background manifest as (quasi-)smooth distributions in these spectra, distinguishing them relies heavily on Monte Carlo simulations, resulting in lower significance and higher uncertainties. Furthermore, while inclusive-tagging methods (reconstructing the "tagged" side from all remaining tracks) offer high efficiency compared to hadronic tagging, they traditionally suffer from poor signal-to-background separation and lower sensitivity.

Methodology
The authors propose an innovative inclusive-tagging scheme based on an asymptotically recursive vector sequence to capture the four-momentum of undetected particles on the signal side. The method operates within the rest frame of the signal ($S$) and tagged ($T$) system, defined by the decay topology $S \to A + B$ and $T \to C + D$, where $B$ is the missing particle (e.g., a neutrino) and $D$ represents unreconstructed particles on the tagged side.

The core of the method is a recursive algorithm that iteratively refines the estimate of the missing particle's momentum, $p(B)$:

1. Initialization ($k=0$): A random vector for the missing tagged momentum $p(D)_{rand}$ is generated. This is scaled to satisfy the mass constraint of the signal particle $S$, yielding an initial estimate $p(B)_0$.
2. Recursive Iteration ($k \geq 1$): The algorithm constructs a sequence where the estimate of the missing momentum at step $k$, denoted $p(B)_k$, is updated using the average of the previous estimate and a correction term derived from the momentum conservation equation:
   $$p(B)_k = p(B)_{k-1} + \frac{p(B)'_{k-1}}{2}$$
   where $p(B)'_{k-1} = -p(A) - p(C) - p(D)_{k-1}$.
3. Convergence: The sequence is designed such that as the number of iterations $n$ approaches infinity, the term $p(B)_n$ asymptotically converges to the true momentum $p(B)$. The "missing" momentum of the tagged side ($p(D)$) is effectively "eaten" by the infinite iterations.
4. Parameterization: To ensure physical validity, the algorithm constrains the invariant masses of narrow-width particles ($S$ and $B$) to their mean values (using Particle Data Group parameters) and calculates the arithmetic mean of two intermediate parameterization methods to determine $p(D)_k$. The authors note that $k=15$ iterations are generally sufficient, as the residual error term ($1/2^{15}$) falls below 0.01%.

Key Results
The scheme was tested using three pseudo-experiment samples generated with ROOT software, simulating data for Belle II, LHCb, and BESIII experiments:

• $\Upsilon(4S) \to B^+B^- \to \mu^+\nu_\mu + X$
• $B^0 \to \tau^-\tau^+ \to \pi^-\pi^+\pi^-\nu_\tau + X$
• $e^+e^- \to (\gamma_{ISR})\Lambda_c^+\Lambda_c^- \to (\gamma_{ISR})\Lambda e^+\nu_e + X$

In these simulations, the reconstructed neutrino momentum distributions showed mean values consistent with theoretical calculations (e.g., $2.6391 \pm 0.0004$ GeV/c for the $B^+$ decay, matching the PDG-based calculation of 2.639 GeV/c). Crucially, the resulting invariant mass squared ($M^2$) distributions for the neutrinos peaked sharply around zero, in stark contrast to the broad, indistinguishable distributions produced by traditional quasi four-momentum transfer squared ($q^2$) methods. The tests confirmed that the method introduces no bias peaks and effectively separates signal from continuum backgrounds.

Significance and Claims
The authors claim this is the first proposed solution to the challenge of extracting neutrino information in events with multiple missing particles. The significance of the work is threefold:

1. Reconstruction Capability: It enables the direct capture of the four-momentum of undetected particles in inclusive-tagging scenarios, a capability previously unavailable for multi-neutrino events.
2. Precision Physics: The method has the potential to significantly improve the precision of Standard Model parameter measurements in the (semi-)leptonic sector using existing data samples, without requiring new data.
3. New Physics Search: By enhancing the ability to distinguish signal from background in (semi-)leptonic decays, the scheme could play a pivotal role in searches for New Physics (NP).

The authors also suggest that the mathematical concept of the asymptotically recursive vector sequence may have applications beyond particle physics, specifically as a filter in machine learning algorithms to process missing or unknown information through infinite iterations. The paper asserts that the scheme is mathematically rigorous and has been validated as effective and reasonable within the context of experimental particle physics simulations.