SEARCH · Engineering Papers

Results for “Vector Processing”

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 1,531 records · Page 85

All sources NASA NTRS DOE OSTI

DOE OSTI2024

Simultaneously improving accuracy and computational cost under parametric constraints in materials property prediction tasks

Abstract Modern data mining techniques using machine learning (ML) and deep learning (DL) algorithms have been shown to excel in the regression-based task of materials property prediction using various materials representations. In an attempt to improve the predictive performance of the deep neural network model, researchers have tried to add more layers as well as develop new architectural components to create sophisticated and deep neural network models that can aid in the training process and improve the predictive ability of the final model. However, usually, these modifications require a lot of computational resources, thereby further increasing the already large model training time, which is often not feasible, thereby limiting usage for most researchers. In this paper, we study and propose a deep neural network framework for regression-based problems comprising of fully connected layers that can work with any numerical vector-based materials representations as model input. We present a novel deep regression neural network, iBRNet, with branched skip connections and multiple schedulers, which can reduce the number of parameters used to construct the model, improve the accuracy, and decrease the training time of the predictive model. We perform the model training using composition-based numerical vectors representing the elemental fractions of the respective materials and compare their performance against other traditional ML and several known DL architectures. Using multiple datasets with varying data sizes for training and testing, We show that the proposed iBRNet models outperform the state-of-the-art ML and DL models for all data sizes. We also show that the branched structure and usage of multiple schedulers lead to fewer parameters and faster model training time with better convergence than other neural networks. Scientific contribution: The combination of multiple callback functions in deep neural networks minimizes training time and maximizes accuracy in a controlled computational environment with parametric constraints for the task of materials property prediction.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2025

Amplitude Analysis of ωπ0 Photoproduction at GlueX

spectrum of light mesons produced from a linearly polarized photon beam. The production and decays of a light meson resonance X such as γp → Xp′ →ωπ0p′ can be modeled with polarized vector-pseudoscalar ampli- tudes, which can describe the contribution of individual amplitudes to the total measured intensity. The status of mass-independent fits to the ωπ0 mass spectrum over a wide range of mass and momentum transfer−twill be presented, with an emphasis on interactions with the b1(1235) meson. We will also present in parallel an analysis of moments of the angular dis- tributions for the same process, as a means of verifying the stability of the amplitude-based results. These results will help to improve the broader knowledge of light meson states and how they are produced.

Scheuer, Kevin [College of William and Mary, Willi↗

NASA NTRS1987

Earthquake source mechanisms and transform fault tectonics in the Gulf of California

The source parameters of 19 large earthquakes in the Gulf of California were determined from inversions of long-period P and SH waveforms. The goal was to understand the recent slip history of this dominantly transform boundary between the Pacific and North American plates as well as the effect on earthquake characteristics of the transition from young oceanic to continental lithosphere. For the better recorded transform events, the fault strike is resolved to + or - 4 deg at 90 percent confidence. The slip vectors thus provide important constraints on the direction of relative plate motion. Most centroid depths are poorly resolved because of tradeoffs between depth and source time function. On the basis of waveform modeling, historical seismicity, and other factors, it is appropriate to divide the Gulf into three distinct zones. The difference in seismic character among the three zones is likely the result of differing levels of maturity of the processes of rifting, generation of oceanic crust, and formation of stable oceanic transform faults. The mechanism of an earthquake on the Tres Marias Escarpment is characterized by thrust faulting and likely indicates the direction of relative motion between the Rivera and North American plates. This mechanism requires revision in plate velocity models which predict strike slip motion at this location.

Goff, John A.↗

NASA NTRS1991

Instability of multi-layer fluid configurations in the presence of time-dependent accelerations in a microgravity environment

The increasing number of research opportunities in a microgravity environment will benefit not only fundamental studies in fluid dynamics, but also technological applications such as those involving materials processing. In particular, fluid configurations which involve fluid-fluid interfaces would occur in a variety of experimental investigations. This work investigates the stability of a configuration involving fluid-fluid interfaces in the presence of a time-dependent forcing. Both periodic (g-jitter) and nonperiodic accelerations are considered. The fluid configuration is multilayered, and infinite in extent. The analysis is linear and inviscid, and the acceleration vector is oriented perpendicular to each interface. A Floquet analysis is employed in the case of the periodic forcing. In the problem of nonperiodic forcing, the resulting system of equations are integrated in time. Specific nondimensional parameters appear in each problem. The configuration behavior is investigated for a range of parameter values.

