SEARCH · Engineering Papers

Results for “Network inference”

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 811 records · Page 45

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Machine learning enhanced predictions of ICRF heating: Overcoming numerical limitations via data curation

In this work, we present the development of robust surrogate models for Ion Cyclotron Range of Frequencies (ICRF) and High-Harmonic Fast Wave (HHFW) heating predictions in fusion plasmas. Building upon our previous efforts to achieve real-time capable models, we identify the cause of the outliers found using TORIC in certain HHFW heating scenarios. The outliers are observed to be spurious ion Bernstein wave (IBW)-like modes caused by a wavelength control algorithm designed to address challenging scenarios with high perpendicular wavenumbers. The effect arises from the modulation in the perpendicular susceptibility, which can induce sign reversal and IBW-like propagation for scenarios featuring normalized ion Larmor radius λ i ≫ 1. We use TORIC with this algorithm disabled to generate a novel HHFW-NSTX database that is free of outliers. Surrogate models trained on this database, including Random Forest Regressor (RFR), Multi-Layer Perceptrons, and Gaussian Process Regressors (GPR), demonstrate the ability to accurately predict HHFW heating profiles, with regression scores of R 2 ∈[0.93−0.99]. Additionally we demonstrate that it is possible to generalize predictions beyond training data by the use of both RFR and GPR models, enabling the prediction of scenarios previously limited to the original model. GPR models also provide uncertainty quantification, offering insights into model confidence. This work introduces a comprehensive Verification, Validation, and Uncertainty Quantification methodology for surrogate modeling, applicable not only to ICRF heating but also to other RF heating challenges and fusion physics problems. Beyond accelerated inference, these models show effective extrapolation capabilities, providing an alternative for addressing numerical challenges.

Artificial neural networks↗

DOE OSTI2021

Investigation of Scale-Dependent Groundwater/Surface-water Exchange in Rivers by Gradient Self-Potential Logging: Numerical Modeling and Field Experiments

Exchanges of groundwater and surface-water are fundamental to a wide range of water-supply and water-quality management issues but challenging to map beyond the reach scale. Waterborne gradient self-potential (SP) measurements are directly sensitive to water flow through riverbed sediments and can be used to infer exchange locations, direction (gain versus loss), scale, and relative changes, but to date applications to river corridor hydrology are limited. Numerical modeling and field experiments were therefore performed herein, each emphasizing waterborne gradient SP logging for identifying and locating focused vertical groundwater discharge (surface-water gain) and recharge (surface-water loss) in a river. Two and three-dimensional numerical models were constructed to simulate the polarities, appearances, and peak amplitudes of streaming-potential and electric-field anomalies on a riverbed and in the surface-water that were attributable to steady-state vertical fluxes of groundwater through high-permeability conduits in the riverbed. Effects of varied hydraulic length-scale of exchange and surface-water depth were tested through numerical modeling. Modeling results aided in data acquisition and interpretation for three repeated field experiments performed along a 1.5–2.0 km reach of the Quashnet River in Cape Cod, Massachusetts, where focused, meter-scale groundwater discharges occur at discrete locations within otherwise ubiquitous and more diffuse groundwater upwelling conditions. Strong gradient SP anomalies were repeatedly measured in the Quashnet River at previously confirmed locations of focused groundwater discharge, showing the efficacy of waterborne gradient SP logging in identifying and characterizing groundwater/surface water exchange dynamics at multiple river network scales.

