Search NASA⌕ Search

SEARCH · Search NASA

Results for “parallel, scaling”

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 397 records · Page 22
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Mode-multiplexed photonic integrated vector dot-product core from inverse design

Photonic computing has the potential to harness the full degrees of freedom (DOFs) of the light field, including the wavelength, spatial mode, spatial location, phase quadrature, and polarization, to achieve a higher level of computing parallelism and scalability than digital electronic processors. While multiplexing using the wavelength and other DOFs can be readily integrated on silicon photonics platforms with compact footprints, conventional mode-division multiplexed (MDM) photonic designs occupy areas exceeding tens to hundreds of microns for a few spatial modes, significantly limiting their scalability. Here, we utilize inverse design to demonstrate an ultracompact photonic computing core that calculates vector dot products based on MDM coherent mixing. Our dot-product core integrates the functionalities of two-mode multiplexers and one multimode coherent mixer within a nominal footprint of 5 μm x 3 μm . We have experimentally demonstrated computing examples on the fabricated dot-product core, including complex number multiplication and motion estimation using optical flow. The compact dot-product core design enables large-scale on-chip integration in a parallel photonic computing primitive cluster for high-throughput scientific computing and computer vision tasks.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Laser-produced plasmas as drivers of laboratory collisionless quasi-parallel shocks

The creation of a repeatable collisionless quasi-parallel shock in the laboratory would provide a valuable platform for experimental studies of space and astrophysical shocks. However, conducting such an experiment presents substantial challenges. Scaling the results of hybrid simulations of quasi-parallel shock formation to the laboratory highlights the experimentally demanding combination of dense, fast, and magnetized background and driver plasmas required. One possible driver for such experiments is high-energy laser-produced plasmas (LPPs). Preliminary experiments at the University of California, Los Angeles, have explored LPPs as drivers of quasi-parallel shocks by combining the Phoenix Laser Laboratory [Niemann et al., J. Instrum. 7, P03010 (2012)] with a large plasma device [Gekelman et al., Rev. Sci. Instrum. 87, 025105 (2016)]. Beam instabilities and waves characteristic of the early stages of shock formation are observed, but spatial dispersion of the laser-produced plasma prematurely terminates the process. This result is illustrated by experimental measurements and Monte Carlo calculations of LPP density dispersion. The experimentally validated Monte Carlo model is then applied to evaluate several possible approaches to mitigating LPP dispersion in future experiments.

Heuer, P. V. (ORCID:0000000150506606)↗
DOE OSTI2022

A second-order distributed memory parallel fast sweeping method for the Eikonal equation

The Eikonal equation is used to calculate wave propagation and distance fields, and due to its complexity requires numerical treatment for its solution. In this work, we present a second-order distributed memory parallel fast sweeping method. The second-order solution switches on a two-point stencil when two upwind points are available, and reverts to first-order otherwise. In all examples, the second-order method improves the solution over the first-order, allowing for significant savings in memory while achieving the same accuracy. Parallelization over distributed memory saw good weak scaling with optimal convergence. The computational time for second-order was approximately 2.5 times slower than first-order, where the largest amount of mesh points ran on 144 cores (512 GB) was ≈20 billion. The savings in memory from the second-order method combined with the distributed memory algorithm result in the ability to solve problems much larger than are possible with the serial first-order method.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2020

Towards improved speed and accuracy of laser powder bed fusion simulations via representation of multiple time scales

Here, we report the result of investigations into alternative representations of the temporal evolution of the laser powder bed fusion (LPBF) process. In order to provide the reader with sufficient context for the following discussion, the modeling challenge and its underlying cause, as well as current attempts to provide a satisfactory solution, will be discussed. Next, the author will describe two numerical methods (multirate time integration and parallel-in-time) to better represent the temporal scales of the problem. Finally, the results of applying these methods to a two-dimensional version of the heat transfer portion of the LPBF problem will be presented, reporting on both convergence and performance behavior, which indicate the possibility of significant speedup of solution calculations (40–100 times), along with well-characterized effects on solution accuracy.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2017

