Search NASA⌕ Search

SEARCH · Search NASA

Results for “algorithmic structures”

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 343 records · Page 19
All sourcesNASA NTRSDOE OSTI
DOE OSTI2016

A rank-exploiting infinite Arnoldi algorithm for nonlinear eigenvalue problems

We consider the nonlinear eigenvalue problem M ( λ ) x = 0, where M ( λ ) is a large parameter-dependent matrix. In several applications, M ( λ ) has a structure where the higher-order terms of its Taylor expansion have a particular low-rank structure. We propose a new Arnoldi-based algorithm that can exploit this structure. More precisely, the proposed algorithm is equivalent to Arnoldi's method applied to an operator whose reciprocal eigenvalues are solutions to the nonlinear eigenvalue problem. The iterates in the algorithm are functions represented in a particular structured vector-valued polynomial basis similar to the construction in the infinite Arnoldi method [Jarlebring, Michiels, and Meerbergen, Numer. Math., 122 (2012), pp. 169–195]. In this paper, the low-rank structure is exploited by applying an additional operator and by using a more compact representation of the functions. This reduces the computational cost associated with orthogonalization, as well as the required memory resources. The structure exploitation also provides a natural way in carrying out implicit restarting and locking without the need to impose structure in every restart. In conclusion, the efficiency and properties of the algorithm are illustrated with two large-scale problems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Systematic comparison of crystalline and amorphous phases: Charting the landscape of water structures and transformations

Systematically resolving different crystalline phases starting from the atomic positions, a mandatory step in algorithms for the prediction of structures or for the simulation of phase transitions, can be a non-trivial task. Extending to amorphous phases and liquids which lack the discrete symmetries, the problem becomes even more difficult, involving subtle topological differences at medium range that, however, are crucial to the physico-chemical and spectroscopic properties of the corresponding materials. Typically, system-tailored order parameters are devised, like global or local symmetry indicators, ring populations, etc. We show that a recently introduced metric provides a simple and general solution to this intricate problem. In particular, we demonstrate that a map can be traced displaying distances among water phases, including crystalline as well as amorphous states and the liquid, consistently with experimental knowledge in terms of phase diagram, structural features, and preparation routes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Scalable All-pairs Shortest Paths for Huge Graphs on Multi-GPU Clusters

We present an optimized Floyd-Warshall (Floyd-Warshall) algorithm that computes the All-pairs shortest path (APSP) for GPU accelerated clusters. The Floyd-Warshall algorithm due to its structural similarities to matrix-multiplication is well suited for highly parallel GPU architectures. To achieve high parallel efficiency, we address two key algorithmic challenges: reducing high communication overhead and addressing limited GPU memory. To reduce high communication costs, we redesign the parallel (a) to expose more parallelism, (b) aggressively overlap communication and computation with pipelined and asynchronous scheduling of operations, and (c) tailored MPI-collective. To cope with limited GPU memory, we employ an offload model, where the data resides on the host and is transferred to GPU on-demand. The proposed optimizations are supported with detailed performance models for tuning. Our optimized parallel Floyd-Warshall implementation is up to 5x faster than a strong baseline and achieves 8.1 PetaFLOPS/sec on 256~nodes of the Summit supercomputer at Oak Ridge National Laboratory. This performance represents 70% of the theoretical peak and 80% parallel efficiency. The offload algorithm can handle 2.5x larger graphs with a 20% increase in overall running time.

Sao, Piyush↗
DOE OSTI2016

Grid-based electronic structure calculations: The tensor decomposition approach

We present a fully grid-based approach for solving Hartree–Fock and all-electron Kohn–Sham equations based on low-rank approximation of three-dimensional electron orbitals. Due to the low-rank structure the total complexity of the algorithm depends linearly with respect to the one-dimensional grid size. Linear complexity allows for the usage of fine grids, e.g. 8192{sup 3} and, thus, cheap extrapolation procedure. We test the proposed approach on closed-shell atoms up to the argon, several molecules and clusters of hydrogen atoms. All tests show systematical convergence with the required accuracy.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2014