Engineering↗

DOE OSTI2020

Seismic tomography of compressional wave velocity and attenuation structure for Makushin Volcano, Alaska

Located in the central portion of the Alaska-Aleutian subduction zone, Makushin Volcano is among the most active volcanoes in the United States and has been classified as a very high threat by the National Volcano Early Warning System based on its eruptive history and proximity to the City of Unalaska and international air routes. In 2015, five standalone seismic stations and three mini seismic arrays of 15 stations each were deployed on Unalaska Island to supplement the Alaska Volcano Observatory permanent seismic network. This temporary array was operational for one year. Taking advantage of the increased azimuthal coverage and the array's increased earthquake detection capability, we developed body-wave Vp and Vp/Vs seismic images of the velocity structure beneath the volcano. Our results show a complex structure with the upper 4 km of the crust. The shallow high-Vp features possibly delineate remnant magma pathways or conduits. Low-Vp regions are found east of the caldera at approximately 4–8 km depth. This is in agreement with previous seismic tomographic work and InSAR-based geodetic models, which had identified this region as a possible long-term source of magma. A three-dimensional compressional wave attenuation model complemented the velocity models. In general, we observe regions of corresponding low P-wave velocity and high attenuation east of the summit caldera between 5 and 7 km depth, supporting the presence of a melt-rich zone. The absolute Vp values (5.5–6.0 km/s) in this region are consistent with the presence of a magma body with a mafic signature. Here, we suggest that a triggering mechanism that involves mafic injections into a more evolved shallow magma chamber is representative of the current magmatic system below the volcano, similarly to what petrological studies have inferred for past eruptions.

58 GEOSCIENCES↗

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↗

DOE OSTI2021

Adoption of image-driven machine learning for microstructure characterization and materials design: A Perspective

Microstructure characterization enables the development of structure-processing-property relationships critical to several research areas within the broad field of materials science, from alloy design to the assessment of corrosion resistance, and failure analysis. Conventional approaches to material characterization have relied on either qualitative inference by the human ex-pert or software applications that can extract high-level features from images, such as boundary segmentation, average grain diameter, etc. Such approaches rely heavily on subject matter expert user intervention and knowledge of what phases or more generally, what microstructural features, are of interest. The recent surge in the adoption of machine learning techniques to address problems in materials engineering has brought with it an increased interest and application of Image Driven Machine Learning (IDML) approaches. In this work, we review the applications of IDML to the field of materials characterization. A canonical hierarchy of stages is defined, which when put sequentially together completes an IDML study: problem definition, dataset building, model selection and training, model evaluation, and integration with existing instrumentation or simulation workflow. The studies reviewed in this work are analyzed from the perspective of each of these stages. Such a review permits agranular assessment of the field, for example the impact of IDML on materials characterization at the nanoscale, the size of a typical dataset required to train a semantic segmentation model on electron microscopy images, ubiquitousness of transfer learning in the domain, etc. Finally, we discuss the importance of interpretability and explainability in the field of IDML for materials characterization, and provide an overview of two emerging techniques in the field: semantic segmentation and generative adversarial networks.

