Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Differentiability”

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 1,567 records · Page 87
All sourcesNASA NTRSDOE OSTI
DOE OSTI2018

Salt-bridge modulates differential calcium-mediated ligand binding to integrin α1- and α2-I domains

Integrins are transmembrane cell-extracellular matrix adhesion receptors that impact many cellular functions. A subgroup of integrins contain an i nserted (I) domain within the α–subunits (αI) that mediate ligand recognition where function is contingent on binding a divalent cation at the m etal i on dependent a dhesion s ite (MIDAS). Ca 2+ is reported to promote α1I but inhibit α2I ligand binding. We co-crystallized individual I-domains with MIDAS-bound Ca 2+ and report structures at 1.4 and 2.15 Å resolution, respectively. Both structures are in the “closed” ligand binding conformation where Ca 2+ induces minimal global structural changes. Comparisons with Mg 2+ -bound structures reveal Mg 2+ and Ca 2+ bind α1I in a manner sufficient to promote ligand binding. In contrast, Ca 2+ is displaced in the α2I domain MIDAS by 1.4 Å relative to Mg 2+ and unable to directly coordinate all MIDAS residues. We identified an E152-R192 salt bridge hypothesized to limit the flexibility of the α2I MIDAS, thus, reducing Ca 2+ binding. A α2I E152A construct resulted in a 10,000-fold increase in Mg 2+ and Ca 2+ binding affinity while increasing binding to collagen ligands 20%. These data indicate the E152-R192 salt bridge is a key distinction in the molecular mechanism of differential ion binding of these two I domains.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2023

Machine-learning-based spectral methods for partial differential equations

Spectral methods are an important part of scientific computing’s arsenal for solving partial differential equations (PDEs). However, their applicability and effectiveness depend crucially on the choice of basis functions used to expand the solution of a PDE. The last decade has seen the emergence of deep learning as a strong contender in providing efficient representations of complex functions. In the current work, we present an approach for combining deep neural networks with spectral methods to solve PDEs. In particular, we use a deep learning technique known as the Deep Operator Network (DeepONet) to identify candidate functions on which to expand the solution of PDEs. We have devised an approach that uses the candidate functions provided by the DeepONet as a starting point to construct a set of functions that have the following properties: (1) they constitute a basis, (2) they are orthonormal, and (3) they are hierarchical, i.e., akin to Fourier series or orthogonal polynomials. We have exploited the favorable properties of our custom-made basis functions to both study their approximation capability and use them to expand the solution of linear and nonlinear time-dependent PDEs. The proposed approach advances the state of the art and versatility of spectral methods and, more generally, promotes the synergy between traditional scientific computing and machine learning.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Differentiating multi-MeV, multi-ion spectra with CR-39 solid-state nuclear track detectors

The development of high intensity petawatt lasers has created new possibilities for ion acceleration and nuclear fusion using solid targets. In such laser-matter interaction, multiple ion species are accelerated with broad spectra up to hundreds of MeV. To measure ion yields and for species identification, CR-39 solid-state nuclear track detectors are frequently used. However, these detectors are limited in their applicability for multi-ion spectra differentiation as standard image recognition algorithms can lead to a misinterpretation of data, there is no unique relation between track diameter and particle energy, and there are overlapping pit diameter relationships for multiple particle species. In this report, we address these issues by first developing an algorithm to overcome user bias during image processing. Second, we use calibration of the detector response for protons, carbon and helium ions (alpha particles) from 0.1 to above 10 MeV and measurements of statistical energy loss fluctuations in a forward-fitting procedure utilizing multiple, differently filtered CR-39, altogether enabling high-sensitivity, multi-species particle spectroscopy. To validate this capability, we show that inferred CR-39 spectra match Thomson parabola ion spectrometer data from the same experiment. Filtered CR-39 spectrometers were used to detect, within a background of ~ 2 × 10 11 sr -1 J -1 protons and carbons, (1.3 ± 0.7) × 10 8 sr -1 J -1 alpha particles from laser-driven proton-boron fusion reactions.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2026

Differentiable multiphase flow model for physics-informed machine learning in reservoir pressure management

