Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “scientific deep learning”

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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

PDBx/mmCIF Ecosystem: Foundational Semantic Tools for Structural Biology

PDBx/mmCIF, Protein Data Bank Exchange (PDBx) macromolecular Crystallographic Information Framework (mmCIF), has become the data standard for structural biology. With its early roots in the domain of small-molecule crystallography, PDBx/mmCIF provides an extensible data representation that is used for deposition, archiving, remediation, and public dissemination of experimentally determined three-dimensional (3D) structures of biological macromolecules by the Worldwide Protein Data Bank (wwPDB, wwpdb.org). Extensions of PDBx/mmCIF are similarly used for computed structure models by ModelArchive (modelarchive.org), integrative/hybrid structures by PDB-Dev (pdb-dev.wwpdb.org), small angle scattering data by Small Angle Scattering Biological Data Bank SASBDB (sasbdb.org), and for models computed generated with the AlphaFold 2.0 deep learning software suite (alphafold.ebi.ac.uk). Community-driven development of PDBx/mmCIF spans three decades, involving contributions from researchers, software and methods developers in structural sciences, data repository providers, scientific publishers, and professional societies. Having a semantically rich and extensible data framework for representing a wide range of structural biology experimental and computational results, combined with expertly curated 3D biostructure data sets in public repositories, accelerates the pace of scientific discovery. Herein, we describe the architecture of the PDBx/mmCIF data standard, tools used to maintain representations of the data standard, governance, and processes by which data content standards are extended, plus community tools/software libraries available for processing and checking the integrity of PDBx/mmCIF data. Use cases exemplify how the members of the Worldwide Protein Data Bank have used PDBx/mmCIF as the foundation for its pipeline for delivering Findable, Accessible, Interoperable, and Reusable (FAIR) data to many millions of users worldwide.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2025

A resolution independent neural operator

The Deep operator network (DeepONet) is a powerful yet simple neural operator architecture that utilizes two deep neural networks to learn mappings between infinite-dimensional function spaces. This architecture is highly flexible, allowing the evaluation of the solution field at any location within the desired domain. However, it imposes a strict constraint on the input space, requiring all input functions to be discretized at the same locations; this limits its practical applications. Here, in this work, we introduce a general framework for operator learning from input–output data with arbitrary number and locations of sensors. This begins by introducing a resolution-independent DeepONet (RI-DeepONet), enabling it to handle input functions that are arbitrarily, but sufficiently finely, discretized. To this end, we propose two dictionary learning algorithms to adaptively learn a set of appropriate continuous basis functions, parameterized as implicit neural representations (INRs), from correlated signals defined on arbitrary point cloud data. These basis functions are then used to project arbitrary input function data as a point cloud onto an embedding space (i.e., a vector space of finite dimensions) with dimensionality equal to the dictionary size, which can be directly used by DeepONet without any architectural changes. In particular, we utilize sinusoidal representation networks (SIRENs) as trainable INR basis functions. The introduced dictionary learning algorithms are then used in a similar way to learn an appropriate dictionary of basis functions for the output function data, which defines a new neural operator architecture referred to as the R esolution I ndependent N eural O perator (RINO). In the RINO, the operator learning task simplifies to learning a mapping from the coefficients of input basis functions to the coefficients of output basis functions. We demonstrate the robustness and applicability of RINO in handling arbitrarily (but sufficiently richly) sampled input and output functions during both training and inference through several numerical examples.