COLA with scale-dependent growth: applications to screened modified gravity models

We present a general parallelized and easy-to-use code to perform numerical simulations of structure formation using the COLA (COmoving Lagrangian Acceleration) method for cosmological models that exhibit scale-dependent growth at the level of first and second order Lagrangian perturbation theory. For modified gravity theories we also include screening using a fast approximate method that covers all the main examples of screening mechanisms in the literature. We test the code by comparing it to full simulations of two popular modified gravity models, namely f ( R ) gravity and nDGP, and find good agreement in the modified gravity boost-factors relative to ΛCDM even when using a fairly small number of COLA time steps.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2017

High Fidelity Simulations of Large-scale Wireless Networks (Part II - FY2017)

The ability to simulate wireless networks at large-scale for meaningful amount of time is considerably lacking in today's network simulators. For this reason, many published work in this area often limit their simulation studies to less than a 1,000 nodes and either over-simplify channel characteristics or perform studies over time scales much less than a day. In this report, we show that one can overcome these limitations and study problems of high practical consequence. This work presents two key contributions to high fidelity simulation of large-scale wireless networks: (a) wireless simulations can be sped up by more than 100X in runtime using ideas from spatial indexing algorithms and clipping of negligible signals and (b) clustering and task-oriented programming paradigm can be used to reduce inter- process communication in a parallel discrete event simulation resulting in a better scaling efficiency.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Scaling neural simulations in STACS

Abstract As modern neuroscience tools acquire more details about the brain, the need to move towards biological-scale neural simulations continues to grow. However, effective simulations at scale remain a challenge. Beyond just the tooling required to enable parallel execution, there is also the unique structure of the synaptic interconnectivity, which is globally sparse but has relatively high connection density and non-local interactions per neuron. There are also various practicalities to consider in high performance computing applications, such as the need for serializing neural networks to support potentially long-running simulations that require checkpoint-restart. Although acceleration on neuromorphic hardware is also a possibility, development in this space can be difficult as hardware support tends to vary between platforms and software support for larger scale models also tends to be limited. In this paper, we focus our attention on Simulation Tool for Asynchronous Cortical Streams (STACS), a spiking neural network simulator that leverages the Charm++ parallel programming framework, with the goal of supporting biological-scale simulations as well as interoperability between platforms. Central to these goals is the implementation of scalable data structures suitable for efficiently distributing a network across parallel partitions. Here, we discuss a straightforward extension of a parallel data format with a history of use in graph partitioners, which also serves as a portable intermediate representation for different neuromorphic backends. We perform scaling studies on the Summit supercomputer, examining the capabilities of STACS in terms of network build and storage, partitioning, and execution. We highlight how a suitably partitioned, spatially dependent synaptic structure introduces a communication workload well-suited to the multicast communication supported by Charm++. We evaluate the strong and weak scaling behavior for networks on the order of millions of neurons and billions of synapses, and show that STACS achieves competitive levels of parallel efficiency.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2017

