Serving Machine Learning Models in a Production Environment
A presentation on serving machine learning models in a production environment
97 MATHEMATICS AND COMPUTING↗
SEARCH · Search NASA
Results for “ENVIRONMENT 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.
Vertebrate development in the environment of space: models, mechanisms, and use of the medaka
With the advent of space travel, it is of immediate interest and importance to study the effects of exposure to various aspects of the altered environment of space, including microgravity, on Earth-based life forms. Initial studies of space travel have focused primarily on the short-term effects of radiation and microgravity on adult organisms. However, with the potential for increased lengths of time in space, it is critical to now address the effects of space on all phases of an organism's life cycle, from embryogenesis to post-natal development to reproduction. It is already possible for certain species to undergo multiple generations within the confines of the Mir Space Station. The possibility now exists for scientists to consider the consequences of even potentially subtle defects in development through multiple phases of an organism's life cycle, or even through multiple generations. In this discussion, we highlight a few of the salient observations on the effects of the space environment on vertebrate development and reproductive function. We discuss some of the many unanswered questions, in particular, in the context of the choice of appropriate models in which to address these questions, as well as an assessment of the availability of hardware already existing or under development which would be useful in addressing these questions.
Flight Experiment↗
Community Coordinated Modeling Center Support of Science Needs for Integrated Data Environment
Space science models are essential component of integrated data environment. Space science models are indispensable tools to facilitate effective use of wide variety of distributed scientific sources and to place multi-point local measurements into global context. The Community Coordinated Modeling Center (CCMC) hosts a set of state-of-the- art space science models ranging from the solar atmosphere to the Earth's upper atmosphere. The majority of models residing at CCMC are comprehensive computationally intensive physics-based models. To allow the models to be driven by data relevant to particular events, the CCMC developed an online data file generation tool that automatically downloads data from data providers and transforms them to required format. CCMC provides a tailored web-based visualization interface for the model output, as well as the capability to download simulations output in portable standard format with comprehensive metadata and user-friendly model output analysis library of routines that can be called from any C supporting language. CCMC is developing data interpolation tools that enable to present model output in the same format as observations. CCMC invite community comments and suggestions to better address science needs for the integrated data environment.
Kuznetsova, M. M.↗
A simple hyperbolic model for communication in parallel processing environments
We introduce a model for communication costs in parallel processing environments called the 'hyperbolic model,' which generalizes two-parameter dedicated-link models in an analytically simple way. Dedicated interprocessor links parameterized by a latency and a transfer rate that are independent of load are assumed by many existing communication models; such models are unrealistic for workstation networks. The communication system is modeled as a directed communication graph in which terminal nodes represent the application processes that initiate the sending and receiving of the information and in which internal nodes, called communication blocks (CBs), reflect the layered structure of the underlying communication architecture. The direction of graph edges specifies the flow of the information carried through messages. Each CB is characterized by a two-parameter hyperbolic function of the message size that represents the service time needed for processing the message. The parameters are evaluated in the limits of very large and very small messages. Rules are given for reducing a communication graph consisting of many to an equivalent two-parameter form, while maintaining an approximation for the service time that is exact in both large and small limits. The model is validated on a dedicated Ethernet network of workstations by experiments with communication subprograms arising in scientific applications, for which a tight fit of the model predictions with actual measurements of the communication and synchronization time between end processes is demonstrated. The model is then used to evaluate the performance of two simple parallel scientific applications from partial differential equations: domain decomposition and time-parallel multigrid. In an appropriate limit, we also show the compatibility of the hyperbolic model with the recently proposed LogP model.
Stoica, Ion↗
Harmonic Infrared and Raman Spectra in Molecular Environments Using the Polarizable Embedding Model
We present a fully analytic approach to calculate infrared (IR) and Raman spectra of molecules embedded in complex molecular environments modeled using the fragment-based polarizable embedding (PE) model. We provide the theory for the calculation of analytic second-order geometric derivatives of molecular energies and first-order geometric derivatives of electric dipole moments and dipole–dipole polarizabilities within the PE model. The derivatives are implemented using a general open-ended response theory framework, thus allowing for an extension to higher-order derivatives. The embedding-potential parameters used to describe the environment in the PE model are derived through first-principles calculations, thus allowing a wide variety of systems to be modeled, including solvents, proteins, and other large and complex molecular environments. Here, we present proof-of-principle calculations of IR and Raman spectra of acetone in different solvents. This work is an important step toward calculating accurate vibrational spectra of molecules embedded in realistic environments.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Numerical models for Halley dust environments
The algebra, parameter values, and results of a reference model for the dust environment of Comet P/Halley are presented. The model employs simplified treatments of the icy conglomerate nucleus and of the hydrodynamic outflow of gas and dust. The model includes a gas production rate derived from observations of Halley's visual magnitude in 1910, gas and dust flux nonuniformly distributed over the nuclear surface, a dust distribution function consistent with large particle inferences from comet antitails, dust density variable with mass, and an approximation to the distribution of particles near the dust envelope. Many parameter values are derived by analogy with the results for other comets. The features are incorporated in a computer program which calculates the concentration, flux, and fluence of dust particles as functions of time, position and particle mass.
