Search NASA⌕ Search

SEARCH · Search NASA

Results for “sparse selection”

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 325 records · Page 18

SOARCA Peach Bottom Atomic Power Station Long-Term Station Blackout Uncertainty Analysis: Convergence of the Uncertainty Results

This paper describes the convergence of MELCOR Accident Consequence Code System, Version 2 (MACCS2) probabilistic results of offsite consequences for the uncertainty analysis of the State-of-the-Art Reactor Consequence Analyses (SOARCA) unmitigated long-term station blackout scenario at the Peach Bottom Atomic Power Station. The consequence metrics evaluated are individual latent-cancer fatality (LCF) risk and individual early fatality risk. Consequence results are presented as conditional risk (i.e., assuming the accident occurs, risk per event) to individuals of the public as a result of the accident. In order to verify convergence for this uncertainty analysis, as recommended by the Nuclear Regulatory Commission’s Advisory Committee on Reactor Safeguards, a ‘high’ source term from the original population of Monte Carlo runs has been selected to be used for: (1) a study of the distribution of consequence results stemming solely from epistemic uncertainty in the MACCS2 parameters (i.e., separating the effect from the source term uncertainty), and (2) a comparison between Simple Random Sampling (SRS) and Latin Hypercube Sampling (LHS) in order to validate the original results obtained with LHS. Three replicates (each using a different random seed) of size 1,000 each using LHS and another set of three replicates of size 1,000 using SRS are analyzed. The results show that the LCF risk results are well converged with either LHS or SRS sampling. The early fatality risk results are less well converged at radial distances beyond 2 miles, and this is expected due to the sparse data (predominance of “zero” results).

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Phase curves of small bodies from the SLOAN Moving Objects Catalog

Extensive photometric surveys continue to produce enormous stores of data on small bodies. These data are typically sparsely obtained at arbitrary (or unknown) rotational phases. Therefore, new methods for processing such data need to be developed to make the most of these vast catalogs. We aim to produce a method of recreating the phase curves of small bodies by considering the uncertainties introduced by the nominal errors in the magnitudes and the effect introduced by rotational variations. Here, we use the SLOAN Moving Objects Catalog data as a benchmark to construct phase curves of all small bodies in u', g', r', i', and z' filters. From the phase curves, we obtain the absolute magnitudes and we use them to set up the absolute colors, which are the colors of the asteroids that are not affected by changes in the phase angle. We selected objects with ≥3 observations taken in at least one filter and spanning over a minimum of 5 degrees in the phase angle. We developed a method that combines Monte Carlo simulations and Bayesian inference to estimate the absolute magnitudes using the HG 12 * photometric system. We obtained almost 15 000 phase curves, with about 12 000 of these including all five filters. The absolute magnitudes and absolute colors are compatible with previously published data that support our method. The method we developed is fully automatic and well suited for a run based on large amounts of data. Moreover, it includes the nominal uncertainties in the magnitudes and the whole distribution of possible rotational states of the objects producing what are possibly less precise values, that is, larger uncertainties, but more accurate, namely, closer to the actual value. To our knowledge, this work is the first to include the effect of rotational variations in such a manner.

79 ASTRONOMY AND ASTROPHYSICS↗

Scalable Techniques for Stochastic Power Flow Problems (Final Report)

The proposed research focuses on developing scalable algorithms for two-stage security-constrained OPF problems with AC power flow constraints, a class of problems complicated by (i) scale arising from a scenario representation; and (ii) the presence of nonlinearity, nonconvexity, and possibly second-stage discreteness or complementarity. Unfortunately, most existing solvers cannot contend with both challenges simultaneously; accordingly, the proposed research focuses on developing solution techniques that can both scale with the number of scenarios and contend with nonconvexity and second-stage complementarity. We consider three avenues for addressing such problems: (i) Variable sample-size SQP (VS-SQP) methods that combine sparse Quasi-Newton updates with a scalable variance-reduced stochastic gradient scheme for stochastic QP subproblems, allowing for contending with second-stage complementarity via regularization; (ii) Variable sample-size stochastic Interior-point (VS-sIP) schemes that propose a sampling-based regularized (to allow for contending with complementarity) interior-point schemes in which a Schur-complement technique is employed for decomposing the Newton direction computation step; (iii) Variable sample-size tractable ADMM (VS-tADMM) schemes combine variable sample-sizes with carefully designed techniques for resolving each of the nonconvex updates (by leveraging the QCQP structures). We intend to compare the three schemes using performance profiles in terms of solution quality, scalability, etc. and then select one scheme which will then be developed and further refined in Python for purposes of the GO competition.

