Search NASA⌕ Search

SEARCH · Search NASA

Results for “computationally expensive function”

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 127 records · Page 7
All sourcesNASA NTRSDOE OSTI
DOE OSTI2019

Surrogate Optimization of Computationally Expensive Black-Box Problems with Hidden Constraints (SHEBO) v1.0

SHEBO (surrogate optimization of problems with hidden constraints and expensive black-box objectives) is an efficient optimization algorithm that employs surrogate models to solve computationally expensive black-box simulation optimization problems that have hidden constraints. Hidden constraints are encountered when the objective function evaluation does not return a value for a parameter vector. These constraints are often encountered in optimization problems in which the objective function is computed by a black-box simulation code. SHEBO uses a combination of local and global search strategies together with an evaluability prediction function and a dynamically adjusted evaluability threshold to iteratively select new sample points.

Mueller, Juliane↗
DOE OSTI2022

Many-fermion simulation from the contracted quantum eigensolver without fermionic encoding of the wave function

Quantum computers potentially have an exponential advantage over classical computers for the quantum simulation of many-fermion quantum systems. Nonetheless, fermions are more expensive to simulate than bosons due to the fermionic encoding—a mapping by which the qubits are encoded with fermion statistics. Here we generalize the contracted quantum eigensolver (CQE) to avoid fermionic encoding of the wave function. In contrast to the variational quantum eigensolver, the CQE solves for a many-fermion stationary state by minimizing the contraction (projection) of the Schrödinger equation onto two fermions. We avoid fermionic encoding of the wave function by contracting the Schrödinger equation onto an unencoded pair of particles. Solution of the resulting contracted equation by a series of unencoded two-body exponential transformations generates an unencoded wave function from which the energy and two-fermion reduced density matrix (2-RDM) can be computed. We apply the unencoded and the encoded CQE algorithms to the hydrogen fluoride molecule, the dissociation of oxygen O 2 , and a series of hydrogen chains. Both algorithms show comparable convergence towards the exact ground-state energies and 2-RDMs, but the unencoded algorithm has computational advantages in terms of state preparation and tomography.

74 ATOMIC AND MOLECULAR PHYSICS↗
DOE OSTI2025

A generalized and adaptable tensor-contraction-based cluster expansion formalism for multicomponent solids

Density functional theory (DFT)-based simulations of materials have first-principles accuracy, but are very computationally expensive. For simulating various properties of multi-component alloys, the cluster expansion (CE) technique has served as the standard workaround to improve computational efficiency. However, the standard CE technique is difficult to extend to exotic and/or low-symmetry lattices, often implemented via iteration over particular cluster types, which must be enumerated per lattice structure. In this work, we introduce the tensor cluster expansion (TCE), implemented in the open-source code tce-lib, which maps correlation functions to mixed tensor contractions, eliminating the need to iterate over cluster types and additionally making the calculation of correlation functions well-suited for massively parallel architectures like GPUs. We show that local interaction energies are an immediate consequence of the TCE formalism, yielding nearly $\mathcal{O}$(1) energy difference calculations. We then use this formalism to fit CE models for the TaW and CoNiCrFeMn systems, and use these models to respectively compute the enthalpy of mixing curve and Cowley short-range order parameters, showing excellent agreement with ground truth data.

Cluster expansion↗
DOE OSTI2017

Final Report of the Project "From the finite element method to the virtual element method"

The Finite Element Method (FEM) is a powerful numerical tool that is being used in a large number of engineering applications. The FEM is constructed on triangular/tetrahedral and quadrilateral/hexahedral meshes. Extending the FEM to general polygonal/polyhedral meshes in straightforward way turns out to be extremely difficult and leads to very complex and computationally expensive schemes. The reason for this failure is that the construction of the basis functions on elements with a very general shape is a non-trivial and complex task. In this project we developed a new family of numerical methods, dubbed the Virtual Element Method (VEM) for the numerical approximation of partial differential equations (PDE) of elliptic type suitable to polygonal and polyhedral unstructured meshes. We successfully formulated, implemented and tested these methods and studied both theoretically and numerically their stability, robustness and accuracy for diffusion problems, convection-reaction-diffusion problems, the Stokes equations and the biharmonic equations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Fast Emulation of Expensive Simulations using Approximate Gaussian Processes [Slides]