Deep operator network (DeepONet)↗
DOE OSTI2025

Machine learning-driven predictive resource management in complex science workflows

Here, the collaborative efforts of large communities in science experiments, often comprising thousands of global members, reflect a monumental commitment to exploration and discovery. Recently, advanced and complex data processing has gained increasing importance in science experiments. Data processing workflows typically consist of multiple intricate steps, and the precise specification of resource requirements is crucial for each step to allocate optimal resources for effective processing. Estimating resource requirements in advance is challenging due to a wide range of analysis scenarios, varying skill levels among community members, and the continuously increasing spectrum of computing options. One practical approach to mitigate these challenges involves initially processing a subset of each step to measure precise resource utilization from actual processing profiles before completing the entire step. While this two-staged approach enables processing on optimal resources for most of the workflow, it has drawbacks such as initial inaccuracies leading to potential failures and suboptimal resource usage, along with overhead from waiting for initial processing completion, which is critical for fast-turnaround analyses. In this context, our study introduces a novel pipeline of machine learning models within a comprehensive workflow management system, the Production and Distributed Analysis (PanDA) system. These models employ advanced machine learning techniques to predict key resource requirements, overcoming challenges posed by limited upfront knowledge of characteristics at each step. Accurate forecasts of resource requirements enable informed and proactive decision-making in workflow management, enhancing the efficiency of handling diverse, complex workflows across heterogeneous resources.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Efficient Training of Deep Neural Operator Networks via Randomized Sampling

Neural operators (NOs) employ deep neural networks to learn the mappings between infinitedimensional function spaces. Deep operator network (DeepONet), a popular NO architecture, has demonstrated success in the real-time prediction of complex dynamics across various scientific and engineering applications. In this work, we introduce a random sampling technique to be adopted during the training of DeepONet, aimed at improving the generalization ability of the model, while significantly reducing the computational time. The proposed approach targets the trunk network of the DeepONet model that outputs the basis functions corresponding to the spatiotemporal locations of the bounded domain on which the physical system is defined. While constructing the loss function, DeepONet training traditionally considers a uniform grid of spatiotemporal points at which all the output functions are evaluated for each iteration. This approach leads to a larger batch size, resulting in poor generalization and increased memory demands, due to the limitations of the stochastic gradient descent (SGD) optimizer. The proposed random sampling over the inputs of the trunk net mitigates these challenges, improving generalization and reducing the memory requirements during training, resulting in significant computational gains. We validate our hypothesis through three benchmark examples, demonstrating substantial reductions in training time while achieving comparable or lower overall test errors relative to the traditional training approach. Our results indicate that incorporating randomization in the trunk network inputs during training enhances the efficiency and robustness of DeepONet, offering a promising avenue for improving the framework’s performance in modeling complex physical systems.

Karumuri, Sharmila [Department of Civil & Systems ↗
DOE OSTI2025

Active Learning Surrogates for Integrating Electron Microscopy and Computational Insights from Simulations in Autonomous Experiments

Artificial Intelligence (AI) combined with simulations and experiments has great potential to accelerate scientific discovery across technology and pharmaceuticals. However, the gap between simulations and experiments is challenging due to disparities in time and scale, making it difficult to estimate properties like energy and electronic states from experiments, and to provide feedback based on theoretical insights.Our research addresses the challenge by developing unique deep kernel based surrogate models that learns from microscopic images, mapping structural features to energy differences from defect formation. We start with full-training using simulated images to determine optimal settings, establishing a baseline for active learning. Using these settings from the baseline, active learning is trained, and predicts structures along simulation trajectories based on uncertainty and energetic stability, thus reducing data requirements, simulation time and computational costs. The results demonstrate that the model achieves a low average error margin of approximately 0.03 meV, indicating good performance. To enhance feature extraction and reconstruction capabilities, we developed an autoencoder-decoder as additional surrogate to create latent space to capture essential features, enabling precise comparisons between simulations and experiments. The results from this model achieved a reconstruction loss of around 0.2 and accurately reconstructed molecular structures.Overall, this work advances the steering of experiments through computational simulations by employing a surrogate models that actively predicts the trajectories of structural evolution, achieving time-to-solution comparable to experimental measurements.

