arXiv subjects

Jeongnim Kim

Publications and source records attributed to Jeongnim Kim.

At least 19 recordsLinked to original sources

arXiv2026

Toward a Unified GPU-Aware OpenSHMEM Specification

Leadership-class HPC systems are now accelerator-centric, with GPUs providing most floating-point throughput and memory bandwidth. As next-generation systems increasingly integrate accelerators through high-speed memory fabrics and system interconnects, exposing larger tightly coupled device domains, \ac{PGAS} models such as OpenSHMEM provide a natural abstraction for expressing fine-grained remote memory operations across these devices. While OpenSHMEM 1.x offers a lean PGAS model for irregular communication, atomics, fine-grained synchronization, and collectives, its memory model lacks portable semantics for accelerator architectures. As a result, existing GPU-enabled OpenSHMEM implementations differ in memory management, capability discovery, and operation semantics, limiting portability and ecosystem cohesion. This risks fracturing the community that OpenSHMEM was originally created to unify. This paper proposes an OpenSHMEM Auxiliary Specification for GPU-Aware Communication, designed as a lightweight, backward-compatible extension to OpenSHMEM 1.x. The auxiliary specification introduces a minimal memory model extension via a GPU-scoped memory space abstraction, along with capability queries and well-defined semantics for using \acs{GPU}-attached buffers in RMA, atomic, synchronization, and collective operations. This is initially conceived through the lens of a host-initiated interface, although it provides a general set of semantics that also allow for optional device-initiated support. A central goal of this effort is to demonstrate that GPU-aware OpenSHMEM semantics can be specified and implemented across GPUs from multiple vendors, providing a practical and rapidly implementable step toward unification under a vendor-neutral specification while informing the design of future OpenSHMEM specifications.

cs.DC↗

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

Pushing the Accuracy Limit of Foundation Neural Network Models with Quantum Monte Carlo Forces and Path Integrals

We propose an end-to-end integrated strategy to produce highly accurate quantum chemistry (QC) synthetic datasets (energies and forces) aimed at deriving Foundation Machine Learning models for molecular simulation. Starting from Density Functional Theory (DFT), a "Jacob's Ladder" approach leverages computationally-optimized layers of massively GPU-accelerated software with increasing accuracy. Thanks to Exascale, this is the first time that the computationally intensive calculation of Quantum Monte Carlo forces (QMC), and the combination of multi-determinant QMC energies and forces with selected-Configuration Interaction wavefunctions, are computed at such scale at the complete basis-set limit. To bridge the gap between accurate QC and condensed-phase Molecular Dynamics, we leverage transfer learning to improve the DFT-based FeNNix-Bio1 foundation model. The resulting approach is coupled to path integrals adaptive sampling quantum dynamics to perform nanosecond reactive simulations at unprecedented accuracy. These results demonstrate the promise of Exascale to deepen our understanding of the inner machinery of complex biosystems.

physics.chem-ph↗

arXiv2025

Performance Analysis of HPC applications on the Aurora Supercomputer: Exploring the Impact of HBM-Enabled Intel Xeon Max CPUs

The Aurora supercomputer is an exascale-class system designed to tackle some of the most demanding computational workloads. Equipped with both High Bandwidth Memory (HBM) and DDR memory, it provides unique trade-offs in performance, latency, and capacity. This paper presents a comprehensive analysis of the memory systems on the Aurora supercomputer, with a focus on evaluating the trade-offs between HBM and DDR memory systems. We explore how different memory configurations, including memory modes (Flat and Cache) and clustering modes (Quad and SNC4), influence key system performance metrics such as memory bandwidth, latency, CPU-GPU PCIe bandwidth, and MPI communication bandwidth. Additionally, we examine the performance of three representative HPC applications -- HACC, QMCPACK, and BFS -- each illustrating the impact of memory configurations on performance. By using microbenchmarks and application-level analysis, we provide insights into how to select the optimal memory system and configuration to maximize performance based on the application characteristics. The findings presented in this paper offer guidance for users of the Aurora system and similar exascale systems.

