arXiv subjects

Xiao-Yong Jin

Publications and source records attributed to Xiao-Yong Jin.

At least 19 recordsLinked to original sources

arXiv2025

Aurora: Architecting Argonne's First Exascale Supercomputer for Accelerated Scientific Discovery

Aurora is Argonne National Laboratory's pioneering Exascale supercomputer, designed to accelerate scientific discovery with cutting-edge architectural innovations. Key new technologies include the Intel(TM) Xeon(TM) Data Center GPU Max Series (code-named Sapphire Rapids) with support for High Bandwidth Memory (HBM), alongside the Intel(TM) Data Center GPU Max Series (code-named Ponte Vecchio) on each compute node. Aurora also integrates the Distributed Asynchronous Object Storage (DAOS), a novel exascale storage solution, and leverages Intel's oneAPI programming environment. This paper presents an in-depth exploration of Aurora's node architecture, the HPE Slingshot interconnect, the supporting software ecosystem, and DAOS. We provide insights into standard benchmark performance and applications readiness efforts via Aurora's Early Science Program and the Exascale Computing Project.

cs.DC↗

arXiv2025

Neural Field Transformations for Hybrid Monte Carlo: Architectural Design and Scaling

Critical slowing down, where autocorrelation grows rapidly near the continuum limit due to Hybrid Monte Carlo (HMC) moving through configuration space inefficiently, still challenges lattice gauge theory simulations. Combining neural field transformations with HMC (NTHMC) can reshape the energy landscape and accelerate sampling, but the choice of neural architectures has yet to be studied systematically. We evaluate NTHMC on a two-dimensional U(1) gauge theory, analyzing how it scales and transfers to larger volumes and smaller lattice spacing. Controlled comparisons let us isolate architectural contributions to sampling efficiency. Good designs can reduce autocorrelation and boost topological tunneling while maintaining favorable scaling. More broadly, our study highlights emerging design guides, such as wider receptive fields and channel-dependent activations, paving the way for systematic extensions to four-dimensional SU(3) gauge theory.

hep-lat↗

arXiv2024

Neural Network Gauge Field Transformation for 4D SU(3) gauge fields

We construct neural networks that work for any Lie group and maintain gauge covariance, enabling smooth, invertible gauge field transformations. We implement these transformations for 4D SU(3) lattice gauge fields and explore their use in HMC. We focus on developing loss functions and optimizing the transformations. We show the effects on HMC's molecular dynamics and discuss the scalability of the approach.

hep-lat↗

arXiv2023

MLMC: Machine Learning Monte Carlo for Lattice Gauge Theory

We present a trainable framework for efficiently generating gauge configurations, and discuss ongoing work in this direction. In particular, we consider the problem of sampling configurations from a 4D $SU(3)$ lattice gauge theory, and consider a generalized leapfrog integrator in the molecular dynamics update that can be trained to improve sampling efficiency. Code is available online at https://github.com/saforem2/l2hmc-qcd.

hep-lat↗

arXiv2023

Moving from continuous to discrete symmetry in the 2D XY model

We study the effects of discretization on the U(1) symmetric XY model in two dimensions using the Higher Order Tensor Renormalization Group (HOTRG) approach. Regarding the $Z_N$ symmetric clock models as specific discretizations of the XY model, we compare those discretizations to ones from truncations of the tensor network formulation of the XY model based on a character expansion, and focus on the differences in their phase structure at low temperatures. We also divide the tensor network formulations into core and interaction tensors and show that the core tensor has the dominant influence on the phase structure. Lastly, we examine a perturbed form of the XY model that continuously interpolates between the XY and clock models. We examine the behavior of the additional phase transition caused by the perturbation as the magnitude of perturbation is taken to zero. We find that this additional transition has a non-zero critical temperature as the perturbation vanishes, suggesting that even small perturbations can have a significant effect on the phase structure of the theory.

hep-lat↗

arXiv2022

Lattice QCD and Particle Physics

Contribution from the USQCD Collaboration to the Proceedings of the US Community Study on the Future of Particle Physics (Snowmass 2021).

