Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “computational efficiency”

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 667 records · Page 37

An algorithm for computing the number of distinct spectral vectors in thematic mapper data

A computationally efficient method was developed to compute the number of distinct spectral vectors and their frequency of occurrence in Landsat-4 Thematic Mapper (TM) data. The algorithm first partitions the image into spectrally disjoint subsets and then computes the frequency distribution of distinct spectral vectors within each subset from a multidimensional histogram. The overall frequency distribution is tabulated by accumulating the results from each subset. The number of distinct spectral vectors could be used as a measure of potential storage compaction of alternate data representations for data compression, or as a measure of information content in the comparison of spectral band combinations and/or spatial resolutions for an image. Results from processing three 512 x 512 pixel Landsat-4 TM images and one Landsat-4 Multispectral Scanner (MSS) image are presented as examples. An algorithm for computing the frequency distribution of distinct spectral vectors in MSS data is given in the Appendix.

Wharton, S. W.↗

The use of the QR factorization in the partial realization problem

The use of the QR factorization of the Hankel matrix in solving the partial realization problem is analyzed. Straightforward use of the QR factorization results in a realization scheme that possesses all of the computational advantages of Rissanen's realization scheme. These latter properties are computational efficiency, recursiveness, use of limited computer memory, and the realization of a system triplet having a condensed structure. Moreover, this scheme is robust when the order of the system corresponds to the rank of the Hankel matrix. When this latter condition is violated, an approximate realization could be determined via the QR factorization. In this second scheme, the given Hankel matrix is approximated by a low-rank non-Hankel matrix. Furthermore, it is demonstrated that column pivoting might be incorporated in this second scheme. The results presented are derived for a single input/single output system, but this does not seem to be a restriction.

Verhaegen, M. H.↗

A two-layer multiple-time-scale turbulence model and grid independence study

A two-layer multiple-time-scale turbulence model is presented. The near-wall model is based on the classical Kolmogorov-Prandtl turbulence hypothesis and the semi-empirical logarithmic law of the wall. In the two-layer model presented, the computational domain of the conservation of mass equation and the mean momentum equation penetrated up to the wall, where no slip boundary condition has been prescribed; and the near wall boundary of the turbulence equations has been located at the fully turbulent region, yet very close to the wall, where the standard wall function method has been applied. Thus, the conservation of mass constraint can be satisfied more rigorously in the two-layer model than in the standard wall function method. In most of the two-layer turbulence models, the number of grid points to be used inside the near-wall layer posed the issue of computational efficiency. The present finite element computational results showed that the grid independent solutions were obtained with as small as two grid points, i.e., one quadratic element, inside the near wall layer. Comparison of the computational results obtained by using the two-layer model and those obtained by using the wall function method is also presented.

Kim, S.-W.↗

Development of a CFD code for internal flows in liquid fueled engines

To support the design efforts of engines in which liquid propellants are used, one is often required to analyze incompressible, two-dimensional or axisymmetric flows within ducts and cavities with rotating walls and complicated geometries. The steady-state solution is of interest in most cases. This code is intended to provide a tool for efficient computational fluid dynamics (CFD) analysis of such flow problems, taking advantage of the artificial compressibility concept and the Beam-Warming numerical scheme modified for second order accurate, implicit boundary conditions. These concepts ensure a stable, robust, and accurate algorithm due to the reduced speed of sound and the accuracy of the boundary conditions. The code is dedicated only to two-dimensional or axisymmetric flows with or without swirl. Three-dimensional computation is excluded to increase efficiency and speed. This paper briefly presents the theory of the code, as well as several benchmark applications with comparison to well known analytical solutions. In all these test cases, the code produced accurate results.

