检索结果-内蒙古大学图书馆

Learning to Parallelize in a Shared-Memory Environment with Transformers

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Harel, Re'em Pinter, Yuval Oren, Gal Department of Computer Science Ben-Gurion University of the Negev Israel Department of Physics Nuclear Research Center - Negev Israel Scientific Computing Center Nuclear Research Center - Negev Israel Department of Computer Science Technion - Israel Institute of Technology Israel

In past years, the world has switched to many-core and multi-core shared memory architectures. As a result, there is a growing need to utilize these architectures by introducing shared memory parallelization schemes to software applications. OpenMP is the most comprehensive API that implements such schemes, characterized by a readable interface. Nevertheless, introducing OpenMP into code, especially legacy code, is challenging due to pervasive pitfalls in management of parallel shared memory. To facilitate the performance of this task, many source-to-source (S2S) compilers have been created over the years, tasked with inserting OpenMP directives into code automatically. In addition to having limited robustness to their input format, these compilers still do not achieve satisfactory coverage and precision in locating parallelizable code and generating appropriate directives. In this work, we propose leveraging recent advances in machine learning techniques, specifically in natural language processing (NLP), to suggest the need for an OpenMP directive. The model also suggests directives complementary to S2S automatic parallelization compilers or provides immediate on-the-fly advice to the developer without compiling or executing the code. We create a database, Open-OMP, specifically for this goal. Open-OMP contains over 17,000 unique code snippets from different domains, half of which contain OpenMP directives while the other half do not need parallelization with high probability. We use the corpus to train systems to automatically classify code segments in need of parallelization, as well as suggest individual OpenMP clauses. We train several transformer models, named PragFormer, for these tasks and show that they outperform statistically-trained baselines and automatic S2S parallelization compilers in both classifying the overall need for an OpenMP directive and the introduction of private and reduction clauses. In the future, our corpus can be used for additional t

关键词： Learning algorithms

Toward exponentially-convergent simulations of extreme-mass-ratio inspirals: A time-domain solver for the scalar Teukolsky equation with singular source terms

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Vishal, Manas Field, Scott E. Rink, Katie Gottlieb, Sigal Khanna, Gaurav Department of Mathematics University of Massachusetts DartmouthMA02747 United States Center for Scientific Computing and Data Science Research University of Massachusetts DartmouthMA02747 United States Department of Physics University of Massachusetts DartmouthMA02747 United States Center for Gravitational Physics University of Texas AustinTX78712 United States Department of Physics Center for Computational Research University of Rhode Island KingstonRI02881 United States

Gravitational wave signals from extreme mass ratio inspirals are a key target for upcoming, space-based gravitational wave detectors. These systems are typically modeled as a distributionally-forced Teukolsky equation, where the smaller black hole is treated as a Dirac delta distribution (i.e., a point-particle). Time-domain solvers often use regularization approaches that approximate the Dirac distribution. Unfortunately, such approaches often introduce small length scales (e.g., when the approximation is by a narrow Gaussian) and are a source of systematic error, especially near the smaller black hole. We describe a multi-domain discontinuous Galerkin (DG) method for solving the distributionally-forced s = 0 Teukolsky equation that describes scalar fields evolving on a Kerr spacetime. To handle the Dirac delta, we expand the solution in spherical harmonics and recast the sourced Teukolsky equation as a first-order, one-dimensional symmetric hyperbolic system. This allows us to derive the method’s numerical flux to correctly account for the Dirac delta. As a result, our method achieves global spectral accuracy even at the source’s location. To connect the near field to future null infinity, we use the hyperboloidal layer method, allowing us to supply trivial outer boundary conditions and providing direct access to the far-field waveform. We document several numerical experiments where we test our method, including convergence tests against exact solutions, energy luminosities for circular orbits, the scheme’s superconvergence properties at future null infinity, and the late-time tail behavior of the scalar field. We also compare two systems that arise from different choices of the first-order reduction variables, finding that certain reasonable choices are numerically problematic in practice. The methods developed here may be beneficial when computing gravitational self-force effects, where the regularization procedure has been developed for the spherical harmonic