Baskaran, Arun↗

DOE OSTI2021

Mechanism and dynamics of fatty acid photodecarboxylase

Photoenzymes are rare biocatalysts driven by absorption of a photon at each catalytic cycle; they inspire development of artificial photoenzymes with valuable activities. Fatty acid photodecarboxylase (FAP) is a natural photoenzyme that has potential applications in the bio-based production of hydrocarbons, yet its mechanism is far from fully understood. RATIONALE To elucidate the mechanism of FAP, we studied the wild-type (WT) enzyme from Chlorella variabilis (CvFAP) and variants with altered active-site residues using a wealth of techniques, including static and time-resolved crystallography and spectroscopy, as well as biochemical and computational approaches. RESULTS A 1.8-Å-resolution CvFAP x-ray crystal structure revealed a dense hydrogen-bonding network positioning the fatty acid carboxyl group in the vicinity of the flavin adenine dinucleotide (FAD) cofactor. Structures solved from free electron laser and low-dose synchrotron x-ray crystal data further highlighted an unusual bent shape of the oxidized flavin chromophore, and showed that the bending angle (14°) did not change upon photon absorption (step 1) or throughout the photocycle. Calculations showed that bending substantially affected the energy levels of the flavin. Structural and spectroscopic analysis of WT and mutant proteins targeting two conserved active-site residues, R451 and C432, demonstrated that both residues were crucial for proper positioning of the substrate and water molecules and for oxidation of the fatty acid carboxylate by 1 FAD* (~300 ps in WT FAP) to form FAD ∙– (step 2). Time-resolved infrared spectroscopy demonstrated that decarboxylation occured quasi-instantaneously upon this forward electron transfer, consistent with barrierless bond cleavage predicted by quantum chemistry calculations and with snapshots obtained by time-resolved crystallography. Transient absorption spectroscopy in H 2 O and D 2 O buffers indicated that back electron transfer from FAD ∙– was coupled to and limited by transfer of an exchangeable proton or hydrogen atom (step 3). Unexpectedly, concomitant with FAD ∙– reoxidation (to a red-shifted form FAD RS ) in 100 ns, most of the CO 2 product was converted, most likely into bicarbonate (as inferred from FTIR spectra of the cryotrapped FAD RS intermediate). Calculations indicated that this catalytic transformation involved an active-site water molecule. Cryo-Fourier transform infrared spectroscopy studies suggested that bicarbonate formation (step 4) was preceded by deprotonation of an arginine residue (step 3). At room temperature, the remaining CO 2 left the protein in 1.5 μs (step 4'). The observation of residual electron density close to C432 in electron density maps derived from time-resolved and cryocrystallography data suggests that this residue may play a role in stabilizing CO 2 and/or bicarbonate. Three routes for alkane formation were identified by quantum chemistry calculations; the one shown in the figure is favored by the ensemble of experimental data. CONCLUSION Finally, we provide a detailed and comprehensive characterization of light-driven hydrocarbon formation by FAP, which uses a remarkably complex mechanism including unique catalytic steps. We anticipate that our results will help to expand the green chemistry toolkit.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

FPGA Acceleration of GCN in Light of the Symmetry of Graph Adjacency Matrix

Graph Convolutional Neural Networks (GCNs) are widely used to process large-scale graph data. Different from deep neural networks (DNNs), GCNs are sparse, irregular, and unstructured, posing unique challenges to hardware acceleration with regular processing elements (PEs). In particular, the adjacency matrix of a GCN is extremely sparse, leading to frequent but irregular memory access, low spatial/temporal data locality and poor data reuse. Furthermore, a realistic graph usually consists of unstructured data (e.g., unbalanced distributions), creating significantly different processing times and imbalanced workload for each node in GCN acceleration. To overcome these challenges, we propose an end-to-end hardware-software co-design to accelerate GCNs on resource-constrained FPGAs with the features including: (1) A custom dataflow that leverages symmetry along the diagonal of the adjacency matrix to accelerate feature aggregation for undirected graphs. We utilize either the upper or the lower triangular matrix of the adjacency matrix to perform aggregation in GCN to improve data reuse. (2) Unified compute cores for both aggregation and transform phases, with full support to the symmetry-based dataflow. These cores can be dynamically reconfigured to the systolic mode for transformation or as individual accumulators for aggregation in GCN processing. (3) Preprocessing of the graph in software to rearrange the edges and features to match the custom dataflow. This step improves the regularity in memory access and data reuse in the aggregation phase. Moreover, we quantize the GCN precision from FP32 to INT8 to reduce the memory footprint without losing the inference accuracy. We implement our accelerator design in Intel Stratix10 MX FPGA board with HBM2, and demonstrate 1.3x-110.5x improvement in end-to-end GCN latency as compared to the state-of the-art FPGA implementations, on the graph datasets of Cora, Pubmed, Citeseer and Reddit.