The metallization and superconductivity of dense hydrogen sulfide

Hydrogen sulfide (H{sub 2}S) is a prototype molecular system and a sister molecule of water (H{sub 2}O). The phase diagram of solid H{sub 2}S at high pressures remains largely unexplored arising from the challenges in dealing with the pressure-induced weakening of S–H bond and larger atomic core difference between H and S. Metallization is yet achieved for H{sub 2}O, but it was observed for H{sub 2}S above 96 GPa. However, the metallic structure of H{sub 2}S remains elusive, greatly impeding the understanding of its metallicity and the potential superconductivity. We have performed an extensive structural study on solid H{sub 2}S at pressure ranges of 10–200 GPa through an unbiased structure prediction method based on particle swarm optimization algorithm. Besides the findings of candidate structures for nonmetallic phases IV and V, we are able to establish stable metallic structures violating an earlier proposal of elemental decomposition into sulfur and hydrogen [R. Rousseau, M. Boero, M. Bernasconi, M. Parrinello, and K. Terakura, Phys. Rev. Lett. 85, 1254 (2000)]. Our study unravels a superconductive potential of metallic H{sub 2}S with an estimated maximal transition temperature of ∼80 K at 160 GPa, higher than those predicted for most archetypal hydrogen-containing compounds (e.g., SiH{sub 4}, GeH{sub 4}, etc.)

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

Morphological analysis of the polarized synchrotron emission with WMAP and Planck

The bright polarized synchrotron emission, away from the Galactic plane, originates mostly from filamentary structures. We implement a filament finder algorithm which allows the detection of bright elongated structures in polarized intensity maps. We analyse the sky at 23 and 30 GHz as observed respectively by WMAP and Planck. We identify 19 filaments, 13 of which have been previously observed. For each filament, we study the polarization fraction, finding values typically larger than for the areas outside the filaments, excluding the Galactic plane, and a fraction of about 30% is reached in two filaments. We study the polarization spectral indices of the filaments, and find a spectral index consistent with the values found in previous analysis (about -3.1) for more diffuse regions. Decomposing the polarization signals into the E and B families, we find that most of the filaments are detected in P E , but not in P B . We then focus on understanding the statistical properties of the diffuse regions of the synchrotron emission at 23 GHz. Using Minkowski functionals and tensors, we analyse the non-Gaussianity and statistical isotropy of the polarized intensity maps. For a sky coverage corresponding to 80% of the fainter emission, and on scales smaller than 6 degrees (ℓ > 30), the deviations from Gaussianity and isotropy are significantly higher than 3σ. The level of deviation decreases for smaller scales, however, it remains significantly high for the lowest analised scale (~ 1.5°). When 60% sky coverage is analysed, we find that the deviations never exceed 3σ. Finally, we present a simple data-driven model to generate non-Gaussian and anisotropic simulations of the synchrotron polarized emission. The simulations are fitted in order to match the spectral and statistical properties of the faintest 80% sky coverage of the data maps.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2024

Evaluating a quantum-classical quantum Monte Carlo algorithm with Matchgate shadows

Solving the electronic structure problem of molecules and solids to high accuracy is a major challenge in quantum chemistry and condensed matter physics. The rapid emergence and development of quantum computers offer a promising route to systematically tackle this problem. Recent work by [Huggins et al ., Nature (London) 603 , 416 (2022)] proposed a hybrid quantum-classical quantum Monte Carlo (QC-QMC) algorithm using Clifford shadows to determine the ground state of a Fermionic Hamiltonian. This approach displayed inherent noise resilience and the potential for improved accuracy compared to its purely classical counterpart. Nevertheless, the use of Clifford shadows introduces an exponentially scaling postprocessing cost. In this work, we investigate an improved QC-QMC scheme utilizing the recently developed Matchgate shadows technique [Commun. Math. Phys. 404 , 629 (2023)], which removes the aforementioned exponential bottleneck. We observe from experiments on quantum hardware that the use of Matchgate shadows in QC-QMC is inherently noise robust. We show that this noise resilience has a more subtle origin than in the case of Clifford shadows. Nevertheless, we find that classical postprocessing, while asymptotically efficient, requires hours of runtime on thousands of classical CPUs for even the smallest chemical systems, presenting a major challenge to the scalability of the algorithm.

