SEARCH · Search NASA

Results for “scientific training”

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 73 records · Page 4

All sources NASA NTRS DOE OSTI

DOE OSTI2023

Evaluation of pre-training large language models on leadership-class supercomputers

Large language models (LLMs) have arisen rapidly to the center stage of artificial intelligence as the foundation models applicable to many downstream learning tasks. However, how to effectively build, train, and serve such models for many high-stake and first-principle-based scientific use cases are both of great interests and of great challenges. Moreover, pre-training LLMs with billions or even trillions of parameters can be prohibitively expensive not just for academic institutions, but also for well-funded industrial and government labs. Furthermore, the energy cost and the environmental impact of developing LLMs must be kept in mind. Here, in this work, we conduct a first-of-its-kind performance analysis to understand the time and energy cost of pre-training LLMs on the Department of Energy (DOE)’s leadership-class supercomputers. Employing state-of-the-art distributed training techniques, we evaluate the computational performance of various parallelization approaches at scale for a range of model sizes, and establish a projection model for the cost of full training. Our findings provide baseline results, best practices, and heuristics for pre-training such large models that should be valuable to HPC community at large. We also offer insights and optimization strategies for using the first exascale computing system, Frontier, to train models of the size of GPT-3 and beyond.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Distributed-Memory Sparse Deep Neural Network Inference Using Global Arrays

Partitioned Global Address Space (PGAS) models exhibit tremendous promise in developing efficient and productive distributed-memory parallel applications. They have been used extensively in scientific computations due to conveniently offering a ``shared-memory''-like model and convenient interfaces that separate communication with synchronization. Traditionally, PGAS communication models have been applied to dense/contiguously distributed data, but most modern applications depict varied levels of sparsity. Existing PGAS models require certain adaptations to support distributed sparse computations, since associated computations often require matrix arithmetic, in addition to data movement. The Global Arrays toolkit from Pacific Northwest National Laboratory (PNNL) is one of the earliest PGAS models to combine one-sided data communication and distributed matrix operations and is still used in the popular NWChem quantum chemistry suite. Recently, we have expanded the Global Arrays toolkit to support common sparse operations, like sparse matrix-dense matrix multiplies (SpMM), sparse matrix-sparse matrix multiplication (SpGEMM) and Sampled Dense-Dense Matrix Multiplication (SDDMM). As it turns out, these operations are the bedrock of sparse Deep Learning (DL); sparse deep neural networks and Graph Neural Networks (GNNs) have gained increasing attention recently in achieving speedups on training and inference with reduced memory footprints. Unlike scientific applications in High Performance Computing (HPC), modern (distributed-memory capable) DL toolkits often rely on non-standardized and closed-source vendor software optimizations, creating challenges in software-hardware co-design at scale. Our goal is to support a variety of distributed-memory sparse matrix operations and helper functions in the newly created Sparse Global Arrays (SGA), such that it is possible to build portable and productive Machine Learning scenarios for algorithm/software and hardware codesign purposes. Contemporary data-parallel schemes for training/inference are undergoing a major overhaul since model replication limits scalability and causes resource inefficiencies. As such, we have adopted tensor parallelism in decomposing the model and inputs, to mitigate memory issues. Current implementation is built on top of MPI and uses CPUs to maximize the portability across the platforms.

Distributed computing, machine learning↗

DOE OSTI2021

A pre-training and self-training approach for biomedical named entity recognition

Named entity recognition (NER) is a key component of many scientific literature mining tasks, such as information retrieval, information extraction, and question answering; however, many modern approaches require large amounts of labeled training data in order to be effective. This severely limits the effectiveness of NER models in applications where expert annotations are difficult and expensive to obtain. In this work, we explore the effectiveness of transfer learning and semi-supervised self-training to improve the performance of NER models in biomedical settings with very limited labeled data (250-2000 labeled samples). We first pre-train a BiLSTM-CRF and a BERT model on a very large general biomedical NER corpus such as MedMentions or Semantic Medline, and then we fine-tune the model on a more specific target NER task that has very limited training data; finally, we apply semi-supervised self-training using unlabeled data to further boost model performance. We show that in NER tasks that focus on common biomedical entity types such as those in the Unified Medical Language System (UMLS), combining transfer learning with self-training enables a NER model such as a BiLSTM-CRF or BERT to obtain similar performance with the same model trained on 3x-8x the amount of labeled data. We further show that our approach can also boost performance in a low-resource application where entities types are more rare and not specifically covered in UMLS.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