42 ENGINEERING↗

Damage imaging using Lamb waves for SHM applications

2-D ultrasonic arrays, due to their beam-steering capability and all azimuth angle coverage are a very promising tool for the inspection of plate-like structures using Lamb waves (LW). Contrary to the classical linear phased arrays (PAs) the 2D arrays enable unequivocal defect localization and they are even capable of mode selectivity of the received LWs . Recently, it has been shown that multistatic synthetic focusing (SF) algorithms applied for 2D arrays are much more effective than the classical phase array mode commonly used in NDT. The multistatic SF assumes multiple transmissions of elements in a transmitting aperture and off-line processing of the data acquired by a receiving aperture. In the simplest implementation of the technique, only a single multiplexed input and a number of output channels are required, which results in significant hardware simplification compared with the PA systems. On the one hand implementation of the multistatic SF to 2D arrays creates additional degrees of freedom during the design of the array topology, which complicates the array design process. On the other hand, it enables designing sparse arrays with performance similar to that of the fully populated dense arrays. In this paper we present a general systematic approach to the design and optimization of imaging systems based on the 2D array operating in the multistatic mode. We start from presenting principles of the SF schemes applied to LW imaging. Then, we outline the coarray concept and demonstrate how it can be used for reducing number of elements of the 2D arrays. Finally, efficient tools for the investigation and experimental verification of the designed 2D array prototypes are presented. The first step in the investigation is theoretical evaluation performed using frequency-dependent structure transfer function (STF), which enables approximate simulation of an array excited with a tone-burst in a dispersive medium. Finally, we show how scanning laser vibrometer, sensing waves in multiple points corresponding to the locations of the 2D receiving array elements, can be used as a tool for rapid experimental verification of the developed topologies. The presented methods are discussed in terms of the beampatterns and sparse versions of the fully populated array topologies are be presented. The effect of apodization applied to the array elements is also investigated. Both simulated and experimental results are included.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Efficient algorithm for locating and sizing series compensation devices in large power transmission grids: I. Model implementation

We explore optimization methods for planning the placement, sizing and operations of Flexible Alternating Current Transmission System (FACTS) devices installed to relieve transmission grid congestion. We limit our selection of FACTS devices to Series Compensation (SC) devices that can be represented by modification of the inductance of transmission lines. Our master optimization problem minimizes the l 1 norm of the inductance modification subject to the usual line thermal-limit constraints. We develop heuristics that reduce this non-convex optimization to a succession of Linear Programs (LP) which are accelerated further using cutting plane methods. The algorithm solves an instance of the MatPower Polish Grid model (3299 lines and 2746 nodes) in 40 seconds per iteration on a standard laptop—a speed up that allows the sizing and placement of a family of SC devices to correct a large set of anticipated congestions. We observe that our algorithm finds feasible solutions that are always sparse, i.e., SC devices are placed on only a few lines. In a companion manuscript, we demonstrate our approach on realistically-sized networks that suffer congestion from a range of causes including generator retirement. In this manuscript, we focus on the development of our approach, investigate its structure on a small test system subject to congestion from uniform load growth, and demonstrate computational efficiency on a realistically-sized network.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Efficient Algorithm for Locating and Sizing Series Compensation Devices in Large Transmission Grids: Model Implementation (PART 1)

We explore optimization methods for planning the placement, sizing and operations of Flexible Alternating Current Transmission System (FACTS) devices installed to relieve transmission grid congestion. We limit our selection of FACTS devices to Series Compensation (SC) devices that can be represented by modification of the inductance of transmission lines. Our master optimization problem minimizes the l 1 norm of the inductance modification subject to the usual line thermal-limit constraints. We develop heuristics that reduce this non-convex optimization to a succession of Linear Programs (LP) which are accelerated further using cutting plane methods. The algorithm solves an instance of the MatPower Polish Grid model (3299 lines and 2746 nodes) in 40 seconds per iteration on a standard laptop—a speed up that allows the sizing and placement of a family of SC devices to correct a large set of anticipated congestions. We observe that our algorithm finds feasible solutions that are always sparse, i.e., SC devices are placed on only a few lines. In a companion manuscript, we demonstrate our approach on realistically-sized networks that suffer congestion from a range of causes including generator retirement. In this manuscript, we focus on the development of our approach, investigate its structure on a small test system subject to congestion from uniform load growth, and demonstrate computational efficiency on a realistically-sized network.

