Search NASA⌕ Search

SEARCH · Search NASA

Results for “accelerated computing”

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 433 records · Page 24

Machine‐learning‐based predictive control of nonlinear processes. Part II: Computational implementation

Abstract Machine learning is receiving more attention in classical engineering fields, and in particular, recurrent neural networks (RNNs) coupled with ensemble regression tools have demonstrated the capability of modeling nonlinear dynamic processes. In Part I of this two‐article series, the Lyapunov‐based model predictive control (LMPC) method using a single RNN model and an ensemble of RNN models, respectively, was rigorously developed for a general class of nonlinear systems. In the present article, computational implementation issues of this new control method ranging from training of the RNN models, ensemble regression of the RNN models, and parallel computation for accelerating the real‐time model calculations are addressed. Furthermore, a chemical reactor example is used to demonstrate the implementation and effectiveness of these machine‐learning tools in LMPC as well as compare them with standard state‐space model identification tools.

Wu, Zhe↗

OpenRBC: A Fast Simulator of Red Blood Cells at Protein Resolution

We present OpenRBC, a coarse-grained molecular dynamics code, which is capable of performing an unprecedented in silico experiment—simulating an entire mammal red blood cell lipid bilayer and cytoskeleton as modeled by multiple millions of mesoscopic particles—using a single shared memory commodity workstation. To achieve this, we invented an adaptive spatial-searching algorithm to accelerate the computation of short-range pairwise interactions in an extremely sparse three-dimensional space. The algorithm is based on a Voronoi partitioning of the point cloud of coarse-grained particles, and is continuously updated over the course of the simulation. The algorithm enables the construction of the key spatial searching data structure in our code, i.e., a lattice-free cell list, with a time and space cost linearly proportional to the number of particles in the system. The position and the shape of the cells also adapt automatically to the local density and curvature. The code implements OpenMP parallelization and scales to hundreds of hardware threads. It outperforms a legacy simulator by almost an order of magnitude in time-to-solution and >40 times in problem size, thus providing, to our knowledge, a new platform for probing the biomechanics of red blood cells.

Biophysics↗

Ten questions on building stock modeling to inform energy efficiency and sustainability

To enhance economic competitiveness and ensure energy efficiency, resilience, and security, cities and governments are adopting technologies and strategies to improve their existing building stocks. This approach aims to reduce energy use, improve energy affordability, and ensure a reliable power supply while safeguarding occupants during extreme weather events that may disrupt energy services. The effectiveness of these solutions will depend on building stock characteristics, use patterns, weather conditions, evolving technologies and their markets, and a city’s socio-economic conditions. This paper presents ten questions and answers that highlight the most important issues regarding the use of building stock modeling as a powerful tool to provide insights for informing stakeholders’ actions and decision-making on energy efficiency, costs reduction, and resilience of buildings in cities. Building stock modeling should build upon the fit-for-purpose framework, balancing the use case accuracy requirements, level of complexity, and needed resources (expertise, compute). The advancements in Artificial Intelligence (AI), the increasingly available open dataset of building stock in cities, and the more affordable powerful computing will accelerate the adoption of building stock modeling across scales by researchers and practitioners to inform decision making on sustainability and efficiency.

AI↗

Flame acceleration in channels with obstacles in the deflagration-to-detonation transition

It was demonstrated recently in Bychkov et al. [Bychkov et al., Phys. Rev. Lett. 101 (2008) 164501], that the physical mechanism of flame acceleration in channels with obstacles is qualitatively different from the classical Shelkin mechanism. The new mechanism is much stronger, and is independent of the Reynolds number. The present study provides details of the theory and numerical modeling of the flame acceleration. It is shown theoretically and computationally that flame acceleration progresses noticeably faster in the axisymmetric cylindrical geometry as compared to the planar one, and that the acceleration rate reduces with increasing Mach number and thereby the gas compressibility. Furthermore, the velocity of the accelerating flame saturates to a constant value that is supersonic with respect to the wall. The saturation state can be correlated to the Chapman-Jouguet deflagration as well as the fast flames observed in experiments. The possibility of transition from deflagration-to-detonation in the obstructed channels is demonstrated. (author)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Learning subgrid-scale models with neural ordinary differential equations