DOE OSTI2021

Laboratory earthquake forecasting: A machine learning competition

Earthquake prediction, the long-sought holy grail of earthquake science, continues to confound Earth scientists. Could we make advances by crowdsourcing, drawing from the vast knowledge and creativity of the machine learning (ML) community? We used Google’s ML competition platform, Kaggle, to engage the worldwide ML community with a competition to develop and improve data analysis approaches on a forecasting problem that uses laboratory earthquake data. The competitors were tasked with predicting the time remaining before the next earthquake of successive laboratory quake events, based on only a small portion of the laboratory seismic data. The more than 4,500 participating teams created and shared more than 400 computer programs in openly accessible notebooks. Complementing the now well-known features of seismic data that map to fault criticality in the laboratory, the winning teams employed unexpected strategies based on rescaling failure times as a fraction of the seismic cycle and comparing input distribution of training and testing data. In addition to yielding scientific insights into fault processes in the laboratory and their relation with the evolution of the statistical properties of the associated seismic data, the competition serves as a pedagogical tool for teaching ML in geophysics. The approach may provide a model for other competitions in geosciences or other domains of study to help engage the ML community on problems of significance.

58 GEOSCIENCES↗

DOE OSTI2025

Investigating performance and variability of NIF ICF experiments with deep learning

The parameter space involved in designing an inertial confinement fusion shot at the National Ignition Facility (NIF) is massively multi-dimensional and the cost of a single shot makes a comprehensive set of sensitivity studies in the laboratory impractical. The use of machine learning to overcome these challenges has gained popularity and has had several successful applications by the scientific community. We extend on these efforts by training a neural network (NN) on information about the experimental design, engineering elements, and drive asymmetry to predict with uncertainty the neutron yield of an experiment. We find the measured and model predicted values are in good agreement, with an R 2 value of 0.91 for a randomly selected test dataset. Almost all the predicted 95% credible intervals contain the corresponding measured value for both training and test datasets. We identify correlations picked up by the NN between the shot design, yield, and variability and use them to motivate shot sensitivity studies. The first shot to exceed the Lawson-like ignition criteria (N210808) was conducted at the NIF and subsequent shots studied the design’s robustness. In a follow-up shot to N210808, our model predicts capsule quality to be the main performance degradation mechanism that prevented the shot from repeating previous performance levels. Shot N221204 was the first shot to exceed a target energy gain of 1. Our model predicts increased yield with reduced coast time for a N221204 study and greater variability for designs with lower peak powers at constant yield. The model’s fast prediction speed and uncertainty prediction are useful for identifying interesting design paths that could warrant further investigation with conventional simulations to search for robust high yield designs.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2024

Effectiveness of denoising diffusion probabilistic models for fast and high-fidelity whole-event simulation in high-energy heavy-ion experiments

Artificial intelligence (AI) generative models, such as generative adversarial networks (GANs), variational autoencoders, and normalizing flows, have been widely used and studied as efficient alternatives for traditional scientific simulations. However, they have several drawbacks, including training instability and inability to cover the entire data distribution, especially for regions where data are rare. This is particularly challenging for whole-event, full-detector simulations in high-energy heavy-ion experiments, such as sPHENIX at the Relativistic Heavy Ion Collider and Large Hadron Collider experiments, where thousands of particles are produced per event and interact with the detector. This work investigates the effectiveness of denoising diffusion probabilistic models (DDPMs) as an AI-based generative surrogate model for the sPHENIX experiment that includes the heavy-ion event generation and response of the entire calorimeter stack. DDPM performance in sPHENIX simulation data is compared with a popular rival, GANs. Results show that both DDPMs and GANs can reproduce the data distribution where the examples are abundant (low-to-medium calorimeter energies). Nonetheless, DDPMs significantly outperform GANs, especially in high-energy regions where data are rare. Additionally, DDPMs exhibit superior stability compared to GANs. The results are consistent between both central and peripheral centrality heavy-ion collision events. Moreover, DDPMs offer a substantial speedup of approximately a factor of 100 compared to the traditional Geant4 simulation method.