关键词： Time domain analysis

Mapping between black-hole perturbation theory and numerical relativity: gravitational-wave energy flux

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Islam, Tousif Department of Physics University of Massachusetts DartmouthMA02747 United States Department of Mathematics University of Massachusetts DartmouthMA02747 United States Center for Scientific Computing and Data Science Research University of Massachusetts DartmouthMA02747 United States

We investigate the α-β mapping, as previously introduced by Islam et al. [1], which relates numerical relativity (NR) and adiabatic point-particle black hole perturbation theory (BHPT) waveforms in the comparable mass regime for quasi-circular, non-spinning binary black holes. This mapping involves scaling the amplitude of individual modes with different values of α and the time (and therefore the phase) with a single parameter, β. In this paper, we demonstrate that this scaling, both in terms of time and orbital frequencies, also extends to the overall gravitational-wave energy flux. This means that we can find a single αF that scales the BHPT flux and a single βF (which matches the value of β) that scales the BHPT time such a way that it aligns with NR flux evolution. We then explore the connection between the scaling parameter αF (βF) and the missing finite size correction for the secondary black hole within the BHPT framework. Copyright © 2023, The Authors. All rights reserved.

关键词： Black holes

Apple Silicon Performance in scientific computing

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Kenyon, Connor Capano, Collin Physics Department The Center for Scientific Computing and Data Science Research The University of Massachusetts Dartmouth North DartmouthMA02747 United States HannoverD-30167 Germany

With the release of the Apple Silicon System-on-a-Chip processors, and the impressive performance shown in general use by both the M1 and M1 Ultra, the potential use for Apple Silicon processors in scientific computing is explored. Both the M1 and M1 Ultra are compared to current state-of-the-art data-center GPUs, including an NVIDIA V100 with PCIe, an NVIDIA V100 with NVLink, and an NVIDIA A100 with PCIe. The scientific performance is measured using the Scalable Heterogeneous computing (SHOC) benchmark suite using OpenCL benchmarks. We find that both M1 processors out perform the GPUs in all benchmarks. Copyright © 2022, The Authors. All rights reserved.

关键词： Fruits

Determining HEDP Foams’ Quality with Multi-View Deep Learning Classification

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Schneider, Nadav Rusanovsky, Matan Gvishi, Raz Oren, Gal Technology Division Soreq Nuclear Research Center Yavne Israel Israel Atomic Energy Commission Scientific Computing Center Nuclear Research Center Negev Israel Department of Physics Nuclear Research Center Negev Israel Applied Physics Division Soreq Nuclear Research Center Yavne Israel Department of Computer Science Technion – Israel Institute of Technology Israel

High energy density physics (HEDP) experiments commonly involve a dynamic wave-front propagating inside a low-density foam. This effect affects its density and hence, its transparency. A common problem in foam production is the creation of defective foams. Accurate information on their dimension and homogeneity is required to classify the foams’ quality. Therefore, those parameters are being characterized using a 3D-measuring laser confocal microscope. For each foam, five images are taken: two 2D images representing the top and bottom surface foam planes and three images of side cross-sections from 3D scannings. An expert has to do the complicated, harsh, and exhausting work of manually classifying the foam’s quality through the image set and only then determine whether the foam can be used in experiments or not. Currently, quality has two binary levels of normal vs. defective. At the same time, experts are commonly required to classify a sub-class of normal-defective, i.e., foams that are defective but might be sufficient for the needed experiment. This sub-class is problematic due to inconclusive judgment that is primarily intuitive. In this work, we present a novel state-of-the-art multi-view deep learning classification model that mimics the physicist’s perspective by automatically determining the foams’ quality classification and thus aids the expert. Our model achieved 86% accuracy on upper and lower surface foam planes and 82% on the entire set, suggesting interesting heuristics to the problem. A significant added value in this work is the ability to regress the foam quality instead of binary deduction and even explain the decision visually. The source code used in this work, as well as other relevant sources, are available at: https://***/scientific-computing-Lab-NRCN/***. © 2022, CC BY.