We propose a new approach to learning the subgrid-scale model when simulating partial differential equations (PDEs) solved by the method of lines and their representation in chaotic ordinary differential equations, based on neural ordinary differential equations (NODEs). Solving systems with fine temporal and spatial grid scales is an ongoing computational challenge, and closure models are generally difficult to tune. Machine learning approaches have increased the accuracy and efficiency of computational fluid dynamics solvers. In this approach neural networks are used to learn the coarse- to fine-grid map, which can be viewed as subgrid-scale parameterization. We propose a strategy that uses the NODE and partial knowledge to learn the source dynamics at a continuous level. Our method inherits the advantages of NODEs and can be used to parameterize subgrid scales, approximate coupling operators, and improve the efficiency of low-order solvers. Numerical results with the two-scale Lorenz 96 ODE, the convection-diffusion PDE, and the viscous Burgers' PDE are used to illustrate this approach.

97 MATHEMATICS AND COMPUTING↗

Optimizing legacy molecular dynamics software with directive-based offload

The directive-based programming models are one solution for exploiting many-core coprocessors to increase simulation rates in molecular dynamics. They offer the potential to reduce code complexity with offload models that can selectively target computations to run on the CPU, the coprocessor, or both. In our paper, we describe modifications to the LAMMPS molecular dynamics code to enable concurrent calculations on a CPU and coprocessor. We also demonstrate that standard molecular dynamics algorithms can run efficiently on both the CPU and an x86-based coprocessor using the same subroutines. As a consequence, we demonstrate that code optimizations for the coprocessor also result in speedups on the CPU; in extreme cases up to 4.7X. We provide results for LAMMAS benchmarks and for production molecular dynamics simulations using the Stampede hybrid supercomputer with both Intel (R) Xeon Phi (TM) coprocessors and NVIDIA GPUs: The optimizations presented have increased simulation rates by over 2X for organic molecules and over 7X for liquid crystals on Stampede. The optimizations are available as part of the "Intel package" supplied with LAMMPS. (C) 2015 Elsevier B.V. All rights reserved.

97 MATHEMATICS AND COMPUTING↗

Tumor-Volume Simulation During Radiotherapy for Head-and-Neck Cancer Using a Four-Level Cell Population Model

Purpose: To develop a fast computational radiobiologic model for quantitative analysis of tumor volume during fractionated radiotherapy. The tumor-volume model can be useful for optimizing image-guidance protocols and four-dimensional treatment simulations in proton therapy that is highly sensitive to physiologic changes. Methods: The analysis is performed using two approximations: (1) tumor volume is a linear function of total cell number and (2) tumor-cell population is separated into four subpopulations: oxygenated viable cells, oxygenated lethally damaged cells, hypoxic viable cells, and hypoxic lethally damaged cells. An exponential decay model is used for disintegration and removal of oxygenated lethally damaged cells from the tumor. Results: We tested our model on daily volumetric imaging data available for 14 head-and-neck cancer patients treated with an integrated computed tomography/linear accelerator system. A simulation based on the averaged values of radiobiologic parameters was able to describe eight cases during the entire treatment and four cases partially (50% of treatment time) with a maximum 20% error. The largest discrepancies between the model and clinical data were obtained for small tumors, which may be explained by larger errors in the manual tumor volume delineation procedure. Conclusions: Our results indicate that the change in gross tumor volume for head-and-neck cancer can be adequately described by a relatively simple radiobiologic model. In future research, we propose to study the variation of model parameters by fitting to clinical data for a cohort of patients with head-and-neck cancer and other tumors. The potential impact of other processes, like concurrent chemotherapy, on tumor volume should be evaluated.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Regional surrogates for predictive control of digital twins

Digital twins of complex systems must involve a model that is fast, generalizable, and usable for real-time control. For example, high-fidelity nonlinear multiphysics simulations can capture laser-material interactions, but are too slow for optimization or model predictive control (MPC). Reduced-order models, used to accelerate such computation, frequently fail to generalize to unseen inputs or control states. We show theoretically that this failure is intrinsic, i.e., that a learned model is non-unique outside the sampled subspace when its low-rank structure arises from limited excitation and clustered eigenvalues, rather than from a user-imposed truncation alone. Motivated by this result, we propose a control-ready regional surrogate-construction framework for both autonomous and nonautonomous dynamics; it employs Koopman lifting to represent nonlinearities, while preserving spatial locality. We illustrate our approach by constructing a control-ready surrogate for the digital twin of a thermal component of additive-manufacturing process. Our surrogate, localized in space through a von Neumann stencil, is learned from noisy high-fidelity simulations that emulate thermal-camera images collected during the manufacturing. It is linear in thermo-physically augmented states so that MPC reduces to a convex quadratic program. The surrogate requires no online correction, generalizes to unseen scan paths and power profiles of the laser, and is more than three orders of magnitude faster than a finite-difference solver. Furthermore, when the MPC sequence computed on the digital twin is applied to this solver, closed-loop temperature regulation is recovered, showing that the surrogate preserves control-relevant input-output behavior.