Divine, N.↗
Physically rigorous reduced-order flow models of fractured subsurface environments without explosive computational cost
Fractured media models comprise discontinuities of multiple lengths (e.g. fracture lengths and apertures, wellbore area) that fall into the relatively insignificant length scales spanning millimeter-scale fractures to centimeter-scale wellbores in comparison to the extensions of the field of interest, and challenge the conventional discretization methods imposing highly-fine meshing and formidably large numerical cost. By utilizing the recent developments in the finite element analysis of electromagnetics that allow to represent material properties on a hierarchical geometry, this project develops computational capabilities to model fluid flow, heat conduction, transport and induced polarization in large-scale geologic environments that possess geometrically-complex fractures and man-made infrastructures without explosive computational cost. The computational efficiency and robustness of this multi-physics modeling tool are demonstrated by considering various highly-realistic complex geologic environments that are common in many energy and national security related engineering problems.
42 ENGINEERING↗
Space - A unique environment for process modeling R&D
Process modeling, the application of advanced computational techniques to simulate real processes as they occur in regular use, e.g., welding, casting and semiconductor crystal growth, is discussed. Using the low-gravity environment of space will accelerate the technical validation of the procedures and enable extremely accurate determinations of the many necessary thermophysical properties. Attention is given to NASA's centers for the commercial development of space; joint ventures of universities, industries, and goverment agencies to study the unique attributes of space that offer potential for applied R&D and eventual commercial exploitation.
Overfelt, Tony↗
Modeling Chandra Space Environment
This paper describes the development of an environmental risk-mitigation tool for the Chandra X-ray Observatory's Advanced CCD Imaging Spectrometer (ACIS). Because exposure to 100-200 keV protons appears to have degraded the front-illuminated CCD's charge transfer inefficiency (CTI), an accurate tool for predicting encounters with magnetospheric regions rich in these particles is required. We implement standard models to predict bow-shock, magnetopause, and plasma-sheet boundaries. Using these models and solar-wind databases compiled from IMP-8 and ACE measurements, we then calculate the probability that Chandra is located in one of these regions, along with predicted particle flux, to arrive at appropriate safing times for the ACIS detector. Finally, we validate this tool by comparing the model's boundary-crossing and proton flux predictions with measurements from Chandra's on-board particle detector and with data from other spacecraft operating in the Earth's magnetosphere.
Blackwell, W. C.↗
Empirical modeling of environment-enhanced fatigue crack propagation in structural alloys for component life prediction
This research aims to develop the methods and understanding needed to incorporate time and loading variable dependent environmental effects on fatigue crack propagation (FCP) into computerized fatigue life prediction codes such as NASA FLAGRO (NASGRO). In particular, the effect of loading frequency on FCP rates in alpha + beta titanium alloys exposed to an aqueous chloride solution is investigated. The approach couples empirical modeling of environmental FCP with corrosion fatigue experiments. Three different computer models have been developed and incorporated in the DOS executable program. UVAFAS. A multiple power law model is available, and can fit a set of fatigue data to a multiple power law equation. A model has also been developed which implements the Wei and Landes linear superposition model, as well as an interpolative model which can be utilized to interpolate trends in fatigue behavior based on changes in loading characteristics (stress ratio, frequency, and hold times).
Richey, Edward, III↗
Modeling the Local Environment within Porous Electrode during Electrochemical Reduction of Bicarbonate
Not Available
30 DIRECT ENERGY CONVERSION↗
Mars 2020 Surface Mission Modeling: Landing Site Thermal Environments
This paper presents work done by the Mars 2020 Mission Planning Team to characterize landing site thermal environments. A process was developed to take in ground temperature simulation data for each landing site and efficiently discretize it into a handful of thermal bin environments. The Mars 2020 Thermal Team then performed detailed heater and energy modeling in each of those environments. The heater modeling results were implemented into a Monte Carlo based surface mission model to understand mission performance impacts. The Mars 2020 Instrument Teams used the results to inform their design and to better understand how the thermal conditions at each landing site affected their instruments.
