SEARCH · Engineering Papers

Results for “model discovery”

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.

SourceResearch connectionFrom yearThrough year

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

At least 451 records · Page 25

All sources NASA NTRS DOE OSTI

DOE OSTI2023

Scientific machine learning for modeling and simulating complex fluids

The formulation of rheological constitutive equations—models that relate internal stresses and deformations in complex fluids—is a critical step in the engineering of systems involving soft materials. While data-driven models provide accessible alternatives to expensive first-principles models and less accurate empirical models in many engineering disciplines, the development of similar models for complex fluids has lagged. The diversity of techniques for characterizing non-Newtonian fluid dynamics creates a challenge for classical machine learning approaches, which require uniformly structured training data. Consequently, early machine-learning based constitutive equations have not been portable between different deformation protocols or mechanical observables. Here, we present a data-driven framework that resolves such issues, allowing rheologists to construct learnable models that incorporate essential physical information, while remaining agnostic to details regarding particular experimental protocols or flow kinematics. These scientific machine learning models incorporate a universal approximator within a materially objective tensorial constitutive framework. By construction, these models respect physical constraints, such as frame-invariance and tensor symmetry, required by continuum mechanics. We demonstrate that this framework facilitates the rapid discovery of accurate constitutive equations from limited data and that the learned models may be used to describe more kinematically complex flows. This inherent flexibility admits the application of these “digital fluid twins” to a range of material systems and engineering problems. We illustrate this flexibility by deploying a trained model within a multidimensional computational fluid dynamics simulation—a task that is not achievable using any previously developed data-driven rheological equation of state.

Science & Technology - Other Topics↗

DOE OSTI2025

Data imbalance in drug response prediction: multi-objective optimization approach in deep learning setting

Abstract Drug response prediction (DRP) methods tackle the complex task of associating the effectiveness of small molecules with the specific genetic makeup of the patient. Anti-cancer DRP is a particularly challenging task requiring costly experiments as underlying pathogenic mechanisms are broad and associated with multiple genomic pathways. The scientific community has exerted significant efforts to generate public drug screening datasets, giving a path to various machine learning models that attempt to reason over complex data space of small compounds and biological characteristics of tumors. However, the data depth is still lacking compared to application domains like computer vision or natural language processing domains, limiting current learning capabilities. To combat this issue and improves the generalizability of the DRP models, we are exploring strategies that explicitly address the imbalance in the DRP datasets. We reframe the problem as a multi-objective optimization across multiple drugs to maximize deep learning model performance. We implement this approach by constructing Multi-Objective Optimization Regularized by Loss Entropy loss function and plugging it into a Deep Learning model. We demonstrate the utility of proposed drug discovery methods and make suggestions for further potential application of the work to achieve desirable outcomes in the healthcare field.

Biochemistry & Molecular Biology↗

DOE OSTI2022

Probing cosmic inflation with the LiteBIRD cosmic microwave background polarization survey

LiteBIRD, the Lite (Light) satellite for the study of B-mode polarization and Inflation from cosmic background Radiation Detection, is a space mission for primordial cosmology and fundamental physics. The Japan Aerospace Exploration Agency (JAXA) selected LiteBIRD in May 2019 as a strategic large-class (L-class) mission, with an expected launch in the late 2020s using JAXA’s H3 rocket. LiteBIRD is planned to orbit the Sun–Earth Lagrangian point L2, where it will map the cosmic microwave background polarization over the entire sky for three years, with three telescopes in 15 frequency bands between 34 and 448 GHz, to achieve an unprecedented total sensitivity of $2.2\, \mu$K-arcmin, with a typical angular resolution of 0.5° at 100 GHz. The primary scientific objective of LiteBIRD is to search for the signal from cosmic inflation, either making a discovery or ruling out well-motivated inflationary models. The measurements of LiteBIRD will also provide us with insight into the quantum nature of gravity and other new physics beyond the standard models of particle physics and cosmology. We provide an overview of the LiteBIRD project, including scientific objectives, mission and system requirements, operation concept, spacecraft and payload module design, expected scientific outcomes, potential design extensions, and synergies with other projects.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2022

Composition-transferable machine learning potential for LiCl-KCl molten salts validated by high-energy x-ray diffraction

Unraveling the liquid structure of multicomponent molten salts is challenging due to the difficulty in conducting and interpreting high-temperature diffraction experiments. Here, motivated by this challenge, we developed composition-transferable Gaussian approximation potential (GAP) for molten LiCl-KCl. A DFT-SCAN accurate GAP is active-learned from only ~1100 training configurations drawn from 10 unique mixture compositions enriched with metadynamics. The GAP-computed structures show strong agreement across high-energy x-ray diffraction experiments, including for a eutectic not explicitly included in model training, thereby opening the possibility of composition discovery.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