Nuclear Computational Low-Energy Initiative (NUCLEI) collaboration uses Density Functional Theory (DFT) simulations to predict the structure and binding energies of nuclei over a wide range of proton (Z) and neutron (N) numbers. The DFT simulations utilize a particular parameterization of a Skyrme energy density functional called UNEDF1 which depends on 12 free parameters that must be fit to data (M Kortelainen et al 2014). Fitting involves comparing (e.g.) predicted binding energies of nuclei to experimentally measured values. We use only binding energies as observables, but DFT with UNEDF1 will predict structure (shape) observables as well. In this work, assessing the capability of approximate GP emulators to balance emulator accuracy with computational speed to facilitate improved UNEDF1 calibration. Sparse GPs are straightforward to train and accurate. Calibration is not straightforward with MCMC (using MH or HMC/NUTS). We produced reusable software for continuing and building on this work as well as accessing and using Darwin cluster compute resources

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

SchrödingerNet: A Universal Neural Network Solver for the Schrödinger Equation

Recent advances in machine learning have facilitated numerically accurate solution of the electronic Schrödinger equation (SE) by integrating various neural network (NN)-based wave function ansatzes with variational Monte Carlo methods. Nevertheless, such NN-based methods are all based on the Born–Oppenheimer approximation (BOA) and require computationally expensive training for each nuclear configuration. In this work, we propose a novel NN architecture, SchrödingerNet, to solve the full electronic-nuclear SE by defining a loss function designed to equalize local energies across the system. This approach is based on a translationally, rotationally and permutationally symmetry-adapted total wave function ansatz that includes both nuclear and electronic coordinates. Furthermore, this strategy not only allows for an efficient and accurate generation of a continuous potential energy surface at any geometry within the well-sampled nuclear configuration space, but also incorporates non-BOA corrections, through a single training process. Comparison with benchmarks of atomic and small molecular systems demonstrates its accuracy and efficiency.

Chemical calculations↗
DOE OSTI2024

Efficient mapping between void shapes and stress fields using Deep Convolutional Neural Networks with sparse data

Establishing fast and accurate structure-to-property relationships is an important component in the design and discovery of advanced materials. Physics-based simulation models like the finite element method (FEM) are often used to predict deformation, stress, and strain fields as a function of material microstructure in material and structural systems. Such models may be computationally expensive and time intensive if the underlying physics of the system is complex. This limits their application to solve inverse design problems and identify structures that maximize performance. In such scenarios, surrogate models are employed to make the forward mapping computationally efficient to evaluate. However, the high dimensionality of the input microstructure and the output field of interest often renders such surrogate models inefficient, especially when dealing with sparse data. Deep convolutional neural network (CNN) based surrogate models have shown great promise in handling such high-dimensional problems. In this paper, a single ellipsoidal void structure under a uniaxial tensile load represented by a linear elastic, high-dimensional and expensive-to-query, FEM model. We consider two deep CNN architectures, a modified convolutional autoencoder framework with a fully connected bottleneck and a UNet CNN, and compare their accuracy in predicting the von Mises stress field for any given input void shape in the FEM model. Additionally, a sensitivity analysis study is performed using the two approaches, where the variation in the prediction accuracy on unseen test data is studied through numerical experiments by varying the number of training samples from 20 to 100.