cs.DC↗

arXiv2024

Distributed Ranges: A Model for Distributed Data Structures, Algorithms, and Views

Data structures and algorithms are essential building blocks for programs, and \emph{distributed data structures}, which automatically partition data across multiple memory locales, are essential to writing high-level parallel programs. While many projects have designed and implemented C++ distributed data structures and algorithms, there has not been widespread adoption of an interoperable model allowing algorithms and data structures from different libraries to work together. This paper introduces distributed ranges, which is a model for building generic data structures, views, and algorithms. A distributed range extends a C++ range, which is an iterable sequence of values, with a concept of segmentation, thus exposing how the distributed range is partitioned over multiple memory locales. Distributed data structures provide this distributed range interface, which allows them to be used with a collection of generic algorithms implemented using the distributed range interface. The modular nature of the model allows for the straightforward implementation of \textit{distributed views}, which are lightweight objects that provide a lazily evaluated view of another range. Views can be composed together recursively and combined with algorithms to implement computational kernels using efficient, flexible, and high-level standard C++ primitives. We evaluate the distributed ranges model by implementing a set of standard concepts and views as well as two execution runtimes, a multi-node, MPI-based runtime and a single-process, multi-GPU runtime. We demonstrate that high-level algorithms implemented using generic, high-level distributed ranges can achieve performance competitive with highly-tuned, expert-written code.

cs.DC↗

arXiv2018

QMCPACK : An open source ab initio Quantum Monte Carlo package for the electronic structure of atoms, molecules, and solids

QMCPACK is an open source quantum Monte Carlo package for ab-initio electronic structure calculations. It supports calculations of metallic and insulating solids, molecules, atoms, and some model Hamiltonians. Implemented real space quantum Monte Carlo algorithms include variational, diffusion, and reptation Monte Carlo. QMCPACK uses Slater-Jastrow type trial wave functions in conjunction with a sophisticated optimizer capable of optimizing tens of thousands of parameters. The orbital space auxiliary field quantum Monte Carlo method is also implemented, enabling cross validation between different highly accurate methods. The code is specifically optimized for calculations with large numbers of electrons on the latest high performance computing architectures, including multicore central processing unit (CPU) and graphical processing unit (GPU) systems. We detail the program's capabilities, outline its structure, and give examples of its use in current research calculations. The package is available at http://www.qmcpack.org .

physics.comp-ph↗

arXiv2017

Embracing a new era of highly efficient and productive quantum Monte Carlo simulations

QMCPACK has enabled cutting-edge materials research on supercomputers for over a decade. It scales nearly ideally but has low single-node efficiency due to the physics-based abstractions using array-of-structures objects, causing inefficient vectorization. We present a systematic approach to transform QMCPACK to better exploit the new hardware features of modern CPUs in portable and maintainable ways. We develop miniapps for fast prototyping and optimizations. We implement new containers in structure-of-arrays data layout to facilitate vectorizations by the compilers. Further speedup and smaller memory-footprints are obtained by computing data on the fly with the vectorized routines and expanding single-precision use. All these are seamlessly incorporated in production QMCPACK. We demonstrate upto 4.5x speedups on recent Intel processors and IBM Blue Gene/Q for representative workloads. Energy consumption is reduced significantly commensurate to the speedup factor. Memory-footprints are reduced by up-to 3.8x, opening the possibility to solve much larger problems of future.

cs.DC↗

arXiv2017

The Nature of Interlayer Binding and Stacking of $sp$-$sp^{2}$ Hybridized Carbon Layers: A Quantum Monte Carlo Study