Lyell, M. J.↗

NASA NTRS2004

Parameter Estimation in Atmospheric Data Sets

In this study the structure tensor technique is used to estimate dynamical parameters in atmospheric data sets. The structure tensor is a common tool for estimating motion in image sequences. This technique can be extended to estimate other dynamical parameters such as diffusion constants or exponential decay rates. A general mathematical framework was developed for the direct estimation of the physical parameters that govern the underlying processes from image sequences. This estimation technique can be adapted to the specific physical problem under investigation, so it can be used in a variety of applications in trace gas, aerosol, and cloud remote sensing. As a test scenario this technique will be applied to modeled dust data. In this case vertically integrated dust concentrations were used to derive wind information. Those results can be compared to the wind vector fields which served as input to the model. Based on this analysis, a method to compute atmospheric data parameter fields will be presented. .

Wenig, Mark↗

DOE OSTI2024

Searching for Strongly-Interacting Dark Matter with the Heavy Photon Search Experiment

The Heavy Photon Search Experiment (HPS) is a fixed-target experiment at Jefferson Lab’s Hall B, designed to explore a hidden sector (HS) of particles containing dark matter and a new force mediator known as the “heavy photon” (A'). The A' is a massive spin-1 gauge boson associated with a new U (1)D symmetry in the HS that kinetically mixes with the Standard Model () photon with a weak coupling strength parameterized by ¿, with ¿2 ~ 10-2 -10-10. HPS utilizes a high-intensity electron beam on a thin tungsten target to produce heavy photons in the MeV-GeV mass range via “dark bremsstrahlung,” a process analogous to SM bremsstrahlung but suppressed by ¿2. The A' can decay resonantly to SM leptons, allowing HPS to conduct both mass resonance searches for prompt decays (large ¿) and displaced vertex searches for long-lived particles (small ¿). In addition to the minimal A' model, HPS probes more complex extensions such as the QCD-like strongly-interacting massive particles (SIMPs) HS containing “dark” pions (pD) and vector mesons (VD), with pD as dark matter candidates. These particles introduce new thermal dark matter freeze- out scenarios and visible signals through long-lived VD decays to SM leptons, which are accessible to HPS. This analysis conducted a displaced vertex search for VD ¿ e-e+ in the mass range 30 MeV to 124 MeV and ¿ between 10-6 < ¿ < 10-2 using data from the 2016 Engineering Run (10.753 nb-1) at 2.3 GeV. Unlike the minimal A' search, SIMP signal kinematics required new approaches to signal normalization and SM background rejection. The strongest signal evidence was a local p-value of 0.01317 for mVD = 119 MeV, corresponding to a global significance of 0.9s. Although no signal was found, this search excluded a region of the SIMP parameter space at 90 % confidence. This work demonstrates HPS’s competitive capability to probe SIMP sectors within cosmologically significant parameters and introduces a new method for HPS displaced vertex searches using track vertical impact parameter cuts

Spellman, Alic [Univ. of California, Santa Cruz, C↗

DOE OSTI2018

OpenMP in VASP: Threading and SIMD

Abstract The Vienna Ab initio Simulation Package (VASP) is a widely used electronic structure code that originally exploits process‐level parallelism through the Message Passing Interface (MPI) for work distribution within and across nodes. Architectural changes of modern parallel processors urge programmers to address thread‐ and data‐level parallelism as well to benefit most from the available compute resources within a node. We describe for VASP how to approach for an MPI + OpenMP parallelization including data‐level parallelism through OpenMP SIMD constructs together with a generic high‐level vector coding scheme. We can demonstrate an improved scalability of VASP and more than 20% gain over the MPI‐only version as well as a 2× increased performance of collective operations using the multiple‐endpoint MPI feature. The high‐level vector coding scheme applied to VASP's general gradient approximation routine gives up 9× performance gain on AVX512 platforms with the Intel compiler.