42 ENGINEERING↗

DOE OSTI2026

Multiphysics and Multiscale Simulation Methods for Electromagnetic Energy Assisted Fossil Fuel to Hydrogen Conversion (Final Scientific/Technical Report)

This report summarizes the technical accomplishments of the four-year research project “Multiphysics and Multiscale Simulation Methods for Electromagnetic Energy Assisted Fossil Fuel to Hydrogen Conversion” (Award No. DE-FE0032092), conducted at Howard University and the University of Houston (subawardee) from September 2021 to August 2025. The project successfully achieved all four major objectives: 1. 3D Structural Characterization – Developed 3D optical imaging and mechanical sectioning methods to characterize catalyst distribution and support morphology in nickel foam substrates. Successfully reconstructed 3D geometries and imported them into COMSOL Multiphysics for electromagnetic simulations. 2. EM Hotspot Simulation – Created all-frequency stable electromagnetic formulations and 3D nodal discontinuous Galerkin (NDG) methods for coupled electromagnetic-thermal-fluid problems in multiscale catalytic media. Demonstrated stable solutions from DC to microwave frequencies. 3. Multiphysics Coupling – Developed multiscale simulation methods coupling FEM electromagnetic solvers with thermal transport equations. Reactive molecular dynamics (ReaxFF MD) simulations were performed to investigate catalytic reaction mechanisms at the atomistic level. Demonstrated electromagnetic-thermal co-simulation capabilities for porous catalyst structures. 4. System Optimization – Designed and optimized EM-assisted catalytic systems using nickel foam and carbon foam structures, demonstrating significant temperature increases due to microwave heating. Observed and characterized plasma generation in carbon fiber catalysts. Investigated multiple reaction chamber geometries for improved microwave energy deposition. The project produced significant scientific contributions including 15+ peer-reviewed publications, trained multiple Ph.D. students and undergraduate researchers, and advanced the understanding of microwave-assisted hydrogen production from fossil fuels.

08 HYDROGEN↗

DOE OSTI2024

Towards Precision Photometric Type ia Supernova Cosmology With Machine Learning

The revolutionary discovery of dark energy and accelerating cosmic expansion was made with just 42 type Ia supernovae (SNe Ia) in 1999. Since then, large synoptic surveys, e.g., Dark Energy Survey (DES), have observed thousands more SNe Ia and the upcoming Rubin Legacy Survey of Space and Time (LSST) and Roman Space Telescope promise to deliver millions in the next decade. This unprecedented data volume can produce the required precision to unambiguously test concordance cosmology which could represent a monumental shift in our understanding of dark energy and its role in cosmic history. However, extracting a pure SN Ia sample with accurate redshifts for such a large dataset will be a challenge. Specifically, spectroscopic classification will not be possible for the vast majority of discovered objects, and only ~25% will have spectroscopic redshifts. This thesis presents a series of observational and methodological studies designed to address the questions associated with this new era of photometric SN Ia cosmology. First, we present a machine learning method for photometric classification of SNe, Supernova Classification with a COnvolutional Neural Network (SCONE). Photometric classification enables SNe with no spectroscopic information to be confidently categorized, and is a critical component of current and future analysis pipelines. SCONE achieves >99% accuracy distinguishing simulated SNe Ia from non-Ia SNe, and has been integrated into DES, LSST, and Roman analysis pipelines. We also demonstrate the efficacy of SCONE on early-time photometric classification, which will be vital for optimal allocation of spectroscopic resources. We show that SCONE can distinguish between 6 SN types with 75% accuracy on the night of initial discovery, comparable to results in the literature for full-phase SNe. Next, we study current methods for estimating SN Ia redshifts and propose a machine learning alternative that uses SN photometry alone to extract redshift information. Most SNe Ia inherit redshift information from their host galaxy, but the process of matching SNe to the correct host galaxy can be challenging. We systematically analyze the impact of incorrect redshifts from host galaxy mismatch on 5 years of DES SN data, and conclude that improved host matching or redshift estimation methods can reduce our systematic errors by ~10%. In response to this finding, we present a SN photometry-only method for estimating redshifts independent of host galaxy information, Photo-zSNthesis. We show that Photo-zSNthesis redshift estimates are accurate to within 2% across the full redshift range of LSST, a first in the literature. Finally, we focus on the robustness of machine learning (ML) algorithms for real-world and scientific applications. ML models generalize poorly beyond their training set and often experience severe performance degradation when deployed on new data. We demonstrate a general method for improving robustness that achieves new state-of-the-art results on astronomical object classification, wildlife identification, and tumor detection.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2021

