Search NASA⌕ Search

SEARCH · Search NASA

Results for “Language model”

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

A Global View Programming Abstraction for Transitioning MPI Codes to PGAS Languages

The multicore generation of scientific high performance computing has provided a platform for the realization of Exascale computing, and has also underscored the need for new paradigms in coding parallel applications. The current standard for writing parallel applications requires programmers to use languages designed for sequential execution. These languages have abstractions that only allow programmers to operate on the process centric local view of data. To provide suitable languages for parallel execution, many research efforts have designed languages based on the Partitioned Global Address Space (PGAS) programming model. Chapel is one of the more recent languages to be developed using this model. Chapel supports multithreaded execution with high-level abstractions for parallelism. With Chapel in mind, we have developed a set of directives that serve as intermediate expressions for transitioning scientific applications from languages designed for sequential execution to PGAS languages like Chapel that are being developed with parallelism in mind.

Global Domain↗

Numerical Investigation of Enhanced Dehumidification Processes By Using Dielectrophoresis Principles in Moist Airflows

Dispersed particle-laden flows are encountered in many building and industrial applications, such as flow in a fluidized bed, hydrocarbon transportation in pipelines, and the fouling of air-cooled heat exchangers (Kuruneru et al., 2016; Ray et al., 2019; Wang et al., 2019). Computational fluid dynamic (CFD) models have been developed in recent years to depict particle-fluid and particle-particle interactions in laminar or turbulent flows with increasing accuracy and stability. One particular particle-laden system of interest for moisture control is electrically-enhanced condensation in air and water droplet flows. Electrically-enhanced condensation consists of the use of highly charged water droplets injected in the moist air. The droplets become electric seeds that attract polar water vapor molecules to their surfaces and promote condensation. The nucleation and growth of the charged droplets deplete the vapor phase near a droplet, which is compensated for by the dielectrophoresis flow and diffusion. Dielectrophoresis flow involves surrounding vapor at a distance of about 10 to 100 nm for droplets charged by an electrospray compared to ~2 nm for a single electron charge in a droplet. As the vapor molecules collapse on the surface of the droplets, their initial electrical charge decreases with time due to the neutralization of the ions. While the physics of this phenomena is well known, engineering models for predicting the condensation rates are not available. This work computationally investigates dehumidification of moist airflow in a converging rectangular duct. The objective is to develop an engineering model that predicts water vapor condensation by employing dielectrophoresis principles. We construct a Computational Fluid Dynamics (CFD) model of the duct with electrically-enhanced condensation. The model is implemented in the open-source software OpenFOAM. We utilize the Multi-Phase Particle-In-Cell (MP-PIC) method coupled with a Population Balance Equation (PBE) approach to simulate the particle-laden system. This methodology is an Eulerian-Lagrangian approach used to simulate the droplets' behavior in the humid air. The MP-PIC approach (Andrews and O'Rourke, 1996) mitigates the computational cost by parceling several fundamental particles with similar properties (such as types, sizes, and temperature) into one computational particle. Thus, the billions of particles can be substituted by millions of computational particles without significant loss of information. The PBE was considered with the Lagrangian frame to combine the particle distribution function used in MP-PIC (Kim et al., 2020). This approach preserves mass and energy conservation between the phases in the Eulerian and Lagrangian structures. The PBE in this procedure was directly linked to the discrete parcels, making the simulation of the particle distribution computationally efficient and robust. The MP-PIC-PBE approach used in the present work was applied to the dehumidification of air. Water droplets were injected in the air stream and forced to grow according to experimentally derived correlation. The experiments were conducted on a converging duct with the same geometry and boundary conditions used to build the CFD model. This approach enabled us to approximate the effect of dielectrophoresis phenomena on the droplet and air interface. This presentation will discuss the details of the new CFD model built for the duct, the implementation of the model in OpenFOAM CFD programming language, and the experimental validation of the newly developed model. The results revealed a moderate yet measurable increase in droplet diameter due to water vapor condensation at the vapor-liquid interface of the electrically charged droplets' surface. The seed water droplet particles grew in size by capturing the water vapor in the surrounding air. The OpenFOAM model predicted reductions of humidity in the air from 5 to 10 percent.