Numerical Prediction of CCV in a PFI Engine using a Parallel LES Approach

Cycle-to-cycle variability (CCV) is detrimental to IC engine operation and can lead to partial burn, misfire, and knock. Predicting CCV numerically is extremely challenging due to two key reasons. Firstly, high-fidelity methods such as large eddy simulation (LES) are required to accurately resolve the incylinder turbulent flowfield both spatially and temporally. Secondly, CCV is experienced over long timescales and hence the simulations need to be performed for hundreds of consecutive cycles. Ameen et al. (Int. J. Eng. Res., 2017) developed a parallel perturbation model (PPM) approach to dissociate this long time-scale problem into several shorter timescale problems. The strategy is to perform multiple single-cycle simulations in parallel by effectively perturbing the initial velocity field based on the intensity of the in-cylinder turbulence. This strategy was demonstrated for motored engine and it was shown that the mean and variance of the in-cylinder flowfield was captured reasonably well by this approach. In the present study, this PPM approach is extended to simulate the CCV in a fired port-fuel injected (PFI) SI engine. Two operating conditions are considered – a medium CCV operating case corresponding to 2500 rpm and 16 bar BMEP and a low CCV case corresponding to 4000 rpm and 12 bar BMEP. The predictions from this approach are also shown to be similar to the consecutive LES cycles. Both the consecutive and PPM LES cycles are observed to under-predict the variability in the early stage of combustion. The parallel approach slightly underpredicts the cyclic variability at all stages of combustion as compared to the consecutive LES cycles. However, it is shown that the parallel approach is able to predict the coefficient of variation (COV) of the in-cylinder pressure and burn rate related parameters with sufficient accuracy, and is also able to predict the qualitative trends in CCV with changing operating conditions. The convergence of the statistics predicted by the PPM approach with respect to the number of consecutive cycles required for each parallel simulation is also investigated. It is shown that this new approach is able to give accurate predictions of the CCV in fired engines in less than one-tenth of the time required for the conventional approach of simulating consecutive engine cycles.

Ameen, Muhsin M↗
DOE OSTI2020

A Scalable Parallel Hypergraph Generator (HyGen)

Graphs are extensively used to model real-world complex systems. An edge in a graph can model pairwise relationships. However, multiway relationships (connections between three or more vertices) are common in many complex systems such as cellular process, image segmentation, and circuit design. A graph edge cannot model multiway relationships. A hypergraph, which can connect more than two vertices, is thus a better option to model multiway relationships. A large-scale hypergraph analysis has the potential to find useful insights from a complex system and assist in knowledge discovery. Currently a limited number of hypergraphs exists that are representative of real-world datasets. Moreover, real-world hypergraph datasets are small in size and inadequate to incorporate future needs. A graph generator that can produce large-scale synthetic hypergraphs can solve the above mentioned problems. In this paper, we present a scalable parallel hypergraph generator (HyGen) based on the Message Passing Interface (MPI) standard. To generate hypergraphs, HyGen takes the following parameter values as inputs: i) number of vertices, ii) number of hyperedges, iii) number of clusters, iv) vertex distribution, v) hyperedge distribution, vi) local cluster cardinality, and vii) global cluster cardinality. We have demonstrated that HyGen can generate hypergraphs of various sizes in a scalable fashion. HyGen takes approximately four minutes to generate a hypergraph with 4.8 million vertices, 1.6 million hyperedges, and 800 clusters using 1,024 processes on a leadership class computing platform. Our strong and weak scaling experiments on supercomputers demonstrate that HyGen can quickly create large-scale hypergraphs in a parallel manner, thus providing a useful capability for hypergraph analysis.

Hasan, S M Shamimul↗
DOE OSTI2024

Optimizing High-Throughput Inference on Graph Neural Networks at Shared Computing Facilities with the NVIDIA Triton Inference Server

Abstract With machine learning applications now spanning a variety of computational tasks, multi-user shared computing facilities are devoting a rapidly increasing proportion of their resources to such algorithms. Graph neural networks (GNNs), for example, have provided astounding improvements in extracting complex signatures from data and are now widely used in a variety of applications, such as particle jet classification in high energy physics (HEP). However, GNNs also come with an enormous computational penalty that requires the use of GPUs to maintain reasonable throughput. At shared computing facilities, such as those used by physicists at Fermi National Accelerator Laboratory (Fermilab), methodical resource allocation and high throughput at the many-user scale are key to ensuring that resources are being used as efficiently as possible. These facilities, however, primarily provide CPU-only nodes, which proves detrimental to time-to-insight and computational throughput for workflows that include machine learning inference. In this work, we describe how a shared computing facility can use the NVIDIA Triton Inference Server to optimize its resource allocation and computing structure, recovering high throughput while scaling out to multiple users by massively parallelizing their machine learning inference. To demonstrate the effectiveness of this system in a realistic multi-user environment, we use the Fermilab Elastic Analysis Facility augmented with the Triton Inference Server to provide scalable and high-throughput access to a HEP-specific GNN and report on the outcome.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2017

