Search NASA⌕ Search

SEARCH · Search NASA

Results for “Optimal sizing method”

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 1,387 records · Page 77
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Simulations of frustrated Ising Hamiltonians using quantum approximate optimization

Novel magnetic materials are important for future technological advances. Theoretical and numerical calculations of ground-state properties are essential in understanding these materials, however, computational complexity limits conventional methods for studying these states. Here we investigate an alternative approach to preparing materials ground states using the quantum approximate optimization algorithm (QAOA) on near-term quantum computers. We study classical Ising spin models on unit cells of square, Shastry-Sutherland and triangular lattices, with varying field amplitudes and couplings in the material Hamiltonian. We find relationships between the theoretical QAOA success probability and the structure of the ground state, indicating that only a modest number of measurements (≲100) are needed to find the ground state of our nine-spin Hamiltonians, even for parameters leading to frustrated magnetism. We further demonstrate the approach in calculations on a trapped-ion quantum computer and succeed in recovering each ground state of the Shastry-Sutherland unit cell with probabilities close to ideal theoretical values. The results demonstrate the viability of QAOA for materials ground state preparation in the frustrated Ising limit, giving important first steps towards larger sizes and more complex Hamiltonians where quantum computational advantage may prove essential in developing a systematic understanding of novel materials.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

Design and Application of CSC-Roll for Heavy Reduction of the Bloom Continuous Casting Process

Center quality control is a key issue affecting large-section bloom production and product quality. The central convex roll could avoid the bloom side part and yield a high reduction amount, which benefit the implement of heavy reduction (HR) for bloom continuous casting process without upgrading the mechanical and hydraulic system of withdrawal and straightening machine. In this work, a curving surface convex roll (CSC-Roll), which could significantly improve bloom center quality, was designed by finite element simulation for a 360 mm × 450 mm #45 steel continuous casting bloom. The optimal roll surface shape of the CSC-Roll was determined based on an analysis of the distribution of the reduction amount in the bloom center and the strain of the bloom centerline and inner surface. The calculation results showed that the optimal size of the CSC-Roll was 200 mm in convex platform region length, 30 mm in convex platform region height, and 80 mm in gradient curvature region length. The bloom heat transfer and deformation behavior in the CSC-Roll reduction process were then analyzed. The results showed that in the CSC-Roll HR process, 61.71 pct of the reduction amount could be transmitted to the inner side of the porosity zone. Macrographs of the 360 mm × 450 mm #45 steel continuous casting bloom proved that the bloom center quality was significantly improved after CSC-Roll HR was applied.

36 MATERIALS SCIENCE↗
DOE OSTI2014

Density functional theory and evolution algorithm calculations of elastic properties of AlON

Different models for aluminum oxynitride (AlON) were calculated using density functional theory and optimized using an evolutionary algorithm. Evolutionary algorithm and density functional theory (DFT) calculations starting from several models of AlON with different Al or O vacancy locations and different positions for the N atoms relative to the vacancy were carried out. The results show that the constant anion model [McCauley et al., J. Eur. Ceram. Soc. 29(2), 223 (2009)] with a random distribution of N atoms not adjacent to the Al vacancy has the lowest energy configuration. The lowest energy structure is in a reasonable agreement with experimental X-ray diffraction spectra. The optimized structure of a 55 atom unit cell was used to construct 220 and 440 atom models for simulation cells using DFT with a Gaussian basis set. Cubic elastic constant predictions were found to approach the experimentally determined AlON single crystal elastic constants as the model size increased from 55 to 440 atoms. The pressure dependence of the elastic constants found from simulated stress-strain relations were in overall agreement with experimental measurements of polycrystalline and single crystal AlON. Calculated IR intensity and Raman spectra are compared with available experimental data.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2017

Towards Highly Scalable Ab Initio Molecular Dynamics (AIMD) Simulations on the Intel Knights Landing Manycore Processor