surrogate modeling; convolutional neural networks;↗
DOE OSTI2021

Machine learning to accelerate screening for Marcus reorganization energies

Understanding and predicting the charge transport properties of π-conjugated materials is an important challenge for designing new organic electronic devices, such as solar cells, plastic transistors, light-emitting devices, and chemical sensors. A key component of the hopping mechanism of charge transfer in these materials is the Marcus reorganization energy which serves as an activation barrier to hole or electron transfer. While modern density functional methods have proven to accurately predict trends in intramolecular reorganization energy, such calculations are computationally expensive. In this work, we outline active machine learning methods to predict computed intramolecular reorganization energies of a wide range of polythiophenes and their use toward screening new compounds with low internal reorganization energies. Our models have an overall root mean square error (RMSE) of ±0.113 eV, but a much smaller RMSE of only ±0.036 eV on the new screening set. Since the larger error derives from high-reorganization energy compounds, the new method is highly effective to screen for compounds with potentially efficient charge transport parameters.

14 SOLAR ENERGY↗
DOE OSTI2020

Predictive Model of Charge Mobilities in Organic Semiconductor Small Molecules with Force-Matched Potentials

Charge mobility of crystalline organic semiconductors (OSC) is limited by local dynamic disorder. Recently, the charge mobility for several high mobility OSCs, including TIPS-pentacene, were accurately predicted from a density functional theory (DFT) simulation constrained by the crystal structure and the inelastic neutron scattering spectrum, which provide direct measures of the structure and the dynamic disorder in the length scale and energy range of interest. However, the computational expense required for calculating all of the atomic and molecular forces is prohibitive. In this work, we demonstrate the use of density functional tight binding (DFTB), a semiempirical quantum mechanical method that is 2 to 3 orders of magnitude more efficient than DFT. We show that force matching a many-body interaction potential to DFT derived forces yields highly accurate DFTB models capable of reproducing the low-frequency intricacies of experimental inelastic neutron scattering (INS) spectra and accurately predicting charge mobility. We subsequently predicted charge mobilities from our DFTB model of a number of previously unstudied structural analogues to TIPS-pentacene using dynamic disorder from DFTB and transient localization theory. The approach we establish here could provide a truly rapid simulation pathway for accurate materials properties prediction, in our vision applied to new OSCs with tailored properties.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2014

Development of EEM based silicon–water and silica–water wall potentials for non-reactive molecular dynamics simulations

Molecular dynamics simulations of heat transfer in gases are computationally expensive when the wall molecules are explicitly modeled. To save computational time, an implicit boundary function is often used. Steele's potential has been used in studies of fluid–solid interface for a long time. In this work, the conceptual idea of Steele's potential was extended in order to simulate water–silicon and water–silica interfaces. A new wall potential model is developed by using the electronegativity-equalization method (EEM), a ReaxFF empirical force field and a non-reactive molecular dynamics package PumMa. Contact angle simulations were performed in order to validate the wall potential model. Contact angle simulations with the resulting tabulated wall potentials gave a silicon–water contact angle of 129°, a quartz–water contact angle of 0°, and a cristobalite–water contact angle of 40°, which are in reasonable agreement with experimental values.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

Kernel fusion in atomistic spin dynamics simulations on Nvidia GPUs using tensor core

In atomistic spin dynamics simulations, the time cost of constructing the space- and time-displaced pair correlation function in real space increases quadratically as the number of spins N, leading to significant computational effort. The GEMM subroutine can be adopted to accelerate the calculation of the dynamical spin-spin correlation function, but the computational cost of simulating large spin systems (>40000 spins) on CPUs remains expensive. In this work, we perform the simulation on the graphics processing unit (GPU), a hardware solution widely used as an accelerator for scientific computing and deep learning. Here we show that GPUs can accelerate the simulation up to 25-fold compared to multi-core CPUs when using the GEMM subroutine on both. To hide memory latency, we fuse the element-wise operation into the GEMM kernel using CUTLASS that can improve the performance by 26% ~ 33% compared to implementation based on cuBLAS. Furthermore, we perform the on-the-fly calculation in the epilogue of the GEMM subroutine to avoid saving intermediate results on global memory, which makes the large-scale atomistic spin dynamics simulation feasible and affordable.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Orbit-modulated transport and sources in time-dependent plasmas