Monte Carlo methods↗
DOE OSTI2023

CP decomposition for tensors via alternating least squares with QR decomposition

The CP tensor decomposition is used in applications such as machine learning and signal processing to discover latent low-rank structure in multidimensional data. Computing a CP decomposition via an alternating least squares (ALS) method reduces the problem to several linear least squares problems. The standard way to solve these linear least squares subproblems is to use the normal equations, which inherit special tensor structure that can be exploited for computational efficiency. However, the normal equations are sensitive to numerical ill-conditioning, which can compromise the results of the decomposition. In this paper, we develop versions of the CP-ALS algorithm using the QR decomposition and the singular value decomposition, which are more numerically stable than the normal equations, to solve the linear least squares problems. Our algorithms utilize the tensor structure of the CP-ALS subproblems efficiently, have the same complexity as the standard CP-ALS algorithm when the input is dense and the rank is small, and are shown via examples to produce more stable results when ill-conditioning is present. Our MATLAB implementation achieves the same running time as the standard algorithm for small ranks, and we show that the new methods can obtain lower approximation error.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2014

Basis functions for electronic structure calculations on spheres

We introduce a new basis function (the spherical Gaussian) for electronic structure calculations on spheres of any dimension D. We find general expressions for the one- and two-electron integrals and propose an efficient computational algorithm incorporating the Cauchy-Schwarz bound. Using numerical calculations for the D = 2 case, we show that spherical Gaussians are more efficient than spherical harmonics when the electrons are strongly localized.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

Insight into Molecular Basis and Dynamics of Full-length CRaf Kinase in Cellular Signaling Mechanism

This study presents the first large-scale simulation using an initial structure predicted by AI/ML algorithms for the 648-amino-acid CRaf kinase, which plays a key role in cellular signaling. Simulation results show the evolution of the predicted structure into much more compact structures with inter-domain interactions that shed insights into auto-inhibition mechanism, paradoxical effect, activation, and recruitment pathways in the CRaf kinase. Newly identified epitopes in the CRaf may suggest additional drug targets. The results were published in Biophysical Journal, DOI:10.1016/j.bpj.2024.06.028.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2018

Toward First-Principles Design of Organic Nonlinear Optical Materials: Crystal Structure Prediction and Halogen Bonding Impact on Hyperpolarizabilities of 2-Iodo-3-hydroxypyridine

Computational methods can potentially accelerate development of more efficient organic materials for second harmonic generation. Here, we test the method that includes the evolutionary algorithm for predicting crystal structure and prognosis of nonlinear optical properties based on the predicted structure. For this test, we selected 2-iodo-3-hydroxypyridine, which exhibits second harmonic generation intensity comparable to that of urea. We performed global minimization of the lattice energy and found the experimental structure when many-body dispersion correction is added to the density functional theory values. We analyzed geometric preferences of the halogen bonding in predicted virtual polymorphs. We also found linear correlation between the lengths of the iodine–iodine halogen bonds and calculated second order susceptibilities.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Binary Complex Neural Network Acceleration on FPGA

Being able to learn from complex data with phase information is imperative for many signal processing applications. Today’s real-valued deep neural networks (DNNs) have shown efficiency in latent information analysis but fall short when applied to the complex domain. Deep complex networks (DCN) , in contrast, can learn from complex data, but have high computational costs; therefore, they cannot satisfy the instant decision making requirements of many deployable systems dealing with short observations or short signal bursts. Recent, Binarized Complex Neural Network (BCNN), which integrates DCNs with binarized neural networks (BNN), shows great potential in classifying complex data in real-time. In this paper, we propose a structural pruning based accelerator of BCNN, which is able to provide more than 5000 frames/s inference throughput on edge devices. The high performance comes from both the algorithm and hardware sides. On the algorithm side, we conduct structural pruning to the original BCNN models and obtain 20 × pruning rates with negligible accuracy loss; on the hardware side, we propose a novel 2D convolution operation accelerator for the binary complex neural network. Experimental results show that the proposed design works with over 90% utilization and is able to achieve the inference throughput of 5882 frames/s and 4938 frames/s for complex NIN-Net and ResNet-18 using CIFAR-10 dataset and Alveo U280 Board.