Accurate subsurface reservoir pressure control is extremely challenging due to geological heterogeneity and multiphase fluid-flow dynamics. Predicting behavior in this setting relies on high-fidelity physics-based simulations that are computationally expensive. Yet, the uncertain, heterogeneous properties that control these flows make it necessary to perform many of these expensive simulations, which is often prohibitive. To address these challenges, we introduce a physics-informed machine learning workflow that couples a fully differentiable multiphase flow simulator, which is implemented in the DPFEHM framework with a convolutional neural network (CNN). The CNN learns to predict fluid extraction rates from heterogeneous permeability fields to enforce pressure limits at critical reservoir locations. By incorporating transient multiphase flow physics into the training process, our method enables more practical and accurate predictions for realistic injection-extraction scenarios compared to previous works. To speed up training, we pretrain the model on single-phase, steady-state simulations and then finetune it on full multiphase scenarios, which dramatically reduces the computational cost. We demonstrate that high-accuracy training can be achieved with fewer than three thousand full-physics multiphase flow simulations – compared to previous estimates requiring up to ten million. This drastic reduction in the number of simulations is achieved by leveraging transfer learning from much less expensive single phase simulations.

25 ENERGY STORAGE↗
DOE OSTI2024

Multi-resolution partial differential equations preserved learning framework for spatiotemporal dynamics

Traditional data-driven deep learning models often struggle with high training costs, error accumulation, and poor generalizability in complex physical processes. Physics-informed deep learning (PiDL) addresses these challenges by incorporating physical principles into the model. Most PiDL approaches regularize training by embedding governing equations into the loss function, yet this depends heavily on extensive hyperparameter tuning to weigh each loss term. To this end, we propose to leverage physics prior knowledge by “baking” the discretized governing equations into the neural network architecture via the connection between the partial differential equations (PDE) operators and network structures, resulting in a PDE-preserved neural network (PPNN). This method, embedding discretized PDEs through convolutional residual networks in a multi-resolution setting, largely improves the generalizability and long-term prediction accuracy, outperforming conventional black-box models. The effectiveness and merit of the proposed methods have been demonstrated across various spatiotemporal dynamical systems governed by spatiotemporal PDEs, including reaction-diffusion, Burgers’, and Navier-Stokes equations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Morphine and high-fat diet differentially alter the gut microbiota composition and metabolic function in lean versus obese mice

There are known associations between opioids, obesity, and the gut microbiome, but the molecular connection/mediation of these relationships is not understood. To better clarify the interplay of physiological, genetic, and microbial factors, this study investigated the microbiome and host inflammatory responses to chronic opioid administration in genetically obese, diet-induced obese, and lean mice. Samples of feces, urine, colon tissue, and plasma were analyzed using targeted LC-MS/MS quantification of metabolites, immunoassays of inflammatory cytokine levels, genome-resolved metagenomics, and metaproteomics. Genetic obesity, diet-induced obesity, and morphine treatment in lean mice each showed increases in distinct inflammatory cytokines. Metagenomic assembly and binning uncovered over 400 novel gut bacterial genomes and species. Morphine administration impacted the microbiome’s composition and function, with the strongest effect observed in lean mice. This microbiome effect was less pronounced than either diet or genetically driven obesity. Based on inferred microbial physiology from the metaproteome datasets, a high-fat diet transitioned constituent microbes away from harvesting diet-derived nutrients and towards nutrients present in the host mucosal layer. Considered together, these results identified novel host-dependent phenotypes, differentiated the effects of genetic obesity versus diet induced obesity on gut microbiome composition and function, and showed that chronic morphine administration altered the gut microbiome.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2022

TbasCO: trait-based comparative ‘omics identifies ecosystem-level and niche-differentiating adaptations of an engineered microbiome

A grand challenge in microbial ecology is disentangling the traits of individual populations within complex communities. Various cultivation-independent approaches have been used to infer traits based on the presence of marker genes. However, marker genes are not linked to traits with complete fidelity, nor do they capture important attributes, such as the timing of gene expression or coordination among traits. To address this, we present an approach for assessing the trait landscape of microbial communities by statistically defining a trait attribute as a shared transcriptional pattern across multiple organisms. Leveraging the KEGG pathway database as a trait library and the Enhanced Biological Phosphorus Removal (EBPR) model microbial ecosystem, we demonstrate that a majority (65%) of traits present in 10 or more genomes have niche-differentiating expression attributes. For example, while many genomes containing high-affinity phosphorus transporter pstABCS display a canonical attribute (e.g. up-regulation under phosphorus starvation), we identified another attribute shared by many genomes where transcription was highest under high phosphorus conditions. Taken together, we provide a novel framework for unravelling the functional dynamics of uncultivated microorganisms by assigning trait-attributes through genome-resolved time-series metatranscriptomics.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2017