$α$-graphyne is a two-dimensional sheet of $sp$-$sp^2$ hybridized carbon atoms in a honeycomb lattice. While the geometrical structure is similar to that of graphene, the hybridized triple bonds give rise to electronic structure that is different from that of graphene. Similar to graphene, $α$-graphyne can be stacked in bilayers with two stable configurations, but the different stackings have very different electronic structures: one is predicted to have gapless parabolic bands and the other a tunable band gap which is attractive for applications. In order to realize applications, it is crucial to understand which stacking is more stable. This is difficult to model, as the stability is a result of weak interlayer van der Waals interactions which are not well captured by density functional theory (DFT). We have used quantum Monte Carlo simulations that accurately include van der Waals interactions to calculate the interlayer binding energy of bilayer graphyne and to determine its most stable stacking mode. Our results show that interlayer bindings of $sp$- and $sp^{2}$-bonded carbon networks are significantly underestimated in a Kohn-Sham DFT approach, even with an exchange-correlation potential corrected to include, in some approximation, van der Waals interactions. Finally, our quantum Monte Carlo calculations reveal that the interlayer binding energy difference between the two stacking modes is only 0.9(4) meV/atom. From this we conclude that the two stable stacking modes of bilayer $α$-graphyne are almost degenerate with each other, and both will occur with about the same probability at room temperature unless there is a synthesis path that prefers one stacking over the other.

cond-mat.mtrl-sci↗

arXiv2016

Optimization and parallelization of B-spline based orbital evaluations in QMC on multi/many-core shared memory processors

B-spline based orbital representations are widely used in Quantum Monte Carlo (QMC) simulations of solids, historically taking as much as 50% of the total run time. Random accesses to a large four-dimensional array make it challenging to efficiently utilize caches and wide vector units of modern CPUs. We present node-level optimizations of B-spline evaluations on multi/many-core shared memory processors. To increase SIMD efficiency and bandwidth utilization, we first apply data layout transformation from array-of-structures to structure-of-arrays (SoA). Then by blocking SoA objects, we optimize cache reuse and get sustained throughput for a range of problem sizes. We implement efficient nested threading in B-spline orbital evaluation kernels, paving the way towards enabling strong scaling of QMC simulations. These optimizations are portable on four distinct cache-coherent architectures and result in up to 5.6x performance enhancements on Intel Xeon Phi processor 7250P (KNL), 5.7x on Intel Xeon Phi coprocessor 7120P, 10x on an Intel Xeon processor E5v4 CPU and 9.5x on BlueGene/Q processor. Our nested threading implementation shows nearly ideal parallel efficiency on KNL up to 16 threads. We employ roofline performance analysis to model the impacts of our optimizations. This work combined with our current efforts of optimizing other QMC kernels, result in greater than 4.5x speedup of miniQMC on KNL.

cs.DC↗

arXiv2015

Structural Stability and Defect Energetics of ZnO from Diffusion Quantum Monte Carlo

We have applied the many-body ab-initio diffusion quantum Monte Carlo (DMC) method to study Zn and ZnO crystals under pressure, and the energetics of the oxygen vacancy, zinc interstitial and hydrogen impurities in ZnO. We show that DMC is an accurate and practical method that can be used to characterize multiple properties of materials that are challenging for density functional theory approximations. DMC agrees with experimental measurements to within 0.3 eV, including the band-gap of ZnO, the ionization potential of O and Zn, and the atomization energy of O$_2$, ZnO dimer, and wurtzite ZnO. DMC predicts the oxygen vacancy as a deep donor with a formation energy of 5.0(2) eV under O-rich conditions and thermodynamic transition levels located between 1.8 and 2.5 eV from the valence band maximum. Our DMC results indicate that the concentration of zinc interstitial and hydrogen impurities in ZnO should be low under n-type, and Zn- and H-rich conditions because these defects have formation energies above 1.4 eV under these conditions. Comparison of DMC and hybrid functionals shows that these DFT approximations can be parameterized to yield a general correct qualitative description of ZnO. However, the formation energy of defects in ZnO evaluated with DMC and hybrid functionals can differ by more than 0.5 eV.

cond-mat.mtrl-sci↗

arXiv2014

Successes and failures of Hubbard-corrected density functional theory: The case of Mg doped LiCoO$_2$

