Search NASA⌕ Search

SEARCH · Search NASA

Results for “Lifting map”

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 19 records

Lift & Learn: Physics-informed machine learning for large-scale nonlinear dynamical systems

In this work, we present Lift & Learn, a physics-informed method for learning low-dimensional models for large-scale dynamical systems. The method exploits knowledge of a system’s governing equations to identify a coordinate transformation in which the system dynamics have quadratic structure. This transformation is called a lifting map because it often adds auxiliary variables to the system state. The lifting map is applied to data obtained by evaluating a model for the original nonlinear system. This lifted data is projected onto its leading principal components, and low-dimensional linear and quadratic matrix operators are fit to the lifted reduced data using a least-squares operator inference procedure. Analysis of our method shows that the Lift & Learn models are able to capture the system physics in the lifted coordinates at least as accurately as traditional intrusive model reduction approaches. This preservation of system physics makes the Lift & Learn models robust to changes in inputs. Numerical experiments on the FitzHugh–Nagumo neuron activation model and the compressible Euler equations demonstrate the generalizability of our model.

97 MATHEMATICS AND COMPUTING↗

Learning the boundary-to-domain mapping using Lifting Product Fourier Neural Operators for partial differential equations

Neural operators such as the Fourier Neural Operator (FNO) have been shown to provide resolution-independent deep learning models that can learn mappings between function spaces. For example, an initial condition can be mapped to the solution of a partial differential equation (PDE) at a future time-step using a neural operator. Despite the popularity of neural operators, their use to predict solution functions over a domain given only data over the boundary (such as a spatially varying Dirichlet boundary condition) remains unexplored. In this paper, we refer to such problems as boundary-to-domain problems; they have a wide range of applications in areas such as fluid mechanics, solid mechanics, heat transfer etc. We present a novel FNO-based architecture, named Lifting Product FNO (or LP-FNO) which can map arbitrary boundary functions defined on the lower-dimensional boundary to a solution in the entire domain. Specifically, two FNOs defined on the lower-dimensional boundary are lifted into the higher dimensional domain using our proposed lifting product layer. We demonstrate the efficacy and resolution independence of the proposed LP-FNO for the 2D Poisson equation.

Kashi, Aditya↗

Operator pencil passing through a given operator

Let Δ be a linear differential operator acting on the space of densities of a given weight λ{sub 0} on a manifold M. One can consider a pencil of operators Π-circumflex(Δ)=(Δ{sub λ}) passing through the operator Δ such that any Δ{sub λ} is a linear differential operator acting on densities of weight λ. This pencil can be identified with a linear differential operator Δ-circumflex acting on the algebra of densities of all weights. The existence of an invariant scalar product in the algebra of densities implies a natural decomposition of operators, i.e., pencils of self-adjoint and anti-self-adjoint operators. We study lifting maps that are on one hand equivariant with respect to divergenceless vector fields, and, on the other hand, with values in self-adjoint or anti-self-adjoint operators. In particular, we analyze the relation between these two concepts, and apply it to the study of diff (M)-equivariant liftings. Finally, we briefly consider the case of liftings equivariant with respect to the algebra of projective transformations and describe all regular self-adjoint and anti-self-adjoint liftings. Our constructions can be considered as a generalisation of equivariant quantisation.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

High critical currents in heavily doped (Gd,Y)Ba 2 Cu 3 O x superconductor tapes

REBa 2 Cu 3 O x superconductor tapes with moderate levels of dopants have been optimized for high critical current density in low magnetic fields at 77 K, but they do not exhibit exemplary performance in conditions of interest for practical applications, i.e., temperatures less than 50 K and fields of 2–30 T. Heavy doping of REBCO tapes has been avoided by researchers thus far due to deterioration in properties. Here, we report achievement of critical current densities (J c ) above 20 MA/cm 2 at 30 K, 3 T in heavily doped (25 mol. % Zr-added) (Gd,Y)Ba 2 Cu 3 O x superconductor tapes, which is more than three times higher than the J c typically obtained in moderately doped tapes. Pinning force levels above 1000 GN/m 3 have also been attained at 20 K. A composition map of lift factor in J c (ratio of J c at 30 K, 3 T to the J c at 77 K, 0 T) has been developed which reveals the optimum film composition to obtain lift factors above six, which is thrice the typical value. A highly c-axis aligned BaZrO 3 (BZO) nanocolumn defect density of nearly 7 × 10 11 cm –2 as well as 2–3nm sized particles rich in Cu and Zr have been found in the high J c films.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Chemical Interaction of Palladium in Uranium Oxycarbide Nuclear Fuel Kernel