The Ab Initio Molecular Dynamics (AIMD) method allows scientists to treat the dynamics of molecular and condensed phase systems while retaining a first-principles-based description of their interactions. This extremely important method has tremendous computational requirements, because the electronic Schr¨odinger equation, approximated using Kohn-Sham Density Functional Theory (DFT), is solved at every time step. With the advent of manycore architectures, application developers have a significant amount of processing power within each compute node that can only be exploited through massive parallelism. A compute intensive application such as AIMD forms a good candidate to leverage this processing power. In this paper, we focus on adding thread level parallelism to the plane wave DFT methodology implemented in NWChem. Through a careful optimization of tall-skinny matrix products, which are at the heart of the Lagrange multiplier and nonlocal pseudopotential kernels, as well as 3D FFTs, our OpenMP implementation delivers excellent strong scaling on the latest Intel Knights Landing (KNL) processor. We assess the efficiency of our Lagrange multiplier kernels by building a Roofline model of the platform, and verify that our implementation is close to the roofline for various problem sizes. Finally, we present strong scaling results on the complete AIMD simulation for a 64 water molecules test case, that scales up to all 68 cores of the Knights Landing processor.

Environmental Molecular Sciences Laboratory↗
DOE OSTI2016

Technical Note: Spot characteristic stability for proton pencil beam scanning

Purpose: The spot characteristics for proton pencil beam scanning (PBS) were measured and analyzed over a 16 month period, which included one major site configuration update and six cyclotron interventions. The results provide a reference to establish the quality assurance (QA) frequency and tolerance for proton pencil beam scanning. Methods: A simple treatment plan was generated to produce an asymmetric 9-spot pattern distributed throughout a field of 16 × 18 cm for each of 18 proton energies (100.0–226.0 MeV). The delivered fluence distribution in air was measured using a phosphor screen based CCD camera at three planes perpendicular to the beam line axis (x-ray imaging isocenter and up/down stream 15.0 cm). The measured fluence distributions for each energy were analyzed using in-house programs which calculated the spot sizes and positional deviations of the Gaussian shaped spots. Results: Compared to the spot characteristic data installed into the treatment planning system, the 16-month averaged deviations of the measured spot sizes at the isocenter plane were 2.30% and 1.38% in the IEC gantry x and y directions, respectively. The maximum deviation was 12.87% while the minimum deviation was 0.003%, both at the upstream plane. After the collinearity of the proton and x-ray imaging system isocenters was optimized, the positional deviations of the spots were all within 1.5 mm for all three planes. During the site configuration update, spot positions were found to deviate by 6 mm until the tuning parameters file was properly restored. Conclusions: For this beam delivery system, it is recommended to perform a spot size and position check at least monthly and any time after a database update or cyclotron intervention occurs. A spot size deviation tolerance of <15% can be easily met with this delivery system. Deviations of spot positions were <2 mm at any plane up/down stream 15 cm from the isocenter.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

Universal method for the optimization of HDC coating uniformity on non-planar, non-stationary substrates for inertial confinement fusion targets

The thickness uniformity of chemical vapor deposited (CVD) diamond coatings on non-planar, non-stationary substrates depends on both the intrinsic instantaneous coating thickness distribution (ICTD) of the coating conditions used and, if applicable, on the frequency of substrate reorientation. While important for many CVD diamond applications, the relative impact of the ICTD and substrate reorientation on the coating thickness uniformity has not been studied. In this work, we systematically investigate the effect of these factors for microwave-plasma chemical vapor deposition (MPCVD) of diamond (referred to as high density carbon (HDC) in the inertial confinement fusion (ICF) community) coatings on spherical, rolling substrates. This coating technique is used to fabricate capsules for ICF experiments, which require extreme coating uniformity with <0.3 % thickness variation (so-called Mode 1 or M1) to ensure symmetric compression of imploding targets. To extract the otherwise unobservable reorientation timescale (Δt), Monte Carlo simulations were performed using experimental ICTD data as input. This combined approach confirms scaling relationships between the substrate reorientation timescale as well as coating thickness and coating uniformity, as expected from a 3D random walk. Simulations confirm that M1 is Rayleigh-distributed and scales as (Δt) 1/2 , consistent with the randomization of two angles that determine orientation of a sphere. We also demonstrate that, under the conditions studied, Δt is the dominant factor in determining thickness uniformity while the intrinsic ICTD has minimal impact. Finally, experiments show that Δt can be affected by total batch size under constant agitation conditions due to space constraints that limit the capsule reorientation kinetics. In conclusion, this study highlights the utility of a combined experiment-simulation approach as a general methodology for understanding and improving coating uniformity on non-planar, non-stationary substrates.