HPC formulations of optimization algorithms for tensor completion

Tensor completion is a powerful tool used to estimate or recover missing values in multi-way data. It has seen great success in domains such as product recommendation and healthcare. Tensor completion is most often accomplished via low-rank sparse tensor factorization, a computationally expensive non-convex optimization problem which has only recently been studied in the context of parallel computing. In this work, we study three optimization algorithms that have been successfully applied to tensor completion: alternating least squares (ALS), stochastic gradient descent (SGD), and coordinate descent (CCD++). We explore opportunities for parallelism on shared- and distributed-memory systems and address challenges such as memory- and operation-efficiency, load balance, cache locality, and communication. Among our advancements are a communication-efficient CCD++ algorithm, an ALS algorithm rich in level-3 BLAS routines, and an SGD algorithm which combines stratification with asynchronous communication. Furthermore, we show that introducing randomization during ALS and CCD++ can accelerate convergence. We evaluate our parallel formulations on a variety of real datasets on a modern supercomputer and demonstrate speedups through 16384 cores. Further, these improvements reduce time-to-solution from hours to seconds on real-world datasets. We show that after our optimizations, ALS is advantageous on parallel systems of small-to-moderate scale, while both ALS and CCD++ provide the lowest time-to-solution on large-scale distributed systems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

Toroidal effect on runaway vortex and avalanche growth rate

The momentum-space dynamics of runaway electrons in a slab geometry, in terms of both the geometry and topological transition of the runaway vortex when synchrotron radiative damping is taking into account, have recently been shown to play a crucial role in runaway mitigation and avoidance. In a tokamak geometry, magnetic trapping arises from parallel motion along the magnetic field that scales as 1/R in strength with R the major radius. Since the transit time for a runaway electron moving along the field is of order 10 -8 sec while the collisional time is of ~ 0.01 sec in ITER-like plasmas, a bounce-averaged formulation can drastically reduce computational cost. Here the LAPS-RFP code’s implementation of a bounce-averaged relativistic Fokker- Planck model, along with the essential physics of synchrotron radiation damping and knock-on collisions, is described. It is found that the magnetic trapping can reduce the volume of the runaway vortex as the momentum-space fluxes are strongly modified inside the trapped-region. As a result, the avalanche growth rate is reduced at off-axis locations. In addition to benchmarking with previous calculations that did not take into account radiation damping, we also clarify how synchrotron radiation damping modifies the avalanche growth rate in a tokamak.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2019

Generation of kinetic Alfvén waves in dayside magnetopause reconnection: A 3-D global-scale hybrid simulation

In this paper, we perform a three-dimensional (3-D) global-scale hybrid simulation to investigate the generation of kinetic Alfvén waves (KAWs) in reconnection at the dayside magnetopause. In the simulation, the magnetopause reconnection takes place due to a self-consistent interaction between the solar wind and Earth's magnetosphere, in which the solar wind is assumed to carry a southward interplanetary magnetic field. As multiple X-line reconnection occurs, shear Alfvénic perturbations are generated from reconnection, which propagate toward polar regions along the magnetic field lines. Our detailed analysis of the electromagnetic characteristics reveals the existence of kinetic Alfvén waves on the ion kinetic scales in these perturbations, in which the parallel electric field is generated. The 3-D structures and the Poynting fluxes in the shear Alfvén waves/KAWs in the global magnetopause are presented. It is shown that these KAWs carry electromagnetic energy into the cusp, which may lead to aurora brightening.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Cross-field electron diffusion due to the coupling of drift-driven microinstabilities