Solving inverse problems in stochastic models using deep neural networks and adversarial training

Inverse problems associated with stochastic models constitute a significant portion of scientific and engineering applications. In such cases the unknown quantities are distributions. The applicability of traditional methods is limited because of their demanding assumptions or prohibitive computational consumption; for example, maximum likelihood methods require closed-form density functions, and Markov Chain Monte Carlo needs a large number of simulations. We propose a new method that estimates the unknown distribution by matching the statistical properties between observed and simulated random processes. We leverage the expressive power of neural networks to approximate the unknown distribution and use a discriminative neural network for computing the statistical discrepancies between the observed and simulated random processes. Here we demonstrated numerically that the proposed methods can estimate both the model parameters and learn complicated unknown distributions.

42 ENGINEERING↗

DOE OSTI2021

NukeLM: Pre-Trained and Fine-Tuned Language Models for the Nuclear and Energy Domains

Natural language processing (NLP) tasks (text classification, named entity recognition, etc.) have seen amazing improvements over the last few years. This is due to models such as BERT that achieve deep knowledge transfer by using a large pre-trained model, then fine-tuning the model on specific tasks. The BERT architecture has shown even better performance on domain-specific tasks when the model is pre-trained using domain-relevant texts. Here, inspired by these recent advancements, we have developed NukeLM, a nuclear-domain BERT model pre-trained on 1.5 million abstracts from the DOE Office of Scientific and Technical Information (OSTI) database. This NukeLM model is then fine-tuned for the classification of research articles into either binary classes (related to the nuclear fuel cycle (NFC) or not) or multiple categories related to the subject of the article. We show that continued pre-training of a BERT-style architecture prior to fine-tuning results in greater performance in both article classification tasks. This information is critical for properly triaging manuscripts, a necessary task for better understanding citation networks that publish in the nuclear space and uncovering new areas of research in the nuclear (or nuclear relevant) domain.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

DOE OSTI2024

Neutrons in Structural Biology: Challenges and Opportunities (Workshop Report)