Nair, Gopikrishnan R.↗

DOE OSTI2024

Identifying Bayesian optimal experiments for uncertain biochemical pathway models

Abstract Pharmacodynamic (PD) models are mathematical models of cellular reaction networks that include drug mechanisms of action. These models are useful for studying predictive therapeutic outcomes of novel drug therapies in silico. However, PD models are known to possess significant uncertainty with respect to constituent parameter data, leading to uncertainty in the model predictions. Furthermore, experimental data to calibrate these models is often limited or unavailable for novel pathways. In this study, we present a Bayesian optimal experimental design approach for improving PD model prediction accuracy. We then apply our method using simulated experimental data to account for uncertainty in hypothetical laboratory measurements. This leads to a probabilistic prediction of drug performance and a quantitative measure of which prospective laboratory experiment will optimally reduce prediction uncertainty in the PD model. The methods proposed here provide a way forward for uncertainty quantification and guided experimental design for models of novel biological pathways.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

CSB-RNN: A Faster-Than-Realtime RNN Acceleration Framework with Compressed Structured Blocks

Recurrent Neural Networks (RNN) is widely applied to temporal sequence analysis, where real-time performance is usually in demand. However, RNN suffers a heavy computational workload as the model comes with a large weight matrix. To alleviate the pain, model compression (pruning) schemes have been proposed for RNN that pruning the redundant (near-zero) weight-values. On the one hand, the non-structured pruning methods achieve a considerable pruning rate while bringing the computational irregularity, which is un-friendly to parallel-hardware. On the other hand, the existing structured pruning methods consider the hardware parallelism; However, they suffer a poor pruning rate due to the restrict constraints on pruning structure. This paper presents CSB-RNN, an optimized full-stack RNN framework with the novel compressed structured block (CSB) technique. The CSB-pruned RNN model comes with both fine-granularity that benefits the pruning rate and regular structure that facilitates the hardware-parallelism. Further, we propose a novel hardware architecture for inferencing the CSB-pruned model. Different from conventional parallel hardware, this architecture solves the block-workload imbalance issue and achieves an over 95% hardware utilization. With the experiments on 10 RNN models in 5 application domains, the CSB-RNN realizes 7×-20× lossless compression and up to 50× acceptable lossy-compression, which is 2×-7× to the prior art. With the addition of the novel hardware, the compressed-RNN inference reaches a super real-time latency of 10-400µs with FPGA implementation.

Shi, Runbin↗

DOE OSTI2025

Quantum optimal control of superconducting qubits based on machine-learning characterization

Implementing fast and high-fidelity quantum operations using open-loop quantum optimal control relies on having an accurate model of the quantum dynamics. Any deviations between this model and the complete dynamics of the device, such as the presence of spurious modes or pulse distortions, can degrade the performance of optimal controls in practice. Here, we propose an experimentally simple approach to realize optimal quantum controls tailored to the device parameters and environment while specifically characterizing this quantum system. Concretely, we use physics-inspired machine learning to infer an accurate model of the dynamics from experimentally available data and then optimize our experimental controls on this trained model. We show the power and feasibility of this approach by optimizing arbitrary single-qubit operations in detailed numerical simulations of a superconducting transmon qubit. Furthermore, we demonstrate that this framework produces an accurate description of the device dynamics under arbitrary controls, together with the precise pulses achieving arbitrary single-qubit gates with a high fidelity of ∼99.99%.

Artificial neural networks↗

DOE OSTI2023

On the influence of over-parameterization in manifold based surrogates and deep neural operators

