arXiv subjects

Maurizio Pierini

Publications and source records attributed to Maurizio Pierini.

At least 19 recordsLinked to original sources

arXiv2026

Scouting Super Resolution: from HLT-Level Reconstruction to Offline-like Quality

Data scouting, introduced by the CMS collaboration in 2011, is a technique to extend the reach of the experiment beyond what full-event storage bandwidth allows: rather than the complete event record, only the physics objects already reconstructed by the High-Level Trigger (HLT) are persisted, which makes it affordable to record O(10) times more data than the standard physics stream. The price for this rate increase is in experimental resolution: online reconstruction has typically a lower resolution and a worse response, mostly due to shortcuts taken to speed up the reconstruction and to the use of suboptimal calibration constants. For this reason, scouting data has mostly been used when the resolution loss is negligible in absolute terms (muons) or with respect to the offline resolution (jets). In any case, the need to derive specific online-to-offline corrections has been a limiting factor. We show that a regression trained on a given object and the surrounding particle cloud can bring the gap between online and offline resolution within the typical uncertainty band of the offline data. This procedure could be applied to the scouting monitoring stream, the subset of scouting data also reconstructed offline. We demonstrate this on CMS Run-1 Open Data, reprocessed through paired HLT and offline passes over the same events. One architecture, applied unchanged to electrons and jets, suppresses residual biases by one to two orders of magnitude on various quantities (kinematics, isolation, energy composition) and achieves a resolution improvement. Measured differentially in p_T, the corrected objects fall inside the offline calibration uncertainties CMS quotes for this dataset, where the uncorrected HLT objects lie outside them: a corrected object can be treated with the standard, centrally provided calibration, bringing scouting data much closer to offline data for practical purposes.

hep-ex↗

arXiv2026

SigmaScale: LLM Compression with SVD-based Low-Rank Decomposition and Learned Scaling Matrices

We present SigmaScale, a method for learning auxiliary scaling matrices $S$ to aid truncated Singular Value Decomposition (SVD) based Large Language Model (LLM) compression. Instead of deriving scaling matrices analytically, SigmaScale optimizes two sets of vectors that define diagonal row and column scaling transformations under an activation-aware compression loss. We show that learned scaling lowers the effective intrinsic rank of weight matrices, as reflected by reductions in effective-rank entropy, and that this reduction is strongly correlated with compression loss. Experiments on Llama 3.1 8B Instruct and Qwen3-8B show that SigmaScale is competitive with closely related state-of-the-art SVD-based compression methods across perplexity and zero-shot benchmarks. By using learned activation-aware transformations, SigmaScale explores a more flexible route to low-rank LLM compression by adapting to the structure of individual model weights. The advantage observed in specific tasks makes our approach a valid option for applications requiring a reduced LLM-inference computing cost.

cs.CL↗

arXiv2026

Quantum enhanced identification of boosted jets with quantum graph neural networks

We present a quantum enhanced tagger to identify jets with large Lorentz boost at colliders. For the first time, a convolutional quantum graph neural network (QGNN) is designed to discriminate boosted jets arising from hadronic decays of the Z boson, against those produced from gluons with large momentum. The network receives data without any physics-driven refinement, relying solely on the dimensionality reduction. The reduction is performed using a convolutional autoencoder whose performance is improved in the presence of added noise. The latent data are put into a graph format and fed to the QGNN of ten qubits. The autoencoder and the QGNN are trained separately, and simultaneously, and the resulting performances are compared with a classic algorithm based on graph networks. The findings indicate a strong potential of quantum graph networks to reproduce the performance of classical methods.

hep-ph↗

arXiv2026

Taming the Exponential: A Fast Softmax Surrogate for Integer-Native Edge Inference

Softmax can become a computational bottleneck in the Transformer model's Multi-Head Attention (MHA) block, particularly in small models under low-precision inference, where exponentiation and normalization incur significant overhead. As such, we suggest using Head-Calibrated Clipped-Linear Softmax (HCCS), a bounded, monotone surrogate to the exponential softmax function, which uses a clipped linear mapping of the max centered attention logits. This approximation produces a stable probability distribution, maintains the ordering of the original logits and has non-negative values. HCCS differs from previous softmax surrogates as it includes a set of lightweight calibration parameters that are optimized offline based on a representative dataset and calibrated for each individual attention head to preserve the statistical properties of the individual heads. We describe a hardware-motivated implementation of HCCS for high-throughput scenarios targeting the AMD Versal AI Engines. The current reference implementations from AMD for this platform rely upon either bfloat16 arithmetic or LUTs to perform the exponential operation, which might limit the throughput of the platform and fail to utilize the high-throughput integer vector processing units of the AI Engine. In contrast, HCCS provides a natural mapping to the AI Engines' int8 multiply accumulate (MAC) units. To the best of our knowledge, this is the first int8 optimized softmax surrogate for AMD AI engines that significantly exceeds the speed performance of other reference implementations while maintaining competitive task accuracy on small or heavily quantized MHA workloads after quantization-aware retraining.