Chemical Interaction of Palladium in Uranium Oxycarbide Nuclear Fuel Kernel Jana Howard1,3, Guang Yang3, Haiyan Zhao1*, Patrick Warren4, Tiankai Yao3, Steven Cavazos4 Elizabeth Sooby Wood4*, Ching-heng Shiau2 1 University of Idaho, Environmental Science, Idaho Falls, Idaho, USA 2 Boise State University, Microscopy and Characterization Suite, Idaho Falls, Idaho, USA 3 Idaho National Laboratory, Idaho Falls, Idaho, USA 4 University of Texas San Antonio, Department of Physics and Astronomy, San Antonio, Texas, USA *Corresponding author: haiyanz@uidaho.edu; elizabeth.soobywood@utsa.edu Tri-structural Isotropic (TRISO) particle fuel is the preferred choice for the newest and next-generation high-temperature nuclear reactors due to its robust construction [1]. The fuel design effectively contains fission products inside the particle at extremely high temperatures [2]. However, the effect of transition metal fission products on fuel performance and structural integrity remains largely unknown, creating uncertainties in predicting fuel performance and potential failure mechanisms [2]. For example, palladium (Pd) has a strong tendency to form intermetallic phases with uranium (U). These new phases tend to be hard and brittle which could lead to fracture of fuel kernel during irradiation [3,4]. Pd is not normally produced in high concentrations in normal fission reactions however, driving the production of Pd into the locality of a Uranium Oxycarbide (UCO) nuclear fuel provides an opportunity to closely observe the diffusion and chemical interactions. This study focuses on detailed transmission electron microscopy characterization of the intermetallic phases formed by the interaction between a UCO kernel and a Pd bar after annealing at 1100 °C for 100 hours. Figure 1 provides an overview of the UCO kernel in contact with the Pd bar. The UCO-Pd sample surface was examined using a Focused Ion Beam (FIB) Quanta 3D FEG FEI in VCD detector mode at 10kV, 83pA to reveal surface features. Several key surface features in the interaction region were observed including: (1) a small area with dendritic microstructure, (2) color contrast near the UCO kernel and Pd boundary, and (3) darker and lighter regions throughout the sample surface. To analyze diffusion behavior, FIB was used to prepare lamellae from five different areas of the UCO-Pd sample, as well as one for the as-received sample. These lamellae were examined using a Scanning Transmission Electron Microscope ThermoFisher Spectra 300 (STEM-Spectra). Energy Dispersive Spectroscopy results confirmed the diffusion between the UCO fuel and Pd bar. Figure 2 shows the identified phases of UC, UO2, and Pd dispersed throughout the interaction region. It was also found that the Pd concentration decreases as the radial distance from the UCO-Pd interface increases. The Pd concentration was 36.93% atomic fraction 11µm from the interaction region then decreased to 7.51% atomic fraction at 257 µm away from the interaction region. Figure 3 shows how concentrations of U and Pd vary across the interaction zone, highlighting the extent of diffusion. This study confirms that Pd interacts with surrounding material to form U-Pd phase and diffusion zones. These diffusion zones vary in composition depending on its radial distance from the point of contact with the Pd bar. These findings contribute to a deeper understanding of the fission product behavior in high temperature nuclear fuels, aiding in the prediction and mitigation of potential fuel degradation mechanisms. A B C Fig. 1. SEM micrographs of UCO fuel kernel in contact with solid Pd bar and the lift out locations. Figure 1A shows UCO fuel kernel in contact with solid Pd bar. Figure 1B shows a higher magnification image of the UCO fuel-Pd interaction zone. Figure 1C shows lift out sites for the lamellae. A B C Fig 2. EDS maps reveal the distribution of U, O, and Pd in location 3. Figure 2A shows the distribution of uranium. Figure 2B shows the distribution of oxygen. Figure 2C shows the distribution of palladium. A B C Fig 3. Palladium and uranium concentration across interaction zone in location 2. Figure 3A shows EDS map of lift out number 2. Figure 3B shows a zoomed in EDS map taken from the interaction zone. Figure 3C shows a line graph of uranium and palladium concentrations across the interaction zone. References: 1. B.E. Wells, N.R. Phillips, K.J. Geelhood. Pacific West Laboratory. (2021). TRISO Fuel: Properties and Failure Modes. https://www.nrc.gov/docs/ML2117/ML21175A152.pdf (Accessed January 16, 2025). 2. American Nuclear Society. TRISO Fuel Development Progresses in INL, ORNL. https://www.gen-4.org/gif/upload/docs/application/pdf/2014-03/nov13nn_fuel_reprint.pdf (Accessed January 16, 2025) 3. Clark R.A., M.A. Con

