Qutrits for physics at the LHC

arXiv:2510.14001 · quant-ph · Submitted 2025-10-15 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.

Kai: Today's paper: "Qutrits for physics at the LHC".

Mira: The identification of anomalous events, not explained by the Standard Model of particle physics, and the possible discovery of exotic physical phenomena pose significant theoretical, experimental and computational challenges.

Kai: First, who's behind it and why it matters.

Title and authors: Kai: Moving on to the title and authors, "Qutrits for physics at the LHC," it immediately tells us this work is directly aimed at next-generation colliders like the HL-LHC, which means they're tackling some pretty demanding data processing challenges.

Mira: The authors are Miranda Carou Laiño, Veronika Chobanova, and Miriam Lucio Martínez from different institutions. It's interesting to see researchers from various places coming together to tackle this problem of identifying exotic physics signals at the LHC.

Lev: I'm thinking about what this implies for running the actual experiments; if these models are meant to be used on real hardware, we need to worry about how they scale up from simulation to the massive datasets generated by an experiment like CMS or ATLAS.

Kai: Precisely, Lev. The paper is proposing a development of a qutrit quantum model and benchmarking it against qubit-based approaches specifically to assess accuracy and scalability for these LHC applications. It’s about seeing if this approach can actually handle the complexity of the data we're talking about.

Mira: What catches my eye is their focus on using the Majorana sphere formalism for encoding qutrit states, which they claim allows for a clearer distinction between pure and mixed states when represented on unit spheres. That geometric representation seems like a key technical step in making this work feasible.

Lev: From an error-correction viewpoint, that geometric constraint helps simplify the state space exploration, but we still have to figure out how to implement those SO(three) transformations mentioned in the text efficiently on any actual quantum processor.

The paper's summary: Kai: So, summarizing what they actually did in "Qutrits for physics at the LHC," they developed a qutrit-based model for anomaly detection and used Majorana encoding to represent those qutrit states on unit spheres to handle the data.

Mira: That summary makes sense when you consider that instead of just mapping things onto a Bloch sphere, which is what qubit models often do, they're using this richer qutrit structure, where the Hilbert space dimension d can be three.

Lev: The core idea seems to be using a Quantum Autoencoder, or QAE model implemented via variational quantum circuits. They compress the input data into a reduced latent representation by discarding what they call "trash states," and an anomaly indicator is derived from the fidelity of the encoder, defined by that expression involving RX, RZ, and RY gates.

Kai: Right, so they are essentially building a system that learns to represent normal physics patterns efficiently using these qutrit encodings, and anything that doesn't fit well gets flagged as anomalous based on how poorly it's reconstructed.

Mira: And the paper highlights their specific improvements in encoding: they adapt the 1P1Q scheme by replacing standard rotation gates with operators suited for qudits, specifically those defined by Gell-Mann matrices, which are used to generate SU(three) symmetry.

Lev: The mention of the Gell-Mann matrices and their role as generators for SU(three) symmetry suggests they're leaning into a structured mathematical framework to define these quantum operations, which is vital if we want any kind of reliable computation.

The paper's improvements: Kai: Now let's look at the specific improvements the authors suggest within this work. They don't just stop at a basic qutrit setup; they introduce novel encoding schemes to make it work better for real data.

Mira: They propose using the Majorana sphere formalism, which is described as representing qutrit states as two pairs of points within a unit sphere, allowing all possible states to be generated from a one-parameter family of canonical states using SO(three) transformations. This seems like a significant way to handle the complexity of the state space.

Lev: From an error-correction standpoint, that geometric constraint simplifies how you generate these states, which is good because it reduces the number of parameters you need to control when trying to build a circuit. But we still have to verify that this representation truly captures all necessary physical degrees of freedom for LHC data.

Kai: On top of the geometry, they extend the input encoding by adding parameters related to jet structure, introducing "N-subjettiness" parameters, like tau N = one/d zero X k pT,k (R 1,k, R 2,k,). This allows the model to look deeper into the substructure of jets rather than just looking at overall kinematics.

Mira: That's a big step because it directly tackles the challenge of separating signal from background jets by incorporating these jet substructure variables, and they even identify specific combinations of angles for phi one and phi two that yield the highest performance.

Lev: Incorporating those N-subjettiness parameters means we're feeding the AI more nuanced information about how particles are clustered inside a jet, which is what separates signal from background noise in these experiments. That adds a layer of complexity to the training data representation.

Conclusion: Kai: Wrapping up on this paper "Qutrits for physics at the LHC," the main conclusion is that their qutrit-based model shows a greater capacity to discern between different types of signals compared to models built on qubits. They quantified this using the Jensen–Shannon distance metric, which showed larger distances between signals for qutrits.