Gaining a thorough understanding of biological systems requires building our knowledge about biological processes from the level of atoms and electrons, and up to whole organisms. Such comprehensive knowledge will allow for a predictive understanding of complex biological systems behavior. It will guide us in the design and development of novel therapeutics and vaccines to tackle existing health threats and to prepare for future pandemics, and it will provide information necessary to create new biomaterials and bio-inspired technologies through manipulation of biological macromolecules, their assemblies, single cells and even microorganisms. Reaching these goals will require a synergistic combination of multiple experimental techniques with molecular calculations and predictive simulations, and the design and development of new techniques and capabilities that bridge current knowledge and technology gaps. Neutron scattering provides unique information about the biomacromolecular structure and function and can play a major role in achieving these goals. A workshop was held to engage the scientific community in identifying pressing challenges in biochemistry, structural biology, enzymology and structure-guided drug design not solved with the current neutron scattering technologies or utilizing other structural biology techniques such as X-ray crystallography, NMR, and cryo-EM. The workshop brought together structural biology, biochemistry and computational experts, as well as early career researchers and students, creating a forum for discussing scientific advancement and collaboration. The workshop included a one-day satellite training workshop where graduate students and postdoctoral researchers were educated in the application of neutron crystallography and small-angle scattering in structural biology. Furthermore, the Instrument Scientific Advisory Board (ISAB) for the development of a macromolecular neutron diffractometer at ORNL’s Second Target Station was introduced at the workshop. The major outcome was that neutrons can provide atomic-level understanding of biomacromolecular structure, function and dynamics which is of paramount importance for addressing the identified challenges. Neutron crystallography, in particular, can resolve long-standing biochemical issues regarding enzyme function by delineating the underlying chemistry and can have a major impact on the design of small-molecule therapeutics, especially in combination with molecular computation (quantum chemistry and molecular dynamics simulations) and the emerging artificial intelligence (AI)-assisted drug design technologies. The unique properties of neutrons, including their high sensitivity to hydrogen and their non-destructive nature, make them ideal probes of biological matter. There is a palpable need in the scientific community to expand and enhance the impact of neutron sciences on biology. Neutron crystallography is the only structural biology method capable of determining positions of all hydrogen atoms in proteins, nucleic acids and their complexes at near-physiological temperatures and of unstable species at cryogenic temperatures. Moreover, neutron analysis is non-ionizing, non-destructive and does not perturb the structure or redox chemistry of active site metal centers and clusters in proteins, which can be invaluable for studying radiation-sensitive metalloprotein complexes. Further, neutron energies used in scattering applications are similar to atomic motions, permitting neutron spectroscopies to characterize the dynamics of biomacromolecules on the picosecond to microsecond timescales. The different sensitivities of neutrons to protium (H) and deuterium (D) isotopes of hydrogen allow enhanced visibility of specific parts of biological complexes through isotopic labeling. The impact of neutrons will be most powerful when neutron scattering is combined with complementary experimental techniques that use photons and electrons, and with high-performance computing. The interconnection and mutuality of the experimental and theoretical capabilities will drive discoveries in biological and health sciences to generate more complete picture of complex biological systems. The major limitation in the field of biological neutron crystallography has been signal-to-noise, demanding large samples that are difficult to produce for the majority of biomacromolecules and limiting the applicability of this technique in biological sciences. A neutron crystallography instrument at the Second Target Station will revolutionize biological science with neutrons by engaging a large scientific community of structural biologists, enabling successful neutron diffraction experiments from radically smaller biomacromolecular crystals, resolving unanswered biochemical questions, and meaningfully contributing to rational drug design. The meeting highlighted 10 grand challenges that will be addressed with this advanced capability over the next decade and beyond, and the recommendations required to help address them are given below.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2023

Materials characterization: Can artificial intelligence be used to address reproducibility challenges?

Material characterization techniques are widely used to characterize the physical and chemical properties of materials at the nanoscale and, thus, play central roles in material scientific discoveries. However, the large and complex datasets generated by these techniques often require significant human effort to interpret and extract meaningful physicochemical insights. Artificial intelligence (AI) techniques such as machine learning (ML) have the potential to improve the efficiency and accuracy of surface analysis by automating data analysis and interpretation. In this perspective paper, we review the current role of AI in surface analysis and discuss its future potential to accelerate discoveries in surface science, materials science, and interface science. We highlight several applications where AI has already been used to analyze surface analysis data, including the identification of crystal structures from XRD data, analysis of XPS spectra for surface composition, and the interpretation of TEM and SEM images for particle morphology and size. We also discuss the challenges and opportunities associated with the integration of AI into surface analysis workflows. These include the need for large and diverse datasets for training ML models, the importance of feature selection and representation, and the potential for ML to enable new insights and discoveries by identifying patterns and relationships in complex datasets. Most importantly, AI analyzed data must not just find the best mathematical description of the data, but it must find the most physical and chemically meaningful results. In addition, the need for reproducibility in scientific research has become increasingly important in recent years. The advancement of AI, including both conventional and the increasing popular deep learning, is showing promise in addressing those challenges by enabling the execution and verification of scientific progress. By training models on large experimental datasets and providing automated analysis and data interpretation, AI can help to ensure that scientific results are reproducible and reliable. Although integration of knowledge and AI models must be considered for the transparency and interpretability of models, the incorporation of AI into the data collection and processing workflow will significantly enhance the efficiency and accuracy of various surface analysis techniques and deepen our understanding at an accelerated pace.