Dakhoul, Youssef↗

Order-Reduced Solution of the Nonlinear High-Fidelity Generalized Method of Cells Micromechanics Relations

The High-Fidelity Generalized Method of Cells (HFGMC) is one technique for accurately simulating nonlinear composite material behavior. The HFGMC uses a higher-order approximation for the subcell displacement field that allows for a more accurate determination of the subcell stressstrain fields at the cost of some computational efficiency. In order to reduce computational costs associated with the solution of the ensuing system of simultaneous equations, the HFGMC global system of equations for doubly-periodic repeating unit cells with nonlinear constituents was reduced in size through the use of a Petrov-Galerkin-based Proper Orthogonal Decomposition order-reduction scheme. A number of cases were presented that address the computational feasibility of using order-reduction techniques to solve solid mechanics problems involving complex microstructures.

order-reduction↗

Hydrogen in energy and information sciences

Beyond its fascinating chemistry as the first element in the Periodic Table, hydrogen is of high societal importance in energy technologies and of growing importance in energy-efficient computing. In energy, hydrogen has reemerged as a potential solution to long-term energy storage and as a carbon-free input for materials manufacturing. Its utilization and production rely on the availability of proton-conducting electrolytes and mixed proton–electron conductors for the components in fuel cells and electrolyzers. In computing, proton mediation of electronic properties has garnered attention for electrochemically controlled energy-efficient neuromorphic computing. Incorporation of substitutional and interstitial hydride ions in oxides, though only recently established, enables tuning of electronic and magnetic properties, inviting a range of possible exotic applications. This article addresses common themes in the fundamental science of hydrogen incorporation and transport in oxides as relevant to pressing technological needs. The content covers (1) lattice (or bulk) mechanisms of hydrogen transport, primarily addressing proton transport, but also touching on hydride ion transport; (2) interfacial transport; (3) exploitation of extreme external drivers to achieve unusual response; and (4) advances in methods to probe the hydrogen environment and transport pathway. The snapshot of research activities in the field of hydrogen-laden materials described here underscores exciting recent breakthroughs, remaining open questions, and breathtaking experimental tools now available for unveiling the nature of hydrogen in solid-state matter.

08 HYDROGEN↗

Partial Derivatives of the Lambert Problem

A procedure for deriving analytic partial derivatives of the Lambert problem is presented. Using the universal, cosine based Lambert formulation; first order partial derivatives of the velocities with respect to the positions and times are developed. Taking advantage of inherent symmetries and intermediate variables, the derivatives are expressed in a computationally efficient form. The added cost of computing these partials is found to be approximately 10% to approximately 60% of the Lambert compute cost. The availability of analytic partial derivatives increases optimization speed, efficiency and allows for trajectory optimization formulations that implicitly enforce continuity constraints via embedded Lambert problems.

trajectory computation↗

Development of Efficient Real-Fluid Model in Simulating Liquid Rocket Injector Flows

The characteristics of propellant mixing near the injector have a profound effect on the liquid rocket engine performance. However, the flow features near the injector of liquid rocket engines are extremely complicated, for example supercritical-pressure spray, turbulent mixing, and chemical reactions are present. Previously, a homogeneous spray approach with a real-fluid property model was developed to account for the compressibility and evaporation effects such that thermodynamics properties of a mixture at a wide range of pressures and temperatures can be properly calculated, including liquid-phase, gas- phase, two-phase, and dense fluid regions. The developed homogeneous spray model demonstrated a good success in simulating uni- element shear coaxial injector spray combustion flows. However, the real-fluid model suffered a computational deficiency when applied to a pressure-based computational fluid dynamics (CFD) code. The deficiency is caused by the pressure and enthalpy being the independent variables in the solution procedure of a pressure-based code, whereas the real-fluid model utilizes density and temperature as independent variables. The objective of the present research work is to improve the computational efficiency of the real-fluid property model in computing thermal properties. The proposed approach is called an efficient real-fluid model, and the improvement of computational efficiency is achieved by using a combination of a liquid species and a gaseous species to represent a real-fluid species.