DOE OSTI2024

Efficient mapping between void shapes and stress fields using Deep Convolutional Neural Networks with sparse data

Establishing fast and accurate structure-to-property relationships is an important component in the design and discovery of advanced materials. Physics-based simulation models like the finite element method (FEM) are often used to predict deformation, stress, and strain fields as a function of material microstructure in material and structural systems. Such models may be computationally expensive and time intensive if the underlying physics of the system is complex. This limits their application to solve inverse design problems and identify structures that maximize performance. In such scenarios, surrogate models are employed to make the forward mapping computationally efficient to evaluate. However, the high dimensionality of the input microstructure and the output field of interest often renders such surrogate models inefficient, especially when dealing with sparse data. Deep convolutional neural network (CNN) based surrogate models have shown great promise in handling such high-dimensional problems. In this paper, a single ellipsoidal void structure under a uniaxial tensile load represented by a linear elastic, high-dimensional and expensive-to-query, FEM model. We consider two deep CNN architectures, a modified convolutional autoencoder framework with a fully connected bottleneck and a UNet CNN, and compare their accuracy in predicting the von Mises stress field for any given input void shape in the FEM model. Additionally, a sensitivity analysis study is performed using the two approaches, where the variation in the prediction accuracy on unseen test data is studied through numerical experiments by varying the number of training samples from 20 to 100.

surrogate modeling; convolutional neural networks;↗

DOE OSTI2023

Novel viruses of the family Partitiviridae discovered in Saccharomyces cerevisiae

It has been 49 years since the last discovery of a new virus family in the model yeast Saccharomyces cerevisiae . A large-scale screen to determine the diversity of double-stranded RNA (dsRNA) viruses in S . cerevisiae has identified multiple novel viruses from the family Partitiviridae that have been previously shown to infect plants, fungi, protozoans, and insects. Most S . cerevisiae partitiviruses (ScPVs) are associated with strains of yeasts isolated from coffee and cacao beans. The presence of partitiviruses was confirmed by sequencing the viral dsRNAs and purifying and visualizing isometric, non-enveloped viral particles. ScPVs have a typical bipartite genome encoding an RNA-dependent RNA polymerase (RdRP) and a coat protein (CP). Phylogenetic analysis of ScPVs identified three species of ScPV, which are most closely related to viruses of the genus Cryspovirus from the mammalian pathogenic protozoan Cryptosporidium parvum . Molecular modeling of the ScPV RdRP revealed a conserved tertiary structure and catalytic site organization when compared to the RdRPs of the Picornaviridae . The ScPV CP is the smallest so far identified in the Partitiviridae and has structural homology with the CP of other partitiviruses but likely lacks a protrusion domain that is a conspicuous feature of other partitivirus particles. ScPVs were stably maintained during laboratory growth and were successfully transferred to haploid progeny after sporulation, which provides future opportunities to study partitivirus-host interactions using the powerful genetic tools available for the model organism S . cerevisiae .

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2025

Protein-ligand binding affinity prediction using multi-instance learning with docking structures

Recent advances in 3D structure-based deep learning approaches demonstrate improved accuracy in predicting protein-ligand binding affinity in drug discovery. These methods complement physics-based computational modeling such as molecular docking for virtual high-throughput screening. Despite recent advances and improved predictive performance, most methods in this category primarily rely on utilizing co-crystal complex structures and experimentally measured binding affinities as both input and output data for model training. Nevertheless, co-crystal complex structures are not readily available and the inaccurate predicted structures from molecular docking can degrade the accuracy of the machine learning methods. We introduce a novel structure-based inference method utilizing multiple molecular docking poses for each complex entity. Our proposed method employs multi-instance learning with an attention network to predict binding affinity from a collection of docking poses. We validate our method using multiple datasets, including PDBbind and compounds targeting the main protease of SARS-CoV-2. The results demonstrate that our method leveraging docking poses is competitive with other state-of-the-art inference models that depend on co-crystal structures. This method offers binding affinity prediction without requiring co-crystal structures, thereby increasing its applicability to protein targets lacking such data.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

Prospects for gravitational wave and ultra-light dark matter detection with binary resonances beyond the secular approximation

Precision observations of orbital systems have recently emerged as a promising new means of detecting gravitational waves and ultra-light dark matter, offering sensitivity in new regimes with significant discovery potential. These searches rely critically on precise modeling of the dynamical effects of these signals on the observed system; however, previous analyses have mainly only relied on the secularly-averaged part of the response. We introduce here a fundamentally different approach that allows for a fully time-resolved description of the effects of oscillatory metric perturbations on orbital dynamics. We find that gravitational waves and ultra-light dark matter can induce large oscillations in the orbital parameters of realistic binaries, enhancing the sensitivity to such signals by orders of magnitude compared to previous estimates.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