The orbit-modulation framework for steady-state ion-orbit boundary fluxes is reformulated to apply to time-dependent plasmas. The new formulation can quantify the extent to which orbit fluxes are affected by the time dependence of the distribution function, and may be applied both to time-dependent physical scenarios and to computationally expensive simulations that are unable to reach a true steady state. As the system tends to a time-independent state, the orbit-loss fluxes approach their steady-state values, which are determined entirely by upstream transport (turbulent and collisional) and upstream sources and sinks. This result holds for the fluxes of any moments that are expressible in terms of the invariants that determine the orbits, including parallel toroidal angular momentum.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
NASA NTRS1991

Markov reward processes

Numerous applications in the area of computer system analysis can be effectively studied with Markov reward models. These models describe the behavior of the system with a continuous-time Markov chain, where a reward rate is associated with each state. In a reliability/availability model, upstates may have reward rate 1 and down states may have reward rate zero associated with them. In a queueing model, the number of jobs of certain type in a given state may be the reward rate attached to that state. In a combined model of performance and reliability, the reward rate of a state may be the computational capacity, or a related performance measure. Expected steady-state reward rate and expected instantaneous reward rate are clearly useful measures of the Markov reward model. More generally, the distribution of accumulated reward or time-averaged reward over a finite time interval may be determined from the solution of the Markov reward model. This information is of great practical significance in situations where the workload can be well characterized (deterministically, or by continuous functions e.g., distributions). The design process in the development of a computer system is an expensive and long term endeavor. For aerospace applications the reliability of the computer system is essential, as is the ability to complete critical workloads in a well defined real time interval. Consequently, effective modeling of such systems must take into account both performance and reliability. This fact motivates our use of Markov reward models to aid in the development and evaluation of fault tolerant computer systems.

Smith, R. M.↗
NASA NTRS2013

Abort Options for Human Missions to Earth-Moon Halo Orbits

Abort trajectories are optimized for human halo orbit missions about the translunar libration point (L2), with an emphasis on the use of free return trajectories. Optimal transfers from outbound free returns to L2 halo orbits are numerically optimized in the four-body ephemeris model. Circumlunar free returns are used for direct transfers, and cislunar free returns are used in combination with lunar gravity assists to reduce propulsive requirements. Trends in orbit insertion cost and flight time are documented across the southern L2 halo family as a function of halo orbit position and free return flight time. It is determined that the maximum amplitude southern halo incurs the lowest orbit insertion cost for direct transfers but the maximum cost for lunar gravity assist transfers. The minimum amplitude halo is the most expensive destination for direct transfers but the least expensive for lunar gravity assist transfers. The on-orbit abort costs for three halos are computed as a function of abort time and return time. Finally, an architecture analysis is performed to determine launch and on-orbit vehicle requirements for halo orbit missions.

Jesick, Mark C.↗
DOE OSTI2025

A machine learning framework for accurate and robust analysis of radiation detector pulses

The microscopic properties of atomic nuclei are used to study various scientific questions. They are essential for understanding the fundamental forces of nature and the chemical evolution of the universe. Detecting decay radiation from radioactive nuclei makes it possible to probe these fundamental nuclear properties. Detector waveform traces may contain additional information about the radiation. Generally, advanced signal processing techniques are needed to extract this additional information, often involving fitting the waveform with model response functions using non-linear least-squares optimization with second-order gradient methods. While this is a powerful technique, it is also computationally expensive, leading to slow processing time, which scales with the volume of data. To address this problem, we have developed a machine learning (ML) approach that infers the characteristics of traces from a model detector response function. In particular, we are interested in classifying whether a single recorded trace consists of one or two pulse constituents and estimating the pulse parameters. Furthermore, our proposed ML method can precisely extract the pulses’ parameters, such as energy and timing information, and accurately classify the pulse multiplicity of a trace. Unlike non-learning-based approaches, our ML approach uses neural networks that are significantly faster at inference, as they do not require any optimization during this stage.