Biomass-derived monomers for performance-differentiated fiber reinforced polymer composites

Nearly all polymer resins used to manufacture critically important fiber reinforced polymer (FRP) composites are petroleum sourced. In particular, unsaturated polyesters (UPEs) are widely used as matrix materials and are often based on maleic anhydride, a four-carbon, unsaturated diacid. Typically, maleic anhydride is added as a reactant in a conventional step-growth polymerization to incorporate unsaturation throughout the backbone of the UPE, which is then dissolved in a reactive diluent (styrene is widely used) infused into a fiber mat and cross-linked. Despite widespread historical use, styrene has come under scrutiny due to environmental and health concerns; in addition, many conceivable UPEs are not soluble in styrene. In this study, we demonstrate that renewably-sourced monomers offer the ability to overcome these issues and improve overall composite performance. The properties of poly(butylene succinate)-based UPEs incorporating maleic anhydride are used as a baseline for comparison against UPEs derived from fumaric acid, cis , cis -muconate, and trans , trans -muconate, all of which can be obtained biologically. The resulting biobased UPEs are combined with styrene, methacrylic acid, or a mixture of methacrylic acid and cinnaminic acid, infused into woven fiberglass and cross-linked with the addition of a free-radical initiator and heat. This process produces a series of partially or fully bio-derived composites. Overall, the muconate-containing UPE systems exhibit a more favorable property suite than the maleic anhydride and fumaric acid counterparts. In all cases at the same olefinic monomer loading, the trans , trans -muconate polymers exhibit the highest shear modulus, storage modulus, and glass transition temperature indicating stronger and more thermally resistant materials. They also exhibit the lowest loss modulus indicating a greater adhesion to the glass fibers. The use of a mixture of methacrylic and cinnaminic acid as the reactive diluent results in a FRP composite with properties that can be matched to reinforced composites prepared with styrene. Significantly, at one-third the monomer loading (corresponding to two-thirds the number of double bonds), trans , trans -muconate produces approximately the same storage modulus and glass transition temperature as maleic anhydride, while exhibiting a superior loss modulus. Altogether, this work demonstrates the novel synthesis of performance-differentiated FRP composites using renewably-sourced monomers.

09 BIOMASS FUELS↗
DOE OSTI2025

SANE: strategic autonomous non-smooth exploration for multiple optima discovery in multi-modal and non-differentiable black-box functions

Both computational and experimental material discovery bring forth the challenge of exploring multidimensional and multimodal parameter spaces, such as phase diagrams of Hamiltonians with multiple interactions, composition spaces of combinatorial libraries, material structure image spaces, and molecular embedding spaces. Often these systems are black-boxes and time-consuming to evaluate, which resulted in strong interest towards active learning methods such as Bayesian optimization (BO). However, these systems are often noisy which make the black box function severely multi-modal and non-differentiable, where a vanilla BO can get overly focused near a single or faux optimum, deviating from the broader goal of scientific discovery. To address these limitations, here we developed Strategic Autonomous Non-Smooth Exploration (SANE) to facilitate an intelligent Bayesian optimized navigation with a proposed cost-driven probabilistic acquisition function to find multiple global and local optimal regions, avoiding the tendency to becoming trapped in a single optimum. To distinguish between a true and false optimal region due to noisy experimental measurements, a human (domain) knowledge driven dynamic surrogate gate is integrated with SANE. We implemented the gate-SANE into pre-acquired piezoresponse spectroscopy data of a ferroelectric combinatorial library with high noise levels in specific regions, and piezoresponse force microscopy (PFM) hyperspectral data. SANE demonstrated better performance than classical BO to facilitate the exploration of multiple optimal regions and thereby prioritized learning with higher coverage of scientific values in autonomous experiments. Our work showcases the potential application of this method to real-world experiments, where such combined strategic and human intervening approaches can be critical to unlocking new discoveries in autonomous research.