Cheng, Gary↗

An Efficient Solution Method for Multibody Systems with Loops Using Multiple Processors

This paper describes a multibody dynamics algorithm formulated for parallel implementation on multiprocessor computing platforms using the divide-and-conquer approach. The system of interest is a general topology of rigid and elastic articulated bodies with or without loops. The algorithm divides the multibody system into a number of smaller sets of bodies in chain or tree structures, called "branches" at convenient joints called "connection points", and uses an Order-N (O (N)) approach to formulate the dynamics of each branch in terms of the unknown spatial connection forces. The equations of motion for the branches, leaving the connection forces as unknowns, are implemented in separate processors in parallel for computational efficiency, and the equations for all the unknown connection forces are synthesized and solved in one or several processors. The performances of two implementations of this divide-and-conquer algorithm in multiple processors are compared with an existing method implemented on a single processor.

Multibody dynamics↗

Ginkgo - A math library designed to accelerate Exascale Computing Project science applications

Large-scale simulations require efficient computation across the entire computing hierarchy. A challenge of the Exascale Computing Project (ECP) was to reconcile highly heterogeneous hardware with the myriad of applications that were required to run on these supercomputers. Mathematical software forms the backbone of almost all scientific applications, providing efficient abstractions and operations that are crucial to harness the performance of computing systems. Ginkgo is one such mathematical software library, nurtured by ECP, providing high-performance, user-friendly, and performance portable interfaces for applications in ECP and beyond. In this paper, we elaborate on Ginkgo’s philosophy of high-performance software that is sustainable, reproducible, and easy to use. We showcase the wide feature set of solvers and preconditioners available in Ginkgo and the central concepts involved in their design. We elaborate on four different ECP software integrations: MFEM, PeleLM + SUNDIALS, XGC, and ExaSGD that use Ginkgo to accelerate their science runs. Performance studies of different problems from these applications highlight the effectiveness of Ginkgo and the benefits incurred by these ECP applications.

Cojean, Terry↗

Energy and Area Efficiency in Neuromorphic Computing for Resource Constrained Devices

Resource constrained devices are the building blocks of the internet of things (IoT) era. Since the idea behind IoT is to develop an interconnected environment where the devices are tiny enough to operate with limited resources, several control systems have been built to maintain low energy and area consumption while operating as IoT edge devices. Several researchers have begun work on implementing control systems built from resource constrained devices using machine learning. However, there are many ways such devices can achieve lower power consumption and area utilization while maximizing application efficiency. Spiky neuromorphic computing (SNC) is an emerging paradigm that can be leveraged in resource constrained devices for several emerging applications. While delivering the benefits of machine learning, SNC also helps minimize power consumption. For example, low energy memory devices (memristors) are often used to achieve low power operation and also help in reducing system area. In total, we anticipate SNC will provide computational efficiency approaching that of deep learning while using low power, resource constrained devices.

Chakma, Gangotree↗

FINETUNA: fine-tuning accelerated molecular simulations

Abstract Progress towards the energy breakthroughs needed to combat climate change can be significantly accelerated through the efficient simulation of atomistic systems. However, simulation techniques based on first principles, such as density functional theory (DFT), are limited in their practical use due to their high computational expense. Machine learning approaches have the potential to approximate DFT in a computationally efficient manner, which could dramatically increase the impact of computational simulations on real-world problems. However, they are limited by their accuracy and the cost of generating labeled data. Here, we present an online active learning framework for accelerating the simulation of atomic systems efficiently and accurately by incorporating prior physical information learned by large-scale pre-trained graph neural network models from the Open Catalyst Project. Accelerating these simulations enables useful data to be generated more cheaply, allowing better models to be trained and more atomistic systems to be screened. We also present a method of comparing local optimization techniques on the basis of both their speed and accuracy. Experiments on 30 benchmark adsorbate-catalyst systems show that our method of transfer learning to incorporate prior information from pre-trained models accelerates simulations by reducing the number of DFT calculations by 91%, while meeting an accuracy threshold of 0.02 eV 93% of the time. Finally, we demonstrate a technique for leveraging the interactive functionality built in to Vienna ab initio Simulation Package (VASP) to efficiently compute single point calculations within our online active learning framework without the significant startup costs. This allows VASP to work in tandem with our framework while requiring 75% fewer self-consistent cycles than conventional single point calculations. The online active learning implementation, and examples using the VASP interactive code, are available in the open source FINETUNA package on Github.