Saranathan, Gayathri [Hewlett-Packard]↗
DOE OSTI2025

Leveraging interpolation models and error bounds for verifiable scientific machine learning

Effective verification and validation techniques for modern scientific machine learning workflows are challenging to devise. Statistical methods are abundant and easily deployed, but often rely on speculative assumptions about the data and methods involved. Error bounds for classical interpolation techniques can provide mathematically rigorous estimates of accuracy, but often are difficult or impractical to determine computationally. Here, in this work, we present a best-of-both-worlds approach to verifiable scientific machine learning by demonstrating that (1) multiple standard interpolation techniques have informative error bounds that can be computed or estimated efficiently; (2) comparative performance among distinct interpolants can aid in validation goals; (3) deploying interpolation methods on latent spaces generated by deep learning techniques enables some interpretability for black-box models. We present a detailed case study of our approach for predicting lift-drag ratios from airfoil images. Code developed for this work is available in a public Github repository.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Deep learning approaches for neural decoding across architectures and recording modalities

Decoding behavior, perception or cognitive state directly from neural signals is critical for brain–computer interface research and an important tool for systems neuroscience. In the last decade, deep learning has become the state-of-the-art method in many machine learning tasks ranging from speech recognition to image segmentation. The success of deep networks in other domains has led to a new wave of applications in neuroscience. In this article, we review deep learning approaches to neural decoding. Here, we describe the architectures used for extracting useful features from neural recording modalities ranging from spikes to functional magnetic resonance imaging. Furthermore, we explore how deep learning has been leveraged to predict common outputs including movement, speech and vision, with a focus on how pretrained deep networks can be incorporated as priors for complex decoding targets like acoustic speech or images. Deep learning has been shown to be a useful tool for improving the accuracy and flexibility of neural decoding across a wide range of tasks, and we point out areas for future scientific development.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2025

jaxhps: An elliptic PDE solver built with machine learning in mind

Elliptic partial differential equations (PDEs) can model many physical phenomena, such as electrostatics, acoustics, wave propagation, and diffusion. In scientific machine learning settings, a high-throughput PDE solver may be required to generate a training dataset, run in the inner loop of an iterative algorithm, or interface directly with a deep neural network. To provide value to machine learning users, such a PDE solver must be compatible with standard automatic differentiation frameworks, scale efficiently when run on graphics processing units (GPUs), and maintain high accuracy for a large range of input parameters. We have designed the jaxhps package with these use-cases in mind by implementing a highly efficient and accurate solver for elliptic problems with native hardware acceleration and automatic differentiation support.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

The Importance of Being Adaptable: An Exploration of the Power and Limitations of Domain Adaptation for Simulation-Based Inference with Galaxy Clusters

The application of deep machine learning methods in astronomy has exploded in the last decade, with new models showing remarkably improved performance on benchmark tasks. Not nearly enough attention is given to understanding the models' robustness, especially when the test data are systematically different from the training data, or "out of domain." Domain shift poses a significant challenge for simulation-based inference, where models are trained on simulated data but applied to real observational data. In this paper, we explore domain shift and test domain adaptation methods for a specific scientific case: simulation-based inference for estimating galaxy cluster masses from X-ray profiles. We build datasets to mimic simulation-based inference: a training set from the Magneticum simulation, a scatter-augmented training set to capture uncertainties in scaling relations, and a test set derived from the IllustrisTNG simulation. We demonstrate that the Test Set is out of domain in subtle ways that would be difficult to detect without careful analysis. We apply three deep learning methods: a standard neural network (NN), a neural network trained on the scatter-augmented input catalogs, and a Deep Reconstruction-Regression Network (DRRN), a semi-supervised deep model engineered to address domain shift. Although the NN improves results by 17% in the Training Data, it performs 40% worse on the out-of-domain Test Set. Surprisingly, the Scatter-Augmented Neural Network (SANN) performs similarly. While the DRRN is successful in mapping the training and Test Data onto the same latent space, it consistently underperforms compared to a straightforward Yx scaling relation. These results serve as a warning that simulation-based inference must be handled with extreme care, as subtle differences between training simulations and observational data can lead to unforeseen biases creeping into the results.