关键词： Classification (of information)

Determining HEDP Foams' Quality with Multi-View Deep Learning Classification

学校读者我要写书评

暂无评论

Determining HEDP Foams' Quality with Multi-View Deep Learnin...

Artificial Intelligence and Machine Learning for scientific Applications (AI4S), IEEE/ACM International Workshop on

作者： Nadav Schneider Matan Rusanovsky Raz Gvishi Gal Oren Technology Division Soreq Nuclear Research Center Yavne Israel Israel Atomic Energy Commission Scientific Computing Center Nuclear Research Center Negev Israel Department of Physics Nuclear Research Center Negev Israel Applied Physics Division Soreq Nuclear Research Center ─ Yavne Israel Department of Computer Science Technion Israel Institute of Technology Israel

ISBN: (纸本)9781665462082

High energy density physics (HEDP) experiments commonly involve a dynamic wave-front propagating inside a lowdensity foam. This effect affects its density and hence, its transparency. A common problem in foam production is the creation of defective foams. Accurate information on their dimension and homogeneity is required to classify the foams' quality. Therefore, those parameters are being characterized using a 3D-measuring laser confocal microscope. For each foam, five images are taken: two 2D images representing the top and bottom surface foam planes and three images of side cross-sections from 3D scannings. An expert has to do the complicated, harsh, and exhausting work of manually classifying the foam's quality through the image set and only then determine whether the foam can be used in experiments or not. Currently, quality has two binary levels of normal vs. defective. At the same time, experts are commonly required to classify a sub-class of normal-defective, i.e., defective foams but might be sufficient for the needed experiment. This sub-class is problematic due to inconclusive judgment that is primarily intuitive. In this work, we present a novel state─of─the─art multi-view deep learning classification model that mimics the physicist's perspective by automatically determining the foams' quality classification and thus aids the expert. Our model achieved 86% accuracy on upper and lower surface foam planes and 82% on the entire set, suggesting interesting heuristics to the problem. A significant added value in this work is the ability to regress the foam quality instead of binary deduction and even explain the decision visually. The source code used in this work, as well as other relevant sources, are available at: https://***/***.

关键词： Deep learning Three-dimensional displays Source coding Microscopy Conferences MIMICs Lasers

Testing eccentric corrections to the radiation-reaction force in the test-mass limit of effective-one-body models

学校读者我要写书评

暂无评论

Physical Review D 2025年第4期111卷 044036-044036页

作者： Guglielmo Faggioli Maarten van de Meent Alessandra Buonanno Aldo Gamboa Mohammed Khalil Gaurav Khanna Max Planck Institute for Gravitational Physics (Albert Einstein Institute) Am Mühlenberg 1 Potsdam 14476 Germany Niels Bohr International Academy Niels Bohr Institute Blegdamsvej 17 2100 Copenhagen Denmark Department of Physics University of Maryland College Park Maryland 20742 USA Perimeter Institute for Theoretical Physics 31 Caroline Street North Waterloo Ontario N2L 2Y5 Canada Department of Physics and Center for Computational Research University of Rhode Island Kingston Rhode Island 02881 USA Department of Physics University of Massachusetts Dartmouth Massachusetts 02747 USA Center for Scientific Computing and Data Science Research University of Massachusetts Dartmouth Massachusetts 02747 USA

In this work, we test an effective-one-body radiation-reaction force for eccentric planar orbits of a test mass in a Kerr background, which contains third-order post-Newtonian (PN) nonspinning and second-order PN spin contributions. We compare the analytical fluxes connected to two different resummations of this force, truncated at different PN orders in the eccentric sector, with the numerical fluxes computed through the use of frequency- and time-domain Teukolsky-equation codes. We find that the different PN truncations of the radiation-reaction force show the expected scaling in the weak gravitational-field regime, and we observe a fractional difference with the numerical fluxes that is <5%, for orbits characterized by eccentricity 0≤e≤0.7, central black-hole spin −0.99M≤a≤0.99M and fixed orbital-averaged quantity x=⟨MΩ⟩2/3=0.06, corresponding to the mildly strong-field regime with semilatera recta 9M