Constructing accurate and generalizable approximators (surrogate models) for complex physico-chemical processes exhibiting highly non-smooth dynamics is challenging. The main question is what type of surrogate models we should construct and should these models be under-parameterized or over-parameterized. In this work, we propose new developments and perform comparisons for two promising approaches: manifold-based polynomial chaos expansion (m-PCE) and the deep neural operator (DeepONet), and we examine the effect of over-parameterization on generalization. While m-PCE enables the construction of a mapping by first identifying low-dimensional embeddings of the input functions, parameters, and quantities of interest (QoIs), a neural operator learns the nonlinear mapping via the use of deep neural networks. Here, we demonstrate the performance of these methods in terms of generalization accuracy by solving the 2D time-dependent Brusselator reaction-diffusion system with uncertainty sources, modeling an autocatalytic chemical reaction between two species. We first propose an extension of the m-PCE by constructing a mapping between latent spaces formed by two separate embeddings of the input functions and the output QoIs. To further enhance the accuracy of the DeepONet, we introduce weight self-adaptivity in the loss function. We demonstrate that the performance of m-PCE and DeepONet is comparable for cases of relatively smooth input-output mappings. However, when highly non-smooth dynamics is considered, DeepONet shows higher approximation accuracy. We also find that for m-PCE, modest over-parameterization leads to better generalization, both within and outside of distribution, whereas aggressive over-parameterization leads to over-fitting. In contrast, an even highly over-parameterized DeepONet leads to better generalization for both smooth and non-smooth dynamics. Furthermore, we compare the performance of the above models with another recently proposed operator learning model, the Fourier Neural Operator, and show that its over-parameterization also leads to better generalization. Taken together, our studies show that m-PCE can provide very good accuracy at very low training cost, whereas a highly over-parameterized DeepONet can provide better accuracy and robustness to noise but at higher training cost. In both methods, the inference cost is negligible.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Machine Learning Models for Binary Molecular Classification using VUV Absorption Spectra

Machine learning methods were combined with differential absorption spectroscopy measurements in the vacuum-ultraviolet region (5.167 – 9.920 eV) in order to develop predictive capabilities for inferring molecular structure from the spectra. Several types of species were analyzed and, for modeling purposes, were defined using a single classification: (1) alkane, (2) conjugation with oxygen (e.g. diacetyl, ethyl vinyl ether), (3) non-conjugated alkene (e.g. 1-butene, 1,4-cyclohexadiene), (4) oxygen-containing (e.g. 1-butanol, tetrahydrofuran), or (5) cyclic (e.g. cyclopentane, cyclohexanone). The latter molecular classification excluded cyclic ethers. Several modeling methods were employed in the analysis of 102 absorption spectra, 24 of which were measured for the first time. The primary objective was to identify suitable methods that enable accurate predictions of molecular structure classifications with minimized statistical uncertainties. Rather than identifying a single, unifying method to reliably predict molecular structure contributions to VUV absorption spectra, coordination is required among a particular method, the type of molecular structure detail (e.g. conjugation), and absorption region of interest. The latter is accomplished using a binning approach, wherein absorption regions of ~0.5 eV were utilized rather than the entire ~4.8 eV range. Photon energy binning enabled analysis of region-specific predictions of accuracy, precision, and recall. The outcome from the binning approach is that, rather than utilizing the entire spectrum, optimal determination of molecular structure using machine learning methods depends on the absorption region. Furthermore, the present work provides separate machine learning models for each molecular classification, which enables the identification of multi-functional species relevant to atmospheric chemistry and combustion chemistry, where isomer-resolved speciation is critical to understanding complex reaction networks.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2025

Comparative Performance Evaluation of Large Language Models for Extracting Molecular Interactions and Pathway Knowledge

Understanding the interactions and regulatory relationships among biomolecules is essential for deciphering complex biological systems and elucidating the mechanisms behind diverse biological functions. Traditionally, the collection of such molecular interaction data has relied on expert curation, a process that is both time-consuming and labor-intensive. To address these limitations, this study explores the use of large language models (LLMs) to automate the genome-scale extraction of molecular interaction knowledge. Here, we evaluate the performance of various LLMs on key biological tasks, including the identification of protein-protein interactions, detection of genes associated with pathways influenced by low-dose radiation, and inference of gene regulatory relationships. Our findings demonstrate that larger LLMs tend to perform better, particularly in extracting intricate gene and protein interactions. Despite their strengths, these models face challenges in recognizing functionally diverse gene groups and highly correlated regulatory relationships. Through a comprehensive analysis using established molecular interaction and pathway databases, we show that LLMs possess the potential to identify relevant biomolecules and predict their interactions, offering valuable insights and marking a significant step toward AI-driven biological knowledge discovery.