We have evaluated the successes and failures of the Hubbard-corrected density functional theory (DFT+U) approach to study Mg doping of LiCoO$_2$. We computed the effect of the U parameter on the energetic, geometric and electronic properties of two possible doping mechanisms: (1) substitution of Mg onto a Co (or Li) site with an associated impurity state and, (2) formation of impurity-state-free complexes of substitutional Mg and point defects in LiCoO$_2$. We find that formation of impurity states results in changes on the valency of Co in LiCoO$_2$. Variation of the Co U shifts the energy of the impurity state, resulting in energetic, geometric and electronic properties that depend significantly on the specific value of U. In contrast, the properties of the impurity-state-free complexes are insensitive to U. These results identify reasons for the strong dependence on the doping properties on the chosen value of U and for the overall difficulty of achieving agreement with the experimentally known energetic and electronic properties of doped transition metal oxides such as LiCoO$_2$.

cond-mat.mtrl-sci↗

arXiv2014

Equations of state and stability of MgSiO$_3$ perovskite and post-perovskite phases from quantum Monte Carlo simulations

We have performed quantum Monte Carlo (QMC) simulations and density functional theory (DFT) calculations to study the equations of state of MgSiO$_3$ perovskite (Pv) and post-perovskite (PPv), up to the pressure and temperature conditions of the base of Earth's lower mantle. The ground state energies were derived using QMC and the temperature dependent Helmholtz free energies were calculated within the quasi-harmonic approximation and density functional perturbation theory. The equations of state for both phases of MgSiO$_3$ agree well with experiments, and better than those from generalized gradient approximation (GGA) calculations. The Pv-PPv phase boundary calculated from our QMC equations of states is also consistent with experiments, and better than previous LDA calculations. We discuss the implications for double crossing of the Pv-PPv boundary in the Earth.

cond-mat.mtrl-sci↗

arXiv2014

Energy density matrix formalism for interacting quantum systems: a quantum Monte Carlo study

We develop an energy density matrix that parallels the one-body reduced density matrix (1RDM) for many-body quantum systems. Just as the density matrix gives access to the number density and occupation numbers, the energy density matrix yields the energy density and orbital occupation energies. The eigenvectors of the matrix provide a natural orbital partitioning of the energy density while the eigenvalues comprise a single particle energy spectrum obeying a total energy sum rule. For mean-field systems the energy density matrix recovers the exact spectrum. When correlation becomes important, the occupation energies resemble quasiparticle energies in some respects. We explore the occupation energy spectrum for the finite 3D homogeneous electron gas in the metallic regime and an isolated oxygen atom with ground state quantum Monte Carlo techniques implemented in the QMCPACK simulation code. The occupation energy spectrum for the homogeneous electron gas can be described by an effective mass below the Fermi level. Above the Fermi level evanescent behavior in the occupation energies is observed in similar fashion to the occupation numbers of the 1RDM. A direct comparison with total energy differences shows a quantitative connection between the occupation energies and electron addition and removal energies for the electron gas. For the oxygen atom, the association between the ground state occupation energies and particle addition and removal energies becomes only qualitative. The energy density matrix provides a new avenue for describing energetics with quantum Monte Carlo methods which have traditionally been limited to total energies.

cond-mat.str-el↗

arXiv2014

Ab initio quantum Monte Carlo calculations of spin superexchange in cuprates: the benchmarking case of Ca$_2$CuO$_3$

In view of the continuous theoretical efforts aimed at an accurate microscopic description of the strongly correlated transition metal oxides and related materials, we show that with continuum quantum Monte Carlo (QMC) calculations it is possible to obtain the value of the spin superexchange coupling constant of a copper oxide in a quantitatively excellent agreement with experiment. The variational nature of the QMC total energy allows us to identify the best trial wave function out of the available pool of wave functions, which makes the approach essentially free from adjustable parameters and thus truly ab initio. The present results on magnetic interactions suggest that QMC is capable of accurately describing ground state properties of strongly correlated materials.