Ntampaka, Michelle [Baltimore, Space Telescope Sci↗
DOE OSTI2023

Decoding structure-spectrum relationships with physically organized latent spaces

Here, a semisupervised machine learning method for the discovery of structure-spectrum relationships is developed and then demonstrated using the specific example of interpreting x-ray absorption near-edge structure (XANES) spectra. This method constructs a one-to-one mapping between individual structure descriptors and spectral trends. Specifically, an adversarial autoencoder is augmented with a rank constraint (RankAAE). The RankAAE methodology produces a continuous and interpretable latent space, where each dimension can track an individual structure descriptor. As a part of this process, the model provides a robust and quantitative measure of the structure-spectrum relationship by decoupling intertwined spectral contributions from multiple structural characteristics. This makes it ideal for spectral interpretation and the discovery of descriptors. The capability of this procedure is showcased by considering five local structure descriptors and a database of >50 000 simulated XANES spectra across eight first-row transition metal oxide families. The resulting structure-spectrum relationships not only reproduce known trends in the literature but also reveal unintuitive ones that are visually indiscernible in large datasets. The results suggest that the RankAAE methodology has great potential to assist researchers in interpreting complex scientific data, testing physical hypotheses, and revealing patterns that extend scientific insight.

36 MATERIALS SCIENCE↗
NASA NTRSDate not supplied

Harnessing Artificial Intelligence for Medical Diagnosis and Treatment During Space Exploration Missions

From May 8th to June 9th, 2023, I had the opportunity to participate in an experiential learning experience at Johnson Space Center in Houston, TX with Exploration Medical Capability (ExMC), an element of the NASA Human Research Program. During this research experience, I was not only able to work on the above titled research project, but also gain an immense exposure to the field of aerospace medicine, make numerous connections within the field, tour NASA facilities, as well as travel to the Aerospace Medical Association Annual Conference (AsMA) in New Orleans. To briefly introduce my project, it is well understood that the medical capabilities available to crew medical officers (CMOs) on the International Space Station will be different than the capabilities available and needed during deep space exploration missions to the Moon, Mars, and beyond. Ground support is particularly limited due to distance, communication delays (or lack of communication), and lack of resupply. Therefore, to support medical care by CMOs on these missions, robust clinical decision support systems (CDSSs) must be designed. The recent publication and public launch of generative artificial intelligence (AI) tools based upon large language models (LLM) such as ChatGPT provides the opportunity to create a smart assistant for onboard triage, diagnosis, and treatment of medical conditions. Ultimately, the overall purpose of the project was to research what AI tools currently exist or are in development, and to see how they might be implemented onboard during exploration class spaceflights of the future. The ExMC element is actively developing several tools to be used in preparation for and during deep space exploration missions. One of those tools, known as IMPACT, is a probabilistic risk assessment model which can be used to propose a desired medical system (based on mass and volume) and suggest the clinical outcomes likely to occur for a design reference mission (DRM). The group recently presented the IMPACT model and a DRM of interest titled “Modified Long Duration Lunar Orbital and Lunar Surface” (mLDLOLS) at the recent AsMA conference. The mLDLOLS mock mission is a 9 month and 6-day deep space exploration mission consisting of time in Moon’s orbit (3 months on the Gateway space station), on the lunar surface (3 months within habitat), and another 3 months on Gateway before return to Earth. For this DRM, IMPACT ultimately outlined a preferred medical system that was then associated with medical conditions considered to be most likely based on frequency, most likely to cause astronaut task time loss (TTL), most likely to cause return to definitive care (RTDC), and most likely cause loss of crew life (LOCL). IMPACT also highlighted the medical capabilities/skills that would be required to care for those medical conditions, such as performing a history of present illness or musculoskeletal exam with ultrasound. The primary objective of the project was to perform a survey of the AI tools and systems applicable to the conditions outlined for the proposed mLDLOLS mission. Using PubMed (including most relevant MeSH terms) and Google Scholar, we then created a robust annotated bibliography organized by condition. The 56-page and over 500 reference annotated bibliography was subsequently used to create a review outline that would become the basis for drafting of a future publication. For the review outline, we took those medical conditions researched within the annotated bibliography (condition-based approach) and deployed a systems-based approach, combining those medical conditions and related tools into ten categories. These categories included general/all-purpose CDSSs, tools to diagnose or manage respiratory, dermatologic, neurologic, auditory and vestibular, ophthalmic, musculoskeletal, infection-associated, and gynecologic conditions, as well as tools that could be deployed in the setting of trauma/emergency. With the completion of the 30-page outline, we then began drafting the review paper. To conclude the research experience, I presented the findings from our survey to the ExMC Clinical and Science team. With these objectives, I ultimately learned about the number of AI tools that exist today to assist medical professionals with the triage, diagnosis, and management of several medical conditions. These tools can span from chatbot assistants to help triage knee pain to vision transformer models that can identify ophthalmic conditions based on ocular surface images captured with a cell phone. We also highlighted the current gaps that exist in the literature alongside the advancements that are needed to make the desired CDSS for deep space exploration missions. With this experience, I certainly confirmed an existing career goal and identified several additional skills needed to become an aerospace medical doctor including knowledge of critical care in an extreme medicine setting, aerospace engineering and human integration systems, artificial intelligence, machine learning, and risk models. I also identified numerous transferable skills for this career goal including the basic knowledge of medicine (MD), deployment of the scientific method for critical thought about new scientific questions (PhD), review of published literature, including creating an annotated bibliography (PhD), as well as detailed scientific writing (PhD). The results of my research will likely guide the design of an all-encompassing onboard medical assistant for use during deep space exploration missions of the future. I plan on sharing the outcomes from this experience with my peers at a student seminar in the Fall semester on August 30th. During the seminar, I will detail the project, my experience at NASA and AsMA, as well as offer best practice guidelines for students entertaining similar experiences or careers. In conclusion, I would like to thank the WVU School of Medicine, Research and Graduate Education office, as well as NASA ExMC for the unwavering support of this life-changing experience.

