Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “parallel, scaling”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,207 records · Page 67

Highly-scalable, physics-informed GANs for learning solutions of stochastic PDEs

Uncertainty quantification for forward and inverse problems is a central challenge across physical and biomedical disciplines. We address this challenge for the problem of modeling subsurface flow at the Hanford Site by combining stochastic computational models with observational data using physics-informed GAN models. The geographic extent, spatial heterogeneity, and multiple correlation length scales of the Hanford Site require training a computationally intensive GAN model to thousands of dimensions. We develop a highly optimized implementation that scales to {\herogpucount} NVIDIA Volta GPUs. We develop a hierarchical scheme based on a multi-player game-theoretic approach for exploiting domain parallelism, map discriminators and generators to multiple GPUs, and employ efficient communication schemes to ensure training stability and convergence. Our implementation scales to {\heronodecount} nodes on the Summit supercomputer with a \weakscalingeff\% scaling efficiency, achieving peak and sustained half-precision rates of {\heropeakperf} PF/s and {\herosustainedperf} PF/s.

physics informed ML, machine learning, generative ↗

Penalty-free method for multi-constrained optimization problems based on parallel simulated annealing

A method is presented for addressing constraints within the context of large-scale combinatorial optimization problems. The method, based on constraint annealing, eliminates the use of traditional constraint penalty factors by treating each constraint as separate and concurrently solved minimization problems within a global optimization search framework. Results are demonstrated for a core loading pattern optimization problem. (authors)

97 MATHEMATICS AND COMPUTING↗

The inviscid incompressible limit of Kelvin–Helmholtz instability for plasmas

The Kelvin–Helmholtz Instability (KHI) is an interface instability that develops between two fluids or plasmas flowing with a common shear layer. KHI occurs in astrophysical jets, solar atmosphere, solar flows, cometary tails, planetary magnetospheres. Two applications of interest, encompassing both space and fusion applications, drive this study: KHI formation at the outer flanks of the Earth’s magnetosphere and KHI growth from non-uniform laser heating in magnetized direct-drive implosion experiments. Here, we study 2D KHI with or without a magnetic field parallel to the flow. We use both the GAMERA code, which solves the compressible Euler equations, and the STRATOSPEC code, which solves the Navier-Stokes equations under the Boussinesq approximation, coupled with the magnetic field dynamics. GAMERA is a global three-dimensional MHD code with high-order reconstruction in arbitrary nonorthogonal curvilinear coordinates, which is developed for a large range of astrophysical applications. STRATOSPEC is a three-dimensional pseudo-spectral code with an accuracy of infinite order (no numerical diffusion). Magnetized KHI is a canonical case for benchmarking hydrocode simulations with extended MHD options. An objective is to assess whether or not, and under which conditions, the incompressibility hypothesis allows to describe a dynamic compressible system. For comparing both codes, we reach the inviscid incompressible regime, by decreasing the Mach number in GAMERA, and viscosity and diffusion in STRATOSPEC. Here, we specifically investigate both single-mode and multi-mode initial perturbations, either with or without magnetic field parallel to the flow. The method relies on comparisons of the density fields, 1D profiles of physical quantities averaged along the flow direction, and scale-by-scale spectral densities. We also address the triggering, formation and damping of filamentary structures under varying Mach number or Atwood number, with or without a parallel magnetic field. Comparisons show very satisfactory results between the two codes. The vortices dynamics is well reproduced, along with the breaking or damping of small-scale structures. We end with the extraction of growth rates of magnetized KHI from the compressible regime to the incompressible limit in the linear regime assessing the effects of compressibility under increasing magnetic field. The observed differences between the two codes are explained either from diffusion or non-Boussinesq effects.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A template-based approach for parallel hexahedral two-refinement

Here, we provide a template-based approach for generating locally refined all-hex meshes. We focus specifically on refinement of initially structured grids utilizing a 2-refinement approach where uniformly refined hexes are subdivided into eight child elements. The refinement algorithm consists of identifying marked nodes that are used as the basis for a set of four simple refinement templates. The target application for 2-refinement is a parallel grid-based all-hex meshing tool for high performance computing in a distributed environment. The result is a parallel consistent locally refined mesh requiring minimal communication and where minimum mesh quality is greater than scaled Jacobian 0.3 prior to smoothing.

