An official website of the United States government
Quantum generative models hold the promise of accelerating or improving machine learning tasks by leveraging the probabilistic nature of quantum states, but the successful optimization of these models remains a difficult challenge. To tackle this challenge, we present a new architecture for quantum generative modeling that combines insights from classical machine learning and quantum phases of matter. In particular, our model utilizes both many-body localized (MBL) dynamics and hidden units to improve the optimization of the model. We demonstrate the applicability of our model on a diverse set of classical and quantum tasks, including a toy version of MNIST handwritten digits, quantum data obtained from quantum many-body states, and nonlocal parity data. Our architecture and algorithm provide novel strategies of utilizing quantum many-body systems as learning resources and reveal a powerful connection between disorder, interaction, and learning in quantum many-body systems. Published by the American Physical Society2024
more »
« less
This content will become publicly available on February 21, 2026
Resimulation-based self-supervised learning for pretraining physics foundation models
Self-supervised learning (SSL) is at the core of training modern large machine learning models, providing a scheme for learning powerful representations that can be used in a variety of downstream tasks. However, SSL strategies must be adapted to the type of training data and downstream tasks required. We propose resimulation-based self-supervised representation learning (RS3L), a novel simulation-based SSL strategy that employs a method of to drive data augmentation for contrastive learning in the physical sciences, particularly, in fields that rely on stochastic simulators. By intervening in the middle of the simulation process and rerunning simulation components downstream of the intervention, we generate multiple realizations of an event, thus producing a set of augmentations covering all physics-driven variations available in the simulator. Using experiments from high-energy physics, we explore how this strategy may enable the development of a foundation model; we show how RS3L pretraining enables powerful performance in downstream tasks such as discrimination of a variety of objects and uncertainty mitigation. In addition to our results, we make the RS3L dataset publicly available for further studies on how to improve SSL strategies. Published by the American Physical Society2025
more »
« less
- Award ID(s):
- 2019786
- PAR ID:
- 10588327
- Publisher / Repository:
- American Physical Society
- Date Published:
- Journal Name:
- Physical Review D
- Volume:
- 111
- Issue:
- 3
- ISSN:
- 2470-0010
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Understanding quantum materials—solids in which interactions among constituent electrons yield a great variety of novel emergent quantum phenomena—is a forefront challenge in modern condensed matter physics. This goal has driven the invention and refinement of several experimental methods, which can spectroscopically determine the elementary excitations and correlation functions that determine material properties. Here we focus on the future experimental and theoretical trends of resonant inelastic x-ray scattering (RIXS), which is a remarkably versatile and rapidly growing technique for probing different charge, lattice, spin, and orbital excitations in quantum materials. We provide a forward-looking introduction to RIXS and outline how this technique is poised to deepen our insight into the nature of quantum materials and of their emergent electronic phenomena. Published by the American Physical Society2024more » « less
-
Heavy-flavor hadron production, in particular bottom hadron production, is difficult to study in deep-inelastic scattering (DIS) experiments due to small production rates and branching fractions. To overcome these limitations, a method for identifying heavy-flavor DIS events based on event topology is proposed. Based on a heavy-flavor jet tagging strategy developed for the LHCb experiment, this algorithm uses displaced vertices to identify decays of heavy-flavor hadrons. The algorithm’s performance at the Electron-Ion Collider is demonstrated using simulation, and it is shown to provide discovery potential for nonperturbative intrinsic bottom quarks in the proton. Published by the American Physical Society2024more » « less
-
Though some LHC searches for new physics exceed the TeV scale, there may be discoveries waiting to be made at much lower masses. We outline a simple quirk model, motivated by models that address the hierarchy problem through neutral naturalness, in which new electroweakly charged states with masses as low as 100 GeV have not yet been probed by the LHC. We also describe a novel search strategy which is complementary to current search methods. In particular, we show its potential to discover natural quirks over regions of parameter space that present methods will leave unexplored, even after the LHC’s high-luminosity run. Published by the American Physical Society2025more » « less