Wende, Florian↗

DOE OSTI2026

Search for Associated Production of a Higgs Boson and Two Vector Bosons via Vector-Boson Scattering at s = 13 TeV

A search for Higgs boson ( H ) production in association with two vector bosons ( V = W , Z ) via vector-boson scattering (VBS) is presented using proton-proton collision data collected at s = 13 TeV by the CMS experiment, corresponding to an integrated luminosity of 138 fb - 1 . Events containing two forward jets consistent with VBS, a large-radius jet from the decay of a boosted H to a pair of b quarks, and zero, one, or two charged leptons coming from V decays are selected. The process is excluded at 95% confidence level for observed (expected) values of the V V H H coupling modifier κ 2 V outside the interval 0.40 < κ 2 V < 1.60 ( 0.34 < κ 2 V < 1.66 ), assuming standard model values for all other couplings, thus establishing a novel probe of the V V H H interaction. Constraints are also set on the individual κ 2 W and κ 2 Z coupling modifiers and on the allowed region in the κ 2 W - κ 2 Z plane.

Hayrapetyan, Aram [Yerevan Phys. Inst.]↗

NASA NTRS1979

Lunar nearside magnetic anomalies

Contour maps of the vector components of the lunar crustal magnetic field measured with the Apollo 16 subsatellite magnetometer at low altitudes across the central near side are presented. A solution was obtained for a minimum mean magnetization level of about 700 times the stable magnetization component of the most magnetic returned lunar crustal samples. Attention is given to the Reiner Gamma anomaly that is found to represent macroscopic evidence for a surficial darkening process that is dominantly solar wind-dependent.

Hood, L. L.↗

NASA NTRS1989

Gravity and animal embryos

Out of more than 4,500 rat hours in space there was only one experimental attempt (Cosmos 1129) at mating with an apparent absence of fertilization, implantation and subsequent development to term and partuition. Portions of this process were successfully flown, however, including the major portion of organogenesis in the rat (Cosmos 1524). These observations show that the cellular and molecular events underlying morphogenesis and differentiation in a small mammal can proceed normally in-utero under microgravity and other conditions encountered during short-duration flight. However, it is not known whether this situation will hold for larger mammals over several generations during extended missions that venture outside of near Earth. Furthermore, it is not understood why the previous attempt at obtaining copulation, fertilization and implantation in orbit failed but may be related to limitations of the rat habitat for meeting the preconditions for reproductive behavior. With respect to mammalian development it is important to appreciate that fertilization and development occur internally within the female and take a long time to complete and their success will, therefore, be contingent upon the maternal response to the space environment. One process central to development (the establishment of cell lines) is initiated prior to implantation by environmental asymmetries preceived by progenitor cells. These asymmetries appear to result from the formation of asymmetric cell-cell contacts and the concommitant development of an electrical axis across the progenitor cells. Other asymmetries were also documented. It is not known whether any of the known asymmetries perceived by progenitor cells are influenced by gravity vectors and/or by the maternal response to microgravity and other conditions encountered in space.

Wiley, Lynn M.↗

NASA NTRS2013

Space Weather Impacts to Conjunction Assessment: A NASA Robotic Orbital Safety Perspective

National Aeronautics and Space Administration (NASA) recognizes the risk of on-orbit collisions from other satellites and debris objects and has instituted a process to identify and react to close approaches. The charter of the NASA Robotic Conjunction Assessment Risk Analysis (CARA) task is to protect NASA robotic (unmanned) assets from threats posed by other space objects. Monitoring for potential collisions requires formulating close-approach predictions a week or more in the future to determine analyze, and respond to orbital conjunction events of interest. These predictions require propagation of the latest state vector and covariance assuming a predicted atmospheric density and ballistic coefficient. Any differences between the predicted drag used for propagation and the actual drag experienced by the space objects can potentially affect the conjunction event. Therefore, the space environment itself, in particular how space weather impacts atmospheric drag, is an essential element to understand in order effectively to assess the risk of conjunction events. The focus of this research is to develop a better understanding of the impact of space weather on conjunction assessment activities: both accurately determining the current risk and assessing how that risk may change under dynamic space weather conditions. We are engaged in a data-- ]mining exercise to corroborate whether or not observed changes in a conjunction event's dynamics appear consistent with space weather changes and are interested in developing a framework to respond appropriately to uncertainty in predicted space weather. In particular, we use historical conjunction event data products to search for dynamical effects on satellite orbits from changing atmospheric drag. Increased drag is expected to lower the satellite specific energy and will result in the satellite's being 'later' than expected, which can affect satellite conjunctions in a number of ways depending on the two satellites' orbits and the geometry of the conjunction. These satellite time offsets can form the basis of a new technique under development to determine whether space weather perturbations, such as coronal mass ejections, are likely to increase, decrease, or have a neutral effect on the collision risk due to a particular close approach.

