high energy physics
PulseAugur coverage of high energy physics — every cluster mentioning high energy physics across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
High-energy physics concepts will be increasingly applied to explain deep learning architectures
Recent evidence shows high-energy physics (HEP) theories like lattice gauge theory and effective field theory being used to frame and analyze deep neural networks. This suggests a growing trend where HEP concepts provide novel theoretical underpinnings for understanding complex AI models, potentially leading to new architectures or training methodologies inspired by physics.
Pairton framework shows promise for advanced particle reconstruction in HEP
The development of the Pairton framework, which models particle reconstruction as a masked prediction process on graph structures, represents a significant advancement. Its success in reconstructing $t ar{t}$ decays indicates a strong potential for improving the accuracy and efficiency of particle identification and analysis in future high-energy physics experiments.
KANs will be adopted for detailed comparative analysis of physics simulation tools
The successful application of additive Kolmogorov-Arnold Networks (KANs) to dissect differences between Pythia and Herwig event generators highlights their utility for fine-grained analysis. This suggests KANs could become a standard tool for understanding discrepancies and validating other complex physics simulation software.
-
Self-supervised learning enhances resonance mass regression in physics
Researchers have developed a self-supervised learning approach using a Transformer encoder pre-trained with VICReg to improve resonance mass regression in high energy physics. This method aims to overcome the limitation…
-
AI tools act as microscopes for human thought processes, enabling self-observation
The author proposes that AI tools, particularly those that record interactions like Claude Code, serve as powerful microscopes for observing one's own thinking processes. This method, akin to the scientific method of ob…
-
Deep learning model accelerates optimal transport calculations for physics data
Researchers have developed a novel deep learning model, the Metric-Aware Particle Flow Network, designed to approximate optimal transport calculations for complex datasets. This model, built using Deep Sets architecture…
-
Deep Neural Networks Framed as Lattice Gauge Theories in Physics Research
Researchers have developed a novel framework that conceptualizes deep neural networks as lattice gauge theories, drawing parallels to high energy physics. This approach modifies existing NN/QFT duality to incorporate th…
-
Physics theory applied to diffusion models for information spread analysis
Researchers have applied principles from effective field theory, a tool from high energy physics, to analyze score-matching diffusion models with convolutional architectures. This approach treats the denoising process a…
-
New KAN method anatomizes Pythia-Herwig differences in physics event generation
Researchers have developed a new method using additive Kolmogorov-Arnold Networks (KANs) to analyze the differences between high-energy physics event generators like Pythia and Herwig. This approach allows for a staged …
-
Neural networks construct spinning conformal fields, recovering Maxwell CFT
Researchers have developed a method to construct spinning conformal fields using neural networks and the embedding formalism. This approach allows for the computation of two-, three-, and four-point functions, building …
-
Pairton framework advances particle reconstruction in high-energy physics
Researchers have developed Pairton, a novel iterative framework designed for reconstructing short-lived particles in high-energy collision events. This method models particle reconstruction as a masked prediction proces…
-
Symbolic regression automates function discovery for high energy physics data
Researchers have developed a new method using symbolic regression to automatically discover parametric functions for modeling high energy physics (HEP) data. This approach automates the previously manual and intuitive p…
-
Quantum Autoencoders Accelerated on FPGAs for Real-Time Anomaly Detection
Researchers have developed a method to accelerate quantum autoencoder models for real-time anomaly detection in high energy physics experiments. These models, capable of processing complex collider data, were synthesize…
-
Machine Learning applied to high-energy physics fits in new lecture notes
Researchers have developed new lecture notes detailing the application of Machine Learning (ML) surrogates for statistical fits in high-energy physics. These notes outline a comprehensive ML workflow, including the use …
-
New AI agent automates complex lattice QCD research workflows
Researchers have developed LQCDMaster, an agentic scientific computing tool designed to automate lattice quantum chromodynamics (LQCD) research. This system converts natural-language research tasks into executable PyQUD…
-
Machine learning framework enhances parameter inference in physics and cosmology
Researchers have developed a new machine learning framework to emulate complex likelihood landscapes in high energy physics and cosmology. This framework utilizes XGBoost to efficiently explore high-dimensional paramete…
-
Machine learning enhances data reconstruction for silicon sensors in high energy physics
Researchers have developed machine learning techniques to improve data reconstruction and compression for resistive silicon sensors used in high energy physics. The study explores recurrent neural networks, specifically…
-
Graph Neural Networks applied to optimization and physics problems · 2 sources tracked
Researchers are exploring the application of graph neural networks (GNNs) beyond their traditional roles in combinatorial optimization and theoretical physics. One study demonstrates that GNNs can function as effective …
-
Factorizable Normalizing Flows introduced for parameter-dependent density morphing · 2 sources tracked
Researchers have introduced Factorizable Normalizing Flows (FNFs), a novel method designed to model how probability densities change with continuous parameters. This approach addresses the intractability of learning sep…
-
HEPTv2 Transformer Achieves State-of-the-Art in Particle Reconstruction
Researchers have developed HEPTv2, an end-to-end point-transformer architecture designed for efficient charged particle reconstruction in high-energy physics. This new model bypasses traditional graph construction and a…
-
Transfer learning boosts AI model efficiency in high-energy physics
Researchers have explored transfer learning techniques to improve machine learning model performance in high-energy physics. By pre-training models on computationally cheaper, fast-simulated data and then adapting them …
-
Neural networks in physics are vulnerable to hidden systematic errors
Researchers have identified a significant vulnerability in neural network models used for high-energy physics analyses. These models, while powerful, can be systematically misled by subtle input perturbations that remai…
-
New frameworks enable asynchronous human-AI collaboration in complex scientific workflows
Researchers have developed HepScript, a Domain-Specific Language (DSL) designed to facilitate human-AI collaboration in high-energy physics data analysis. This DSL abstracts complex analysis logic into a formal syntax t…