Lange, Robert D.↗
Towards Plume Impingement Modeling in Space Environments
After 30 years human presence in low-earth orbit, NASA is returning to the moon and eventually will go to Mars. To this end, NASA is constructing the Lunar Gateway, an ISS-like space station to act as a home base for Lunar exploration. A large space station must be assembled while in orbit, using a “piecemeal” approach. In this context, it means that the different modules will arrive at different times and attach to what is already in service. The ISS provides an excellent example of this approach, and the proposed Lunar Gateway will undergo a similar assembly process. This assembly is achieved via “docking” maneuvers between modules, which are made possible by sequential firings of the onboard reaction control system (RCS) thrusters. They work by firing hot gases to produce adverse thrust and the needed change in velocity to safely finish the docking approach. The issue is that the gas from these thrusters' forms flow structures described as “plumes” and can impinge onto the outer surfaces of the space station, causing unwanted forces and moments, heat loads, sediment deposition, and in extreme cases, even surface erosion. All mechanisms that can damage the space station and must be avoided. Both permanent and visiting modules will have these RCS thruster exhaust impingement problems. Accurately and efficiently modeling these plume is an involved multi-physics calculation but also an important tool when designing the control algorithms of approaching modules. This poster presents progress towards this simulation on two fronts. First is the verification of OpenFOAM for rarefied plume impingement calculations by direct comparison to published DAC cases. Second is the estimation of plume impingement strikes over the time scale of an entire docking event. This is done by using a simple plume source flow model and a prescribed motion visualizer—coded in Python. Together these tools help push NASA's capabilities for simulating these plume impingement effects.
Rarefied Flows↗
Characterization of the Scale Model Acoustic Test Overpressure Environment using Computational Fluid Dynamics
The Scale Model Acoustic Test (SMAT) is a 5% scale test of the Space Launch System (SLS), which is currently being designed at Marshall Space Flight Center (MSFC). The purpose of this test is to characterize and understand a variety of acoustic phenomena that occur during the early portions of lift off, one being the overpressure environment that develops shortly after booster ignition. The pressure waves that propagate from the mobile launcher (ML) exhaust hole are defined as the ignition overpressure (IOP), while the portion of the pressure waves that exit the duct or trench are the duct overpressure (DOP). Distinguishing the IOP and DOP in scale model test data has been difficult in past experiences and in early SMAT results, due to the effects of scaling the geometry. The speed of sound of the air and combustion gas constituents is not scaled, and therefore the SMAT pressure waves propagate at approximately the same speed as occurs in full scale. However, the SMAT geometry is twenty times smaller, allowing the pressure waves to move down the exhaust hole, through the trench and duct, and impact the vehicle model much faster than occurs at full scale. The DOP waves impact portions of the vehicle at the same time as the IOP waves, making it difficult to distinguish the different waves and fully understand the data. To better understand the SMAT data, a computational fluid dynamics (CFD) analysis was performed with a fictitious geometry that isolates the IOP and DOP. The upper and lower portions of the domain were segregated to accomplish the isolation in such a way that the flow physics were not significantly altered. The Loci/CHEM CFD software program was used to perform this analysis.
Nielsen, Tanner↗
Mars 2020 Surface Mission Modeling Landing Site Thermal Environments
This paper presents work done by the Mars 2020 Mission Planning Team to characterize landing site thermal environments. A process was developed to take in ground temperature simulation data for each landing site and efficiently discretize it into six thermal environment bins. The Mars 2020 Thermal Team then performed detailed heater and energy modeling in each of those environments. The heater modeling results were implemented into a Monte Carlo based surface mission model to understand mission performance impacts. The Mars 2020 Instrument Teams used the results to inform their design and to better understand how the thermal conditions at each landing site affected their instruments.
Lange, Robert D.↗
Mars 2020 Surface Mission Modeling Landing Site Thermal Environments
This paper presents work done by the Mars 2020 Mission Planning Team to characterize landing site thermal environments. A process was developed to take in ground temperature simulation data for each landing site and efficiently discretize it into six thermal environment bins. The Mars 2020 Thermal Team then performed detailed heater and energy modeling in each of those environments. The heater modeling results were implemented into a Monte Carlo based surface mission model to understand mission performance impacts. The Mars 2020 Instrument Teams used the results to inform their design and to better understand how the thermal conditions at each landing site affected their instruments.
Lange, Robert D.↗
Shadow modeling in urban environments for solar harvesting devices with freely defined positions and orientations
Not Available
Arias-Rosales, Andrés (ORCID:0000000292668950)↗