97 MATHEMATICS AND COMPUTING↗

Efficient Method for Optimizing Placement of Sensors

A computationally efficient method has been developed to enable optimization of the placement of sensors for the purpose of diagnosis of a complex engineering system (e.g., an aircraft or spacecraft). The method can be used both in (1) designing a sensor system in which the number and positions of sensors are initially not known and must be determined and (2) adding sensors to a pre-existing system to increase the diagnostic capability. The optimal-sensor-placement problem can be summarized as involving the following concepts, issues, and subproblems: a) Degree of Diagnosability - This is a concept for characterizing the set of faults that can be discriminated by use of a given set of sensors. b) Minimal Sensor Set - The idea is one of finding a minimal set of sensors that guarantees a specific degree of diagnosability. c) Minimal-Cost Sensors - In a case in which different sensors are assigned with different costs, it is desired to choose the least costly set of sensors that affords a specific degree of diagnosability.

Fijany, Amir↗

Towards a more general understanding of the algorithmic utility of recurrent connections

Lateral and recurrent connections are ubiquitous in biological neural circuits. Yet while the strong computational abilities of feedforward networks have been extensively studied, our understanding of the role and advantages of recurrent computations that might explain their prevalence remains an important open challenge. Foundational studies by Minsky and Roelfsema argued that computations that require propagation of global information for local computation to take place would particularly benefit from the sequential, parallel nature of processing in recurrent networks. Such “tag propagation” algorithms perform repeated, local propagation of information and were originally introduced in the context of detecting connectedness, a task that is challenging for feedforward networks. Here, we advance the understanding of the utility of lateral and recurrent computation by first performing a large-scale empirical study of neural architectures for the computation of connectedness to explore feedforward solutions more fully and establish robustly the importance of recurrent architectures. In addition, we highlight a tradeoff between computation time and performance and construct hybrid feedforward/recurrent models that perform well even in the presence of varying computational time limitations. We then generalize tag propagation architectures to propagating multiple interacting tags and demonstrate that these are efficient computational substrates for more general computations of connectedness by introducing and solving an abstracted biologically inspired decision-making task. Our work thus clarifies and expands the set of computational tasks that can be solved efficiently by recurrent computation, yielding hypotheses for structure in population activity that may be present in such tasks.

59 BASIC BIOLOGICAL SCIENCES↗

Numerical computation of transonic flow about wing-fuselage configurations on a vector computer

The transonic wing analysis code TWING, which uses the AF2 relaxation algorithm, has been vectorized to run on the Cray-1S computer. Vectorization of this code improved computational efficiency over that of the CDC 7600 computer by factors of 11 to 13. The improvement compares favorably with the prediction of a theoretical performance model. A convenient generalization now permits the treatment of rudimentary wing-fuselage combinations. Flow predictions for a transport configuration in both isolated-wing and wing-fuselage modes show the expected trends in shock strength and position when compared with wind-tunnel results. An isolated fighter wing is examined in terms of execution time on three different computers and in comparison with experimental data. The computational fluid dynamics code produced during this study is a careful union of an efficient three-dimensional, transonic, numerical algorithm and the vector features presently available on modern computers.

Thomas, S. D.↗

Quantum Branch-and-Bound Algorithm and its Application to the Travelling Salesman Problem

We propose a quantum branch-and-bound algorithm based on the general scheme of the branch-and-bound method and the quantum nested searching algorithm and examine its computational efficiency. We also compare this algorithm with a similar classical algorithm on the example of the travelling salesman problem. We show that in the vast majority of problems, the classical algorithm is quicker than the quantum algorithm due to greater adaptability. However, the operation time of the quantum algorithm is constant for all problem, whereas the classical algorithm runs very slowly for certain problems. In the worst case, the quantum branch-and-bound algorithm is proved to be several times more efficient than the classical algorithm.