Yel Mahi, Maliha↗

QLiG: Query Like a Graph For Subgraph Matching

A graph is a natural and flexible modeling approach to represent entities and relationships between them in real-world. A Knowledge Graphs (KG) is a specialized graph with formal and structured representation of facts, relationships, annotated with semantic descriptions. Subgraph matching is one of the fundamental graph problems to identify relationships, interactions and activities of interest within a large graph. A query specification is a collection of abstract components, operations, and constraints to express a pattern. The specification can be implemented in different ways based on underlying data model. Various graph query specifications have been developed over the years and have led to the development of different open-sourced and vendor-specific query languages. Such specification are modeled as an extension of relational algebra used to develop relational query languages such as SQL. Such relational concepts do not inherently support graph queries. There is a need to represent graph queries in terms on graph-based components to expedite query construction by non-database experts. We present a graph-based query approach QLiG (pronounced cleeg), to perform subgraph matching in Labeled Property Graph. We present the query specifications, salient features, and a use case to show functional examples.

Purohit, Sumit↗

Plug-and -Play Model Architecture and Development Environment for Powertrain/Propulsion System - Final CRADA Report

Several tools already exist to develop detailed plant model, including GT-Power, AMESim, CarSim, and SimScape. The objective of Autonomie is not to provide a language to develop detailed models; rather, Autonomie supports the assembly and use of models from design to simulation to analysis with complete plug-and-play capabilities. Autonomie provides a plug-and-play architecture to support this ideal use of modeling and simulation for math-based automotive control system design. Models in the standard format create building blocks, which are assembled at runtime into a simulation model of a vehicle, system, subsystem, or component to simulate. All parts of the graphical user interface (GUI) are designed to be flexible to support architectures, systems, components, and processes not yet envisioned. This allows the software to be molded to individual uses, so it can grow as requirements and technical knowledge expands. This flexibility also allows for implementation of legacy code, including models, controller code, processes, drive cycles, and post-processing equations. A library of useful and tested models and processes is included as part of the software package to support a full range of simulation and analysis tasks, immediately. Autonomie also includes a configuration and database management front end to facilitate the storage, versioning, and maintenance of all required files, such as the models themselves, the model’s supporting files, test data, and reports. During the duration of the CRADA, Argonne has worked closely with GM to implement and demonstrate each one of their requirements. A use case was developed by GM for every requirement and demonstrated by Argonne. Each of the new features were verified by GM experts through a series of Gate. Once all the requirements were validated they were presented to the directors as part of GM Gate process.

33 ADVANCED PROPULSION SYSTEMS↗

Dynamic Modeling of Full Converter Adjustable-speed Pumped Storage Hydropower (FC AS-PSH)

Full converter adjustable-speed pumped storage hydropower (FC AS-PSH) technology, as one of advanced-PSH technology, is developed from wind turbine technology. By making the synchronous machine connect to the grid through a full-size converter, FC AS-PSH has a wider adjustment range of speed and a better reactive power control capability compared with a doubly-fed asynchronous generator AS-PSH technology. When it plays as an energy backup in the power system, FC AS-PSH can provide a much faster response than conventional-PSH (C-PSH) which makes this technology provide better ancillary service for a high renewable penetrated system. In this paper, the dynamic modeling of FC AS-PSH is fully studied. We develop a detailed model of this technology in the IEEE 14-bus system based on GE Positive Sequence Load Flow (PSLF) platform. Especially, the first governor model is developed based on the Engineer’s Program Control Language (EPCL) user-defined model in this platform. All operation modes are validated and studied under a system contingency. Besides, comparison cases between FC AS-PSH and C-PSH are studied to show advantages providing from FC AS-PSH when it works with renewable energy.

50 EE - Wind and Water Power Program - Water (EE-4↗

Hardware In the Loop for Demand Flexibility (HIL4DF) v1.0

The software package in question is a collection of simulation models in the Modelica language, representing a variety of mechanical system designs and envelope conditions related to LBL's FLEXLAB facility. The collection of models also features multiple controls sequences that can be simulated with the FLEXLAB model to simulate different demand flexibility scenarios. Additionally, this package will feature datasets from 3 experimental tests, used for calibration, validation and comparison against the Modelica models, this includes weather data that can be used to replicate different scenarios in simulation across the same weather conditions experienced in real experiments. Given FLEXLAB high level of instrumentation and available data, the models are calibrated across multiple measurement points, and thus results from the extension of this model to other climate zones or control sequences, would provide high level of confidence.

Huang, Weiping↗

Scaling communication-intensive applications on BlueGene/P using one-sided communication and overlap

In earlier work, we showed that the one-sided communication model found in PGAS languages (such as UPC) offers significant advantages in communication efficiency by decoupling data transfer from processor synchronization. We explore the use of the PGAS model on IBM Blue-Gene/P, an architecture that combines low-power, quad-core processors with extreme scalability. We demonstrate that the PGAS model, using a new port of the Berkeley UPC compiler and GASNet one-sided communication layer, outperforms two-sided (MPI) communication in both microbenchmarks and a case study of the communication-limited benchmark, NAS FT. We scale the benchmark up to 16,384 cores of the BlueGene/P and demonstrate that UPC consistently outperforms MPI by as much as 66% for some processor configurations and an average of 32%. Additionally, the results demonstrate the scalability of the PGAS model and the Berkeley implementation of UPC, the viability of using it on machines with multicore nodes, and the effectiveness of the BG/P communication layer for supporting one-sided communication and PGAS languages.

