Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “efficient exploration”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 127 records · Page 7

Enabling R&D for accurate simulation of non-ideal explosives.

We implemented two numerical simulation capabilities essential to reliably predicting the effect of non-ideal explosives (NXs). To begin to be able to treat the multiple, competing, multi-step reaction paths and slower kinetics of NXs, Sandia's CTH shock physics code was extended to include the TIGER thermochemical equilibrium solver as an in-line routine. To facilitate efficient exploration of reaction pathways that need to be identified for the CTH simulations, we implemented in Sandia's LAMMPS molecular dynamics code the MSST method, which is a reactive molecular dynamics technique for simulating steady shock wave response. Our preliminary demonstrations of these two capabilities serve several purposes: (i) they demonstrate proof-of-principle for our approach; (ii) they provide illustration of the applicability of the new functionality; and (iii) they begin to characterize the use of the new functionality and identify where improvements will be needed for the ultimate capability to meet national security needs. Next steps are discussed.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

Testbed Environment for Space Situational Awareness: The world’s most accurate simulation of space debris and orbital object tracking

The Testbed Environment for Space Situational Awareness (TESSA) is a modeling and simulation environment designed for efficient exploration of scenarios involving satellites and space debris and their detection by both ground and space-based radars and optical sensors. TESSA is a Parallelized Discrete Event Simulator optimized for deployment on a large cluster of high-performance computers running Linux. TESSA includes capabilities for predicting the future location of a satellite using the satellite’s current location and velocity, determining an object’s orbit from sensor data, finding new orbital objects, predicting close approaches, determining the probability of collisions, modeling radar and optical and dynamic sensors, and modeling the generation of debris. TESSA also includes a scalable visualization environment implemented as a post-processor on workstation-class platforms.

42 ENGINEERING↗

High-Dimensional Spectral Sampling

A wide variety of applications in science and engineering depend on effectively and efficiently exploring high- dimensional parameter spaces. Common examples are uncertainty quantification, inverse problems, or various optimization approaches. In such applications, the first step is to create an initial uniform, random sampling of the space to create a baseline of knowledge. The simulations evaluated at these samples are then used for subsequent processing. In its most generic form, the goal of sampling is to produce the maximal amount of information with the minimal number of samples. However, despite being the first step and thus influencing the performance of the entire workflow, the quality of the initial sample designs is rarely analyzed or optimized. This project develops a new spectral sampling theory for analyzing and creating experiment designs in high-dimensional spaces. Further, new algorithms for estimating sample properties and sample synthesis are designed. Finally, applications of the proposed spectral sampling tools in surrogate modeling, scientific machine learning and image analysis are presented.

97 MATHEMATICS AND COMPUTING↗

Optimizing Distributed Training on Frontier for Large Language Models

Large language models (LLMs) have demonstrated remarkable success as foundational models, benefiting various downstream applications through fine-tuning. Loss scaling studies have demonstrated the superior performance of larger LLMs compared to their smaller counterparts. Nevertheless, training LLMs with billions of parameters poses significant challenges and requires considerable computational resources. For example, training a one trillion parameter GPT-style model on 20 trillion tokens requires a staggering 120 million exaflops. This research explores efficient distributed training strategies to extract this computation from Frontier, the world's first exascale supercomputer. We enable and investigate various model and data parallel training techniques, such as tensor parallelism, pipeline parallelism, and sharded data parallelism, to facilitate training a trillion-parameter model on Frontier. We empirically assess these techniques and their associated parameters to determine their impact on memory footprint, communication latency, and GPU's computational efficiency. We analyze the complex interplay among these techniques and find a strategy to combine them to achieve high throughput through hyperparameter tuning. We have identified efficient strategies for training large LLMs of varying sizes through empirical analysis and hyperparameter tuning. For 22 Billion, 175 Billion, and 1 Trillion parameters, we achieved GPU throughputs of 38.38%, 36.14%, and 31.96%, respectively. For the training of the 175 Billion parameter model and the 1 Trillion parameter model, we achieved 100% weak scaling efficiency on 1024 and 3072 Mi250X GPUs, respectively. We also achieved strong scaling efficiencies of 89% and 87% for these two models. We trained these models only tens of iterations instead of training till completion.

