Searcharxiv⌕ Search

arXiv · 2611.17354

Hybrid Quantum-Classical Sequence Models for Sample-Efficient Energy Regression in Simulated Calorimeter Showers

Abstract

Energy reconstruction from electromagnetic calorimeter data underpins particle identification, event reconstruction and precision measurements in high-energy physics. Machine learning already performs well on this task, yet two questions remain open: how much the choice of data representation matters, and whether hybrid quantum-classical models add value in practice. We compare classical and quantum-enhanced models for incident positron energy reconstruction on the CaloGAN dataset, a public benchmark of 100,000 simulated electromagnetic shower events spanning 1 to 100 GeV. Eight architectures, namely SVR, ANN, CNN and LSTM together with their quantum counterparts, are trained on three calorimeter representations: engineered physical features, calorimeter images, and longitudinal shower sequences. Models that follow the longitudinal development of the shower reach the highest accuracy, which points to the value of physically informed representations. The hybrid QLSTM gives the lowest reconstruction error, with an RMSE of 0.294 GeV. Under matched-data conditions with only 1,500 training events, the QLSTM improves on the classical LSTM by 52.7% and also surpasses an LSTM trained on the full set. These observations indicate that longitudinal representations are central to calorimeter energy reconstruction and give preliminary evidence that hybrid quantum-classical sequence models can be more sample-efficient for regression in high-energy physics.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Tariq Mahmood, Muhammad Ahsan, Talab Hussain, Grecia Itzel Colín-González, Alfredo Raya, Muhammad Awais Rafique, HassabAlla M. A. Mahmoud. 2026-10-11. Hybrid Quantum-Classical Sequence Models for Sample-Efficient Energy Regression in Simulated Calorimeter Showers. https://searcharxiv.com/abs/2611.17354

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

Improved Pion-Kaon Identification in Heavy-Ion Collisions with a Two-Dimensional Transformation

Accurate identification of charged pions and kaons is essential for precision measurements in relativistic heavy-ion collisions, but becomes increasingly challenging at intermediate and high transverse momentum because of the increasing overlap of the TPC and TOF particle identification responses. In this work, we quantitatively benchmark a two-dimensional shift-and-rotation PID procedure that combines the $nσ_π$ and mass-squared $(m^{2})$ information. The benchmark is performed using Au+Au collision events generated with the AMPT model, where the detector response is modeled using $p_{T}$-dependent PID parameters extracted from STAR data. Since the true particle identities and input yields are known in AMPT but are not used during the reconstruction, the procedure can be evaluated through a direct closure test that is not possible with experimental data alone. The two-dimensional procedure is compared with a reference one-dimensional method consisting of an $|nσ_π|<$ 3 preselection followed by a Gaussian fit to the $m^{2}$ distribution. While both methods perform well when pion and kaon PID responses are well separated, the two-dimensional procedure provides improved yield closure as the PID distributions increasingly overlap at high $p_{T}$. The reconstructed-to-input yield ratio remains within 2\% of unity up to $p_T =$ 3 GeV/$c$, whereas the reference one-dimensional procedure shows a larger deviation. Reaction-plane $v_{2}$ closure tests show that both the reference one-dimensional and two-dimensional PID procedures reproduce the AMPT input $v_{2}$ within the statistical uncertainties over the studied $p_{T}$ range. These results demonstrate that the two-dimensional procedure substantially improves the high-$p_{T}$ yield reconstruction without introducing a detectable bias in the extracted $v_{2}$ under the STAR-like detector response considered here.

physics.ins-det↗

Cryogenic Mechanical Loss of GaAs/AlGaAs Crystalline Coatings

Optical coatings with minimal thermal noise are essential for high-precision, laser-based metrology experiments. Substrate-transferred crystalline GaAs/Al$_{0.92}$Ga$_{0.08}$As coatings have the lowest thermal noise for high-reflectivity mirror coatings with $λ=900$ nm -- 12 $μ$m. Time-frequency stabilization experiments have demonstrated that GaAs/AlGaAs mirror coatings have a lower noise floor than amorphous mirror coatings. These mirrors are now being developed at sizes $\ge 300$ mm for gravitational-wave detectors. In order to understand the coating's fundamental loss mechanisms, which cause thermal noise, we have performed the first measurement of the mechanical loss at cryogenic temperatures using a cryogenic, multimodal, gentle nodal suspension system, which operates from 12 K to room temperature. This initial measurement used a silicon substrate whose asymmetry produced excess friction with the nodal support. Accounting for this friction and Akhiezer loss in the substrate and coating, the mechanical loss agrees well with the thermal noise observed in experiments using cryogenic fixed-cavities with GaAs/AlGaAs mirrors.

physics.ins-det↗

Rotating Capacitive System for Measurement of a Third Order Gravity Gradient Based on a Single-Axis Mechanical Oscillator. Part I, Theory

This paper is a feasibility study on a single-axis mechanical oscillator, mounted on a miniature circular rotating platform, that is proposed to be used for measuring a quantity proportional to the second vertical derivative of gravitational acceleration projected onto the oscillator's sensitivity axis. The latter is aligned with the axis of rotation. Part I establishes theoretical feasibility only. Design constraints, quantitative sensitivity, error budgets, and mobile platform performance will be addressed in Part II.

physics.ins-det↗