97 MATHEMATICS AND COMPUTING↗

Hierarchical computation in the SPMD programming model

Large-scale parallel machines are programmed mainly with the single program, multiple data (SPMD) model of parallelism. While this model has advantages of scalability and simplicity, it does not fit well with divide-and-conquer parallelism or hierarchical machines that mix shared and distributed memory. In this paper, we define the recursive single program, multiple data model (RSPMD) that extends SPMD with a hierarchical team mechanism to support hierarchical algorithms and machines. We implement this model in the Titanium language and describe how to eliminate a class of deadlocks by ensuring alignment of collective operations. We present application case studies evaluating the RSPMD model, showing that it enables divide-and-conquer algorithms such as sorting to be elegantly expressed and that team collective operations increase performance of conjugate gradient by up to a factor of two. The model also facilitates optimizations for hierarchical machines, improving scalability of particle in cell by 8x and performance of sorting and a stencil code by up to 40% and 14%, respectively.

Kamil, A↗

pvplr-python: Python package implementation of PVplr for Performance Loss Rate (PLR) analysis

Due to software fragmentation, PV system modeling teams can be limited to language specific packages, preventing cross-sectional analysis of different modeling techniques and workflows. To this end, PVplr, a popular PV performance modeling R software package, has been ported to the Python programming language. To verify and test the robustness of the port, NSRDB data has been used to simulated PV installations at native resolution (~2 million Sites), with a variety of degradation rates, degradation patterns, and modules. Performance Ratios were calculated using the ported functions from pvplr-python and compared against Rdtools YoY values. Due to the complicated nature of degradation, a new metric has been proposed to quantify the performance loss of a system. The cumulative production loss, is the total amount of energy lost due to the degrading performance of the system. Cumulative production loss alleviates the problems with fitting linear functions to non-linear degradation. Cumulative Production loss was shown to better estimate the total lost revenue for non-linear degradation patterns. $XbX + UTC$ was found to most accurately predict the total lost revenue in simulated systems.

Kumar, Suraj↗

Condition-Based Maintenance of a Circulating Water System of a Canadian Nuclear Power Plant using Machine Learning and Statistical Tools

Canada Deuterium Uranium pressurized-heavy-water reactors (PHWR) are a type of nuclear power plant that generate clean and reliable energy. The scope of this work is to automate data analysis methodologies to inform a condition-based maintenance strategy of a circulating water system (CWS) of a PHWR. The multiunit CWS provides a continuous supply of water to cool steam condensers, even during transient scenarios, thereby improving the thermal efficiency. This work aims to develop a machine learning (ML) based approach to detect anomalies in heterogeneous data of a CWS in a PHWR to help inform a predictive maintenance strategy. The heterogeneous data include textual and numeric time series data for a PHWR. Natural-language-processing (NLP)-based models are used to analyze textual data contained in work orders and operator logs and an event-timeseries correlation detection method is applied to assist anomalies diagnoses for CWS. An ML model Robust Linear Model (RLM) is also used to remove the seasonal variations in the system variable distributions based on distributions of environmental variables. A machine learning model, Density-Based Spatial Clustering of Applications with Noise (DBSCAN), trained on both original data and data without any seasonal variations will then be used to detect if an anomaly exists. Thus, by moving to an automated methodology to detect, classify, and forecast anomalies, the maintenance strategy would be based on component condition instead of a time-based schedule.

97 - MATHEMATICS AND COMPUTING↗

PhysBERT: A text embedding model for physics scientific literature

The specialized language and complex concepts in physics pose significant challenges for information extraction through Natural Language Processing (NLP). Central to effective NLP applications is the text embedding model, which converts text into dense vector representations for efficient information retrieval and semantic analysis. In this work, we introduce PhysBERT, the first physics-specific text embedding model. Pre-trained on a curated corpus of 1.2 × 106 arXiv physics papers and fine-tuned with supervised data, PhysBERT outperforms leading general-purpose models on physics-specific tasks, including the effectiveness in fine-tuning for specific physics subdomains.