2-refinement↗

Efficient exascale discretizations: High-order finite element methods

Efficient exploitation of exascale architectures requires rethinking of the numerical algorithms used in many large-scale applications. These architectures favor algorithms that expose ultra fine-grain parallelism and maximize the ratio of floating point operations to energy intensive data movement. One of the few viable approaches to achieve high efficiency in the area of PDE discretizations on unstructured grids is to use matrix-free/partially assembled high-order finite element methods, since these methods can increase the accuracy and/or lower the computational time due to reduced data motion. In this paper we provide an overview of the research and development activities in the Center for Efficient Exascale Discretizations (CEED), a co-design center in the Exascale Computing Project that is focused on the development of next-generation discretization software and algorithms to enable a wide range of finite element applications to run efficiently on future hardware. CEED is a research partnership involving more than 30 computational scientists from two US national labs and five universities, including members of the Nek5000, MFEM, MAGMA and PETSc projects. We discuss the CEED co-design activities based on targeted benchmarks, miniapps and discretization libraries and our work on performance optimizations for large-scale GPU architectures. We also provide a broad overview of research and development activities in areas such as unstructured adaptive mesh refinement algorithms, matrix-free linear solvers, high-order data visualization, and list examples of collaborations with several ECP and external applications.

97 MATHEMATICS AND COMPUTING↗

Numerical Analysis of the Dynamics of Two- and Three-Dimensional Fluidized Bed Reactors using an Euler-Lagrange Approach

Biomass thermochemical conversion, often done in fluidized beds, recently gained a lot of attention due to its potential to efficiently produce renewable liquid fuels. Optimization of reactor design and operating conditions, however, requires a fundamental understanding of bed dynamics. In this work, a numerical framework based on an Euler-Lagrange approach is developed and used to perform and analyze large-scale simulations of two- and three-dimensional periodic fluidized beds. Collisions are handled using a soft-sphere model. An efficient parallel implementation allows one to explicitly track over 30 million particles, which is representative of the number of particles found in lab-scale reactor, therefore demonstrating the capability of Lagrangian approaches to simulate realistic systems at that scale. An on-the-fly bubble identification and tracking algorithm is used to characterize bubble dynamics for inlet velocities up to 9 times the minimum fluidization velocity. Statistics for gas volume fraction, gas and particle velocities, bed expansion, and bubble size and velocity, is compared across the two- and three-dimensional configurations, and comparison with literature data generally shows good agreement. The wide distribution of gas residence times observed in the simulations is linked to the different gas hold-up characteristics of the gas-solid system.

09 BIOMASS FUELS↗

Turbulent Regimes in Collisions of 3D Alfvén-wave Packets

Using three-dimensional gyrofluid simulations, we revisit the problem of Alfvén-wave (AW) collisions as building blocks of the Alfvénic turbulent cascade and their interplay with magnetic reconnection at magnetohydrodynamic (MHD) scales. Depending on the large-scale value of the nonlinearity parameter χ 0 (the ratio between the AW linear propagation time and nonlinear turnover time), different regimes are observed. For strong nonlinearities (χ 0 ~ 1), turbulence is consistent with a dynamically aligned, critically balanced cascade—fluctuations exhibit a scale-dependent alignment $\sin {\theta }_{{k}_{\perp }}\propto {k}_{\perp }^{-1/4}$, resulting in a ${k}_{\perp }^{-3/2}$ spectrum and ${k}_{\parallel }\propto {k}_{\perp }^{1/2}$ spectral anisotropy. At weaker nonlinearities (small χ 0 ), a spectral break marking the transition between a large-scale weak regime and a small-scale ${k}_{\perp }^{-11/5}$ tearing-mediated range emerges, implying that dynamic alignment occurs also for weak nonlinearities. At χ 0 < 1 the alignment angle ${\theta }_{{k}_{\perp }}$ shows a stronger scale dependence than in the χ 0 ~ 1 regime, namely $\sin {\theta }_{{k}_{\perp }}\propto {k}_{\perp }^{-1/2}$ at χ 0 ~ 0.5, and $\sin {\theta }_{{k}_{\perp }}\propto {k}_{\perp }^{-1}$ at χ 0 ~ 0.1. Dynamic alignment in the weak regime also modifies the large-scale spectrum, scaling approximately as ${k}_{\perp }^{-3/2}$ for χ 0 ~ 0.5 and as ${k}_{\perp }^{-1}$ for χ 0 ~ 0.1. A phenomenological theory of dynamically aligned turbulence at weak nonlinearities that can explain these spectra and the transition to the tearing-mediated regime is provided; at small χ 0 , the strong scale dependence of the alignment angle combines with the increased lifetime of turbulent eddies to allow tearing to onset and mediate the cascade at scales that can be larger than those predicted for a critically balanced cascade by several orders of magnitude. Such a transition to tearing-mediated turbulence may even supplant the usual weak-to-strong transition.