63 RADIATION, THERMAL, AND OTHER ENVIRON. POLLUTAN↗

DOE OSTI2025

Graph reinforcement learning for exploring model spaces beyond the standard model

We present a methodology for performing scans of beyond the standard model (BSM) parameter spaces with reinforcement learning. We identify a novel procedure using graph neural networks that is capable of exploring spaces of models without the user specifying a fixed particle content, allowing broad classes of BSM models to be explored—in theory, the technique is applicable to nearly any model space with a prespecified gauge group. We provide a generic procedure by which a suitable graph grammar can be developed for any BSM model that features user-specified symmetry groups and a finite number of different possible particle species, the use of which is applicable to a variety of machine learning tasks over the actions of BSM theories beyond our particular reinforcement learning use case. As a proof of concept, we construct the graph grammar for theories with vectorlike leptons that may or may not be charged under a dark U ( 1 ) group, inspired by portal matter extensions of the sub-GeV vector portal/kinetic mixing simplified dark matter models. We then use this graph grammar to create a reinforcement learning environment tasked with creating models with these vectorlike leptons that are consistent with a list of a variety of precision observables. The reinforcement learning agent succeeds in developing models that can address the observed muon anomalous magnetic moment discrepancy while remaining consistent with flavor violation and electroweak precision observables, including both constructions that have previously been studied as well as new models that have not, to our knowledge, previously been identified. By inspecting the resulting ensembles of models that the agent produces and experimenting with different configurations for our reinforcement learning environment and graph grammar, we also infer various lessons about the development of these environments that can be transferable to reinforcement learning scans of more complicated model spaces and comment on future directions for the development of this technique into a more mature tool. Published by the American Physical Society 2025

Wojcik, George N.↗

DOE OSTI2021

Physics guided machine learning using simplified theories

Recent applications of machine learning, in particular deep learning, motivate the need to address the generalizability of the statistical inference approaches in physical sciences. In this Letter, we introduce a modular physics guided machine learning framework to improve the accuracy of such data-driven predictive engines. The chief idea in our approach is to augment the knowledge of the simplified theories with the underlying learning process. To emphasize their physical importance, our architecture consists of adding certain features at intermediate layers rather than in the input layer. To demonstrate our approach, we select a canonical airfoil aerodynamic problem with the enhancement of the potential flow theory. We include the features obtained by a panel method that can be computed efficiently for an unseen configuration in our training procedure. By addressing the generalizability concerns, our results suggest that the proposed feature enhancement approach can be effectively used in many scientific machine learning applications, especially for the systems where we can use a theoretical, empirical, or simplified model to guide the learning module.

42 ENGINEERING↗

DOE OSTI2025

Karhunen–Loève deep learning method for surrogate modeling and approximate Bayesian parameter estimation

We evaluate the performance of the Karhunen-Loève Deep Neural Network (KL-DNN) framework for surrogate modeling and approximate Bayesian parameter estimation in partial differential equation models. In the surrogate model, the Karhunen-Loève (KL) expansions are used for the dimensionality reduction of the number of unknown parameters and variables, and a deep neural network is employed to relate the reduced space of parameters to that of the state variables. The KL-DNN surrogate model is used to formulate a maximum-a-posteriori-like least-squares problem, which is randomized to draw samples of the posterior distribution of the parameters. We test the proposed framework for a hypothetical unconfined aquifer via comparison with the forward MODFLOW and inverse PEST++ iterative ensemble smoother (IES) solutions as well as the state-of-the-art Fourier neural operator (FNO) and deep operator networks (DeepONets) operator learning surrogate models. Our results show that the KL-DNN surrogate model outperforms FNO and DeepONet for forward predictions. For solving inverse problems, the randomized algorithm provides the same or more accurate Bayesian predictions of the parameters than IES as evidenced by the higher log-predictive probability of both the estimated parameter field and the forecast hydraulic head. The posterior mean obtained from the randomized algorithm is closer to the reference parameter field than that obtained with FNO as the maximum a posteriori estimate.