36 - MATERIALS SCIENCE↗

Chemical Interaction of Palladium in Uranium Oxycarbide Nuclear Fuel Kernel

Chemical Interaction of Palladium in Uranium Oxycarbide Nuclear Fuel Kernel Jana Howard1,3, Guang Yang3, Haiyan Zhao1*, Patrick Warren4, Tiankai Yao3, Steven Cavazos4 Elizabeth Sooby Wood4*, Ching-heng Shiau2 1 University of Idaho, Environmental Science, Idaho Falls, Idaho, USA 2 Boise State University, Microscopy and Characterization Suite, Idaho Falls, Idaho, USA 3 Idaho National Laboratory, Idaho Falls, Idaho, USA 4 University of Texas San Antonio, Department of Physics and Astronomy, San Antonio, Texas, USA *Corresponding author: haiyanz@uidaho.edu; elizabeth.soobywood@utsa.edu Tri-structural Isotropic (TRISO) particle fuel is the preferred choice for the newest and next-generation high-temperature nuclear reactors due to its robust construction [1]. The fuel design effectively contains fission products inside the particle at extremely high temperatures [2]. However, the effect of transition metal fission products on fuel performance and structural integrity remains largely unknown, creating uncertainties in predicting fuel performance and potential failure mechanisms [2]. For example, palladium (Pd) has a strong tendency to form intermetallic phases with uranium (U). These new phases tend to be hard and brittle which could lead to fracture of fuel kernel during irradiation [3,4]. Pd is not normally produced in high concentrations in normal fission reactions however, driving the production of Pd into the locality of a Uranium Oxycarbide (UCO) nuclear fuel provides an opportunity to closely observe the diffusion and chemical interactions. This study focuses on detailed transmission electron microscopy characterization of the intermetallic phases formed by the interaction between a UCO kernel and a Pd bar after annealing at 1100 °C for 100 hours. Figure 1 provides an overview of the UCO kernel in contact with the Pd bar. The UCO-Pd sample surface was examined using a Focused Ion Beam (FIB) Quanta 3D FEG FEI in VCD detector mode at 10kV, 83pA to reveal surface features. Several key surface features in the interaction region were observed including: (1) a small area with dendritic microstructure, (2) color contrast near the UCO kernel and Pd boundary, and (3) darker and lighter regions throughout the sample surface. To analyze diffusion behavior, FIB was used to prepare lamellae from five different areas of the UCO-Pd sample, as well as one for the as-received sample. These lamellae were examined using a Scanning Transmission Electron Microscope ThermoFisher Spectra 300 (STEM-Spectra). Energy Dispersive Spectroscopy results confirmed the diffusion between the UCO fuel and Pd bar. Figure 2 shows the identified phases of UC, UO2, and Pd dispersed throughout the interaction region. It was also found that the Pd concentration decreases as the radial distance from the UCO-Pd interface increases. The Pd concentration was 36.93% atomic fraction 11µm from the interaction region then decreased to 7.51% atomic fraction at 257 µm away from the interaction region. Figure 3 shows how concentrations of U and Pd vary across the interaction zone, highlighting the extent of diffusion. This study confirms that Pd interacts with surrounding material to form U-Pd phase and diffusion zones. These diffusion zones vary in composition depending on its radial distance from the point of contact with the Pd bar. These findings contribute to a deeper understanding of the fission product behavior in high temperature nuclear fuels, aiding in the prediction and mitigation of potential fuel degradation mechanisms. A B C Fig. 1. SEM micrographs of UCO fuel kernel in contact with solid Pd bar and the lift out locations. Figure 1A shows UCO fuel kernel in contact with solid Pd bar. Figure 1B shows a higher magnification image of the UCO fuel-Pd interaction zone. Figure 1C shows lift out sites for the lamellae. A B C Fig 2. EDS maps reveal the distribution of U, O, and Pd in location 3. Figure 2A shows the distribution of uranium. Figure 2B shows the distribution of oxygen. Figure 2C shows the distribution of palladium. A B C Fig 3. Palladium and uranium concentration across interaction zone in location 2. Figure 3A shows EDS map of lift out number 2. Figure 3B shows a zoomed in EDS map taken from the interaction zone. Figure 3C shows a line graph of uranium and palladium concentrations across the interaction zone. References: 1. B.E. Wells, N.R. Phillips, K.J. Geelhood. Pacific West Laboratory. (2021). TRISO Fuel: Properties and Failure Modes. https://www.nrc.gov/docs/ML2117/ML21175A152.pdf (Accessed January 16, 2025). 2. American Nuclear Society. TRISO Fuel Development Progresses in INL, ORNL. https://www.gen-4.org/gif/upload/docs/application/pdf/2014-03/nov13nn_fuel_reprint.pdf (Accessed January 16, 2025) 3. Clark R.A., M.A. Con