Peng, Hongwu↗
DOE OSTI2018

Solving protein structure from sparse serial microcrystal diffraction data at a storage-ring synchrotron source

In recent years, the success of serial femtosecond crystallography and the paucity of beamtime at X-ray free-electron lasers have motivated the development of serial microcrystallography experiments at storage-ring synchrotron sources. However, especially at storage-ring sources, if a crystal is too small it will have suffered significant radiation damage before diffracting a sufficient number of X-rays into Bragg peaks for peak-indexing software to determine the crystal orientation. As a consequence, the data frames of small crystals often cannot be indexed and are discarded. Introduced here is a method based on the expand–maximize–compress (EMC) algorithm to solve protein structures, specifically from data frames for which indexing methods fail because too few X-rays are diffracted into Bragg peaks. The method is demonstrated on a real serial microcrystallography data set whose signals are too weak to be indexed by conventional methods. In spite of the daunting background scatter from the sample-delivery medium, it was still possible to solve the protein structure at 2.1 Å resolution. The ability of the EMC algorithm to analyze weak data frames will help to reduce sample consumption. Furthermore, it will also allow serial microcrystallography to be performed with crystals that are otherwise too small to be feasibly analyzed at storage-ring sources.

77 NANOSCIENCE AND NANOTECHNOLOGY↗
DOE OSTI2013

A variational multi-symplectic particle-in-cell algorithm with smoothing functions for the Vlasov-Maxwell system

Smoothing functions are commonly used to reduce numerical noise arising from coarse sampling of particles in particle-in-cell (PIC) plasma simulations. When applying smoothing functions to symplectic algorithms, the conservation of symplectic structure should be guaranteed to preserve good conservation properties. In this paper, we show how to construct a variational multi-symplectic PIC algorithm with smoothing functions for the Vlasov-Maxwell system. The conservation of the multi-symplectic structure and the reduction of numerical noise make this algorithm specifically suitable for simulating long-term dynamics of plasmas, such as those in the steady-state operation or long-pulse discharge of a super-conducting tokamak. The algorithm has been implemented in a 6D large scale PIC code. Numerical examples are given to demonstrate the good conservation properties of the multi-symplectic algorithm and the reduction of the noise due to the application of smoothing function.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

Correlated pair ansatz with a binary tree structure

We develop an efficient algorithm to implement the recently introduced binary tree state (BTS) ansatz on a classical computer. BTS allows a simple approximation to permanents arising from the computationally intractable antisymmetric product of interacting geminals and respects size-consistency. We show how to compute BTS overlap and reduced density matrices efficiently. We also explore two routes for developing correlated BTS approaches: Jastrow coupled cluster on BTS and linear combinations of BT states. Here, the resulting methods show great promise in benchmark applications to the reduced Bardeen–Cooper–Schrieffer Hamiltonian and the one-dimensional XXZ Heisenberg Hamiltonian.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2018

Predicting kinetics of polymorphic transformations from structure mapping and coordination analysis

To extend materials design and discovery into the space of metastable polymorphs, rapid and reliable assessment of transformation kinetics to lower energy structures is essential. Herein we focus on diffusionless polymorphic transformations and investigate routes to assess their kinetics using solely crystallographic arguments. As part of this investigation we developed a general algorithm to map crystal structures onto each other, and ascertain the low-energy (fast-kinetics) transformation pathways between them. Pathways with minimal dissociation of chemical bonds, along which the number of bonds (in ionic systems the first-shell coordination) does not decrease below that in the end structures, are shown to always be the fast-kinetics pathways. These findings enable the rapid assessment of the kinetics of polymorphic transformation and the identification of long-lived metastable structures. In conclusion, the utility is demonstrated on a number of transformations including those between high-pressure SnO 2 phases, which lack a detailed atomic-level understanding.

36 MATERIALS SCIENCE↗
DOE OSTI2025

NE-COST plug-in: Expanding ACCERT's Capabilities for Life-Cycle Cost Modeling

The Algorithm for the Capital Cost Estimation of Reactor Technologies (ACCERT) is a structured methodology and software tool designed to simplify and standardize cost estimation for nuclear reactor technologies [1]. By utilizing a relational database structure and modular cost estimation algorithms, ACCERT delivers a robust, flexible, and scalable framework for evaluating costs across various reactor types and configurations [2]. The recent integration of the NE-COST plugin further expands ACCERT’s scope by introducing detailed life-cycle cost modeling and probabilistic analysis of uncertainties. This addition enables users to evaluate costs across front-end processes such as uranium enrichment and fabrication, as well as back-end activities including waste disposal and geologic storage. Through Monte Carlo statistical cost simulations, the plugin provides probabilistic insights into cost ranges, offering critical decision-making support for stakeholders including reactor developers, policymakers, and researchers.

Zhou, Jia↗