Capsule↗
DOE OSTI2015

Electronic and magnetic properties of small rhodium clusters

We report a theoretical study of the electronic and magnetic properties of rhodium-atomic clusters. The lowest energy structures at the semi-empirical level of rhodium clusters are first obtained from a novel global-minimum search algorithm, known as PTMBHGA, where Gupta potential is used to describe the atomic interaction among the rhodium atoms. The structures are then re-optimized at the density functional theory (DFT) level with exchange-correlation energy approximated by Perdew-Burke-Ernzerhof generalized gradient approximation. For the purpose of calculating the magnetic moment of a given cluster, we calculate the optimized structure as a function of the spin multiplicity within the DFT framework. The resultant magnetic moments with the lowest energies so obtained allow us to work out the magnetic moment as a function of cluster size. Rhodium atomic clusters are found to display a unique variation in the magnetic moment as the cluster size varies. However, Rh{sub 4} and Rh{sub 6} are found to be nonmagnetic. Electronic structures of the magnetic ground-state structures are also investigated within the DFT framework. The results are compared against those based on different theoretical approaches available in the literature.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

Medial axis and local thickness computation using the Fast Sweeping Method

This report describes an efficient and robust voxel-based methodology for computing the medial axis, local thickness, and distance-to-skeleton of arbitrary three-dimensional geometries. It is assumed that the object can be represented by an exact or approximate signed distance function on a discrete grid. The gradient of such function is used to formulate a hyperbolic partial differential equation (PDE) that models the collapse of the position vector in space. By exploiting the causality property of the PDE, the Fast Sweeping Method is able to obtain the solution in a finite number of sweeps independent of the mesh resolution. The intersection of characteristic lines leads to the formation of shocks and a discrete bisector function is used to identify the medial axis. The same PDE approach is used to compute the local thickness inside the object and obtain the distance-to-skeleton field. Multiple examples are given in two and three dimensions along with a resolution study. The methodology has optimal complexity and yields subsecond computational times for geometries with over a million zones on a single core. The methodology is also capable of parallelization across shared and distributed memory architectures.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Learning to Optimize Variational Quantum Circuits to Solve Combinatorial Problems

Quantum computing is a computational paradigm with the potential to outperform classical methods for a variety of problems. Proposed recently, the Quantum Approximate Optimization Algorithm (QAOA) is considered as one of the leading candidates for demonstrating quantum advantage in the near term. QAOA is a variational hybrid quantum-classical algorithm for approximately solving combinatorial optimization problems. The quality of the solution obtained by QAOA for a given problem instance depends on the performance of the classical optimizer used to optimize the variational parameters. In this paper, we formulate the problem of finding optimal QAOA parameters as a learning task in which the knowledge gained from solving training instances can be leveraged to find high-quality solutions for unseen test instances. To this end, we develop two machine-learning-based approaches. Our first approach adopts a reinforcement learning (RL) framework to learn a policy network to optimize QAOA circuits. Our second approach adopts a kernel density estimation (KDE) technique to learn a generative model of optimal QAOA parameters. In both approaches, the training procedure is performed on small-sized problem instances that can be simulated on a classical computer; yet the learned RL policy and the generative model can be used to efficiently solve larger problems. Furthermore, extensive simulations using the IBM Qiskit Aer quantum circuit simulator demonstrate that our proposed RL- and KDE-based approaches reduce the optimality gap by factors up to 30.15 when compared with other commonly used off-the-shelf optimizers.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Understanding the Effect of Na in Improving the Performance of CuInSe 2 Based Photovoltaics