Yin, Junqi↗

Streamlining Ocean Dynamics Modeling with Fourier Neural Operators: A Multiobjective Hyperparameter and Architecture Optimization Approach

Training an effective deep learning model to learn ocean processes involves careful choices of various hyperparameters. We leverage DeepHyper’s advanced search algorithms for multiobjective optimization, streamlining the development of neural networks tailored for ocean modeling. The focus is on optimizing Fourier neural operators (FNOs), a data-driven model capable of simulating complex ocean behaviors. Selecting the correct model and tuning the hyperparameters are challenging tasks, requiring much effort to ensure model accuracy. DeepHyper allows efficient exploration of hyperparameters associated with data preprocessing, FNO architecture-related hyperparameters, and various model training strategies. We aim to obtain an optimal set of hyperparameters leading to the most performant model. Moreover, on top of the commonly used mean squared error for model training, we propose adopting the negative anomaly correlation coefficient as the additional loss term to improve model performance and investigate the potential trade-off between the two terms. The numerical experiments show that the optimal set of hyperparameters enhanced model performance in single timestepping forecasting and greatly exceeded the baseline configuration in the autoregressive rollout for long-horizon forecasting up to 30 days. Utilizing DeepHyper, we demonstrate an approach to enhance the use of FNO in ocean dynamics forecasting, offering a scalable solution with improved precision.

97 MATHEMATICS AND COMPUTING↗

An interactive computer code for calculation of gas-phase chemical equilibrium (EQLBRM)

A user friendly, menu driven, interactive computer program known as EQLBRM which calculates the adiabatic equilibrium temperature and product composition resulting from the combustion of hydrocarbon fuels with air, at specified constant pressure and enthalpy is discussed. The program is developed primarily as an instructional tool to be run on small computers to allow the user to economically and efficiency explore the effects of varying fuel type, air/fuel ratio, inlet air and/or fuel temperature, and operating pressure on the performance of continuous combustion devices such as gas turbine combustors, Stirling engine burners, and power generation furnaces.

Pratt, B. S.↗

On the next generation of reliability analysis tools

The current generation of reliability analysis tools concentrates on improving the efficiency of the description and solution of the fault-handling processes and providing a solution algorithm for the full system model. The tools have improved user efficiency in these areas to the extent that the problem of constructing the fault-occurrence model is now the major analysis bottleneck. For the next generation of reliability tools, it is proposed that techniques be developed to improve the efficiency of the fault-occurrence model generation and input. Further, the goal is to provide an environment permitting a user to provide a top-down design description of the system from which a Markov reliability model is automatically constructed. Thus, the user is relieved of the tedious and error-prone process of model construction, permitting an efficient exploration of the design space, and an independent validation of the system's operation is obtained. An additional benefit of automating the model construction process is the opportunity to reduce the specialized knowledge required. Hence, the user need only be an expert in the system he is analyzing; the expertise in reliability analysis techniques is supplied.

Babcock, Philip S., IV↗

Lunar He-3, fusion propulsion, and space development

The recent identification of a substantial lunar resource of the fusion energy fuel He-3 may provide the first terrestrial market for a lunar commodity and, therefore, a major impetus to lunar development. The impact of this resource-when burned in D-He-3 fusion reactors for space power and propulsion-may be even more significant as an enabling technology for safe, efficient exploration and development of space. One possible reactor configuration among several options, the tandem mirror, illustrates the potential advantages of fusion propulsion. The most important advantage is the ability to provide either fast, piloted vessels or high-payload-fraction cargo vessels due to a range of specific impulses from 50 sec to 1,000,000 sec at thrust-to-weight ratios from 0.1 to 5x10(exp -5). Fusion power research has made steady, impressive progress. It is plausible, and even probable, that fusion rockets similar to the designs presented here will be available in the early part of the twenty-first century, enabling a major expansion of human presence into the solar system.