Approximate Bayesian inference↗

DOE OSTI2023

Fast 2D Bicephalous Convolutional Autoencoder for Compressing 3D Time Projection Chamber Data

High-energy large-scale particle colliders produce data at high speed in the order of 1 terabytes per second in nuclear physics and petabytes per second in high energy physics. Developing real-time data compression algorithms to reduce such data at high throughput to fit permanent storage has drawn increasing attention. Specifically, at the newly constructed sPHENIX experiment at the Relativistic Heavy Ion Collider (RHIC), a time projection chamber is used as the main tracking detector, which records particle trajectories in a volume of three-dimensional (3D) cylinder. The resulting data are usually very sparse with occupancy around 10.8%. Such sparsity presents a challenge to conventional learning-free lossy compression algorithms, such as SZ, ZFP, and MGARD. The 3D convolutional neural network (CNN)-based approach, Bicephalous Convolutional Autoencoder (BCAE), outperforms traditional methods both in compression rate and reconstruction accuracy. BCAE can also utilize the computation power of graphical processing units suitable for deployment in a modern heterogeneous highperformance computing environment. This work introduces two BCAE variants: BCAE++ and BCAE-2D. BCAE++ achieves a 15% better compression ratio and a 77% better reconstruction accuracy measured in mean absolute error compared with BCAE. BCAE-2D treats the radial direction as the channel dimension of an image, resulting in a 3× speedup in compression throughput. In addition, we demonstrate an unbalanced autoencoder with a larger decoder can improve reconstruction accuracy without significantly sacrificing throughput. Lastly, we observe both the BCAE++ and BCAE-2D can benefit more from using half-precision mode in throughput (76 - 79% increase) without loss in reconstruction accuracy. The source code and links to data and pretrained models can be found at https://github.com/BNL-DAQ-LDRD/NeuralCompression_v2

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2020

NEON AOP foliar trait maps, maps of model uncertainty estimates, and conifer map, East River, CO 2018

This data package contains mapped trait estimates and their uncertainties, and conifer map, for the National Ecological Observatory Network's Airborne Observation Platform survey data acquired over the Upper East River, Colorado in 2018. For full details, please see associated reference. in brief, trait models were developed independently for needle and non-needle leaf species using partial least squares regression (PLSR) using ground data from additional datasets: doi:10.15485/1618130, doi:10.15485/1618132, and doi:10.15485/1631278, merged with extracted spectral data from doi:10.15485/1618131. We separated vegetated pixels into needle and non-needle classes in order to generate a classification map based on the spectral differences between these leaf types (conifer.tif). We trained a deep learning model with custom architecture, detailed in Chadwick et al. In Press. The model performed with 0.998 true positive rate and 0.982 true negative rate, with ‘positives’ being non-needle identification. We then utilized PLSR to generate models of foliar traits for each leaf type. So that we could also map uncertainty in these predictions, we generated ten different models for needle and non-needle leaf species using different testing holdout sets of discrete sites. Each of these models was developed with a 100-fold cross validation procedure that utilized a 70% training set and 30% validation set with each fold, and then assessed based on the 10% of testing sites that were not included in that model’s development. The mean predicted value across the 10 models is used for the trait estimate in each pixel across the study area. The models are applied according to the leaf type designation in the conifer.tif map. The errors are the standard deviation across the 10 different models developed, with high error suggesting instability in model prediction and areas where values may not be reliable for ecological inference. These maps are only applied to areas with a NDVI > 0.5 to exclude non-vegetated areas. Shade masks could be applied to these data (doi:10.15485/1618131), but have not been for this data package. These data are also available on Google Earth Engine: https://code.earthengine.google.com/?asset=users/kdc/ER_NEON

54 ENVIRONMENTAL SCIENCES↗