97 MATHEMATICS AND COMPUTING↗

Optimal Control of the Energy-Saving Hybrid Hydraulic-Electric Architecture (HHEA) for Off-Highway Mobile Machines

Most off-highway constructions and agriculture equipment use hydraulics, which has unmatched power density, for power transmission and throttling as a means for control. A novel hybrid hydraulic-electric architecture (HHEA) has recently been proposed to improve efficiency for high-power machines that would have been cost-prohibitive to electrify directly. HHEA uses a set of common pressure rails (CPRs) to transmit the majority of power hydraulically and small electric motor drives to modulate that power and to achieve precise control. This article proposes a computationally efficient Lagrange multiplier method (LMM) for computing the optimal sequence of pressure rail selections to minimize energy use. This is needed to evaluate HHEA's energy-saving potential and for iterative architecture design and sizing. An interesting complication is that the cost function is not fully defined until the candidate control sequence is fully specified. This issue is dealt with by decomposing the original problem into a set of sub-problems with additional constraints that can be solved efficiently. Computational effort can be further reduced if actuators are optimized individually instead of together. However, additional steps are required to prevent the constraint functions from becoming discontinuous with respect to the Lagrange multipliers, which is necessary for meeting the constraints. Lastly, a case study of a construction machine demonstrates the efficacy of the method and shows that the HHEA reduces energy consumption by 68%-73% compared to the baseline load-sensing architecture.

Lagrange multiplier↗

SMALE: Enhancing Scalability of Machine Learning Algorithms on Extreme-Scale Computing Platforms

Deployment and execution of machine learning tasks on extreme-scale computing platforms face several significant technical challenges: 1) High computing cost incurred by dense networks – The computing workload of deep networks with densely-connected topology increases rapidly with the network size, imposing a non-scalable computing model of extreme-scale computing platforms; 2) Non-optimized workload distribution – Many advanced deep learning algorithms, e.g., sparsification and irregular net-work topology, produce very unbalanced workload distribution on extreme-scale computing platforms. The computation efficiency is greatly hindered by the incurred data and computation redundancies as well as long tails of the node with extensive workload; 3) Constraints in data movement and I/O bottle-neck – Inter-node data movement in extreme-scale computing platforms are associated with high energy and latency costs, and subject to the constraints of I/O bandwidth; and 4) Generalization of algorithm realization and acceleration on computing platforms – The large varieties of machine learning algorithms and structures of extreme-scale computing platforms make the derivation of a generalized algorithm realization and acceleration method very challenging, which, however, is the requirement by domain scientists and interested users. We call the above challenges Smale’s Problems in Machine Learning and Understanding for High-Performance Computing Scientific Discovery. The objective of our three-year research project is to develop a holistic innovation set at structure, assembly, and acceleration layers of machine learning algorithms to address the above challenges in algorithm deployment and execution. Three tasks are particularly performed, including: At the algorithm structure level, we investigate the techniques that can structurally sparsify on the topology of deep networks for computing workload reduction. We also study clustering and pruning techniques that can optimize the workload distributions over the extreme-scale computing platforms; At the algorithm assembly level, we derive a unified learning framework for unsupervised transfer learning and dynamic growing capabilities. Novel training methods are also exploited to enhance the training efficiency of the proposed framework; At the algorithm acceleration level, we will develop a series of techniques that can accelerate the computation of sparse matrix operations, which are one of the core executions in deep learning and optimize memory access of the concerned platforms. Our proposed techniques attack the fundamental problems in machine learning algorithms running on extreme-scale computing platforms by vertically integrating the solutions at three closely entangled layers, paving the long-term scaling path of machine learning applications under DOE context. Three tasks corresponding to the above respective research orientations are performed during the three-year project period with our collaborators at ORNL. The outcome of the proposed project is anticipated to form a holistic solution set of novel algorithms and network topologies, efficient training techniques, and fast acceleration methods to promote the computing scalability of the machine learning applications of particular interest to DOE.

97 MATHEMATICS AND COMPUTING↗