关键词：

Aretakis hair for extreme Kerr black holes with axisymmetric scalar perturbations

学校读者我要写书评

暂无评论

Physical Review D 2023年第12期107卷 124023-124023页

作者： Lior M. Burko Gaurav Khanna Subir Sabharwal Theiss Research La Jolla California 92037 USA Department of Physics University of Rhode Island Kingston Rhode Island 02881 USA Center for Computational Research University of Rhode Island Kingston Rhode Island 02881 USA Physics Department and Center for Scientific Computing and Data Science Research University of Massachusetts Dartmouth Dartmouth Massachusetts 02747 USA

We study the evolution of axially symmetric scalar field perturbations on an extreme Kerr spacetime for initial data with multipole moments ℓ′ higher than the least radiative mode, and we measure modes ℓ—and, for the first time, also horizon charges—that are excited by mode-coupling interactions. We then find the Ori-Sela prefactors, a certain quantity that can be evaluated at finite distances, and the Aretakis constant along the event horizon of the extreme Kerr black hole for a sequence of initial data preparations that differ only by their distance from the event horizon. We find that for initial data in the near field there is a linear relationship of the Aretakis constant and the Ori-Sela prefactor. For initial data farther than these, the linear relationship is not universal, and we propose that stronger numerical simulations would be needed to regain linearity. The linear relationship suggests that the Aretakis charge along the event horizon can be measured at a finite distance, thereby extending this type of violation of the no-hair theorems from the least radiative axisymmetric mode also to situations that involve mode coupling.

关键词： Classical black holes Gravitational wave sources

Quantifying variabilities in cardiac digital twin models of the electrocardiogram

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zappon, Elena Gsell, Matthias A.F. Gillette, Karli Plank, Gernot Division of Biophysics and Medical Physics Gottfried Schatz Research Center Medical University of Graz Graz Austria Scientific Computing and Imaging Institute University of Utah SLC UT United States Department of Biomedical Engineering University of Utah SLC UT United States BioTechMed-Graz Graz Austria

Cardiac digital twins (CDTs) of human cardiac electrophysiology (EP) are digital replicas of patient hearts that match like-for-like clinical observations. The electro-cardiogram (ECG), as the most prevalent noninvasive observation of cardiac electrophysiology, is considered an ideal target for CDT calibration. Recent advanced CDT calibration methods have demonstrated their ability to minimize discrepancies between simulated and measured ECG signals, effectively replicating all key morphological features relevant to diagnostics. However, due to the inherent nature of clinical data acquisition and CDT model generation pipelines, discrepancies inevitably arise between the real physical electrophysiology in a patient and the simulated virtual electrophysiology in a CDT. In this study, we aim to qualitatively and quantitatively analyze the impact of these uncertainties on ECG morphology and diagnostic markers. We analyze residual beat-to-beat variability in ECG recordings obtained from healthy subjects and patients. Using a biophysically detailed and anatomically accurate computational model of whole-heart electrophysiology combined with a detailed torso model calibrated to closely replicate measured ECG signals, we vary anatomical factors (heart location, orientation, size), heterogeneity in electrical conductivities in the heart and torso, and electrode placements across ECG leads to assess their qualitative impact on ECG morphology. Our study demonstrates that diagnostically relevant ECG features and overall morphology appear relatively robust against the investigated uncertainties. This resilience is consistent with the narrow distribution of ECG due to residual beat-to-beat variability observed in both healthy subjects and *** Codes G110, I65, I64, J3 Copyright © 2024, The Authors. All rights reserved.

关键词： Electrocardiograms