36 - MATERIALS SCIENCE↗

Enabling probabilistic learning on manifolds through double diffusion maps

Here, we present a generative learning framework for probabilistic sampling that extends Probabilistic Learning on Manifolds (PLoM), which is designed to generate statistically consistent realizations of a random vector in a finite-dimensional Euclidean space, informed by a (representative) set of observations. In its original form, PLoM constructs a reduced-order probabilistic model by combining three main components: (a) kernel density estimation to approximate the underlying probability measure, (b) Diffusion Maps to characterize the manifold of the data, and (c) a reduced-order Itô Stochastic Differential Equation (ISDE) to sample from the learned distribution. However, its sampling dynamics are posed in the ambient space and the retained number of reduced coordinates is chosen by projection-reconstruction error. In practice, this often (i) requires more coordinates than the data’s intrinsic dimension to achieve stable sampling and (ii) lacks a smooth, basis-independent lifting back to the data domain; moreover, standard Diffusion Maps emphasize harmonic eigenfunctions and can miss non-harmonic latent structure. We address these limitations by decoupling geometry learning from sampling: a first Diffusion Maps pass identifies non-harmonic coordinates on which we formulate a full-order ISDE directly in the latent space, while Double Diffusion Maps captures multiscale geometric features and Geometric Harmonics (GH) learns a smooth lifting map to the ambient variables that is independent of the particular diffusion basis. This hybrid design preserves the system’s dynamical richness with a compact geometric representation and enables principled out-of-sample inference. The effectiveness and robustness of the proposed method are illustrated through two numerical studies: one based on data generated from two-dimensional Hermite polynomial functions and another based on high-fidelity simulations of a detonation wave in a reactive flow.

Double diffusion maps↗

Multi-level optimization with the koopman operator for data-driven, domain-aware, and dynamic system security

Cyber-Physical Systems (CPSs) like the power grid are critically important but also increasingly vulnerable; ensuring reliable system operation in the face of disruptions is becoming more and more challenging. Multi-Level Optimization (MLO) is a powerful way to model adversarial interactions, which naturally makes it applicable to studying CPS security. However, MLO typically does not address underlying system dynamics, and incorporating nonlinear dynamics is generally infeasible. In this paper, we show how to combine MLO with the Koopman Operator (KO) to remedy this. The KO maps nonlinear dynamics to a lifted space in which those dynamics are linear, thus making it ideal for use with MLO. Moreover, the structure of the KO also provides convenient ways to incorporate domain knowledge into the data-driven process of learning the KO representation of a given system. Here we then demonstrate the use of MLO-KO on a small example problem taken from the power grid domain, discuss the scalability and computational cost of MLO-KO, and identify future research directions for this work.