Ryan A. Lacinski↗
NASA NTRS2005

Learning from the Mars Rover Mission: Scientific Discovery, Learning and Memory

Purpose: Knowledge management for space exploration is part of a multi-generational effort. Each mission builds on knowledge from prior missions, and learning is the first step in knowledge production. This paper uses the Mars Exploration Rover mission as a site to explore this process. Approach: Observational study and analysis of the work of the MER science and engineering team during rover operations, to investigate how learning occurs, how it is recorded, and how these representations might be made available for subsequent missions. Findings: Learning occurred in many areas: planning science strategy, using instrumen?s within the constraints of the martian environment, the Deep Space Network, and the mission requirements; using software tools effectively; and running two teams on Mars time for three months. This learning is preserved in many ways. Primarily it resides in individual s memories. It is also encoded in stories, procedures, programming sequences, published reports, and lessons learned databases. Research implications: Shows the earliest stages of knowledge creation in a scientific mission, and demonstrates that knowledge management must begin with an understanding of knowledge creation. Practical implications: Shows that studying learning and knowledge creation suggests proactive ways to capture and use knowledge across multiple missions and generations. Value: This paper provides a unique analysis of the learning process of a scientific space mission, relevant for knowledge management researchers and designers, as well as demonstrating in detail how new learning occurs in a learning organization.

Linde, Charlotte↗
DOE OSTI2021

Train Like a (Var)Pro: Efficient Training of Neural Networks with Variable Projection