Mira: That result is interesting because it confirms what we suspected from the theory: the higher dimension of qutrits gives them a better feature space for distinguishing between signal classes in this context. They also mentioned that when trained on simulated data like JetClass, the AUC scores were comparable to or higher than those achieved by QAE Qubits.

Lev: If we look at it from a practical implementation angle, even with the current simulator limitations they acknowledge regarding memory consumption, they found that qutrit-based models achieve performance equivalent to qubit-based models or sometimes even higher. That’s encouraging for scaling up the underlying physics idea.

Kai: So, to summarize for everyone, this paper suggests that using qutrits with a Majorana sphere representation provides a way to handle the complexity of LHC data better than qubit models do, leading to better separation of signals and identifying specific decay signatures like the t to bqq decay as the most anomalous.

Mira: Indeed, it shows that the richer structure inherent in qutrits allows for a more detailed classification based on kinematic information derived from jet substructure parameters. This suggests that exploring these higher-dimensional quantum systems could be a viable path for searching for physics beyond the Standard Model at future colliders.

Lev: I just reiterate that while the performance is good, we still need to address those practical hurdles related to memory consumption when trying to run this on actual hardware, but the theoretical framework presented here is solid.

Kai: We've covered a lot about how they built this model and what it showed in "Qutrits for physics at the LHC." It really shows that exploring these alternative quantum architectures can lead to better data analysis techniques for high-energy physics.

Mira: It’s definitely something worth keeping an eye on as we look toward the next phase of experimental data processing. We've got a lot of interesting territory here before we move on.

Lev: I agree; the theoretical framework is solid, and it gives us a concrete direction for what to test with our error-correction simulations.

University of A Coruña · Instituto de Física Corpuscular (IFIC), University of Valencia

quant-ph

Submitted: 2025-10-15

Updated: 2026-10-01

Code: https://github.com/MirandaCarou/Qutritsfor-physics-at-LHC

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 73/100

The gist: The identification of anomalous events, not explained by the Standard Model of particle physics, and the possible discovery of exotic physical phenomena pose significant theoretical, experimental and

Key concepts

Qutrit-based Model
This is a quantum machine learning framework that uses qutrits instead of standard qubits to analyze complex particle data. Qutrits have three possible states, offering more expressive power than qubits, which helps the model better separate different physical signals in high-energy experiments.
Majorana Encoding
This is a specific mathematical method used to represent qutrit states on a unit sphere. It involves using two pairs of points within the sphere to define a state, allowing all possible states to be generated through simple transformations. This encoding simplifies how complex qutrit information is handled.
N-subjettiness
This is an extension of the input data used in the model. It involves calculating parameters related to jet structure, specifically how many particles are present within a certain angular distance from a central particle in a jet. This adds more detail about the particle arrangement for better anomaly detection.
Jensen–Shannon (JS) Distance
This is a metric used to quantify the difference or distance between two different signals detected by the model. A larger JS distance indicates that the model has successfully separated those two signals into distinct categories, showing improved discrimination power.

Terminology

Summary

The identification of anomalous events, not explained by the Standard Model of particle physics, and the possible discovery of exotic physical phenomena pose significant theoretical, experimental and computational challenges. This work explores the use of qutrit-based Quantum Machine Learning models for anomaly detection in high-energy physics data, with a focus on LHC applications.

The gist

A qutrit-based model for anomaly detection in CMS experiment data has been developed, and Majorana encoding for qutrit representation on unit spheres has been proven to be an effective way to represent the information on a unitary sphere for qutrit systems.

Reference Model Structure and Encoding

The study builds upon the One Particle-One Qubit (1P1Q) scheme, which encodes particle kinematics into a single qubit by mapping pseudo-rapidity and azimuthal angle parameters onto the spherical coordinates of the Bloch sphere. The Quantum Autoencoder (QAE) model utilizes an encoder and a decoder implemented through variational quantum circuits. The encoder compresses input data into a reduced latent representation, discarding redundant qubits referred to as trash states. An anomaly indicator is derived from the fidelity of the encoder, defined as:

"U(Θ) = O N i=1 RX(ϕi)RZ(θi)RY (ωi)! ⊗ (O 1≤i<j≤N Cij)"

Qutrit Representation and Majorana Encoding

The qutrit system generalizes the qubit by having a Hilbert space dimension of d, where a register of n qudits corresponds to dimension d n. For qutrits (d=3), this offers greater expressive power. The paper proposes using the Majorana sphere formalism to represent qutrit states, where states are represented as two pairs of points within the unit sphere, allowing all possible states to be generated from a one-parameter family of canonical states using SO(3) transformations. A general pure state is characterized by four degrees of freedom:

any pure state can be specified in its most general form with four degrees of freedom [14]