In this paper, the nonlinear interaction between kinetic instabilities driven by multiple ion beams and magnetized electrons is investigated. Electron diffusion across magnetic field lines is enhanced by the coupling of plasma instabilities. Here, a two-dimensional collisionless particle-in-cell simulation is performed accounting for singly and doubly charged ions in a cross-field configuration. Consistent with prior linear kinetic theory analysis and observations from coherent Thomson scattering experiments, the present simulations identify an ion-ion two-stream instability due to multiply charged ions (flowing in the direction parallel to the applied electric field) which coexists with the electron cyclotron drift instability (propagating perpendicular to the applied electric field and parallel to the ExB drift). Small-scale fluctuations due to the coupling of these naturally driven kinetic modes are found to be a mechanism that can enhance cross-field electron transport and contribute to the broadening of the ion velocity distribution functions.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2018

Scalable Distributed Memory Community Detection Using Vite

Graph clustering, popularly known as community detection, is a fundamental graph operation used in several applications of relevance for network analysis and cybersecurity. Community detection is an operation of partitioning a network into an arbitrary number of “communities” such that each com munity represents a tightly-knit group of nodes with relatively sparser connections to the rest of the nodes in the network. The need to compute clustering on large-scale networks necessitates the development of efficient parallel algorithms capable of exploiting modern parallel architectures. However, due to their irregular and inherently sequential nature, many of the current algorithms for community detection are challenging to parallelize. In response to the 2018 Graph Challenge, we present Vite which is a distributed-memory parallel community detection method. Vite is a parallelization of the algorithmic template provided by a widely-used serial method (Louvain). Alongside, it uses several heuristics aimed at significantly improving parallelism and performance, while preserving output quality. Using the inputs from the 2018 Graph Challenge (static and streaming), we demonstrate superior performance and high quality solutions based on four parallelization heuristics.

community detection↗
DOE OSTI2021

Butterfly Factorization Via Randomized Matrix-Vector Multiplications

This paper presents an adaptive randomized algorithm for computing the butterfly factorization of an m × n matrix with m ≈ n provided that both the matrix and its transpose can be rapidly applied to arbitrary vectors. The resulting factorization is composed of O(log n) sparse factors, each containing O(n) nonzero entries. The factorization can be attained using O(n 3/2 log n) computation and O(n log n) memory resources. Furthermore, the proposed algorithm can be implemented in parallel and can apply to matrices with strong or weak admissibility conditions arising from surface integral equation solvers as well as multi-frontal-based finite-difference, finite-element, or finite-volume solvers. A distributed-memory parallel implementation of the algorithm demonstrates excellent scaling behavior.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Ensuring statistical reproducibility of ocean model simulations in the age of hybrid computing

Novel high performance computing systems that feature hybrid architectures require large scale code refactoring to unravel underlying exploitable parallelism. Such redesign can often be accompanied with machine-precision changes as the order of computation cannot always be maintained. For chaotic systems like climate models, these round-off level differences can grow rapidly. Systematic errors may also manifest initially as machine-precision differences. Isolating genuine round off level differences from such errors remains a challenge. Here, we apply two-sample equality of distribution tests to evaluate statistical reproducibility of the ocean model component of US Department of Energy's Energy Exascale Earth System Model (E3SM). A 2-year control simulation ensemble is compared to a modified ensemble as a test case - after a known non-bit-for-bit change in a model component is introduced - to evaluate the null hypothesis that the two ensembles are statistically indistinguishable. To quantify the false negative rates of these tests, we conduct a formal power analysis using a targeted suite of short simulation ensembles. The ensemble suite contains several perturbed ensembles, each with a progressively different climate than the baseline ensemble - obtained by perturbing the magnitude of a single model tuning parameter, the Gent and McWilliams κ, in a controlled manner. The null hypothesis is evaluated for each of perturbed ensembles using these tests. The power analysis informs on the detection limits of the tests for given ensemble size allowing model developers to evaluate the impact of an introduced non-bit-for-bit change to the model.