Deep neural networks (DNNs) have achieved state-of-the-art performance across a variety of traditional machine learning tasks, e.g., speech recognition, image classification, and segmentation. The ability of DNNs to efficiently approximate high-dimensional functions has also motivated their use in scientific applications, e.g., to solve partial differential equations and to generate surrogate models. In this paper, we consider the supervised training of DNNs, which arises in many of the above applications. We focus on the central problem of optimizing the weights of the given DNN such that it accurately approximates the relation between observed input and target data. Devising effective solvers for this optimization problem is notoriously challenging due to the large number of weights, nonconvexity, data sparsity, and nontrivial choice of hyperparameters. To solve the optimization problem more efficiently, we propose the use of variable projection (VarPro), a method originally designed for separable nonlinear least-squares problems. Our main contribution is the Gauss--Newton VarPro method (GNvpro) that extends the reach of the VarPro idea to nonquadratic objective functions, most notably cross-entropy loss functions arising in classification. These extensions make GNvpro applicable to all training problems that involve a DNN whose last layer is an affine mapping, which is common in many state-of-the-art architectures. In our four numerical experiments from surrogate modeling, segmentation, and classification, GNvpro solves the optimization problem more efficiently than commonly used stochastic gradient descent (SGD) schemes. Finally, GNvpro finds solutions that generalize well, and in all but one example better than well-tuned SGD methods, to unseen data points.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

The role of international collaboration in the United States geologic disposal research program

Abstract. More than a decade ago, the United States disposal program discontinued all research activities focused on the unsaturated fractured tuff formation at Yucca Mountain as the geologic disposal site for spent fuel and high-level radioactive waste. A new research and development (R&D) program was initiated to provide a sound technical basis for alternative disposal options across clay, crystalline, and salt rocks. The goals of this broad program were (and still are) to (1) increase confidence in the robustness of generic disposal concepts, (2) develop the science and engineering tools needed to support disposal concept implementation, and (3) conduct R&D on the direct disposal of existing dual-purpose (storage and transportation) canisters. Recognizing the benefits of international collaboration toward the common goal of safely and efficiently managing the back end of the nuclear fuel cycle, the program emphasized international cooperation as an effective strategy for sharing information and knowledge. In a multi-laboratory effort coordinated by Lawrence Berkeley National Laboratory, the United States Department of Energy (DOE) program established formal and informal cooperation partnerships with several international initiatives and institutions and developed a number of collaborative R&D activities in important research areas, such as engineered barrier integrity, near-field perturbations, radionuclide transport, performance assessment, and methods for characterization and monitoring of engineered and natural barriers. This presentation gives an overview of these R&D activities, with a specific focus on activities that improve our current understanding of the coupled thermal–hydrological–mechanical–chemical (THMC) processes occurring in engineered and natural barriers. We start with a brief review of selected international collaboration initiatives and then describe a few specific collaboration projects. We focus specifically on such studies that use experimental data sets provided by international research cooperation for joint modeling work to increase confidence in performance-relevant predictions of coupled processes. Overall, the focus on international collaboration has allowed deep engagement of US researchers with the international waste management R&D community in terms of best practices, new scientific advances, state-of-the-art simulation tools, new monitoring and performance confirmation approaches, and lessons learned. The joint R&D with international researchers, worldwide sharing of knowledge and experience, and access to relevant data and experiments from a variety of host rocks have helped our researchers to significantly improve their understanding of the current technical basis for disposal in a range of potential host rock environments. International collaboration also provides ample opportunity for training and educating junior staff that are well suited to move the United States disposal research program forward into the next decades, a promising avenue for developing a next-generation workforce of disposal scientists.

Birkholzer, Jens T.↗
DOE OSTI2024

Learning dynamical systems from data: An introduction to physics-guided deep learning

Modeling complex physical dynamics is a fundamental task in science and engineering. Traditional physics-based models are first-principled, explainable, and sample-efficient. However, they often rely on strong modeling assumptions and expensive numerical integration, requiring significant computational resources and domain expertise. While deep learning (DL) provides efficient alternatives for modeling complex dynamics, they require a large amount of labeled training data. Furthermore, its predictions may disobey the governing physical laws and are difficult to interpret. Physics-guided DL aims to integrate first-principled physical knowledge into data-driven methods. It has the best of both worlds and is well equipped to better solve scientific problems. Recently, this field has gained great progress and has drawn considerable interest across discipline Here, we introduce the framework of physics-guided DL with a special emphasis on learning dynamical systems. We describe the learning pipeline and categorize state-of-the-art methods under this framework. We also offer our perspectives on the open challenges and emerging opportunities.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Make the Fastest Faster: Importance Mask Synthesis for Interactive Volume Visualization using Reconstruction Neural Networks