Biswas, Arpan [University of Tennessee, Knoxville,↗
DOE OSTI2025

Strain-driven oxygen vacancy ordering in LaNiO 3 thin films revealed by integrated differential phase contrast imaging in scanning transmission electron microscopy

Rare-earth nickelates, such as LaNiO 3 (LNO), exhibit complex electronic properties, with ordered oxygen vacancies (OOV) influencing conductivity and magnetic behavior. We investigate the structural stability of strain-induced OOV phases in LNO thin films grown on SrTiO 3 substrates and the impact of Ruddlesden–Popper (RP) faults. Using high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and integrated differential phase contrast (iDPC) STEM imaging, we conducted atomic-scale structural and compositional analyses of OOV. Geometric phase analysis (GPA) was employed to measure the strain in fault-free and RP fault regions, while density functional theory (DFT) calculations explored different OOV arrangements in the LNO phase. Simulated iDPC-STEM imaging of energy-stabilized structures was performed to correlate with experimental results. Here, our findings reveal superstructure modulation in the chemical composition and atomic-scale lattice structure in LNO, primarily due to the formation of the OOV in Ni–O layers of the LaNiO 2.5 phase. The out-of-plane compressive strain of about 2% stabilizes this phase, reducing the strain, diminishing OOV, and transforming them into LNO.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Active Learning of Microgrid Frequency Dynamics Using Neural Ordinary Differential Equations

Accurate frequency modelling of inverter‐based resource (IBR)‐dominated power systems is crucial for ensuring stable, reliable and resilient operations, particularly given their inherent low‐inertia characteristics and fast dynamics that traditional swing equation‐based models inadequately capture. This paper explores neural ordinary differential equations (Neural ODEs) as a computationally efficient, data‐driven framework for modelling power system frequency dynamics, specifically within microgrids integrating high penetrations of distributed energy resources (DERs). The developed neural ODEs framework incorporates a neural network architecture designed to capture input dynamics. By actively perturbing the system with a known signal, the Python‐based neural ODEs framework was trained using measured system states and inputs, without the need for detailed system information. The framework, tested on a model of the Cordova, AK, microgrid, achieved a goodness of fit ranging from 60% to 99% across different state variables and maintained a mean square error in the 10 -6 p.u. range under square and step excitation signals. The proposed approach demonstrated robustness to measurement noise and initial condition variations while maintaining low computational complexity suitable for real‐time power system control applications. Furthermore, transfer learning enabled the neural ODEs model to adapt to the following changes in system topology or generator dispatch, highlighting its effectiveness for dynamic microgrids with frequently evolving configurations and diverse DERs.

Aryal, Tara [South Dakota State Univ., Brookings, ↗
DOE OSTI2021

Probabilistic error estimation for non-intrusive reduced models learned from data of systems governed by linear parabolic partial differential equations

This work derives a residual-based a posteriori error estimator for reduced models learned with non-intrusive model reduction from data of high-dimensional systems governed by linear parabolic partial differential equations with control inputs. It is shown that quantities that are necessary for the error estimator can be either obtained exactly as the solutions of least-squares problems in a non-intrusive way from data such as initial conditions, control inputs, and high-dimensional solution trajectories or bounded in a probabilistic sense. Here, the computational procedure follows an offline/online decomposition. In the offline (training) phase, the high-dimensional system is judiciously solved in a black-box fashion to generate data and to set up the error estimator. In the online phase, the estimator is used to bound the error of the reduced-model predictions for new initial conditions and new control inputs without recourse to the high-dimensional system. Numerical results demonstrate the workflow of the proposed approach from data to reduced models to certified predictions.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2016

Highly repeatable room temperature negative differential resistance in AlN/GaN resonant tunneling diodes grown by molecular beam epitaxy

AlN/GaN resonant tunneling diodes grown on low dislocation density semi-insulating bulk GaN substrates via plasma-assisted molecular-beam epitaxy are reported. The devices were fabricated using a six mask level, fully isolated process. Stable room temperature negative differential resistance (NDR) was observed across the entire sample. The NDR exhibited no hysteresis, background light sensitivity, or degradation of any kind after more than 1000 continuous up-and-down voltage sweeps. The sample exhibited a ∼90% yield of operational devices which routinely displayed an average peak current density of 2.7 kA/cm{sup 2} and a peak-to-valley current ratio of ≈1.15 across different sizes.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2016

State-to-state differential cross sections for D 2 + OH → D + DOH reaction: Influence of vibrational excitation of OH reactant

Here, state-to-state differential cross sections (DCSs) are computed quantum mechanically in full dimensionality for the title reaction using a reactant-product decoupling scheme. The DCSs are calculated at three collision energies of 0.25, 0.28, and 0.34 eV, corresponding to the existing experimental results. In good agreement with experiment, the calculated DCSs are dominated by backward scattering, thanks to the direct rebound mechanism, and the DOH product has two quanta of OD stretching vibration in the newly formed OD bond. In addition, the vibrational excitation of the OH reactant is found to result in a very different but predictable vibrational distribution of the DOH product. It is further shown at the state-to-state level that the DCSs of the DOH( v OD , v b , v OH ) product state from the OH( v = 1) reactant state resemble the ones of the DOH( v OD , v b , v OH -1) product state from the OH( v = 0) reactant state, thanks to the spectator nature of the OH moiety.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2016