hep-lat↗

arXiv2022

Neural Network Field Transformation and Its Application in HMC

We propose a generic construction of Lie group agnostic and gauge covariant neural networks, and introduce constraints to make the neural networks continuous differentiable and invertible. We combine such neural networks and build gauge field transformations that is suitable for Hybrid Monte Carlo (HMC). We use HMC to sample lattice gauge configurations in the transformed space by the neural network parameterized gauge field transformations. Tested with 2D U(1) pure gauge systems at a range of couplings and lattice sizes, compared with direct HMC sampling, the neural network transformed HMC (NTHMC) generates Markov chains of gauge configurations with improved tunneling of topological charges, while allowing less force calculations as the lattice coupling increases.

hep-lat↗

arXiv2022

Applications of Machine Learning to Lattice Quantum Field Theory

There is great potential to apply machine learning in the area of numerical lattice quantum field theory, but full exploitation of that potential will require new strategies. In this white paper for the Snowmass community planning process, we discuss the unique requirements of machine learning for lattice quantum field theory research and outline what is needed to enable exploration and deployment of this approach in the future.

hep-lat↗

arXiv2022

LeapfrogLayers: A Trainable Framework for Effective Topological Sampling

We introduce LeapfrogLayers, an invertible neural network architecture that can be trained to efficiently sample the topology of a 2D $U(1)$ lattice gauge theory. We show an improvement in the integrated autocorrelation time of the topological charge when compared with traditional HMC, and look at how different quantities transform under our model. Our implementation is open source, and is publicly available on github at https://github.com/saforem2/l2hmc-qcd.

hep-lat↗

arXiv2021

HMC with Normalizing Flows

We propose using Normalizing Flows as a trainable kernel within the molecular dynamics update of Hamiltonian Monte Carlo (HMC). By learning (invertible) transformations that simplify our dynamics, we can outperform traditional methods at generating independent configurations. We show that, using a carefully constructed network architecture, our approach can be easily scaled to large lattice volumes with minimal retraining effort. The source code for our implementation is publicly available online at https://github.com/nftqcd/fthmc.

cs.LG↗

arXiv2021

Deep Learning Hamiltonian Monte Carlo

We generalize the Hamiltonian Monte Carlo algorithm with a stack of neural network layers and evaluate its ability to sample from different topologies in a two dimensional lattice gauge theory. We demonstrate that our model is able to successfully mix between modes of different topologies, significantly reducing the computational cost required to generated independent gauge field configurations. Our implementation is available at https://github.com/saforem2/l2hmc-qcd .

hep-lat↗

arXiv2021

Near-conformal dynamics in a chirally broken system

Composite Higgs models must exhibit very different dynamics from quantum chromodynamics (QCD) regardless whether they describe the Higgs boson as a dilatonlike state or a pseudo-Nambu-Goldstone boson. Large separation of scales and large anomalous dimensions are frequently desired by phenomenological models. Mass-split systems are well-suited for composite Higgs models because they are governed by a conformal fixed point in the ultraviolet but are chirally broken in the infrared. In this work we use lattice field theory calculations with domain wall fermions to investigate a system with four light and six heavy flavors. We demonstrate how a nearby conformal fixed point affects the properties of the four light flavors that exhibit chiral symmetry breaking in the infrared. Specifically we describe hyperscaling of dimensionful physical quantities and determine the corresponding anomalous mass dimension. We obtain $y_m=1+γ^*= 1.47(5)$ suggesting that $N_f=10$ lies inside the conformal window. Comparing the low energy spectrum to predictions of dilaton chiral perturbation theory, we observe excellent agreement which supports the expectation that the 4+6 mass-split system exhibits near-conformal dynamics with a relatively light $0^{++}$ isosinglet scalar.