Ghrist, Richard↗

NASA NTRS2015

Rao-Blackwellization for Adaptive Gaussian Sum Nonlinear Model Propagation

When dealing with imperfect data and general models of dynamic systems, the best estimate is always sought in the presence of uncertainty or unknown parameters. In many cases, as the first attempt, the Extended Kalman filter (EKF) provides sufficient solutions to handling issues arising from nonlinear and non-Gaussian estimation problems. But these issues may lead unacceptable performance and even divergence. In order to accurately capture the nonlinearities of most real-world dynamic systems, advanced filtering methods have been created to reduce filter divergence while enhancing performance. Approaches, such as Gaussian sum filtering, grid based Bayesian methods and particle filters are well-known examples of advanced methods used to represent and recursively reproduce an approximation to the state probability density function (pdf). Some of these filtering methods were conceptually developed years before their widespread uses were realized. Advanced nonlinear filtering methods currently benefit from the computing advancements in computational speeds, memory, and parallel processing. Grid based methods, multiple-model approaches and Gaussian sum filtering are numerical solutions that take advantage of different state coordinates or multiple-model methods that reduced the amount of approximations used. Choosing an efficient grid is very difficult for multi-dimensional state spaces, and oftentimes expensive computations must be done at each point. For the original Gaussian sum filter, a weighted sum of Gaussian density functions approximates the pdf but suffers at the update step for the individual component weight selections. In order to improve upon the original Gaussian sum filter, Ref. [2] introduces a weight update approach at the filter propagation stage instead of the measurement update stage. This weight update is performed by minimizing the integral square difference between the true forecast pdf and its Gaussian sum approximation. By adaptively updating each component weight during the nonlinear propagation stage an approximation of the true pdf can be successfully reconstructed. Particle filtering (PF) methods have gained popularity recently for solving nonlinear estimation problems due to their straightforward approach and the processing capabilities mentioned above. The basic concept behind PF is to represent any pdf as a set of random samples. As the number of samples increases, they will theoretically converge to the exact, equivalent representation of the desired pdf. When the estimated qth moment is needed, the samples are used for its construction allowing further analysis of the pdf characteristics. However, filter performance deteriorates as the dimension of the state vector increases. To overcome this problem Ref. [5] applies a marginalization technique for PF methods, decreasing complexity of the system to one linear and another nonlinear state estimation problem. The marginalization theory was originally developed by Rao and Blackwell independently. According to Ref. [6] it improves any given estimator under every convex loss function. The improvement comes from calculating a conditional expected value, often involving integrating out a supportive statistic. In other words, Rao-Blackwellization allows for smaller but separate computations to be carried out while reaching the main objective of the estimator. In the case of improving an estimator's variance, any supporting statistic can be removed and its variance determined. Next, any other information that dependents on the supporting statistic is found along with its respective variance. A new approach is developed here by utilizing the strengths of the adaptive Gaussian sum propagation in Ref. [2] and a marginalization approach used for PF methods found in Ref. [7]. In the following sections a modified filtering approach is presented based on a special state-space model within nonlinear systems to reduce the dimensionality of the optimization problem in Ref. [2]. First, the adaptive Gaussian sum propagation is explained and then the new marginalized adaptive Gaussian sum propagation is derived. Finally, an example simulation is presented.

state estimation↗

DOE OSTI2018

Learning Memento archive routing with Character-based Artificial Neural Networks

This white paper describes a series of tests that were performed to determine if a neural network could learn patterns from a service that maintains a cache of routing decisions for the discovery of version information for discrete web-at-large URLs. Labeled training data was derived from a log file that records the by-archive availability of Memento availability for a given URL. Training data sets were generated from this log file on a by-archive basis (thus making it a binary classification problem). Each training data set consisted of equal numbers of hit and miss URLs, selected at random. The URLs were converted to a normalized numeric representation where each integer in a URL training vector represents a character in that URL. The corresponding label indicates whether or not Mementos were available for that URL in the selected archive. The training data matrix and the label vector became input to a neural network. A number of neural network architectures and network hyperparameters were explored, however the log entries themselves were used as-is, without any feature engineering, beyond the aforementioned normalization.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Self-Supervised Cloud Classification