Data-driven model↗

Significant acceleration of solid-state NMR simulations via three-angle powder averaging

The anisotropic frequency shifts imparted onto the NMR resonance frequency depend on the spherical angular coordinates that describe the orientations of the NMR interaction tensors with respect to the applied magnetic field direction. Experiments performed using magic-angle spinning, however, gain a dependence on a third angle: the rotor phase γ. Traditionally, a carousel average is performed to integrate over γ, which leads to a slow convergence of intensities without contributing to the underlying powder patterns. Herein, we show an order of magnitude acceleration in computation time may be obtained by including the γ-averaging into the main powder average to eliminate redundant calculation of resonance frequencies.

magic-angle spinning↗

Towards grounding nuclear physics in QCD

Exascale computing could soon enable a predictive theory of nuclear structure and reactions rooted in the Standard Model, with quantifiable and systematically improvable uncertainties. Such a predictive theory will help exploit experiments that use nucleons and nuclei as laboratories for testing the Standard Model and its limitations. Examples include direct dark matter detection, neutrinoless double beta decay, and searches for permanent electric dipole moments of the neutron and atoms. It will also help connect QCD to the properties of cold neutron stars and hot supernova cores. We discuss how a quantitative bridge between QCD and the properties of nuclei and nuclear matter will require a synthesis of lattice QCD (especially as applied to two- and three- nucleon interactions), effective field theory, and ab initio methods for solving the nuclear many-body problem. While there are significant challenges that must be addressed in developing this triad of theoretical tools, the rapid advance of computing is accelerating progress. In particular, we focus this review on the anticipated advances from lattice QCD and how these advances will impact few-body effective theories of nuclear physics by providing critical input, such as constraints on unknown low-energy constants of the effective (field) theories. We also review particular challenges that must be overcome for the successful application of lattice QCD for low-energy nuclear physics. We describe progress in developing few-body effective (field) theories of nuclear physics, with an emphasis on HOBET, a non-relativistic effective theory of nuclear physics, which is less common in the literature. We use the examples of neutrinoless double beta decay and the nuclear-matter equation of state to illustrate how the coupling of lattice QCD to effective theory might impact our understanding of symmetries and exotic astrophysical environments.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Moderate deviations-based importance sampling for stochastic recursive equations

Subsolutions to the Hamilton–Jacobi–Bellman equation associated with a moderate deviations approximation are used to design importance sampling changes of measure for stochastic recursive equations. Analogous to what has been done for large deviations subsolution-based importance sampling, these schemes are shown to be asymptotically optimal under the moderate deviations scaling. Here, we present various implementations and numerical results to contrast their performance, and also discuss the circumstances under which a moderate deviation scaling might be appropriate.

97 MATHEMATICS AND COMPUTING↗

On the convergence of an IEQ-based first-order semi-discrete scheme for the Beris-Edwards system

We present a convergence analysis of an unconditionally energy-stable first-order semi-discrete numerical scheme designed for a hydrodynamic Q-tensor model, the so-called Beris-Edwards system, based on the Invariant Energy Quadratization Method (IEQ). The model consists of the Navier–Stokes equations for the fluid flow, coupled to the Q-tensor gradient flow describing the liquid crystal molecule alignment. By using the Invariant Energy Quadratization Method, we obtain a linearly implicit scheme, accelerating the computational speed. However, this introduces an auxiliary variable to replace the bulk potential energy and it is a priori unclear whether the reformulated system is equivalent to the Beris-Edward system. In this work, we prove stability properties of the scheme and show its convergence to a weak solution of the coupled liquid crystal system. We also demonstrate the equivalence of the reformulated and original systems in the weak sense.