42 ENGINEERING↗

Flow control around a circular cylinder using pulsed dielectric barrier discharge surface plasma

Dielectric barrier discharge (DBD) plasma actuators have been used to control the flow around a circular cylinder at Re=15 000, where the near-wake structure was studied using time-resolved particle image velocimetry with simultaneous measurements of the dynamic lift and drag forces. It was shown that the vortex shedding was suppressed when the surface plasma placed near the natural separation point was activated in a pulsed mode at nondimensional frequency, f{sub p}{sup +}, above 0.6 with a force coefficient, C{sub p}, greater than 0.05%. Plasma actuator performance on flow control was summarized by mapping the changes in drag and lift fluctuations as a function of the forcing frequency and the force coefficient. They showed that more than 70% reduction in lift fluctuations was obtained with up to 32% drag reduction at f{sub p}{sup +}=2.0 and C{sub p}=0.32%. Here, narrowing of the wake was observed as the plasma promoted shear-layer roll-ups at the forcing frequency. This, however, did not affect the shear layer on the opposite side of the wake. At nondimensional forcing frequencies less than 0.6, the vortex shedding locked onto a multiple of the plasma frequency to amplify the wake oscillations. This caused more than 85% increase in lift fluctuations with 8% drag increase at f{sub p}{sup +}=0.2 and C{sub p}=0.01%. The efficiency of flow control using DBD plasma was found to be 1%-2% for drag reduction while around 6% for drag increase.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

LIFTING THE DUSTY VEIL WITH NEAR- AND MID-INFRARED PHOTOMETRY. III. TWO-DIMENSIONAL EXTINCTION MAPS OF THE GALACTIC MIDPLANE USING THE RAYLEIGH-JEANS COLOR EXCESS METHOD

We provide new, high-resolution A(K{sub s} ) extinction maps of the heavily reddened Galactic midplane based on the Rayleigh-Jeans Color Excess ({sup R}JCE{sup )} method. RJCE determines star-by-star reddening based on a combination of near- and mid-infrared photometry. The new RJCE-generated maps have 2' Multiplication-Sign 2' pixels and span some of the most severely extinguished regions of the Galaxy-those covered with Spitzer/IRAC imaging by the GLIMPSE-I, -II, -3D, and Vela-Carina surveys, from 256 Degree-Sign < l < 65 Degree-Sign and, in general, for |b| {<=} 1 Degree-Sign -1.{sup 0}5 (extending up to |b| {<=} 4 Degree-Sign in the bulge). Using RJCE extinction measurements, we generate dereddened color-magnitude diagrams and, in turn, create maps based on main sequence, red clump, and red giant star tracers, each probing different distances and thereby providing coarse three-dimensional information on the relative placement of dust cloud structures. The maps generated from red giant stars, which reach to {approx}18-20 kpc, probe beyond most of the Milky Way extinction in most directions and provide close to a 'total Galactic extinction' map-at minimum they provide high angular resolution maps of lower limits on A(K{sub s} ). Because these maps are generated directly from measurements of reddening by the very dust being mapped, rather than inferred on the basis of some less direct means, they are likely the most accurate to date for charting in detail the highly patchy differential extinction in the Galactic midplane. We provide downloadable FITS files and an IDL tool for retrieving extinction values for any line of sight within our mapped regions.

79 ASTRONOMY AND ASTROPHYSICS↗

Evidence of lithium mobility under neutron irradiation

Understanding the evolution of intermetallic materials in high-radiation environments is of great importance for fusion science and national security. Tritium ( 3 H) and lithium ( 6 Li, 7 Li) transport within neutron irradiated claddings coated with aluminide (FeAl3) has been investigated using state-of-the-art multimodal imaging. Specifically, scanning electron microscopy – focused ion beam (SEM-FIB) was used to prepare irradiated coating lift-out samples for follow-on microanalysis. Here, scanning transmission electron microscopy (STEM) was used to acquire atomic-scale information on the carbonaceous structure and elemental mapping. Atomic force microscopy (AFM) was used to determine lift-out dimensions nondestructively. Time-of-flight secondary ion mass spectrometry (ToF-SIMS) spectral and depth profiling show unexpected lithium isotopic distributions in the irradiated cladding, raising an important evidence of possible light isotope mobility within the cladding under elevated temperature and irradiation. ToF-SIMS three-dimensional chemical mappings of mid- and bottom-cladding coatings show light isotopic (e.g., 3 H, 6 Li, 7 Li) distributions in the irradiated coating and give new insights into the fundamental mechanism related to transport mechanisms within the cladding. Multimodal imaging is power to link microstructure, chemistry, and nanoscale defects that impact reliability of these materials. Furthermore, chemical mapping offers observations of the microstructural evolution due to irradiation and provides insights into unexpected material transport under extreme conditions.