Mahajan, Salil↗
DOE OSTI2025

Knowledge Oriented Graph Unified Transformer (KOGUT) v0.1

KOGUT — Knowledge Oriented Graph Unified Transformer KOGUT implements the Relational Graph Transformer (RelGT) architecture for knowledge graph link prediction in biological domains, with a primary focus on microbial growth media prediction. While the original RelGT (arXiv:2505.10960) targets relational tables, time series, and multi-table databases, KOGUT adapts this architecture for heterogeneous biological knowledge graphs, providing first-in-class AI predictive models for microbial cultivation. Key Adaptations Beyond Original RelGT: - Knowledge Graph Focus: Applied to biological KGs with semantic node types (taxa, chemicals, media, phenotypes, environments) versus generic relational database tables, trained on the KG-Microbe knowledge graph (1.3M entities, 2.9M edges, 24 relation types). - Multimodal Node Encoding: Integrates node labels, categories, descriptions, and synonyms from KG metadata through learned embedding layers—adapting relational column features to graph node attributes with textual semantics. - Extended K-Hop Subgraph Strategy: Optimized neighborhood sampling (3-hop default, configurable up to 200 nodes) tuned for sparse biological networks, building on the original local-global attention framework with biological relation preservation. - Biolink Predicate Preservation: Type-specific transformations for 24 biological edge semantics (occurs_in, consumes, produces, has_phenotype, subclass_of) beyond standard relational foreign keys, enabling multi-relation link prediction. - Inductive Learning Support: Enables zero-shot predictions for novel taxa through feature-based embeddings (temperature, oxygen requirements, gram stain, cell shape), extending the original transductive relational benchmark scope to uncultured microorganisms. CheapSOTA Performance Optimizations (This Distribution): - VQ-EMA Centroid Attention: Vector quantization with exponential moving average for improved global context modeling (+5-10% MRR improvement). - HDF5 Precomputed Data Loading: One-time preprocessing of k-hop subgraphs to eliminate redundant graph traversals (2-5× training speedup). - Distributed Data Parallel Training: Multi-GPU support for scaling to larger knowledge graphs (tested on 4× NVIDIA A100 GPUs at NERSC Perlmutter). - Mixed Precision Training: Automatic mixed precision (AMP) for memory efficiency and faster training. Advantages Over Standard Knowledge Graph Embedding Models: Combines RelGT's proven multi-element tokenization (features, type, hop, structure) with graph-native biological representations, enabling interpretable link prediction across heterogeneous entities that standard embedding models (TransE, RotatE, ComplEx) and table-based transformers cannot directly model. Achieves near-perfect performance on microbial growth media prediction (MRR: 0.9966, Precision@1: 0.9932, Hit@10: 1.0000) while maintaining explainability through attention-based reasoning over biological pathways. Training Data: - KG-Microbe merged knowledge graph: 1,379,337 nodes, 2,960,472 edges - 24 biological relation types including taxonomic hierarchies, metabolic interactions, phenotype associations, and environmental relationships - Primary prediction task: Growth media suitability for microbial taxa (biolink:occurs_in, 50K edges) - Multi-relation capability: Predicts links for any of the 24 relation types, including chemical consumption/production, phenotype associations, and taxonomic classification Citation: Original RelGT Architecture: Dwivedi et al., "Relational Graph Transformer", arXiv:2505.10960, 2025 KOGUT Implementation: Knowledge Oriented Graph Unified Transformer for Microbial Growth Media Prediction Developed at Lawrence Berkeley National Laboratory (LBNL) Trained on NERSC Perlmutter supercomputer

Joachimiak, Marcin [Lawrence Berkeley National Lab↗