Santarius, John F.↗

Radiation Shielding for Space Flight

A safe and efficient exploration of space requires an understanding of space radiations so that human life and sensitive equipment can be protected. On the way to these sensitive sites, the radiation is modified in both quality and quantity. Many of these modifications are thought to be due to the production of pions and muons in the interactions between the radiation and intervening matter. A method to predict the effects of the presence of these particles on the transport of radiation through materials is presented.

Blattnig, Steve R.↗

MESTRN: A Deterministic Meson-Muon Transport Code for Space Radiation

A safe and efficient exploration of space requires an understanding of space radiations, so that human life and sensitive equipment can be protected. On the way to these sensitive sites, the radiation fields are modified in both quality and quantity. Many of these modifications are thought to be due to the production of pions and muons in the interactions between the radiation and intervening matter. A method used to predict the effects of the presence of these particles on the transport of radiation through materials is developed. This method was then used to develop software, which was used to calculate the fluxes of pions and muons after the transport of a cosmic ray spectrum through aluminum and water. Software descriptions are given in the appendices.

Blattnig, Steve R.↗

Development of Testing Station for Prototype Rover Thermal Subsystem

In order to successfully and efficiently explore the moon or other planets, a vehicle must be built to assist astronauts as they travel across the surface. One concept created to meet this need is NASA's Space Exploration Vehicle (SEV). The SEV, a small pressurized cabin integrated onto a 12-wheeled chassis, can support two astronauts up to 14 days. Engineers are currently developing the second generation of the SEV, with the goal of being faster, more robust, and able to carry a heavier payload. In order to function properly, the rover must dissipate heat produced during operation and maintain an appropriate temperature profile inside the rover. If these activities do not occur, components of the rover will start to break down, eventually leading to the failure of the rover. On the rover, these requirements are the responsibility of the thermal subsystem. My project for the summer was to design and build a testing station to facilitate the design and testing of the new thermal subsystem. As the rover develops, initial low fidelity parts can be interchanged for the high fidelity parts used on the rover. Based on a schematic of the proposed thermal system, I sized and selected parts for each of the components in the thermal subsystem. For the components in the system that produced heat but had not yet been finalized or fabricated, I used power resistors to model their load patterns. I also selected all of the fittings to put the system together and a mounting platform to support the testing station. Finally, I implemented sensors at various points in the system to measure the temperature, pressure, and flow rate, and a data acquisition system to collect this information. In the future, the information from these sensors will be used to study the behavior of the subsystem under different conditions and select the best part for the rover.

Burlingame, Kaitlin↗

The Representation of Tropical Cyclones Within the Global William Putman Non-Hydrostatic Goddard Earth Observing System Model (GEOS-5) at Cloud-Permitting Resolutions

The Goddard Earth Observing System Model (GEOS-S), an earth system model developed in the NASA Global Modeling and Assimilation Office (GMAO), has integrated the non-hydrostatic finite-volume dynamical core on the cubed-sphere grid. The extension to a non-hydrostatic dynamical framework and the quasi-uniform cubed-sphere geometry permits the efficient exploration of global weather and climate modeling at cloud permitting resolutions of 10- to 4-km on today's high performance computing platforms. We have explored a series of incremental increases in global resolution with GEOS-S from irs standard 72-level 27-km resolution (approx.5.5 million cells covering the globe from the surface to 0.1 hPa) down to 3.5-km (approx. 3.6 billion cells).