36 MATERIALS SCIENCE↗

AI‐Driven Robot Enables Synthesis‐Property Relation Prediction for Metal Halide Perovskites in Humid Atmosphere

Materials Acceleration Platforms (MAPs) – also known as self-driving laboratories– present a new paradigm for materials science and promise an order of magnitude accelerated materials discovery compared to the traditional trial-and-error approach. Metal halide perovskites (MHPs) are an emerging class of materials for optoelectronic applications but are plagued by irreproducible optoelectronic quality, particularly for films fabricated in a humid atmosphere. Here, in this work, a machine learning (ML)-guided closed-loop platform is developed with a multimodal data fusion approach to predict synthesis–property relations for the optical quality of MHP thin films in relative humidities (RHs) ranging from 5–55%. The efficiency of this approach is confirmed by the fast-dropping learning rate to 2% after experimentally sampling less than 1% of the possible 5,000+ combinations. The prediction of synthesis–property relations is done by optical and imaging characterizations. In situ photoluminescence characterization revealed the origin of thin film quality variation at different RH. These insights provide an avenue for controlling the MHP crystallization by fine-tuning the synthesis parameters and RH for a given chemistry, thus lifting the need for stringent atmosphere control. The MAP enables an accelerated screening and understanding of the synthesis design space, facilitating rational synthesis recipe choice for a wide range of materials.

AI-driven robot↗

Nano-scale orientation mapping of graphite in cast irons

A diametrical section of a graphite spheroid from a ductile iron sample was prepared using the focused ion beam-lift out technique. Characterization of this section was carried out through automated crystal orientation mapping in a transmission electron microscope. This new technique automatically collects electron diffraction patterns and matches them with precalculated templates. The results of this investigation are crystal orientation and phase maps of the specimen, which bring new light to the understanding of growth mechanisms of this peculiar graphite morphology. This article shows that mapping the orientation of carbon-based materials such as graphite, which is difficult to achieve with conventional techniques, can be performed automatically and at high spatial resolution using automated crystal orientation mapping in a transmission electron microscope. - Highlights: • ACOM/TEM can be used to study the crystal orientation of carbon-based materials. • A spheroid is formed by conical sectors radiating from a central nuclei. • Misorientations exist within the conical sectors, defining various orientation domains.

36 MATERIALS SCIENCE↗

Spin-orbit enabled quantum transport channels in a two-hole double quantum dot

We analyze experimentally and theoretically the transport spectra of a gated lateral GaAs double quantum dot containing two holes. The strong spin-orbit interaction present in the hole subband lifts the Pauli spin blockade and allows to map out the complete spectra of the two-hole system. By performing measurements in both source-drain voltage directions, at different detunings and magnetic fields, we carry out quantitative fitting to a Hubbard two-site model accounting for the tunnel coupling to the leads and the spin-flip relaxation process. We extract the singlet-triplet gap and the magnetic field corresponding to the singlet-triplet transition in the double-hole ground state. Additionally, at the singlet-triplet transition we find a resonant enhancement (in the blockaded direction) and suppression of current (in the conduction direction). The current enhancement stems from the multiple resonance of two-hole levels, opening several conduction channels at once. The current suppression arises from the quantum interference of spin-conserving and spin-flipping tunneling processes.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Generative learning of densities on manifolds

A generative modeling framework is proposed that combines diffusion models and manifold learning to efficiently sample data densities on manifolds. The approach utilizes Diffusion Maps to uncover possible low-dimensional underlying (latent) spaces in the high-dimensional data (ambient) space. Two approaches for sampling from the latent data density are described. The first is a score-based diffusion model, which is trained to map a standard normal distribution to the latent data distribution using a neural network. The second one involves solving an Itô stochastic differential equation in the latent space. Additional realizations of the data are generated by lifting the samples back to the ambient space using Double Diffusion Maps , a recently introduced technique typically employed in studying dynamical system reduction; here the focus lies in sampling densities rather than system dynamics. The proposed approaches enable sampling high dimensional data densities restricted to low-dimensional, a priori unknown manifolds. The efficacy of the proposed framework is demonstrated through a benchmark problem and a material with multiscale structure.