25 ENERGY STORAGE↗

All-purpose, all-sky photometric redshifts for the Legacy Imaging Surveys Data Release 8

ABSTRACT In this paper, we present photometric redshift (photo-z) estimates for the Dark Energy Spectroscopic Instrument (DESI) Legacy Imaging Surveys, currently the most sensitive optical survey covering the majority of the extragalactic sky. Our photo-z methodology is based on a machine-learning approach, using sparse Gaussian processes augmented with Gaussian mixture models (GMMs) that allow regions of parameter space to be identified and trained separately in a purely data-driven way. The same GMMs are also used to calculate cost-sensitive learning weights that mitigate biases in the spectroscopic training sample. By design, this approach aims to produce reliable and unbiased predictions for all parts of the parameter space present in wide area surveys. Compared to previous literature estimates using the same underlying photometry, our photo-zs are significantly less biased and more accurate at z > 1, with negligible loss in precision or reliability for resolved galaxies at z < 1. Our photo-z estimates offer accurate predictions for rare high-value populations within the parent sample, including optically selected quasars at the highest redshifts (z > 6), as well as X-ray or radio continuum selected populations across a broad range of flux (densities) and redshift. Deriving photo-z estimates for the full Legacy Imaging Surveys Data Release 8, the catalogues provided in this work offer photo-z estimates predicted to be of high quality for ≳9 × 108 galaxies over ${\sim}19\, 400\, \text{deg}^{2}$ and spanning 0 < z ≲ 7, offering one of the most extensive samples of redshift estimates ever produced.

79 ASTRONOMY AND ASTROPHYSICS↗

Isotope ratio characteristics and sensitivity for uranium determinations using a liquid sampling-atmospheric pressure glow discharge ion source coupled to an Orbitrap mass analyzer

Abstract The continued development of the liquid sampling-atmospheric pressure glow discharge (LS-APGD) microplasma as an ion source for diverse, elemental/isotopic analysis applications continues. To this end, characterization of the capabilities in performing precise and accurate isotope ratio (IR) determinations is essential. Based on past experience with the Thermo Exactive Orbitrap mass analyzer, the LS-APGD was interfaced with this instrument for these tests. While the Orbitrap platform has demonstrated excellent mass resolution and accuracy in “organic” mass spectrometry (MS) applications, work using an Orbitrap for IR analysis is very sparse. These efforts build off previous work in this coupling, where the importance of a few of the LS-APGD discharge parameters and Orbitrap data acquisition methods on IR precision and accuracy were probed. Presented here are the results of a study that evaluated the analytical precision for natural uranium sample (assumed 235U/238U = 0.0072) determinations. The instrumental parameters evaluated include the number of microscans and scans making up a data acquisition set, uranium concentration/signal level, sample make-up, and Fourier transform digitization window. Ultimately, a precision of 0.41% relative standard deviation (RSD) can be achieved for a single determination, with a reproducibility of 1.63 %RSD over 10 separate analytical measurements. A preliminary study of matrix effects on IR measurements of U is presented, highlighting the importance of pre-mass selection before injection into the Orbitrap. The analytical system sensitivity is suggested with the ability to produce a calibration function having an R2 value of >0.99 over a range of 4 orders of magnitude of concentration (~1 – 1000 ng mL-1). These efforts demonstrate the very promising pairing of the LS-APGD ionization source and the Orbitrap, pointing as well to definitive paths forward to better utilize both components in high quality isotope ratio mass spectrometry (IRMS).

Environmental Molecular Sciences Laboratory↗

Expediting field-effect transistor chemical sensor design with neuromorphic spiking graph neural networks

Improving the sensitive and selective detection of analytes in a variety of applications requires accelerating the rational design of field-effect transistor (FET) chemical sensors. Achieving high-performance detection relies on identifying optimal probe materials that can effectively interact with target analytes, a process traditionally driven by chemical intuition and time-consuming trial-and-error methods. To address the difficulties in probe screening for FET sensor development, this work presents a methodology that combines neuromorphic machine learning (ML) architectures, specifically a hybrid spiking graph neural network (SGNN), with an enriched dataset of physicochemical properties through semi-automated data extraction using large language models. Achieving a classification accuracy of 0.89 in predicting sensor sensitivity categories, the SGNN model outperformed traditional ML techniques by leveraging its ability to capture both global physicochemical properties and sparse topological features through a hybrid modeling framework. Next-generation sensor design was informed by the actionable insights into the connections between material properties and sensing performance offered by the SGNN framework. Through virtual screening for the detection of per- and polyfluoroalkyl substances (PFAS) as a use case, the effectiveness of the SGNN model was further validated. Density functional theory simulations confirmed graphene as a promising active material for PFAS detection as suggested by the SGNN framework. By bridging gaps in predictive modeling and data availability, this integrated approach provides a strong foundation for accelerating advancements in FET sensor design and innovation.