79 ASTRONOMY AND ASTROPHYSICS↗

Parallel performance of algebraic multigrid domain decomposition

Algebraic multigrid (AMG) is a widely used scalable solver and preconditioner for large-scale linear systems resulting from the discretization of a wide class of elliptic PDEs. While AMG has optimal computational complexity, the cost of communication has become a significant bottleneck that limits its scalability as processor counts continue to grow on modern machines. This article examines the design, implementation, and parallel performance of a novel algorithm, algebraic multigrid domain decomposition (AMG-DD), designed specifically to limit communication. The goal of AMG-DD is to provide a low-communication alternative to standard AMG V-cycles by trading some additional computational overhead for a significant reduction in communication cost. Numerical results show that AMG-DD achieves superior accuracy per communication cost compared with AMG, and speedup over AMG is demonstrated on a large GPU cluster.

97 MATHEMATICS AND COMPUTING↗

A parallel discrete dislocation dynamics/kinetic Monte Carlo method to study non-conservative plastic processes

Non-conservative processes play a fundamental role in plasticity and are behind important macroscopic phenomena such as creep, dynamic strain aging, loop raft formation, etc. In the most general case, vacancy-induced dislocation climb is the operating unit mechanism. While dislocation/vacancy interactions have been modeled in the literature using a variety of methods, the approaches developed rely on continuum descriptions of both the vacancy population and its fluxes. However, there are numerous situations in physics where point defect populations display heterogeneous concentrations and/or non-smooth kinetics. Here, a kinetic Monte Carlo (kMC) approach for modeling vacancy transport in response to arbitrary stress fields is used. Vacancies are treated as point particles and are coupled to the dislocation substructure representing a deformed material via an advection term defined by the local stress gradients. The stress fields and the dislocation substructure are evolved using a discrete dislocation dynamics (DDD) module. To extend the coupled model to the treatment of large systems, we have implemented it in the massively-parallel DDD code ParaDiS. To avoid numerical incompatibilities associated with merging deterministic (DDD) and stochastic (kMC) integration algorithms, we cast the entire elasto-plastic-diffusive problem within a single stochastic framework, taking advantage of a parallel kMC algorithm to evolve the system as a single event-driven process. The large-scale implementation enables the study of the evolution of a variety of dislocation-defect scenarios governed by non-conservative transport kinetics. After carrying out an exhaustive numerical and computational analysis of our parallel algorithm, we show results that emphasize situations where inhomogeneous vacancy dynamics are of relevance, and compare discrete kinetics to continuum solutions for several cases.

36 MATERIALS SCIENCE↗

Feshbach resonances and the unitary limit in a model of strongly correlated nucleons

A model of strongly interacting and correlated hadrons is developed. The interaction used contains a long-range attraction and short-range repulsive hard core. Using this interaction and various limiting situations of it, a study of the effect of bound states and Feshbach resonances is given. The limiting situations are a pure square-well interaction, a delta-shell potential, and a pure hard-core potential. The limit of a pure hard-core potential are compared with the corresponding well-known results for a spinless Bose and Fermi gas. The limit of many partial waves for a pure hard-core interaction is also considered and results in expressions involving the hard-core volume. This feature arises from a scaling relation similar to that for hard-sphere scattering with diffractive corrections. Features that parallel the van der Waals gas and results based on a Skyrme interaction are discussed. Thermodynamic features associated with a unitary limit of very large scattering length are given. The ratio of effective range r{sub 0} to thermal wavelength lambda{sub T} appears as a limiting scale in various thermodynamic properties. The effective-range corrections to the entropy vary as the cube of this ratio for low temperatures and are therefore considerably reduced compared to the corrections to the interaction energy, Helmholtz free energy, compressibility, and chemical potential, which vary linearly with this ratio.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Multichannel Analysis of Surface Waves Accelerated (MASWAccelerated): Software for efficient surface wave inversion using MPI and GPUs

Multichannel Analysis of Surface Waves (MASW) is a technique frequently used in geotechnical engineering and engineering geophysics to infer 1D layered models of seismic shear wave velocities in the top tens to hundreds of meters of the subsurface. We aim to accelerate MASW calculations by capitalizing on modern computer hardware available in the workstations of most engineers: multiple cores and graphics processing units (GPUs). We propose new parallel and GPU accelerated algorithms for computing 1D MASW inversion, and provide software implementations in C using Message Passing Interface (MPI) and CUDA. These algorithms take advantage of sparsity that arises in the problem, and the work balance between processes considers typical data trends. We compare our methods to an existing open source Matlab MASW tool. Our serial C implementation achieves a 2x speedup over the Matlab software, and we continue to see improvements by parallelizing the problem with MPI. Here we see nearly perfect strong and weak scaling for uniform data, and improve strong scaling for realistic data by repartitioning the problem to process mapping. By utilizing GPUs available on most modern workstations, we observe an additional 1.3x speedup over the serial C implementation on the first use of the method. We typically repeatedly evaluate theoretical dispersion curves as part of an optimization procedure, and on the GPU the kernel can be cached for faster reuse on later runs. We observe a 3.2x speedup on the cached GPU runs compared to the serial C runs. This work is the first open-source parallel or GPU-accelerated software tool for MASW imaging, and should enable geotechnical engineers to fully utilize all computer hardware at their disposal.