NASA NTRS2008

NASA's THEMIS Mission: Multipoint Observations of Substorms, the Foreshock, and the Magnetopause

From launch on February 17 through the repositioning to final orbits that began in September 2007, the five-spacecraft of the THEMIS mission operated nominally in nearly identical 14.6 RE apogee near-equatorial orbits. On March 23, while aligned from east to west in the duskside magnetotail, the spacecraft observed two substorm sequences in fast survey mode. Timing the motion of these signatures served as an early proof of concept for the main phase of the mission: particle injection and dipolarization signatures propagated duskward from one probe to another, as did auroral intensifications seen by the dedicated array of ground-based observatories. During the summer of 2007, the spacecraft were on the dayside, where the three inner spacecraft (C, D, E) were separated by 100-500 km and the two outer probes (B, -4) by 5,000 - 10,000 km. Here the THEMIS probes repeatedly encountered the magnetopause and bow shock, dissecting flux transfer events (FTEs), determining the instantaneous width of the low-latitude boundary layer, and simultaneously observing hot flow anomalies upstream and downstream from the bow shock at the moment of their inception. From January to March 2008, the spacecraft were in the Earths magnetotail with apogees of 31.0, 19.5, 11.8 (2) and 10.0 RE corresponding to periods of 4, 2, and 1 days. Radial alignments once each four days offered an opportunity to pinpoint when and where substorms begin. This talk reviews THEMIS discoveries to date, with an emphasis on model-data comparisons of FTE characteristics

Sibeck, D. G.↗

NASA NTRS2019

Using UMM-Var and E2E to Improve the User Experience for Accessing NASA EOSDIS Data Sets

The UMM-Variables (Var) Metadata Model has been evolved to support an End-to-End Services (E2E) capability, which enables variable level subsetting, data transformation, and data reformatting. This talk will discuss what is new with the model, how users can get their metadata ready for the E2E capability, and include a demo of how the model is being used to drive and improve the user experience in Earthdata Search Client when accessing EOSDIS data sets.

Metadata Modeling↗

NASA NTRSDate not supplied

Interpretable Tree-Based and Graph Neural Network Approaches for Novel Solid State Electrolyte Design

All-solid-state batteries with Li metal anode can address the safety issues surrounding traditional Li-ion batteries as well as the demand for higher energy densities. However, the development of solid electrolytes simultaneously possessing high ionic conductivity and good chemical and electrochemical stabilities has proven to be a challenge. I will present our informatics approach to explore the Li compound space for promising solid electrolytes using high-throughput multi-property screening and interpretable machine learning. This is accomplished through the generation of a large database of battery-related materials properties of Li compounds. We use tree-based ensemble learning methods and graph neural network approaches to accurately learn relationships between crystal structures and corresponding thermodynamic and kinetic properties, with interpretability being a major focus. Our models give us the ability to enable rapid discovery and design of novel solid-state battery chemistries.

Materials discovery↗

NASA NTRSDate not supplied

Interpretable ML Approaches for Novel Solid State Electrolyte Design

Shreyas J Honrao↗

DOE OSTI2026

US Participation in the Higgs Discovery

On July 4, 2012, scientists at CERN announced the discovery of an essential part of the Standard Model of particle physics: the Higgs boson. The search for the Higgs at experiments at the Large Hadron Collider was an international effort involving thousands of people, with physicists and engineers from US institutions playing a significant role throughout.

Fermilab, Fermilab↗

NASA NTRSDate not supplied

Modeling and Observations of OH UV-emissions in the Lunar Exosphere

The discovery of water on the Moon has led to several questions about lunar water’s origin, evolution, and total amount. One of the technique used to better understand the origin of the water, notably by the NASA LADEE/UVS instrument, is the observation of OH fluorescence. However, to better interpret the data, it is important to understand the processes leading to that emission. We present an application of the Aeroplanets model to the simulation of the OH emissions from the solar fluorescence and the dissociation of H2O from photon and electron impact in function of the solar activity. We applied the model to several cases of Lunar exosphere affected or not by meteor impacts that are known to release water in the exosphere.We show how we can retrieve the amount of OH and H2O on the line of sight of the different instruments and how the solar activity can affect that retrieval.