Materials Science↗

DOE OSTI2024

Physics-Guided, Physics-Informed, and Physics-Encoded Neural Networks and Operators in Scientific Computing: Fluid and Solid Mechanics

Abstract Advancements in computing power have recently made it possible to utilize machine learning and deep learning to push scientific computing forward in a range of disciplines, such as fluid mechanics, solid mechanics, materials science, etc. The incorporation of neural networks is particularly crucial in this hybridization process. Due to their intrinsic architecture, conventional neural networks cannot be successfully trained and scoped when data are sparse, which is the case in many scientific and engineering domains. Nonetheless, neural networks provide a solid foundation to respect physics-driven or knowledge-based constraints during training. Generally speaking, there are three distinct neural network frameworks to enforce the underlying physics: (i) physics-guided neural networks (PgNNs), (ii) physics-informed neural networks (PiNNs), and (iii) physics-encoded neural networks (PeNNs). These methods provide distinct advantages for accelerating the numerical modeling of complex multiscale multiphysics phenomena. In addition, the recent developments in neural operators (NOs) add another dimension to these new simulation paradigms, especially when the real-time prediction of complex multiphysics systems is required. All these models also come with their own unique drawbacks and limitations that call for further fundamental research. This study aims to present a review of the four neural network frameworks (i.e., PgNNs, PiNNs, PeNNs, and NOs) used in scientific computing research. The state-of-the-art architectures and their applications are reviewed, limitations are discussed, and future research opportunities are presented in terms of improving algorithms, considering causalities, expanding applications, and coupling scientific and deep learning solvers.

Computer Science↗

DOE OSTI2024

Theranostic Radiopnictogens: 71 As, 72 As, and 119 Sb (Final Technical Report)

This project has developed new methods for the cyclotron production of medically relevant radionuclides 71 As and 119 Sb. Arsenic and antimony are chemically homologous elements (group 5A, also known as the pnictogens) that have radionuclides that are of considerable interest within nuclear medicine. Such radiopnictogens include the potentially therapeutic radionuclides 119 Sb (t 1/2 = 38 h) that decays with the emission of 24.5 low energy, high potency electrons per decay with little concomitant photon radiation and 77 As (t 1/2 = 39 h) that decays with average beta energy of 230 keV and diagnostic nuclides 71 As (t 1/2 = 65 h, 28% β+) and 72 As (t 1/2 = 26 h, 80% β+) for positron emission tomography (PET). This work has had major success developing new methods for the cyclotron production and radiochemical isolation of 71 As, supporting parallel developments for 119 Sb, and assessing the chemical similarities between these two homologous radionuclides. This project brought into collaboration two universities with complimentary skill sets, proficiencies, expertise, and facilities: the University of Wisconsin (UWisc) and the University of Missouri (Mizzou). Professors Ellison and Engle have experience in the small cyclotron production and radiochemical isolation of radionuclides, including 72 As and 119 Sb. Their recently developed metallurgic methods for fabricating cyclotron targets have great potential to expand and allow for the biomedical cyclotron production of long- lived, lower positron energy 71 As. Professors Hennkens and Jurisson have significant experience in the reactor production, radiochemical isolation, and biological functionalization of radioarsenic. Recent development of trithiol-based chelator molecules for functionalizing radioarsenic provide a platform for the investigation of the fundamental challenges of the promising low-energy-electron emitter, 119 Sb. Through their positions within their respective University’s graduate schools, the PIs and Co-Is effectively trained of graduate students and postdoctoral researchers in nuclear and radiochemistry, sub-specialties specifically identified in the Department of Energy (DOE) Office of Science Isotope Program long-range plan. Annual laboratory research visits for students between UWisc and Mizzou provided essential broad-field experience and scientific networking that is critical for maintaining their path along the training pipeline to productive careers in isotope production. This research collaboration has provided significant benefits to the DOE University Isotope Network and radionuclide-using researchers around the country.