The qutrit generators are the eight Gell-Mann matrices, denoted by λi (i ∈ 1 to 8), which form a complete hermitian set of generators for the SU(3) symmetry group. The density matrix is represented as:

ρ(⃗n) = 1/3 (13×3 + √3 ⃗n ·⃗λ

Implementation with Qutrits and Novel Features

The adaptation of the 1P1Q scheme to qutrits requires replacing standard rotation gates with operators suited to the new paradigm, such as those defined by Eq. (9) based on Gell-Mann matrices. The initial encoding is replaced by the Majorana encoding scheme, which extends the input by two more parameters related to jet structure, introducing N-subjettiness parameters:

τN = 1/d0 X k pT,k min(∆R1,k, ∆R2,k,..., ∆RN,k), where k runs over the constituent particles in a given jet

The most effective combination of variables for qutrits is identified as those where the angles remain fixed according to Eq. (1c), with specific combinations for ϕ1 and ϕ2:

the combinations ϕ1 ∈ τ12, τ23, τ34, ϕ2 = ε and ϕ1 = ϱ0, ϕ2 = ϱz prove to yield the highest performance

Results and Performance Comparison

The results demonstrate that the qutrit-based model exhibits a greater capacity to discern between the three types of signals compared to qubit-based models. The discrimination capability was quantified using the Jensen–Shannon (JS) distance as a metric, showing larger distances between signals for qutrits:

the qutrit-based model ultimately exhibited larger distances between the signals, as expected and as illustrated in Table II

Furthermore, when trained on simulated data (JetClass), the AUC scores for QAE Qutrits (A, B, C, D) were comparable to or higher than those for QAE Qubits. The model was found to identify the decay signal of interest:

the t → bqq¯ decay is identified as the most anomalous signal by all models

Model Robustness and Future Work

The study validates the implementation by analyzing analytical calculations, such as recovering encoding angles through reverse processes involving QR decomposition and a second-degree Majorana polynomial. The performance, even with current simulator limitations in memory consumption, shows that qutrit-based models achieve similar performance equivalent to the qubit-based model - or in some cases even higher - has been achieved. Future research should involve testing the new encoding and model behavior with data from other LHC experiments like ATLAS and LHCb.

Improvements for AI systems

Based on the provided scientific paper, here are specific improvements that could be made to an AI system (specifically a Quantum Machine Learning model for anomaly detection) using qutrit architectures and Majorana representations:

  1. Improve data encoding efficiency and density:

  2. Enhance pattern discrimination between physical signals:

  3. Increase robustness against background noise and jet scale variations:

  4. Enable more complex kinematic feature extraction from particle data:

Improving the AI system using this paper can result in the following specific capabilities:

  1. A quantum-enhanced encoder will be able to represent jet kinematics (transversal momentum, azimuthal angle, pseudo-rapidity) using a compact set of qutrit states (Majorana encoding), allowing for a higher information density compared to qubit models, potentially reducing circuit depth and decoherence.

  2. The system will gain superior ability to distinguish between different physical processes—specifically separating signal events like top quark decays from dominant QCD background jets—by exploiting the richer feature space of qutrit states, leading to higher Jensen–Shannon (JS) distances between fidelity distributions for different signals.

  3. The model can be trained and perform better on real LHC data by incorporating multi-body kinematic variables (N-subjettiness parameters like τ2/τ1, τ3/τ2) and jet energy into the variational layer, making the detection of anomalous events more sensitive to subtle substructural differences.

  4. The system can be developed to provide a more complete kinematic description of particles by incorporating longitudinal and transverse impact parameters (d0, dz), offering geometric discrimination that is not encoded in standard four-momentum, leading to a more robust anomaly detection mechanism across both simulated and real data scenarios.

Abstract

The identification of anomalous events that are not explained by the Standard Model of particle physics, and the possible discovery of exotic physical phenomena, pose significant theoretical, experimental and computational challenges. The task will intensify at the High-Luminosity Large Hadron Collider and next-generation colliders, such as the proposed Future Circular Collider. Consequently, considerable challenges are expected concerning data processing, signal reconstruction, and analysis. This work explores the previously unstudied application of qutrit-based Quantum Machine Learning models for anomaly detection in high-energy physics data, with a focus on LHC scenarios. Motivated by the potential of higher-dimensional quantum systems (qudits) to enhance state space capacity and expressive power, we benchmark a qutrit-based quantum autoencoder against a standard qubit baseline to evaluate its feasibility and resource requirements. Our results show that the qutrit-based model achieves competitive anomaly detection performance and higher expressive power in terms of anomaly discrimination capacity, while requiring fewer physical units, thereby demonstrating the practical viability of Majorana-encoded ternary representations for high-energy physics data compression within current and near-future quantum frameworks.

Sources

Related papers