cond-mat.str-el↗

arXiv2014

Cohesion Energetics of Carbon Allotropes : Quantum Monte Carlo Study

We have performed quantum Monte Carlo calculations to study the cohesion energetics of carbon allotropes, including $sp^3$-bonded diamond, $sp^2$-bonded graphene, $sp$-$sp^2$ hybridized graphynes, and $sp$-bonded carbyne. The computed cohesive energies of diamond and graphene are found to be in excellent agreement with the corresponding values determined experimentally for diamond and graphite, respectively, when the zero-point energies, along with the interlayer binding in the case of graphite, are included. We have also found that the cohesive energy of graphyne decreases systematically as the ratio of $sp$-bonded carbon atoms increases. The cohesive energy of $γ$-graphyne, the most energetically-stable graphyne, turns out to be 6.766(6) eV/atom, which is smaller than that of graphene by 0.698(12) eV/atom. Experimental difficulty in synthesizing graphynes could be explained by their significantly smaller cohesive energies. Finally we conclude that the cohesive energy of a newly-proposed graphyne can be accurately estimated with the carbon-carbon bond energies determined from the cohesive energies of graphene and three different graphynes considered here.

cond-mat.mtrl-sci↗

arXiv2013

The Transition to the Metallic State in Low Density Hydrogen

Solid atomic hydrogen is one of the simplest systems to undergo a metal-insulator transition. Near the transition, the electronic degrees of freedom become strongly correlated and their description provides a difficult challenge for theoretical methods. As a result, the order and density of the phase transition are still subject to debate. In this work we use diffusion quantum Monte Carlo to benchmark the transition between paramagnetic and anti-ferromagnetic body centered cubic atomic hydrogen in its ground state. We locate the density of the transition by computing the equation of state for these two phases and identify the phase transition order by computing the band gap near the phase transition. These benchmark results show that the phase transition is continuous and occurs at a Wigner-Seitz radius of $r_s=2.27(3) a_0$. We compare our results to previously reported density functional theory, Hedin's GW approximation, and dynamical mean field theory results.

cond-mat.str-el↗

arXiv2013

Generalizing the self-healing diffusion Monte Carlo approach to finite temperature: a path for the optimization of low-energy many-body bases

A statistical method is derived for the calculation of thermodynamic properties of many-body systems at low temperatures. This method is based on the self-healing diffusion Monte Carlo method for complex functions [F. A. Reboredo J. Chem. Phys. 136, 204101 (2012)] and some ideas of the correlation function Monte Carlo approach [D. M. Ceperley and B. Bernu, J. Chem. Phys. 89, 6316 (1988)]. In order to allow the evolution in imaginary time to describe the density matrix, we remove the fixed-node restriction using complex antisymmetric guiding wave functions. In the process we obtain a parallel algorithm that optimizes a small subspace of the many-body Hilbert space to have maximum overlap with the subspace spanned by the lowest-energy eigenstates of a many-body Hamiltonian. We show in a model system that the partition function is progressively maximized within this subspace. We show that the subspace spanned by the small basis systematically converges towards the subspace spanned by the lowest energy eigenstates. Possible applications of this method to calculate the thermodynamic properties of many-body systems near the ground state are discussed. The resulting basis can be also used to accelerate the calculation of the ground or excited states with Quantum Monte Carlo.

cond-mat.other↗

arXiv2013

The Quantum Energy Density: Improved Efficiency for Quantum Monte Carlo

We establish a physically meaningful representation of a quantum energy density for use in Quantum Monte Carlo calculations. The energy density operator, defined in terms of Hamiltonian components and density operators, returns the correct Hamiltonian when integrated over a volume containing a cluster of particles. This property is demonstrated for a helium-neon "gas," showing that atomic energies obtained from the energy density correspond to eigenvalues of isolated systems. The formation energies of defects or interfaces are typically calculated as total energy differences. Using a model of delta-doped silicon (where dopant atoms form a thin plane) we show how interfacial energies can be calculated more efficiently with the energy density, since the region of interest is small. We also demonstrate how the energy density correctly transitions to the bulk limit away from the interface where the correct energy is obtainable from a separate total energy calculation.

cond-mat.mtrl-sci↗