Cu(In,Ga)Se 2 (CIGS) thin film photovoltaic technology is in the early stages of commercialization with an annual manufacturing capacity over 1 GW and has demonstrated the highest module efficiency of any of the thin film technologies. However there still is a lack of fundamental understanding of the relationship between the material properties and solar cell device operation. It is well known that the incorporation of a small amount of Na into the CIGS film during processing is essential for high efficiency devices. However, there are conflicting explanations for how Na behaves at the atomic scale. This report investigates how Na is incorporated into the CIGS device structure and evaluates the diffusion of Na into CIGS grain boundaries (GBs) and bulk crystallites. Participants: This project was carried out at the Institute of Energy Conversion at the University of Delaware, collaborating with the Rockett group at the University of Illinois Urbana-Champagne. Significant Findings: The significant outcomes of this project for each task include; Task 1.0: Effect of Na in Devices Fabricated on PVD Deposited CIGS; Na diffusion occurs through the Mo back contact via GBs driven by the presence of oxygen; Na reversibly compensates donor defects in CIGS GBs,Task 2.0: Na Incorporation in Single Crystal CIGS; and bulk Na diffusion proceeds rapidly such that grains are Na-saturated immediately following CIGS thin film manufacture. Industry Guidance: The presented results offer interesting concepts for modification of manufacturing processes of CIGS-based PV modules. Possible approaches to improve control of Na uptake and uniformly increase levels in CIGS films are highlighted for processes that employ either soda-lime glass or NaF as the Na source. Concepts include the potential of O 2 or oxidative based treatments of Mo back contacts to improve Na diffusion through the metal film and increase Na uptake into the growing CIGS. This project has also offered fundamental understanding of the behavior of Na in CIGS grains and GBs, particularly the confirmation that CIGS grains will be saturated with Na immediately following manufacture Summary of Results: Most commercially available CIGS modules are fabricated on soda-lime glass coated with Mo as the back electric contact, and Na in the glass diffuses through the Mo layer into the CIGS during film growth. In Task 1 the transport of Na through Mo was evaluated using x-ray photoelectron surface spectroscopy along with diffusion modeling to obtain diffusion coefficients at several temperatures. It was determined that Na diffusion in Mo only occurs along GBs and that oxygen provides an additional driving force to enhance Na transport. Device data revealed that older Mo substrates with a greater amount of surface oxide resulted in slightly higher efficiencies due to enhanced Na incorporation caused by the oxide. This finding shows that Mo substrates could potentially undergo an oxidation treatment prior to CIGS deposition to further improve and control the incorporation of Na. To determine if in-grain Na affects device performance, in Task 1 Na was selectively removed from GBs using heat/rinse cycles. Due to the low temperature of this treatment, Na at GBS remained mobile while diffusion within the bulk was too slow for Na removal from the grain interiors. Changes in electrical properties were evaluated using conductivity and Seebeck coefficient measurements, with both decreasing as Na was removed to reach values similar to Na-free controls samples. This can be explained by the compensation of donor defects by Na, causing an increase in the free carrier concentration. Devices showed a decrease in open-circuit voltage after Na removal confirming that the beneficial effects of GB Na. The findings of this project will provide guidance for rational optimization of Na incorporation procedures in the manufacturing of CIGS solar cells. While it is known that Na segregates at CIGS GBs, the nature and role of Na diffusion into grain interiors was less clear. In Task 2, single crystal CuInSe 2 was used as a model system to represent the grain interiors of CIGS. Crystals processed by two different methods of different compositions and dislocation densities, were evaluated. Diffusion coefficients were obtained at two temperatures after Na diffusion, giving near identical values, ~2x10 11 cm 2 /s and ~6x10 11 cm2/s at 420°C and 480°C, respectively, for each crystal. Characterization confirmed that dislocation densities were too low to significantly impact the effective diffusion coefficient. The Cu-poor crystal had a higher solubility suggesting that Na diffusion is mediated by Cu-vacancies, but was not accompanied by an expected increase in diffusion coefficient. The activation energy for diffusion was similar to values expected for interstitial diffusion, but the large size of Na + ions should result in a solubility that is much lower than what was experimentally measured. A hybrid interstitial-substitutional mechanism is proposed that combines the fast diffusion of interstitial atoms with the high solubility common for substitutional impurities. Lattice diffusion of Na proceeds fast enough that CIGS grain interiors will have Na concentrations near the solubility limit of 1018 cm -3 when manufactured under standard conditions. Na and K treated epitaxial CIS films showed a significant increase in cathodoluminescence emission intensity, indicating a reduction of non-radiative recombination pathways, which is consistent with improvements in CIGS device performance, though the mechanism is not clear. Pathways forward: Despite the success of this project, there are a number of questions remaining related to further the understanding of the chemistry of Na in CIGS films and devices. These include further elucidation of the mechanisms of Na passivation in CIGS GBs, with identification of which defects are involved and confirmation of the possible effects of in-grain Na on device performance. To complete analysis of the cell structure, conformation of the presence and possible chemistries of Na at the CIGS/CdS junction and/or in the front transparent contacts, and its effects on device performance, is needed.