cs.LG↗

arXiv2026

PQuantML: A Tool for End-to-End Hardware-aware Model Compression

PQuantML is a new open-source, hardware-aware neural network model compression library tailored to end-to-end workflows. Motivated by the need to deploy performant models to environments with strict latency constraints, PQuantML simplifies training of compressed models by providing a unified interface to apply pruning and quantization, either jointly or individually. The library implements multiple pruning methods with different granularities, as well as fixed-point quantization with support for High-Granularity Quantization. We evaluate PQuantML on representative tasks such as the jet substructure classification, so-called jet tagging, an on-edge problem related to real-time LHC data processing. Using various pruning methods with fixed-point quantization, PQuantML achieves substantial parameter and bit-width reductions while maintaining accuracy. The resulting compression is further compared against existing tools, such as QKeras and HGQ.

cs.LG↗

arXiv2026

CHLU: The Causal Hamiltonian Learning Unit as a Symplectic Primitive for Deep Learning

Current deep learning primitives dealing with temporal dynamics suffer from a fundamental dichotomy: they are either discrete and unstable (LSTMs) \citep{pascanu_difficulty_2013}, leading to exploding or vanishing gradients; or they are continuous and dissipative (Neural ODEs) \citep{dupont_augmented_2019}, which destroy information over time to ensure stability. We propose the \textbf{Causal Hamiltonian Learning Unit} (pronounced: \textit{clue}), a novel Physics-grounded computational learning primitive. By enforcing a Relativistic Hamiltonian structure and utilizing symplectic integration, a CHLU strictly conserves phase-space volume, as an attempt to solve the memory-stability trade-off. We show that the CHLU is designed for infinite-horizon stability, as well as controllable noise filtering. We then demonstrate a CHLU's generative ability using the MNIST dataset as a proof-of-principle.

cs.LG↗

arXiv2026

Towards Tensor Network Models for Low-Latency Jet Tagging on FPGAs

We present a systematic study of Tensor Network (TN) models $\unicode{x2013}$ Matrix Product States (MPS) and Tree Tensor Networks (TTN) $\unicode{x2013}$ for real-time jet tagging in high-energy physics, with a focus on low-latency deployment on Field Programmable Gate Arrays (FPGAs). Motivated by the strict requirements of the HL-LHC Level-1 trigger system, we explore TNs as compact and interpretable alternatives to deep neural networks. Using low-level jet constituent features, our models achieve competitive performance compared to state-of-the-art deep learning classifiers. We investigate post-training quantization to enable hardware-efficient implementations without degrading classification performance or latency. The best-performing models are synthesized to estimate FPGA resource usage, latency, and memory occupancy, demonstrating sub-microsecond latency and supporting the feasibility of online deployment in real-time trigger systems. Overall, this study highlights the potential of TN-based models for fast and resource-efficient inference in low-latency environments.

cs.LG↗

arXiv2026

AIE4ML: An End-to-End Framework for Compiling Neural Networks for the Next Generation of AMD AI Engines

Efficient AI inference on AMD's Versal AI Engine (AIE) is challenging due to tightly coupled VLIW execution, explicit datapaths, and local memory management. Prior work focused on first-generation AIE kernel optimizations, without tackling full neural network execution across the 2D array. In this work, we present AIE4ML, the first comprehensive framework for converting AI models automatically into optimized firmware targeting the AIE-ML generation devices, also with forward compatibility for the newer AIE-MLv2 architecture. At the single-kernel level, we attain performance close to the architectural peak. At the graph and system levels, we provide a structured parallelization method that can scale across the 2D AIE-ML fabric and exploit its dedicated memory tiles to stay entirely on-chip throughout the model execution. As a demonstration, we designed a generalized and highly efficient linear-layer implementation with intrinsic support for fused bias addition and ReLU activation. Also, as our framework necessitates the generation of multi-layer implementations, our approach systematically derives deterministic, compact, and topology-optimized placements tailored to the physical 2D grid of the device through a novel graph placement and search algorithm. Finally, the framework seamlessly accepts quantized models imported from high-level tools such as hls4ml or PyTorch while preserving bit-exactness. In layer scaling benchmarks, we achieve up to 98.6% efficiency relative to the single-kernel baseline, utilizing 296 of 304 AIE tiles (97.4%) of the device with entirely on-chip data movement. With evaluations across real-world model topologies, we demonstrate that AIE4ML delivers GPU-class throughput under microsecond latency constraints, making it a practical companion for ultra-low-latency environments such as trigger systems in particle physics experiments.