58 GEOSCIENCES↗

Reionization on large scales. IV. Predictions for the 21 cm signal incorporating the light cone effect

We present predictions for the 21 cm brightness temperature power spectrum during the Epoch of Reionization (EoR). We discuss the implications of the 'light cone' effect, which incorporates evolution of the neutral hydrogen fraction and 21 cm brightness temperature along the line of sight. Using a novel method calibrated against radiation-hydrodynamic simulations, we model the neutral hydrogen density field and 21 cm signal in large volumes (L = 2 Gpc h {sup –1}). The inclusion of the light cone effect leads to a relative decrease of about 50% in the 21 cm power spectrum on all scales. We also find that the effect is more prominent at the midpoint of reionization and later. The light cone effect can also introduce an anisotropy along the line of sight. By decomposing the 3D power spectrum into components perpendicular to and along the line of sight, we find that in our fiducial reionization model, there is no significant anisotropy. However, parallel modes can contribute up to 40% more power for shorter reionization scenarios. The scales on which the light cone effect is relevant are comparable to scales where one measures the baryon acoustic oscillation. We argue that due to its large comoving scale and introduction of anisotropy, the light cone effect is important when considering redshift space distortions and future application to the Alcock-Paczyński test for the determination of cosmological parameters.

79 ASTRONOMY AND ASTROPHYSICS↗

Enabling Parallel Performance and Portability of Solid Mechanics Simulations Across CPU and GPU Architectures

Efficiently simulating solid mechanics is vital across various engineering applications. As constitutive models grow more complex and simulations scale up in size, harnessing the capabilities of modern computer architectures has become essential for achieving timely results. This paper presents advancements in running parallel simulations of solid mechanics on multi-core CPUs and GPUs using a single-code implementation. This portability is made possible by the C++ matrix and array (MATAR) library, which interfaces with the C++ Kokkos library, enabling the selection of fine-grained parallelism backends (e.g., CUDA, HIP, OpenMP, pthreads, etc.) at compile time. MATAR simplifies the transition from Fortran to C++ and Kokkos, making it easier to modernize legacy solid mechanics codes. We applied this approach to modernize a suite of constitutive models and to demonstrate substantial performance improvements across different computer architectures. This paper includes comparative performance studies using multi-core CPUs along with AMD and NVIDIA GPUs. Results are presented using a hypoelastic–plastic model, a crystal plasticity model, and the viscoplastic self-consistent generalized material model (VPSC-GMM). The results underscore the potential of using the MATAR library and modern computer architectures to accelerate solid mechanics simulations.

Morgan, Nathaniel (ORCID:0000000276118449)↗

Near-continuum, hypersonic oxygen flow over a double cone simulated by direct simulation Monte Carlo informed from quantum chemistry

A large-scale, fully resolved direct simulation Monte Carlo (DSMC) computation of a non-equilibrium, reactive flow of pure oxygen over a double cone is presented. Under the simulated near-continuum conditions, the computational demands are shown to be significant because of the wide range of length scales that must be resolved. Therefore, robust grid adaption capabilities and efficient parallelization of the Stochastic PArallel Rarefied-gas Time-accurate Analyzer (SPARTA) code that is utilized in this work are essential. The thermochemical and transport collision models were selected for efficiency and simplicity. First-principles data, obtained from the highly accurate direct molecular simulation method, were used to inform the collision models’ parameters. Importantly, because SPARTA implements molecular collision models using collision-specific energies, the resulting macroscopic relaxation rates were evaluated a posteriori via zero-dimensional heat bath simulations. The comparisons of surface properties, namely heat flux and pressure, show very close agreement with previous computational fluid dynamics (CFD) results. Differences with the measurements were found to be similar to the CFD simulations. The unresolved discrepancy with the measurements could be due to inconsistent free stream conditions with the actual experimental data or missing physical phenomena altogether, for example atomic and molecular oxygen electronically excited states, three-dimensional effects, or more complex gas–surface interactions. As shown in this work, the advantages of obtaining a DSMC particle solution for these flows reside in the method's ability to be directly informed from first principles and to seamlessly describe internal energy non-equilibrium for all modes. With the advent of exascale computing and beyond, particle methods will be an increasingly important tool to verify the validity of physical assumptions in reduced-order models via fully resolved, experimental-scale simulations, down to the level of molecular-level distributions.