Abstract Low-level marine clouds play a pivotal role in Earth’s weather and climate through their interactions with radiation, heat and moisture transport, and the hydrological cycle. These interactions depend on a range of dynamical and microphysical processes that result in a broad diversity of cloud types and spatial structures, and a comprehensive understanding of cloud morphology is critical for continued improvement of our atmospheric modeling and prediction capabilities moving forward. Deep learning has recently accelerated our ability to study clouds using satellite remote sensing, and machine learning classifiers have enabled detailed studies of cloud morphology. A major limitation of deep learning approaches to this problem, however, is the large number of hand-labeled samples that are required for training. This work applies a recently developed self-supervised learning scheme to train a deep convolutional neural network (CNN) to map marine cloud imagery to vector embeddings that capture information about mesoscale cloud morphology and can be used for satellite image classification. The model is evaluated against existing cloud classification datasets and several use cases are demonstrated, including training cloud classifiers with very few labeled samples, interrogation of the CNN’s learned internal feature representations, cross-instrument application, and resilience against sensor calibration drift and changing scene brightness. The self-supervised approach learns meaningful internal representations of cloud structures and achieves comparable classification accuracy to supervised deep learning methods without the expense of creating large hand-annotated training datasets. Significance Statement Marine clouds heavily influence Earth’s weather and climate, and improved understanding of marine clouds is required to improve our atmospheric modeling capabilities and physical understanding of the atmosphere. Recently, deep learning has emerged as a powerful research tool that can be used to identify and study specific marine cloud types in the vast number of images collected by Earth-observing satellites. While powerful, these approaches require hand-labeling of training data, which is prohibitively time intensive. This study evaluates a recently developed self-supervised deep learning method that does not require human-labeled training data for processing images of clouds. We show that the trained algorithm performs competitively with algorithms trained on hand-labeled data for image classification tasks. We also discuss potential downstream uses and demonstrate some exciting features of the approach including application to multiple satellite instruments, resilience against changing image brightness, and its learned internal representations of cloud types. The self-supervised technique removes one of the major hurdles for applying deep learning to very large atmospheric datasets.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2018

Generalized trigonometric Fourier-series method with automatic time step control for non-linear point kinetics equations

The trigonometric Fourier-series method (TFS) is generalized to provide approximate solutions for non-linear point kinetics equations with feedback using varying step sizes. This method can provide a very stable solution against the size of the discrete time step allowing much larger step sizes to be used. Systems of the point kinetics equations are solved using Fourier-series expansion over a partition of the total time interval. The approximate solution requires determining the series coefficients over each time step in that partition. These coefficients are determined using the high-order derivatives of the solution vector at the beginning of the time step introducing a system of linear algebraic equations to be solved at each step. This system is similar to the Vandermonde system. Two successive orders of the partial sums are used to estimate the local truncation error. This error and some other constrains are used to produce the largest step size allowable at each step while keeping the error within a specific tolerance. The process of calculating suitable step sizes should be automatic and inexpensive. Convergence and stability of the proposed method are discussed and a new formula is introduced to maintain stability. The proposed method solves the general linear and non-linear kinetics problems. The method has been applied to five different types of reactivities including step/ramp insertions with temperature feedback. The method is seemed to be valid for larger time intervals than those used in the conventional numerical integration, and is thus useful in reducing computing time. Computational results are found to be consistent with the analysis, they demonstrate that the convergence of the iteration scheme can be accelerated and the resulting computing times can be greatly reduced while maintain computational accuracy.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2019

Tensile properties, strain rate sensitivity, and activation volume of additively manufactured 316L stainless steels [Tensile properties of 316L stainless steels made by laser powder-bed-fusion additive manufacturing]