14 SOLAR ENERGY↗
DOE OSTI2024

Electron cyclotron emission detection of neoclassical tearing modes for control for ITER

Successful operation of ITER requires control of magnetic instabilities including neoclassical tearing modes (NTMs) that can degrade confinement and lead to disruption. Low latency detection by electron cyclotron emission (ECE) diagnostics has been demonstrated in a few current experiments. Using a synthetic diagnostic, we demonstrate low latency NTM detection for ITER with plasmas described by ITER IMAS database scenarios and with realistic limitations imposed on the instrumentation by these high temperature scenarios. 2/1 NTMs are detected 430 ms after magnetic island seeding and before island locking. The radiometer configuration was optimized using simulation, and the smallest detectable island size was explored. Island sizes of ∼3 cm are detectable at the 2/1 surface. The simulated signals incorporate recent physics models for island growth and rotation, which show early locking and continued island growth after locking and before disruption. This work determines limits for ITER ECE spatial resolution imposed by relativistic broadening of channels, which informs hardware design. Real-time detection is demonstrated in hardware that is required by ITER, including on an NI PXI-7853R FPGA system. Development of a synthetic diagnostic and details of the hardware will be discussed.

Cyclotron radiation↗
DOE OSTI2017

Molecular weight and uniformity define the mechanical performance of lignin-based carbon fiber

Lignin-based carbon fiber represents a renewable and low-cost alternative to petroleum-based carbon fiber. However, the poor mechanical performance of current lignin carbon fiber hinders its application. We hypothesized that the lower optimal mechanical performance is caused by the inherent heterogeneity of lignin. It is still unknown how the molecular weights (MWs) and lignin uniformity will impact the performance of lignin carbon fiber. We thereby addressed this hypothesis by fractionating lignin into fractions with different MWs and polydispersity indices (PDIs). An enzyme–mediator-based method and a dialysis method were developed to derive lignin fractions with increased MW and decreased PDI. Lignin fractions were electro-spun into fibers after blending with polyacrylonitrile (PAN) at 1 : 1 (w/w) ratio. The fractionation in general improved the spinnability of lignin to allow us to obtain finer lignin-based carbon fibers. The elastic modulus of lignin carbon fibers, as measured by nanoindentation, was increased as the lignin MW increased and as PDI decreased. The scatter plot and linear regression revealed very good correlation between the elastic modulus and PDI, as well as certain correlation between the elastic modulus and MW. XRD and Raman spectroscopy revealed that the crystallite size and the content of the pre-graphitic turbostratic carbon increased with higher lignin MW and lower PDI, revealing the mechanism of the improvement in carbon fiber mechanical performance. Finally, this study elucidated the impacts of lignin MW and uniformity on the mechanical properties of carbon fiber, and could thus guide the development of lignin processing technologies for quality lignin-based carbon fiber.

09 BIOMASS FUELS↗
DOE OSTI2015

Limits of carrier mobility in Sb-doped SnO{sub 2} conducting films deposited by reactive sputtering

Electron transport in Sb-doped SnO{sub 2} (ATO) films is studied to unveil the limited carrier mobility observed in sputtered films as compared to other deposition methods. Transparent and conductive ATO layers are deposited from metallic tin targets alloyed with antimony in oxygen atmosphere optimized for reactive sputtering. The carrier mobility decreases from 24 cm{sup 2} V{sup −1} s{sup −1} to 6 cm{sup 2} V{sup −1} s{sup −1} when increasing the doping level from 0 to 7 at. %, and the lowest resistivity of 1.8 × 10{sup −3} Ω cm corresponding to the mobility of 12 cm{sup 2} V{sup −1} s{sup −1} which is obtained for the 3 at. % Sb-doped ATO. Temperature-dependent Hall effect measurements and near-infrared reflectance measurements reveal that the carrier mobility in sputtered ATO is limited by ingrain scattering. In contrast, the mobility of unintentionally doped SnO{sub 2} films is determined mostly by the grain boundary scattering. Both limitations should arise from the sputtering process itself, which suffers from the high-energy-ion bombardment and yields polycrystalline films with small grain size.