cs.LG↗

arXiv2025

Fast Jet Tagging with MLP-Mixers on FPGAs

We explore the innovative use of MLP-Mixer models for real-time jet tagging and establish their feasibility on resource-constrained hardware like FPGAs. MLP-Mixers excel in processing sequences of jet constituents, achieving state-of-the-art performance on datasets mimicking Large Hadron Collider conditions. By using advanced optimization techniques such as High-Granularity Quantization and Distributed Arithmetic, we achieve unprecedented efficiency. These models match or surpass the accuracy of previous architectures, reduce hardware resource usage by up to 97%, double the throughput, and half the latency. Additionally, non-permutation-invariant architectures enable smart feature prioritization and efficient FPGA deployment, setting a new benchmark for machine learning in real-time data processing at particle colliders.

physics.ins-det↗

arXiv2025

hls4ml: A Flexible, Open-Source Platform for Deep Learning Acceleration on Reconfigurable Hardware

We present hls4ml, a free and open-source platform that translates machine learning (ML) models from modern deep learning frameworks into high-level synthesis (HLS) code that can be integrated into full designs for field-programmable gate arrays (FPGAs) or application-specific integrated circuits (ASICs). With its flexible and modular design, hls4ml supports a large number of deep learning frameworks and can target HLS compilers from several vendors, including Vitis HLS, Intel oneAPI and Catapult HLS. Together with a wider eco-system for software-hardware co-design, hls4ml has enabled the acceleration of ML inference in a wide range of commercial and scientific applications where low latency, resource usage, and power consumption are critical. In this paper, we describe the structure and functionality of the hls4ml platform. The overarching design considerations for the generated HLS code are discussed, together with selected performance results.

cs.AR↗

arXiv2025

Knowledge is Overrated: A zero-knowledge machine learning and cryptographic hashing-based framework for verifiable, low latency inference at the LHC

Low latency event-selection (trigger) algorithms are essential components of Large Hadron Collider (LHC) operation. Modern machine learning (ML) models have shown great offline performance as classifiers and could improve trigger performance, thereby improving downstream physics analyses. However, inference on such large models does not satisfy the $40\text{MHz}$ online latency constraint at the LHC. In this work, we propose \texttt{PHAZE}, a novel framework built on cryptographic techniques like hashing and zero-knowledge machine learning (zkML) to achieve low latency inference, via a certifiable, early-exit mechanism from an arbitrarily large baseline model. We lay the foundations for such a framework to achieve nanosecond-order latency and discuss its inherent advantages, such as built-in anomaly detection, within the scope of LHC triggers, as well as its potential to enable a dynamic low-level trigger in the future.

hep-ex↗

arXiv2025

It's not a FAD: first results in using Flows for unsupervised Anomaly Detection at 40 MHz at the Large Hadron Collider

We present the first implementation of a Continuous Normalizing Flow (CNF) model for unsupervised anomaly detection within the realistic, high-rate environment of the Large Hadron Collider's L1 trigger systems. While CNFs typically define an anomaly score via a probabilistic likelihood, calculating this score requires solving an Ordinary Differential Equation, a procedure too complex for FPGA deployment. To overcome this, we propose a novel, hardware-friendly anomaly score defined as the squared norm of the model's vector field output. This score is based on the intuition that anomalous events require a larger transformation by the flow. Our model, trained via Flow Matching on Standard Model data, is synthesized for an FPGA using the hls4ml and da4ml libraries. We demonstrate that our approach effectively identifies a variety of beyond-the-Standard-Model signatures with performance comparable to existing machine learning-based triggers. The algorithm achieves a latency of a few hundred nanoseconds, or even less when using advanced quantization techniques, and requires minimal FPGA resources, establishing CNFs as a viable new tool for real-time, data-driven discovery at 40 MHz.

hep-ex↗

arXiv2025

$\texttt{HEPfit}$: a Code for the Combination of Indirect and Direct Constraints on High Energy Physics Models