Mechanics↗

The ion-ion hybrid Alfvén resonator in a fusion environment

An investigation is made of a shear Alfvén wave resonator for burning plasma conditions expected in the ITER device. For small perpendicular scale-lengths the shear mode, which propagates predominantly along the magnetic field direction, experiences a parallel reflection where the wave frequency matches the local ion-ion hybrid frequency. In a tokamak device operating with a deuterium–tritium fuel, this effect can form a natural resonator because of the variation in local field strength along a field line. The relevant kinetic dispersion relation is examined to determine the relative importance of Landau and cyclotron damping over the possible resonator parameter space. A WKB model based on the kinetic dispersion relation is used to determine the eigenfrequencies and the quality factors of modes trapped in the resonator. The lowest frequency found has a value slightly larger than the ion-ion hybrid frequency at the outboard side of a given flux surface. The possibility that the resonator modes can be driven unstable by energetic alpha particles is considered. It is found that within a bandwidth of roughly 600 kHz above the ion-ion hybrid frequency on the outboard side of the flux surface, the shear modes can experience significant spatial amplification. An assessment is made of the form of an approximate global eigenmode that possesses the features of a resonator. It is identified that magnetic field shear combined with large ion temperature can cause coupling to an ion-Bernstein wave, which can limit the instability.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

SLoG: Large-Scale Logging Middleware for HPC and Big Data Convergence

Cloud developers traditionally rely on purpose-specific services to provide the storage model they need for an application. In contrast, HPC developers have a much more limited choice, typically restricted to a centralized parallel file system for persistent storage. Unfortunately, these systems often offer low performance when subject to highly concurrent, conflicting I/O patterns. This makes difficult the implementation of inherently concurrent data structures such as distributed shared logs. Yet, this data structure is key to applications such as computational steering, data collection from physical sensor grids, or discrete event generators. In this paper we tackle this issue. We present SLoG, shared log middleware providing a shared log abstraction over a parallel file system, designed to circumvent the aforementioned limitations. We evaluate SLoG’s design on up to 100,000 cores of the Theta supercomputer: the results show high append velocity at scale while also providing substantial benefits for other persistent backend storage systems.

big data↗

Testing predictions of electron scale turbulent pedestal transport in two DIII-D ELMy H-modes

In this study, we present linear and nonlinear gyrokinetic analyses in the pedestal region of two DIII-D ELMy H-mode discharges using the CGYRO code. The otherwise matched discharges employ different divertor configurations to investigate the impact of varying recycling and particle source on pedestal profiles. Linear gyrokinetic simulations find electrostatic ion-scale instabilities (ion temperature gradient and trapped electron modes, ITG-TEM) are present just inside the top of the pedestal with growth rates that are enhanced significantly by parallel velocity shear. In the sharp gradient region, E x B shearing rates are comparable or larger than ion scale growth rates, suggesting the suppression of ITG-TEM modes in this region. Instead, the electron temperature profiles are found to be correlated with and just above the electron temperature gradient (ETG) instability thresholds. Using gradients varied within experimental uncertainties, nonlinear electron-scale gyrokinetic simulations predict electron heat fluxes from ETG turbulence, that when added to neoclassical ion thermal transport simulated by NEO, account for 30-60% of the total experimental heat flux. In addition, the neoclassical electron particle flux is found to contribute significantly to the experimental fluxes inferred from SOLPS-ITER analysis. Additional nonlinear gyrokinetic simulations are run varying input gradients to develop a threshold-based reduced model for ETG transport, finding a relatively simple dependence on η e = L ne /L Te . Predictive transport simulations are used to validate this pedestal-specific ETG model, in conjunction with a model for neoclassical particle transport. In both discharges, the predicted electron temperatures are always overpredicted, indicative of the insufficient stiffness in the ETG pedestal model to account for all of the experimental electron thermal transport. In the case of the closed divertor discharge with lower particle source, the predicted electron density is close to the experiment, consistent with the magnitude of neoclassical particle transport in that discharge. However, the density profiles are overpredicted in the open divertor discharge (larger particle source), due to insufficient model transport. In conclusion, the implications for other mechanisms accounting for the remainder of transport in the sharp gradient region in the two discharges are discussed.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