36 MATERIALS SCIENCE↗
DOE OSTI2019

Transfer of autocollimator calibration for use with scanning gantry profilometers for accurate determination of surface slope and curvature of state-of-the-art x-ray mirrors

X-ray optics, desired for beamlines at free-electron-laser and diffraction-limited-storage-ring x-ray light sources, must have almost perfect surfaces, capable of delivering light to experiments without significant degradation of brightness and coherence. To accurately characterize such optics at an optical metrology lab, two basic types of surface slope profilometers are used: The long trace profilers (LTPs) and nanometer optical measuring (NOM) like angular deflectometers, based on electronic autocollimator (AC) ELCOMAT-3000. The inherent systematic errors of the instrument's optical sensors set the principle limit to their measuring performance. Where autocollimator of a NOM-like profiler may be calibrated at a unique dedicated facility, this is for a particular configuration of distance, aperture size, and angular range that does not always match the exact use in a scanning measurement with the profiler. Here we discuss the developed methodology, experimental set-up, and numerical methods of transferring the calibration of one reference AC to the scanning AC of the Optical Surface Measuring System (OSMS), recently brought to operation at the ALS Xray Optics Laboratory. We show that precision calibration of the OSMS performed in three steps, allows us to provide high confidence and accuracy low-spatial-frequency metrology and not â€print into' measurements the inherent systematic error of tool in use. Here, with the examples of the OSMS measurements with a state-of-The-Art x-ray aspherical mirror, available from one of the most advanced vendors of X-ray optics, we demonstrate the high efficacy of the developed calibration procedure. The results of our work are important for obtaining high reliability data, needed for sophisticated numerical simulations of beamline performance and optimization of beamline usage of the optics. This work was supported by the U. S. Department of Energy under contract number DE-AC02-05CH11231.

47 OTHER INSTRUMENTATION↗
DOE OSTI2016

PolyCheck: Dynamic Verification of Iteration Space Transformations on Affine Programs

High-level compiler transformations, especially loop transformations, are widely recognized as critical optimizations to restructure programs to improve data locality and expose parallelism. Guaranteeing the correctness of program transformations is essential, and to date three main approaches have been developed: proof of equivalence of affine programs, matching the execution traces of programs, and checking bit-by-bit equivalence of the outputs of the programs. Each technique suffers from limitations in either the kind of transformations supported, space complexity, or the sensitivity to the testing dataset. In this paper, we take a novel approach addressing all three limitations to provide an automatic bug checker to verify any iteration reordering transformations on affine programs, including non-affine transformations, with space consumption proportional to the original program data, and robust to arbitrary datasets of a given size. We achieve this by exploiting the structure of affine program control- and data-flow to generate at compile-time lightweight checker code to be executed within the transformed program. Experimental results assess the correctness and effectiveness of our method, and its increased coverage over previous approaches.