07 ISOTOPE AND RADIATION SOURCES↗

DOE OSTI2026

3D Gaussian Splatting for Volume Compression

This codebase uses machine learning to train a collection of 3D Gaussian distributions to approximate scientific volume data. Because this collection uses less memory than the original dataset, it can be used as a compressed model of the original data for applications such as visualization.

Dyken, Landon↗

DOE OSTI2025

Reliable and Efficient Machine Learning (Final Technical Report)

Modern scientific experiments generate massive amounts of data at a pace much faster than humans can manually analyze. While machine learning has revolutionized commercial data analysis (such as recommending movies or recognizing faces), applying these tools to complex scientific discovery is challenging because scientific answers must be precise, interpretable, and adhere to physical laws. The research under this project aims to develop new mathematical tools and computer algorithms specifically designed for scientific applications. Major progress has been made in automatically cleaning and deconstructing messy experimental data, analyzing the visual information of physical phenomena, determining the underlying physical variables, and providing rig orous mathematical analysis of interesting algorithms and concepts widely used in machine learning. This project addressed the critical gap between our ability to generate massive scientific data and our ability to extract interpretable information from it. We established mathematical foundations for Scientific Machine Learning (SciML) aimed at effective data analytics and automated discovery. Our work focused on three core objectives: (1) developing reliable feature extraction methods for dynamic high-dimensional data, (2) establishing mathematical foundations for discovering dynamics via neural networks, and (3) creating rigorous optimization techniques for these models. Key outcomes come from two fronts. On the practical side, they include the development of algorithms that significantly enhance the extraction of signals from field data, as well as the capability to handle situations that exhibit smooth variations or physical stretching due to temperature changes. They also include the creation of an automated framework for discovering fundamental state variables from raw experimental data, demonstrating the ability to identify intrinsic physical dimensions without prior knowledge of the governing laws. On the theoretical front, the research results in theoretical advances in Optimal Transport, a widely used notion in SciML, specifically regarding functions with fixed-size nodal sets, provide sharp bounds relevant to uncertainty quantification. Meanwhile, the outcomes also include the establishment of convergence theories for nonlocal gradient descent methods, enabling robust optimization with noisy data in high-dimensional settings commonly encountered in scientific modeling. The project also helps creating opportunities to train the next generation of researchers, equipping them with the necessary technical skills for today’s workplace and preparing them for future advances.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

Combinatorial Algorithms in Scientific Computing

We provide the final report for this grant, detailing the publications, software produced, students trained who have joined the DOE workforce, and the impact our work has had on computational mathematics and related disciplines.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Tokenized Data for FORGE Foundation Models

This dataset comprises a vast corpus of 257 billion tokens, accompanied by the corresponding vocabulary file employed in the pre-training of FORGE foundation models. The primary data source for this corpus is scientific documents derived from diverse origins, and they have been tokenized using the Hugging Face BPE tokenizer. Further details about this research can be found in the publication titled FORGE: Pre-Training Open Foundation Models for Science authored by Junqi Yin, Sajal Dash, Feiyi Wang, and Mallikarjun (Arjun) Shankar, presented at SC'23. The data tokenization pipeline and resulting artifacts use CORE data [Ref: Knoth, P., and Zdrahal, Z. (2012). CORE: three access levels to underpin open access. D-Lib Magazine, 18(11/12)]. For use of these data sets for any purpose, please follow the guidelines provided in https://core.ac.uk/terms .

Yin, Junqi↗