hep-ph↗

arXiv2019

Ensemble Quasi-Newton HMC

We present a modification of the Hybrid Monte Carlo algorithm for tackling the critical slowing down of generating Markov chains of lattice gauge configurations towards the continuum limit. We propose a new method to exchange information within an ensemble of Markov chains, and use it to construct an approximate inverse Hessian matrix of the action inspired from quasi-Newton algorithms for optimization. The kinetic term of the molecular dynamics evolution includes the approximate Hessian for long distance couplings among the momenta. We show the result of applying the new algorithm to the $U(1)$ gauge theory in two dimensions, and discuss our future plans.

hep-lat↗

arXiv2019

Nonperturbative investigations of SU(3) gauge theory with eight dynamical flavors

We present our lattice studies of SU(3) gauge theory with $N_f$ = 8 degenerate fermions in the fundamental representation. Using nHYP-smeared staggered fermions we study finite-temperature transitions on lattice volumes as large as $L^3 \times N_t = 48^3 \times 24$, and the zero-temperature composite spectrum on lattice volumes up to $64^3 \times 128$. The spectrum indirectly indicates spontaneous chiral symmetry breaking, but finite-temperature transitions with fixed $N_t \leq 24$ enter a strongly coupled lattice phase as the fermion mass decreases, which prevents a direct confirmation of spontaneous chiral symmetry breaking in the chiral limit. In addition to the connected spectrum we focus on the lightest flavor-singlet scalar particle. We find it to be degenerate with the pseudo-Goldstone states down to the lightest masses reached so far by non-perturbative lattice calculations. Using the same lattice approach, we study the behavior of the composite spectrum when the number of light fermions is changed from eight to four. A heavy flavor-singlet scalar in the 4-flavor theory affirms the contrast between QCD-like dynamics and the low-energy behavior of the 8-flavor theory.

hep-lat↗

arXiv2017

Continuum extrapolation of critical point for finite temperature QCD with Nf=3

We study the critical point for finite temperature Nf=3 QCD using several temporal lattice sizes up to 10. In the study, the Iwasaki gauge action and non-perturbatively O(a) improved Wilson fermions are employed. We estimate the critical temperature and the upper bound of the critical pseudo-scalar meson mass.

hep-lat↗

arXiv2017

Critical point phase transition for finite temperature 3-flavor QCD with non-perturbatively O($a$) improved Wilson fermions at $N_{\rm t}=10$

We study the finite temperature phase structure for three-flavor QCD with a focus on locating the critical point which separates crossover and first order phase transition region in the chiral regime of the Columbia plot. In this study, we employ the Iwasaki gauge action and the non-perturvatively O($a$) improved Wilson-Clover fermion action. We discuss the finite size scaling analysis including the mixing of magnetization-like and energy-like observables. We carry out the continuum extrapolation of the critical point using newly generated data at $N_{\rm t}=8$, $10$ and estimate the upper bound of the critical pseudo-scalar meson mass $m_{\rm PS,E} \lesssim 170 {\rm MeV}$ and the critical temperature $T_{\rm E}=134(3){\rm MeV}$. Our estimate of the upper bound is derived from the existence of the critical point as an edge of the 1st order phase transition while that of the staggered-type fermions is based on its absence.

hep-lat↗

arXiv2016

Update on Nf=3 finite temperature QCD phase structure with Wilson-Clover fermion action

We present an update of the finite temperature phase structure analysis for three flavor QCD. In the study the Iwasaki gauge action and non-perturvatively O($a$) improved Wilson-Clover fermion action are employed. We discuss finite size scaling analysis including mixings of magnetization-like and energy-like observables. Preliminary results are shown of the continuum limit of the critical point using newly generated data at Nt=8,10, including estimates of the critical pseudo-scalar meson mass and critical temperature.

hep-lat↗

arXiv2016

QEX: a framework for lattice field theories

We present a new software framework for simulating lattice field theories. It features an intuitive programming interface, while simultaneously achieving high performance supercomputing, all in one programming language, Nim. With a macro system based on its abstract syntax tree, the language enables us to check and optimize our code at compile time. It also allows us to code intrinsics that map directly to machine instructions, and generates efficient native code. We show how we use Nim's metaprogramming features in our code, and present the current status of the code and future plans.

hep-lat↗