affine programs↗
DOE OSTI2016

Further Investigation of Employing Binomial Small-Sample Estimator in M-LROM

Reduced Order Modeling (ROM) is essential to enable the identification of the so-called active parameter subspace for high-fidelity models that are employed in computationally intensive analyses such as uncertainty characterization and data assimilation. Given the complexity of models for real-world application, brute force methods to identify the active subspace may not be computationally efficient to realize the value of reduction, since in realistic problems, the full scale model must be executed few hundred times, which may be computationally taxing. Earlier work has introduced a new approach, which represents a further development of our previous work on ROM, referred to as multi-level ROM (M-LROM). In M-LROM, approximate models are employed to identify the active subspace instead of the full scale model. In this summary, we present our recent results on M-LROM, specifically in regard to the construction of reliable error metrics that can bound the errors resulting from the reduction. The idea is to employ a lower fidelity model, i.e., an approximation of the full scale or high fidelity model, to build the active subspace, and via the proposed error bound, the quality of the active subspace can be determined by calculating an upper bound on the errors resulting from the reduction. For this approach to be practical, the error bound must be realistic, implying that it should be close to the actual errors expected, and not much higher. Therefore, an important goal of this summary is to compare different error bounds in terms of the relation between their estimated bounds to the actual errors. In particular, we compare the proposed error bound which numerically construct an estimator using binomial distribution to the widely used small-sample norm estimators. Unlike most probabilistic error estimation methods, the proposed bound employs a binomial distribution to construct the estimator whereas in other conventional methods, the distribution employed is typically the standard Gaussian, normalized over a unit hypersphere as initially proposed by the seminal paper of J. Dixon on random matrix theory. Previous publications provided evidence that the binomial distribution calculates more realistic bounds as compared to the Gaussian distribution, which tends to be conservative, implying that the bound is much bigger than the actual errors, which diminishes the value of the reduction, as now must increase the size of the active subspace to reach the desired upper bound on the reduction errors. This work has provided a rigorous approach by which error bounds on the reduction errors can be established without having to execute the high fidelity model during the construction of the active subspace. The high fidelity model is only need to test the error bound using small number of executions. This methodology paves the way for a more efficient application of ROM to complex engineering models, where standard ROM techniques may be too taxing to optimally realize their value in practical situations. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2023

An automated and portable method for selecting an optimal GPU frequency

Power consumption poses a significant challenge in current and emerging graphics processing unit (GPU) enabled high-performance computing systems. In modern GPUs, dynamic voltage frequency scaling (DVFS) appears to be a reliable control to regulate power consumption and performance. However, the DVFS design space is large - hence, brute-force approaches are infeasible to select the optimal frequency. Furthermore, no single frequency can be universally optimal for applications with varying computational intensities. Thus, the application's complexity and the availability of a wide range of frequency settings are a challenge in selecting the optimal frequency configuration for a given GPU workload. To that end, this paper proposes a systematic approach that consists of three steps. The feature characterization study identifies the fine-grain GPU utilization metrics that influence the power consumption and execution time of a given workload. To understand the performance, power, and energy consumption behaviors of a workload across GPU's DVFS design space, we derived analytical power and performance models using the identified fine-grain features. Here, it is shown that the same set of GPU utilization metrics can estimate both the power consumption and execution time while being agnostic of changes to frequency and input sizes. Applying a power control with the single objective of reducing power may cause performance degradation, leading to more energy consumption. A multi-objective approach is proposed to select the optimal GPU DVFS configuration for a workload that reduces power consumption with negligible degradation in performance. The evaluation was conducted using SPEC ACCEL benchmarks and three real applications - NAMD LAMMPS, and LSTM on NVIDIA GV100, GA100, and AMD MI210 GPUs. On average, real applications showed 29.6% energy savings with a performance loss of 5.2% on GA100 and 22.6% energy savings with a performance loss of 4.7% on GV100. Moreover, the proposed models are portable to real applications, GPU architectures, and vendors, and require metric collection at only the default frequency rather than all supported DVFS configurations. Additionally, we conducted a comparison between our models and the GPU assembly instructions (PTX)-based static models. The results revealed a significant reduction in the average error rates, with a decrease from 19.7% to 3.1% for power models and from 29.4% to 5.2% for performance models.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Inverse methods for design of soft materials

Functional soft materials, comprising colloidal and molecular building blocks that self-organize into complex structures as a result of their tunable interactions, enable a wide array of technological applications. Inverse methods provide a systematic means for navigating their inherently high-dimensional design spaces to create materials with targeted properties. Furthermore, while multiple physically motivated inverse strategies have been successfully implemented in silico, their translation to guiding experimental materials discovery has thus far been limited to a handful of proof-of-concept studies. In this perspective, we discuss recent advances in inverse methods for design of soft materials that address two challenges: (1) methodological limitations that prevent such approaches from satisfying design constraints and (2) computational challenges that limit the size and complexity of systems that can be addressed. Strategies that leverage machine learning have proven particularly effective, including methods to discover order parameters that characterize complex structural motifs and schemes to efficiently compute macroscopic properties from the underlying structure. We also highlight promising opportunities to improve the experimental realizability of materials designed computationally, including discovery of materials with functionality at multiple thermodynamic states, design of externally directed assembly protocols that are simple to implement in experiments, and strategies to improve the accuracy and computational efficiency of experimentally relevant models.

36 MATERIALS SCIENCE↗