Guillaume Gronoff↗

NASA NTRS2011

A Quantitative, Time-Dependent Model of Oxygen Isotopes in the Solar Nebula: Step one

The remarkable discovery that oxygen isotopes in primitive meteorites were fractionated along a line of slope I rather than along the typical slope 0,52 terrestrial fractionation line occurred almost 40 years ago, However, a satisfactory, quantitative explanation for this observation has yet to be found, though many different explanations have been proposed, The first of these explanations proposed that the observed line represented the final product produced by mixing molecular cloud dust with a nucleosynthetic component, rich in O-16, possibly resulting from a nearby supernova explosion, Donald Clayton suggested that Galactic Chemical Evolution would gradually change the oxygen isotopic composition of the interstellar grain population by steadily producing O-16 in supernovae, then producing the heavier isotopes as secondary products in lower mass stars, Thiemens and collaborators proposed a chemical mechanism that relied on the availability of additional active rotational and vibrational states in otherwise-symmetric molecules, such as CO2, O3 or SiO2, containing two different oxygen isotopes and a second, photochemical process that suggested that differential photochemical dissociation processes could fractionate oxygen , This second line of research has been pursued by several groups, though none of the current models is quantitative,

Nuth, J. A.↗

NASA NTRS2012

The Evolution of Friction Stir Welding Theory at Marshall Space Flight Center

From 1995 to the present the friction stir welding (FSW) process has been under study at Marshall Space Flight Center (MSFC). This is an account of the progressive emergence of a set of conceptual tools beginning with the discovery of the shear surface, wiping metal transfer, and the invention of a kinematic model and making possible a treatment of both metallurgical structure formation and process dynamics in friction stir welding from a unified point of view. It is generally observed that the bulk of the deformation of weld metal around the FSW pin takes place in a very narrow, almost discontinuous zone with high deformation rates characteristic of metal cutting. By 1999 it was realized that this zone could be treated as a shear surface like that in simple metal cutting models. At the shear surface the seam is drawn out and compressed and pressure and flow conditions determine whether or not a sound weld is produced. The discovery of the shear surface was followed by the synthesis of a simple 3- flow kinematic model of the FSW process. Relative to the tool the flow components are: (1) an approaching translational flow at weld speed V, (2) a rotating cylindrical plug flow with the angular velocity of the tool , and (3) a relatively slow ring vortex flow (like a smoke ring) encircling the tool and driven by shoulder scrolls and pin threads. The rotating plug flow picks up an element of weld metal, rotates it around with the tool, and deposits it behind the tool ( wiping metal transfer ); it forms plan section loops in tracers cut through by the tool. Radially inward flow from the ring vortex component retains metal longer in the rotating plug and outward flow expels metal earlier; this interaction forms the looping weld seam trace and the tongue and groove bimetallic weld contour. The radial components of the translational and ring vortex flows introduce parent metal intrusions into the small grained nugget material close to the tool shoulder; if this feature is pronounced, nugget collapse may result. Certain weld features, in particular internal banding seen in transverse section as onion rings and associated surface ridges called tool marks , have long implied an oscillation flow component, but have only recently been attributed in the literature to tool eccentricity. Rotating plug shape, typically a hollow cylinder flared at the end where it sticks to the shoulder, varies as pressure distribution on the tool determines where sticking occurs. Simplified power input estimates balanced against heat loss estimates give reasonable temperature estimates, explain why the power requirement changes hardly at all over a wide range of RPM s, and yield isotherms that seem to fall along boundaries of parameter windows of operation.

Nunes, Arthur C.↗

DOE OSTI2025

Critical impact of experimentally-driven strut level anisotropic material models in advanced stress analysis of additively manufactured lattice structures

The rapid acceleration in materials discovery may overshadow the importance of thoroughly understanding the mechanical performance of newly developed materials in demanding environments. The recent interest in combining parametric studies with machine learning techniques to explore how changes in specific processing parameters or model inputs affect the overall behavior of a material system can only be truly beneficial if the governing constitutive relations describing material behavior are accurately established. In this study, we demonstrate the critical impact of accurately representing strut-level anisotropic material behavior in advanced stress analysis of additively manufactured lattice structures (AMLS). We introduce a systematic experimental and modeling approach for developing strut-level anisotropic elastoplastic material models that account for the influence of microstructural features such as porosity, texture, and surface roughness on the development of local anisotropic mechanical properties, which vary with strut orientation relative to the build direction (BD). As a result the presented material model captures and relates the statistics of spatially varying struts’ microstructural features to the local stress distribution. Our findings suggest that incorporating strut-level anisotropic material behavior into unit cell analysis significantly influences the load distribution and evolution of local stresses within the structure. Therefore, accounting for this anisotropy is critical for developing an understanding of unit cell behavior and performance, including subsequent topology/component design optimization based on this analysis.

Sahoo, Subhadip [University of Arizona]↗

DOE OSTI2024

A Causal Approach to Model Validation and Calibration

This poster presents a novel method for validation and verification that focuses on identifying causal relationships between data elements, moving beyond traditional statistical and machine learning approaches. These methods employ causal discovery techniques to reveal the underlying mechanisms of data generation. The research utilizes structural causal models and directed acyclic graphs to depict causal relationships. This approach assists in achieving alignment between simulation models and reality.

97 MATHEMATICS AND COMPUTING↗