The tensile properties of additively manufactured (AM) metals and alloys are among the most important variables that impact the potential applications of these materials. Here we examine and report on the tensile properties of AM 316L stainless steels fabricated by the laser powder-bed-fusion (L-PBF) technique, via twelve sets of optimized laser processing parameters that produce materials with density >98.8 ± 0.10%. A heterogeneous microstructure is observed in all L-PBF samples, including microscopic features such as dislocations, cellular walls, elemental segregations, local misorientations, impurities, precipitates, and a large fraction of low-angle grain boundaries (2-10°, ~40–60%). The derived average grain size defined by high-angle grain boundaries (>10°) is ~30–50 μm. Tensile testing reveals a yield strength ranging from 552 to 635 MPa and a tensile-elongation-to-failure (TEF) of 0.09–0.42 for directly-printed samples, whereas these values are 592–690 MPa and 0.29–0.50 for samples machined from the as-built rectangular thin plates. In all samples, we observe a variation of tensile yield strength within ~15% but not the TEF, suggesting marginal microstructural changes despite a wide range of laser processing parameters. The large scatter of TEF in directly-printed samples originates from the sensitivity of thin gauge geometry (~2 mm 2 cross-section area) to the built-in flaws. We measured a substantially higher strain rate sensitivity (m~0.02–0.03) of L-PBF 316L compared to the coarse-grained counterparts (~0.006), together with a small activation volume of ~20–30b3 (where b is the Burgers vector of 316L). In conclusion, these deformation kinetics parameters suggest that the tensile plasticity of L-PBF 316L is controlled by a much finer microstructural length scale than the measured grain size, consistent with the high strength and juxtaposed nano- to macro-structures seen in these materials. Strategies to optimize the tensile properties of AM materials are discussed.

36 MATERIALS SCIENCE↗

DOE OSTI2023

Characterizing the performance of node-aware strategies for irregular point-to-point communication on heterogeneous architectures

Supercomputer architectures are trending toward higher computational throughput due to the inclusion of heterogeneous compute nodes. These multi-GPU nodes increase on-node computational efficiency, while also increasing the amount of data to be communicated and the number of potential data flow paths. In this work, we characterize the performance of irregular point-to-point communication with MPI on heterogeneous compute environments through performance modeling, demonstrating the limitations of standard communication strategies for both device-aware and staging-through-host communication techniques. Presented models suggest staging communicated data through host processes then using node-aware communication strategies for high inter-node message counts. Notably, the models also predict that node-aware communication utilizing all available CPU cores to communicate inter-node data leads to the most performant strategy when communicating with a high number of nodes. Furthermore, model validation is provided via a case study of irregular point-to-point communication patterns in distributed sparse matrix–vector products. Importantly, we include a discussion on the implications model predictions have on communication strategy design for emerging supercomputer architectures.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

A case study in contrastive learning information combination: Application to technical forensics of additive manufacturing filament source identification

Combination of information from disparate data sources into a single decision is a core challenge in many fields, including the field of technical forensics. Technical forensics (TF) utilizes technical characterization of questioned samples to determine properties of that sample; these properties are then used to infer information of forensic interest, such as provenance, age, or attribution. TF is utilized in traditional forensic applications, such as the attribution of material fragments from an explosive, and in nuclear forensic applications, such as the attribution of actinides which have been interdicted out of regulatory control. The challenge of combining information from disparate sources, described alternately by many terms including “Data Fusion” and “Data Integration”, is exacerbated in the technical forensics domain due to at least two factors: the challenge of interpreting each information source singularly, and the relatively small data set sizes available. Extensive literature exists attempting to combine technical forensics information sources, both in manual and automated processes. These attempts are often bespoke to the specific information sources (such as the bi-, tri-, or quad-isotope chart (Moody, Grant, and Hutcheon 2005)), with some emerging examples of simple early- and late- fusion (, respectively). Simultaneous to the information combination efforts described in the previous paragraph, the field of natural language processing attempted (and largely succeeded) in combining information from multiple non-technical information sources. The ecosystem of “multi-modal” language models, which can take text and images as input, and generate text and images as output, became large and diverse by 2025 (Khan et al. 2025). In a generalized sense, many of these methods are trained by learning neural networks which can convert raw text or images into a vector of numbers describing the text or image, hereafter called “embeddings” and the neural networks performing the conversion are called “embedders”. By using a separate embedder for text and images, finding coincident text and images (such as images with their captions), and optimizing the parameters of the embedders such that the embeddings for the text and the image are similar, the field has found a bridge between text and images (Girdhar et al. 2023). It is the contention of the authors of this report that this insight is not limited to text and images but instead can be extended to any modality which can be found coincidently. The subject of the rest of this report is the application of this method to example multi-modal technical forensic data. Some details about the data used in this report are not appropriate for this report, and are included in a companion report (PNNL-38669).

36 MATERIALS SCIENCE↗