Visualizing a large-scale volumetric dataset with high resolution is challenging due to the substantial computational time and space complexity. Recent deep learning-based image inpainting methods significantly improve rendering latency by reconstructing a high-resolution image for visualization in constant time on GPU from a partially rendered image where only a portion of pixels go through the expensive rendering pipeline. However, existing solutions need to render every pixel of either a predefined regular sampling pattern or an irregular sample pattern predicted from a low-resolution image rendering. Both methods require a significant amount of expensive pixel-level rendering. In this work, we provide Importance Mask Learning (IML) and Synthesis (IMS) networks, which are the first attempts to directly synthesize important regions of the regular sampling pattern from the user’s view parameters, to further minimize the number of pixels to render by jointly considering the dataset, user behavior, and the downstream reconstruction neural network. Our solution is a unified framework to handle various types of inpainting methods through the proposed differentiable compaction/decompaction layers. Experiments show our method can further improve the overall rendering latency of state-of-the-art volume visualization methods using reconstruction neural network for free when rendering scientific volumetric datasets. Our method can also directly optimize the off-the-shelf pre-trained reconstruction neural networks without elongated retraining.

Large-scale data↗
DOE OSTI2025

SAGIPS: a physics-inspired scalable asynchronous generative inverse-problem solver

Abstract Solving large-scale inverse problems using deep-learning algorithms have become an essential part of modern research and industrial applications. The complexity of the underlying inverse problem may require the utilization of high performance computing systems which poses a challenge on the algorithmic design of the inverse problem solver. Most deep learning algorithms require, due to their design, custom parallelization techniques in order to be resource efficient while showing a reasonable convergence. In this paper we introduce a S calable A synchronous G enerative I nverse P roblem S olver (SAGIPS) on high-performance computing systems. We present a workflow that utilizes an asynchronous ring-allreduce algorithm to transfer the gradients of the generator network across multiple GPUs. Experiments with a scientific proxy application demonstrate that SAGIPS shows near linear weak scaling, together with a convergence quality that is comparable to traditional methods. The approach presented here allows leveraging Generative Adverserial Network across multiple GPUs, promising advancements in solving complex inverse problems at scale.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Latent Twins

Over the past decade, scientific machine learning has transformed the development of mathematical and computational frameworks for analyzing, modeling, and predicting complex systems. From inverse problems to numerical partial differential equations (PDEs), dynamical systems, and model reduction, these advances have pushed the boundaries of what can be simulated. Yet they have often progressed in parallel, with representation learning and algorithmic solution methods evolving largely as separate pipelines. With Latent Twins, we propose a unifying mathematical framework that creates a hidden surrogate in latent space for the underlying equations. Whereas digital twins mirror physical systems in the digital world, Latent Twins mirror mathematical systems in a learned latent space governed by operators. Through this lens, classical modeling, inversion, model reduction, and operator approximation all emerge as special cases of a single principle. We establish the fundamental approximation properties of Latent Twins for both ordinary differential equations (ODEs) and PDEs and demonstrate the framework across three representative settings: (i) canonical ODEs, capturing diverse dynamical regimes; (ii) a PDE benchmark using the shallow-water equations, contrasting Latent Twin simulations with deep operator network and forecasts with a four-dimensional variational method baseline; and (iii) a challenging real-data geopotential reanalysis dataset, reconstructing and forecasting from sparse, noisy observations. Latent Twins provide a compact, interpretable surrogate for solution operators that evaluate across arbitrary time gaps in a single-shot, while remaining compatible with scientific pipelines such as assimilation, control, and uncertainty quantification. Looking forward, this framework offers scalable, theory-grounded surrogates that bridge data-driven representation learning and classical scientific modeling across disciplines.

Latent Twins↗