Hellert, Thorsten (ORCID:0000000227970926)↗

Exponential Scaling of Water Exchange Rates with Ion Interaction Strength from the Perspective of Dynamic Facilitation Theory

Water exchange reactions around ionic solutes are ubiquitous in aqueous solution-phase chemistry. Yet, the extreme sensitivity of exchange rates to perturbations in the chemistry of an ionic solute is not well understood. We explore water exchange around model ions within the language of dynamic facilitation theory, typically used to describe glassy and other systems with collective, facilitated dynamics. Through the development of a coarse-grained, kinetically constrained lattice model of water exchange, we show that the time scale for water exchange scales exponentially with the strength of the solute–solvent interactions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Accelerating Transformer-based Deep Learning Models on FPGAs using Column Balanced Block Pruning

Although Transformer-based language representations achieve state-of-the-art accuracy on various natural language processing (NLP) tasks, the large model size has been challenging the resource constrained computing platforms. Weight pruning, as a popular and effective technique in reducing the number of weight parameters and accelerating the Transformer, has been investigated on GPUs. However, the Transformer acceleration using weight pruning on field-programmable gate array (FPGAs) remains unexplored. This paper investigates the column balanced block-wise pruning on Transformer and designs an FPGA acceleration engine to customize the balanced blockwise matrix multiplication. We implement the Transformer model with proper hardware scheduling, and the experiments show that the Transformer inference on FPGA achieves 10.35 ms latency with the batch size of 32, which is 10.96 × speed up comparing to CPU platform and 2.08 × speed up comparing to GPU platform.

Peng, Hongwu↗

Foundation models for atomistic simulation of chemistry and materials

Conventional computational methods for modeling chemical and materials systems are limited by system size and timescale, forcing a trade-off between quantum-mechanical accuracy and the sampling needed for realistic observables. Large language and vision foundation models — pre-trained on massive datasets using transformer architectures — have revolutionized many fields. It is thus interesting to ask whether a foundation model — subject to suitable data, parameter scaling and training — could enable learned simulations of chemistry and materials. Here, in this study, we review the field of machine-learned interatomic potentials (MLIPs) and posit that scaling up large and diverse chemical and materials datasets and highly expressive architectures using advanced training strategies should result in models that are: more efficient, transferable, robust to out-of-distribution scenarios, and easier to fine-tune to a variety of downstream physical observables than models trained from scratch on small datasets corresponding to specific, targeted atomistic simulation tasks. We provide specific criteria for creating such large-scale MLIP foundation models, coordinated strategies for their development, evaluation and deployment, and highlight potential emergent capabilities that could transform predictive simulations in chemistry and materials science and accelerate discovery across multiple technological domains.

Yuan, Eric C.-Y. [University of California, Berkel↗

Parallel Programming Models and Systems for High Performance Computing. In: Emerging Research in Cloud Distributed Computing Systems

A parallel programming model is an abstraction of a parallel system that allows expression of both algorithms and shared data structures. To accommodate the diversity in parallel system architectures and user requirements, there are a variety of programming models including the models providing a shared memory view or a distributed memory view of the system. The programming models are implemented as libraries, language extensions, or compiler directives. This chapter provides a discussion on programming models and its implementations aimed at application developers, system software researchers, and hardware architects. The first part provides an overview of the programming models. The second part is an in-depth discussion on high-performance networking interface to implement the programming model. The last part of the chapter discusses implementation of a programming model with a case study. Each part of the chapter concludes with a discussion on current research trends and its impact on future architectures.

Venkata, Manjunath Gorentia↗

Preprocessing Prediction of Advanced Algorithms for Medical Imaging

Advanced medical imaging algorithms (such as bone removal, vessel segmentation, or a lung nodule detection) can provide extremely valuable information to the radiologists, but they might sometimes be very time consuming. Being able to run the algorithms in advance can be a possible solution. However, we do not know which algorithm to run on a given dataset before it is actually used. It is possible to manually insert matching rules for preprocessing algorithms, but it requires high maintenance and does not work well in practice. This paper presents a dynamic machine learning solution for predicting which advanced visualization (AV) algorithm needs to be applied on a given series. The system gets a handful of free text DICOM tags as an input and builds a model in the clinical setting. It incorporates a Bag of Words (BOW) feature extractor and a Random Forest classifier. The approach was tested on two datasets from clinical sites which use different languages and varying scanner models. We show that even without feature extraction, sensitivity of above 90% can be reached on both of them. By using BOW feature extractor, precision and sensitivity can usually be further improved. Even on a noisy and highly unbalanced dataset, only around 100 samples were needed to reach sensitivity of above 80% and specificity of above 97%. We show how the solution can be part of a Smart Preprocessing mechanism in a viewing software. Using such a system will ultimately minimize the time to launch studies and improve radiologists reading time efficiency.

62 RADIOLOGY AND NUCLEAR MEDICINE↗