Micro-differential scanning calorimeter for liquid biological samples

Here, we developed an ultrasensitive micro-DSC (differential scanning calorimeter) for liquid protein sample characterization. Our design integrated vanadium oxide thermistors and flexible polymer substrates with microfluidics chambers to achieve a high sensitivity (6 V/W), low thermal conductivity (0.7 mW/K), high power resolutions (40 nW), and well-defined liquid volume (1 μl) calorimeter sensor in a compact and cost-effective way. Furthermore, we demonstrated the performance of the sensor with lysozyme unfolding. The measured transition temperature and enthalpy change were in accordance with the previous literature data. This micro-DSC could potentially raise the prospect of high-throughput biochemical measurement by parallel operation with miniaturized sample consumption.

42 ENGINEERING↗
DOE OSTI2016

A reactant-coordinate-based approach to state-to-state differential cross sections for tetratomic reactions

A new algorithm is proposed to compute quantum mechanically state-to-state differential cross sections for reactions involving four atoms in full dimensionality. This algorithm, which is based on the propagation of an initial state specific wave packet exclusively in reactant coordinates, extracts the S-matrix elements in the product channel by first interpolating the time-dependent wave packet using a collocation method at selected time intervals on the product coordinate grid and then projecting out the contributions of all final product states. Here, this approach is efficient and accurate, particularly for reactions that are dominated by a product well or long-range interactions. Validation of this approach is demonstrated for the H 2 + OH → H + H 2 O reaction.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2018

Differential pumping requirements for the light-ion helicon source and heating systems of Proto-MPEX

The physics of electron and ion heating of high-density deuterium helicon plasmas (>3 × 10 19 m –3 ) in the Proto-Material Plasma Exposure Experiment linear device are under investigation. Theoretical estimates indicate that for efficient heating, discharges with very low neutral gas content ( &laquo; 0.1 Pa) in the heating sections are required to minimize collisional losses and charge exchange interactions with neutrals. However, this requirement is typically not compatible with the neutral gas pressures (1–2 Pa) commonly used in high-density, light-ion helicon sources. To satisfy these competing requirements, differential pumping techniques are needed. Here, results are presented that demonstrate the production of high-density discharges (2–6 × 10 19 m –3 ) with very low neutral gas content (<0.01 Pa) and high degree of ionization (>75%) in the heating sections. Results indicate that the best fueling location is upstream of the plasma source. We elaborate on the key aspects that must be considered to produce these discharges: (1) fueling location, radio-frequency pulse length, and magnetic field configuration, (2) flow rate and timing of the gas injection, and (3) use of conductance-limiting elements.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2017

A 3D-printed microbial cell culture platform with in situ PEGDA hydrogel barriers for differential substrate delivery

Additive manufacturing, or 3D-printing techniques have recently begun to enable simpler, faster, and cheaper production of millifluidic devices at resolutions approaching 100–200 μ m. At this resolution, cell culture devices can be constructed that more accurately replicate natural environments compared with conventional culturing techniques. A number of microfluidics researchers have begun incorporating additive manufacturing into their work, using 3D-printed devices in a wide array of chemical, fluidic, and even some biological applications. In this work, we describe a 3D-printed cell culture platform and demonstrate its use in culturing Pseudomonas putida KT2440 bacteria for 44 h under a differential substrate gradient. Polyethylene glycol diacrylate (PEGDA) hydrogel barriers are patterned in situ within a 3D-printed channel. Transport of the toluidine blue tracer dye through the hydrogel barriers is characterized. Nutrients and oxygen were delivered to cells in the culture region by diffusion through the PEGDA hydrogel barriers from adjacent media or saline perfusion channels. Expression of green fluorescent protein by P. putida KT2440 enabled real time visualization of cell density within the 3D-printed channel, and demonstrated cells were actively expressing protein over the course of the experiment. Cells were observed clustering near hydrogel barrier boundaries where fresh substrate and oxygen were being delivered via diffusive transport, but cells were unable to penetrate the barrier. The device described here provides a versatile and easy to implement platform for cell culture in readily controlled gradient microenvironments. By adjusting device geometry and hydrogel properties, this platform could be further customized for a wide variety of biological applications.

42 ENGINEERING↗