Double diffusion maps↗

Double Diffusion Maps and their Latent Harmonics for scientific computations in latent space

In this work, we introduce a data-driven approach to building reduced dynamical models through manifold learning; the reduced latent space is discovered using Diffusion Maps (a manifold learning technique) on time series data. A second round of Diffusion Maps on those latent coordinates allows the approximation of the reduced dynamical models. This second round enables mapping the latent space coordinates back to the full ambient space (what is called lifting); it also enables the approximation of full state functions of interest in terms of the reduced coordinates. In our work, we develop and test three different reduced numerical simulation methodologies, either through pre-tabulation in the latent space and integration on the fly or by going back and forth between the ambient space and the latent space. The data-driven latent space simulation results, based on the three different approaches, are validated through (a) the latent space observation of the full simulation through the Nyström Extension formula, or through (b) lifting the reduced trajectory back to the full ambient space, via Latent Harmonics. Latent space modeling often involves additional regularization to favor certain properties of the space over others, and the mapping back to the ambient space is then constructed mostly independently from these properties; here, we use the same data-driven approach to construct the latent space and then map back to the ambient space.

97 MATHEMATICS AND COMPUTING↗

Learning Optimal Aerodynamic Designs

This project created a framework for efficient, accurate, and scalable deep neural network representations of design optimization problem solutions. The inputs to these DNN representations are the vector of design requirement parameters, the outputs are the optimal design variables, and the goal is to learn the map from inputs to outputs (i.e., inverse design). The team addressed the problem of the optimal shape design of aerodynamic lifting surfaces—in particular aircraft wings—using a Reynolds-Average Navier Stokes model to govern the CFD-based aerodynamic shape optimization. The inverse design map for such problems is very complex and high-dimensional, involving inputs and outputs on the order of 1000s. To approximate this inverse design map, the team developed algorithms to construct parsimonious DNN architectures, which automatically identify low-dimensional manifolds in which design requirements affect optimal shape parameters, and trained these architectures with multifidelity optimization methods. The resulting methodology accurately and automatically designs optimal aerodynamic lifting surfaces with very high accuracy (99%) at interactive speeds, of the order of milliseconds, resulting in factors of one million or more speedup relative to CFD-based design optimization.

97 MATHEMATICS AND COMPUTING↗

Data-Driven Modeling and Correction of Vehicle Dynamics

We develop a data-driven framework for learning and correcting nonautonomous vehicle dynamics. Physics-based vehicle models are often simplified for tractability and therefore exhibit inherent model-form uncertainty, motivating the need for data-driven correction. Moreover, nonautonomous dynamics are governed by time-dependent control inputs, which pose challenges in learning predictive models directly from temporal snapshot data. To address these, we reformulate the vehicle dynamics via a local parameterization of the time-dependent inputs, yielding a modified system composed ofa sequence of local parametric dynamical systems. Here, we approximate these parametric systems using two complementary approaches. First, we employ the dimension reduction and interpolation in parameter space (DRIPS) methodology to construct efficient linear surrogate models, equipped with lifted observable spaces and manifold-based operator interpolation. This enables data-efficient learning of vehicle models whose dynamics admit accurate linear representations in the lifted spaces. Second, for more strongly nonlinear systems, we employ flow map learning (FML), a deep neural network (DNN) approach that approximates the parametric evolution map without requiring special treatment of nonlinearities. We further extend FML with a transfer-learning-based model correction procedure, enabling the correction of misspecified prior models using only a sparse set of high-fidelity or experimental measurements, without assuming a prescribed form for the correction term. Through a suite of numerical experiments on unicycle, simplified bicycle, and slip-based bicycle models, we demonstrate that DRIPS offers robust and highly data-efficient learning of nonautonomous vehicle dynamics, while FML provides expressive nonlinear modeling and effective correction of model-form errors under severe data scarcity.

data-driven modeling↗