Mia Liu

This study introduces a novel transformer model optimized for large-scale point cloud processing in scientific domains such as high-energy physics (HEP) and astrophysics. Addressing the limitations of graph neural networks and standard transformers, our model integrates local inductive bias and achieves near-linear complexity with hardware-friendly regular operations. One contribution of this work is the quantitative analysis of the error-complexity tradeoff of various sparsification techniques for building efficient transformers. Our findings highlight the superiority of using locality-sensitive hashing (LSH), especially OR \& AND-construction LSH, in kernel approximation for large-scale point cloud data with local inductive bias. Based on this finding, we propose LSH-based Efficient Point Transformer (\textbf{HEPT}), which combines ELSH with OR \& AND constructions and is built upon regular computations. HEPT demonstrates remarkable performance in two critical yet time-consuming HEP tasks, significantly outperforming existing GNNs and transformers in accuracy and computational speed, marking a significant advancement in geometric deep learning and large-scale scientific data processing. Our code is available at \url{https://github.com/Graph-COM/HEPT}.

We will present a program to establish the first development and manufacturing of HEP-specific sensors monolithically integrated into a standard CMOS process using a US-based foundry. In collaboration with several US universities the project aims to develop Monolithic Active Pixel Sensors (MAPS) designs implemented in the 90 nm technology node, including simple test structures and multi-pixel arrays, and monolithic CMOS sensors with readout integrated circuits, perform detailed characterization of the detector prototypes and quantify their performance for HEP applications.

A search for pair production of doubly charged Higgs bosons (), each decaying into a pair of prompt, isolated, and highly energetic leptons with the same electric charge, is presented. The search uses a proton–proton collision data sample at a centre-of-mass energy of 13 TeV corresponding to an integrated luminosity of 139 fb recorded by the ATLAS detector during Run 2 of the Large Hadron Collider (LHC). This analysis focuses on same-charge leptonic decays, where , in two-, three-, and four-lepton channels, but only considers final states which include electrons or muons. No evidence of a signal is observed. Corresponding upper limits on the production cross-section of a doubly charged Higgs boson are derived, as a function of its mass , at 95% confidence level. Assuming that the branching ratios to each of the possible leptonic final states are equal, B(H±±→e±e±)=B(H±±→e …

Frontiers | Corrigendum: Applications and techniques for fast machine learning in science Skip to main content Download Article Download Article Download PDF ReadCube EPUB XML (NLM) Share on Export citation EndNote Reference Manager Simple TEXT file BibTex View article impact View altmetric score SHARE ON TABLE OF CONTENTS Correction Publisher's note Export citation EndNote Reference Manager Simple TEXT file BibTex Check for updates People also looked at CORRECTION article Front. Big Data, 16 October 2023 Sec. Big Data and AI in High Energy Physics Volume 6 - 2023 | https://doi.org/10.3389/fdata.2023.1301942 Corrigendum: Applications and techniques for fast machine learning in science Allison McCarn Deiana 1 * Nhan Tran 2,3 * Joshua Agar 4 Michaela Blott 5 Giuseppe Di Guglielmo 6 Javier Duarte 7 Philip Harris 8 Scott Hauck 9 Mia Liu 10 Mark S. Neubauer 11 Jennifer …

Modern large-scale physics experiments create datasets with sizes and streaming rates that can exceed those from industry leaders such as Google Cloud and Netflix. Fully processing these datasets requires both sufficient compute power and efficient workflows. Recent advances in Machine Learning (ML) and Artificial Intelligence (AI) can either improve or replace existing domain-specific algorithms to increase workflow efficiency. Not only can these algorithms improve the physics performance of current algorithms, but they can often be executed more quickly, especially when run on coprocessors such as GPUs or FPGAs. In the winter of 2023, MIT hosted the Accelerating Physics with ML at MIT workshop, which brought together researchers from gravitational-wave physics, multi-messenger astrophysics, and particle physics to discuss and share current efforts to integrate ML tools into their workflows. The following white paper highlights examples of algorithms and computing frameworks discussed during this workshop and summarizes the expected computing needs for the immediate future of the involved fields.

The HL-LHC run is anticipated to start at the end of this decade and will pose a significant challenge for the scale of the HEP software and computing infrastructure. The mission of the US CMS Software and Computing Operations Program is to develop and operate the software and computing resources necessary to process CMS data expeditiously and to enable US physicists to fully participate in the physics of CMS. We have developed a strategic plan to prioritize R and D efforts to reach this goal for the HL-LHC. This plan includes four grand challenges modernizing physics software and improving algorithms, building infrastructure for exabyte-scale datasets, transforming the scientific data analysis process and transitioning from R and D to operations. We are involved in a variety of R and D projects that fall within these grand challenges. In this talk, we will introduce our four grand challenges and outline the R and D program of the US CMS Software and Computing Operations Program.

The high instantaneous luminosity of the CERN Large Hadron Collider leads to multiple proton–proton interactions in the same or nearby bunch crossings (pileup). Advanced pileup mitigation algorithms are designed to remove this noise from pileup particles and improve the performance of crucial physics observables. This study implements a semi-supervised graph neural network for particle-level pileup noise removal, by identifying individual particles produced from pileup. The graph neural network is firstly trained on charged particles with known labels, which can be obtained from detector measurements on data or simulation, and then inferred on neutral particles for which such labels are missing. This semi-supervised approach does not depend on the neutral particle pileup label information from simulation, and thus allows us to perform training directly on experimental data. The performance of this approach …

This paper presents the muon momentum calibration and performance studies for the ATLAS detector based on the pp collisions data sample produced at  = 13 TeV at the LHC during Run 2 and corresponding to an integrated luminosity of 139 . An innovative approach is used to correct for potential charge-dependent momentum biases related to the knowledge of the detector geometry, using the resonance. The muon momentum scale and resolution are measured using samples of and events. A calibration procedure is defined and applied to simulated data to match the performance measured in real data. The calibration is validated using an independent sample of events. At the Z peak, the momentum scale is measured with an uncertainty at the 0.05% (0.1%) level, and the resolution is measured with an uncertainty at the 1.5% (2%) level. The charge-dependent …