Putman, William M.↗

Facilitating NASA Earth Science Data Processing Using Nebula Cloud Computing

Cloud Computing has been implemented in several commercial arenas. The NASA Nebula Cloud Computing platform is an Infrastructure as a Service (IaaS) built in 2008 at NASA Ames Research Center and 2010 at GSFC. Nebula is an open source Cloud platform intended to: a) Make NASA realize significant cost savings through efficient resource utilization, reduced energy consumption, and reduced labor costs. b) Provide an easier way for NASA scientists and researchers to efficiently explore and share large and complex data sets. c) Allow customers to provision, manage, and decommission computing capabilities on an as-needed bases

Pham, Long↗

LEGION: Lightweight Expandable Group of Independently Operating Nodes

LEGION is a lightweight C-language software library that enables distributed asynchronous data processing with a loosely coupled set of compute nodes. Loosely coupled means that a node can offer itself in service to a larger task at any time and can withdraw itself from service at any time, provided it is not actively engaged in an assignment. The main program, i.e., the one attempting to solve the larger task, does not need to know up front which nodes will be available, how many nodes will be available, or at what times the nodes will be available, which is normally the case in a "volunteer computing" framework. The LEGION software accomplishes its goals by providing message-based, inter-process communication similar to MPI (message passing interface), but without the tight coupling requirements. The software is lightweight and easy to install as it is written in standard C with no exotic library dependencies. LEGION has been demonstrated in a challenging planetary science application in which a machine learning system is used in closed-loop fashion to efficiently explore the input parameter space of a complex numerical simulation. The machine learning system decides which jobs to run through the simulator; then, through LEGION calls, the system farms those jobs out to a collection of compute nodes, retrieves the job results as they become available, and updates a predictive model of how the simulator maps inputs to outputs. The machine learning system decides which new set of jobs would be most informative to run given the results so far; this basic loop is repeated until sufficient insight into the physical system modeled by the simulator is obtained.

Burl, Michael C.↗

Improving operations: Metrics to Results

As a result of the mission failure of the Mars Climate Orbiter (MCO) spacecraft in 1999, the Jet Propulsion Laboratory (JPL) initiated the development of a Mission Operations Assurance (MOA) program to be implemented across all flight projects managed by JPL. One of the initiatives undertaken in 2001 was the collection of data on command file errors occurring in the operational phase of the mission. This paper defines command file errors and how and where they occur in the operations process. It also describes the problem reporting system (PRS) in use for mission operations at JPL. We examine the recent modifications to the PRS that enable the collection of metrics, specifically on command file errors. This paper discusses what the data show us since metrics have been collected for the operational missions conducted by JPL. We examine the evolution of an operational working group initiative to evaluate proximate, contributing, and root causes for the errors. As part of this discussion we see what the metrics have indicated over a decade. At the macro level, we can say that the aggregate command file error rate has been cut to roughly one third of the initial 2001 level by the end of 2011. Additionally, we explore efficient and innovative means to continually integrate the findings and recommendations from the working group back into the flight operations environment.

command file errors↗

Global Optimization of Low-Thrust Interplanetary Trajectories Subject to Operational Constraints

Low-thrust interplanetary space missions are highly complex and there can be many locally optimal solutions. While several techniques exist to search for globally optimal solutions to low-thrust trajectory design problems, they are typically limited to unconstrained trajectories. The operational design community in turn has largely avoided using such techniques and has primarily focused on accurate constrained local optimization combined with grid searches and intuitive design processes at the expense of efficient exploration of the global design space. This work is an attempt to bridge the gap between the global optimization and operational design communities by presenting a mathematical framework for global optimization of low-thrust trajectories subject to complex constraints including the targeting of planetary landing sites, a solar range constraint to simplify the thermal design of the spacecraft, and a real-world multi-thruster electric propulsion system that must switch thrusters on and off as available power changes over the course of a mission.

Design↗

MARGInS Model-Based Analysis of Realizable Goals in Systems

The high complexity of modern aircraft and spacecraft requires elaborate Verification and Validation (V&V) approaches to make sure that such complex systems work properly and reliably. MARGInS is a framework for the analysis, understanding, and prediction of the behavior of a complex, hybrid system. MARGInS contains a set of machine learning and statistical algorithms for multivariate clustering, treatment learning, critical factor determination, time-series analysis, event prediction, and safety-boundary detection and characterization. The framework supports system testing and can be configured to find novel features in test suites, determine classes of behavior, propose new experiments that can efficiently explore and characterize the boundaries between classes of system behavior, and to create visualizations and reports.

He, Yuning↗