Weber, Franziska↗

A study on leakage radiation dose at ELV-4 electron accelerator bunker

Shielding is an important aspect in the safety of an accelerator and the most important aspects of a bunker shielding is the door. The bunker’s door should be designed properly to minimize the leakage radiation and shall not exceed the permitted limit of 2.5μSv/hr. In determining the leakage radiation dose that passed through the door and gaps between the door and the wall, 2-dimensional manual calculations are often used. This method is hard to perform because visual 2-dimensional is limited and is also very difficult in the real situation. Therefore estimation values are normally performed. In doing so, the construction cost would be higher because of overestimate or underestimate which require costly modification to the bunker. Therefore in this study, two methods are introduced to overcome the problem such as simulation using MCNPX Version 2.6.0 software and manual calculation using 3-dimensional model from Autodesk Inventor 2010 software. The values from the two methods were eventually compared to the real values from direct measurements using Ludlum Model 3 with Model 44-9 probe survey meter.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Efficient two-component relativistic method for large systems

This paper reviews a series of theoretical studies to develop efficient two-component (2c) relativistic method for large systems by the author’s group. The basic theory is the infinite-order Douglas-Kroll-Hess (IODKH) method for many-electron Dirac-Coulomb Hamiltonian. The local unitary transformation (LUT) scheme can effectively produce the 2c relativistic Hamiltonian, and the divide-and-conquer (DC) method can achieve linear-scaling of Hartree-Fock and electron correlation methods. The frozen core potential (FCP) theoretically connects model potential calculations with the all-electron ones. The accompanying coordinate expansion with a transfer recurrence relation (ACE-TRR) scheme accelerates the computations of electron repulsion integrals with high angular momenta and long contractions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Experimental demonstration of externally driven millimeter-wave particle accelerator structure

We report the experimental demonstration of a mm-wave electron accelerating structure powered by a high-power rf source. We demonstrate reliable coupling of an unprecedented rf power—up to 575kW into the mm-wave accelerator structure using a quasi-optical setup. This standing wave accelerating structure consists of a single-cell copper cavity and a Gaussian to TM 01 mode converter. The accelerator structure is powered by 110 GHz, 10-ns long rf pulses. These pulses are chopped from 3 microsecond pulses from a gyrotron oscillator using a laser-driven silicon switch. We show an unprecedented high gradient up to 230MV/m that corresponds to a peak surface electric field of more than 520 MV/m. We have achieved these results after conditioning the cavity with more than 100,000 pulses. We also report preliminary measurements of rf breakdown rates, which are important for understanding rf breakdown physics in the millimeter-wave regime. Furthermore, these results open up many frontiers for applications not only limited to the next generation particle accelerators but also x-ray generation, probing material dynamics, and nonlinear light-matter interactions at mm-wave frequency.

43 PARTICLE ACCELERATORS↗

Consistent coupling algorithms for coupled core-edge simulations of plasma turbulence

Two schemes for coupling gyrokinetic simulations of microturbulence in tokamaks are proposed. The first scheme is based on an additive Schwarz domain decomposition. In this work, we show that, because the goal of turbulence is long-time averages of the dynamics rather than temporal accuracy, the iteration to self-consistency across domains, which is typically required by Schwarz schemes, can be avoided, thereby accelerating the computation. Second, we propose a coupling scheme that relies entirely on the addition of source terms, leaving the boundary conditions arbitrary. The practical motivations for such a scheme are discussed, and forms of the source terms that ensure consistency and stability are derived. The schemes are tested on a nonlinear, one-dimensional model problem, and the first scheme is further tested on the Hasegawa–Wakatani model.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Revealing the loss mechanisms of a 3D superconducting microwave cavity for use in a dark matter search

Superconducting microwave cavities have found applications in many areas including quantum computing, particle accelerators, and dark matter searches. Their extremely high quality factors translate to very narrow bandwidth, which makes them key components of sensitive detectors. In this study, we aim to understand the loss mechanisms of an aluminium cavity, with a fundamental mode frequency of about 2.5123 GHz, and how they change as the cavity material transitions from the superconducting to normal state. We found that at temperatures not much lower than the transition temperature $T_c$, losses are dominated by quasiparticle excitations and are well described by the Bardeen–Cooper–Schrieffer theory. The exponential decrease of the quasiparticle density below $T_c$ results in a 1000-fold increase of the quality factor, as well as a shift of the resonance frequency due to the change of the kinetic inductance of the superconductor. At very low temperatures, losses due to two-level systems begin to dominate giving a peak in the quality factor of about 2.76 x 10 7 at 130 mK. Understanding the loss mechanisms is invaluable, as the working temperature of the cavity may vary during operation regardless of its application.

Physics - Physics of elementary particles and fiel↗