$\texttt{HEPfit}$ is a flexible open-source tool which, given the Standard Model or any of its extensions, allows to $\textit{i)}$ fit the model parameters to a given set of experimental observables; $\textit{ii)}$ obtain predictions for observables. $\texttt{HEPfit}$ can be used either in Monte Carlo mode, to perform a Bayesian Markov Chain Monte Carlo analysis of a given model, or as a library, to obtain predictions of observables for a given point in the parameter space of the model, allowing $\texttt{HEPfit}$ to be used in any statistical framework. In the present version, around a thousand observables have been implemented in the Standard Model and in several new physics scenarios. In this paper, we describe the general structure of the code as well as models and observables implemented in the current release.

hep-ph↗

arXiv2025

Tensor Network for Anomaly Detection in the Latent Space of Proton Collision Events at the LHC

The pursuit of discovering new phenomena at the Large Hadron Collider (LHC) demands constant innovation in algorithms and technologies. Tensor networks are mathematical models on the intersection of classical and quantum machine learning, which present a promising and efficient alternative for tackling these challenges. In this work, we propose a tensor network-based strategy for anomaly detection at the LHC and demonstrate its superior performance in identifying new phenomena compared to established quantum methods. Our model is a parametrized Matrix Product State with an isometric feature map, processing a latent representation of simulated LHC data generated by an autoencoder. Our results highlight the potential of tensor networks to enhance new-physics discovery.

hep-ph↗

arXiv2025

tn4ml: Tensor Network Training and Customization for Machine Learning

Tensor Networks have emerged as a prominent alternative to neural networks for addressing Machine Learning challenges in foundational sciences, paving the way for their applications to real-life problems. This paper introduces tn4ml, a novel library designed to seamlessly integrate Tensor Networks into optimization pipelines for Machine Learning tasks. Inspired by existing Machine Learning frameworks, the library offers a user-friendly structure with modules for data embedding, objective function definition, and model training using diverse optimization strategies. We demonstrate its versatility through two examples: supervised learning on tabular data and unsupervised learning on an image dataset. Additionally, we analyze how customizing the parts of the Machine Learning pipeline for Tensor Networks influences performance metrics.

cs.LG↗

arXiv2025

Electroweak precision observables and Higgs-boson signal strengths in the Standard Model and beyond: present and future

We present results from a state-of-the-art fit of electroweak precision observables and Higgs-boson signal-strength measurements performed using 7 and 8 TeV data from the Large Hadron Collider. Based on the HEPfit package, our study updates the traditional fit of electroweak precision observables and extends it to include Higgs-boson measurements. As a result we obtain constraints on new physics corrections to both electroweak observables and Higgs-boson couplings. We present the projected accuracy of the fit taking into account the expected sensitivities at future colliders.

hep-ph↗

arXiv2024

Quantum anomaly detection in the latent space of proton collision events at the LHC

The ongoing quest to discover new phenomena at the LHC necessitates the continuous development of algorithms and technologies. Established approaches like machine learning, along with emerging technologies such as quantum computing show promise in the enhancement of experimental capabilities. In this work, we propose a strategy for anomaly detection tasks at the LHC based on unsupervised quantum machine learning, and demonstrate its effectiveness in identifying new phenomena. The designed quantum models, an unsupervised kernel machine and two clustering algorithms, are trained to detect new-physics events using a latent representation of LHC data, generated by an autoencoder designed to accommodate current quantum hardware limitations on problem size. For kernel-based anomaly detection, we implement an instance of the model on a quantum computer, and we identify a regime where it significantly outperforms its classical counterparts. We show that the observed performance enhancement is related to the quantum resources utilised by the model.

quant-ph↗

arXiv2024

Triggering Dark Showers with Conditional Dual Auto-Encoders

We present a family of conditional dual auto-encoders (CoDAEs) for generic and model-independent new physics searches at colliders. New physics signals, which arise from new types of particles and interactions, are considered in our study as anomalies causing deviations in data with respect to expected background events. In this work, we perform a normal-only anomaly detection, which employs only background samples, to search for manifestations of a dark version of strong force applying (variational) auto-encoders on raw detector images, which are large and highly sparse, without leveraging any physics-based pre-processing or strong assumption on the signals. The proposed CoDAE has a dual-encoder design, which is general and can learn an auxiliary yet compact latent space through spatial conditioning, showing a neat improvement over competitive physics-based baselines and related approaches, therefore also reducing the gap with fully supervised models. It is the first time an unsupervised model is shown to exhibit excellent discrimination against multiple dark shower models, illustrating the suitability of this method as an accurate, fast, model-independent algorithm to deploy, e.g., in the real-time event triggering systems of Large Hadron Collider experiments such as ATLAS and CMS.

hep-ex↗