Curve fitting↗
DOE OSTI2019

HLE17: An Efficient Way To Predict Band Gaps of Complex Materials

The recently developed HLE 17 exchange-correlation functional, which is a local functional with high local exchange, is tested for determining the band gaps of various kinds of semiconductors and insulators, including metal organic frameworks (MOFs), covalent organic frameworks (COFs), perovskites, zeolites, and others. The calculated band gaps are compared to those calculated by more conventional functionals, in particular PBE, PBEsol, PBE+U, and HSE06. The HLE 17 functional is found to be more accurate PBE, PBEsol, and PBE+U and almost as accurate as the much more expensive HSE06. For the MOFs, HLE 17 is as accurate as HSE06 with almost 100 times less computational cost. As a result, the HLE 17 functional is found to be more accurate than the hybrid functional for covalent organic frameworks (COFs), β-GaSe, and BC 3 .

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2016

Calibration of an agricultural-hydrological model (RZWQM2) using surrogate global optimization

Robust calibration of an agricultural-hydrological model is critical for simulating crop yield and water quality and making reasonable agricultural management. However, calibration of the agricultural-hydrological system models is challenging because of model complexity, the existence of strong parameter correlation, and significant computational requirements. Therefore, only a limited number of simulations can be allowed in any attempt to find a near-optimal solution within an affordable time, which greatly restricts the successful application of the model. The goal of this study is to locate the optimal solution of the Root Zone Water Quality Model (RZWQM2) given a limited simulation time, so as to improve the model simulation and help make rational and effective agricultural-hydrological decisions. To this end, we propose a computationally efficient global optimization procedure using sparse-grid based surrogates. We first used advanced sparse grid (SG) interpolation to construct a surrogate system of the actual RZWQM2, and then we calibrate the surrogate model using the global optimization algorithm, Quantum-behaved Particle Swarm Optimization (QPSO). As the surrogate model is a polynomial with fast evaluation, it can be efficiently evaluated with a sufficiently large number of times during the optimization, which facilitates the global search. We calibrate seven model parameters against five years of yield, drain flow, and NO 3 -N loss data from a subsurface-drained corn-soybean field in Iowa. Results indicate that an accurate surrogate model can be created for the RZWQM2 with a relatively small number of SG points (i.e., RZWQM2 runs). Compared to the conventional QPSO algorithm, our surrogate-based optimization method can achieve a smaller objective function value and better calibration performance using a fewer number of expensive RZWQM2 executions, which greatly improves computational efficiency.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2024

A road map to cosmological parameter analysis with third-order shear statistics: III. Efficient estimation of third-order shear correlation functions and an application to the KiDS-1000 data

Context. Third-order lensing statistics contain a wealth of cosmological information that is not captured by second-order statistics. However, the computational effort it takes to estimate such statistics in forthcoming stage IV surveys is prohibitively expensive. Aims. We derive and validate an efficient estimation procedure for the three-point correlation function (3PCF) of polar fields such as weak lensing shear. We then use our approach to measure the shear 3PCF and the third-order aperture mass statistics on the KiDS-1000 survey. Methods We constructed an efficient estimator for third-order shear statistics that builds on the multipole decomposition of the 3PCF. We then validated our estimator on mock ellipticity catalogs obtained from N -body simulations. Finally, we applied our estimator to the KiDS-1000 data and presented a measurement of the third-order aperture statistics in a tomographic setup. Results. Our estimator provides a speedup of a factor of ∼100–1000 compared to the state-of-the-art estimation procedures. It is also able to provide accurate measurements for squeezed and folded triangle configurations without additional computational effort. We report a significant detection of tomographic third-order aperture mass statistics in the KiDS-1000 data (S/N = 6.69). Conclusions. Our estimator will make it computationally feasible to measure third-order shear statistics in forthcoming stage IV surveys. Furthermore, it can be used to construct empirical covariance matrices for such statistics.

Astronomy & Astrophysics↗