Ferreira, Rodrigo Pires [Univ. of Chicago, IL (Uni↗

SU-D-201-03: Imaging Cellular Pharmacokinetics of 18F-FDG in Inflammatory/Stem Cells

Purpose: Atherosclerosis is a progressive inflammatory condition that underlies coronary artery disease (CAD)—the leading cause of death in the USA. Thus, understating the metabolism of inflammatory cells can be a valuable tool for investigating CAD. To the best of our knowledge, we are the first to successfully investigate the pharmacokinetics of [18F]fluoro-deoxyglucose (18F-FDG) uptake in a single macrophages and compared with induced pluripotent stem cells (iPSCs) and mesenchymal stem cells (MSCs) with a novel imaging technique, radioluminescence microscopy, initially developed for cancer imaging. Methods: Live cells were cultured sparsely on Matrigel in a glass-bottom dish and starved for 1 hour before incubation with 250 microCi of 18F-FDG for 45 minutes. Excess radiotracer was removed using DMEM medium without glucose. Before imaging, DMEM (1 mL) was added to the cell culture and a 100 microm-thin CdWO4 scintillator plate was placed on top of the cells. Light produced following beta decay was imaged with a highly sensitive inverted microscope (LV200, Olympus) fitted with a 40x/1.3 high-NA oil objective, and an EM-CCD camera. The images were collected over 18,000 frames with 4×4 binning (1200 MHz EM Gain, 300ms exposure). Custom-written software was developed in MATLAB for image processing (Figure 1). For statistical analysis 10 different region-of-interests (ROIs) were selected for each cell type. Results: Figures 2A–2B show bright-field/fusion images for all three different cell types. The relationship between cell-to-cell comparisons was found to be linear for macrophages unlike iPSCs and MSCs, which were best fitted with moving or rolling average (Figure 2C). The average observed decay of 18F-FDG in a single cell of MSCs per second (0.067) was 20% and 36% higher compared to iPSCs (0.054) and macrophages (0.043), respectively (Figure 2D). Conclusion: MSCs was found to be 2–3x more sensitive to glucose molecule despite constant parameters for each cell type examined.

60 APPLIED LIFE SCIENCES↗

Application of Reconfigurable Computing Technology to Multi-KiloHertz Micro-Laser Altimeter (MMLA) Data Processing

The Multi-KiloHertz Micro-Laser Altimeter (MMLA) is an aircraft based instrument developed by NASA Goddard Space Flight Center with several potential spaceflight applications. This presentation describes how reconfigurable computing technology was employed to perform MMLA signal extraction in real-time under realistic operating constraints. The MMLA is a "single-photon-counting" airborne laser altimeter that is used to measure land surface features such as topography and vegetation canopy height. This instrument has to date flown a number of times aboard the NASA P3 aircraft acquiring data at a number of sites in the Mid-Atlantic region. This instrument pulses a relatively low-powered laser at a very high rate (10 kHz) and then measures the time-of-flight of discrete returns from the target surface. The instrument then bins these measurements into a two-dimensional array (vertical height vs. horizontal ground track) and selects the most likely signal path through the array. Return data that does not correspond to the selected signal path are classified as noise returns and are then discarded. The MMLA signal extraction algorithm is very compute intensive in that a score must be computed for every possible path through the two dimensional array in order to select the most likely signal path. Given a typical array size with 50 x 6, up to 33 arrays must be processed per second. And for each of these arrays, roughly 12,000 individual paths must be scored. Furthermore, the number of paths increases exponentially with the horizontal size of the array, and linearly with the vertical size. Yet, increasing the horizontal and vertical sizes of the array offer science advantages such as improved range, resolution, and noise rejection. Due to the volume of return data and the compute intensive signal extraction algorithm, the existing PC-based MMLA data system has been unable to perform signal extraction in real-time unless the array is limited in size to one column, This limits the ability of the MMLA to operate in environments with sparse signal returns and a high number of noise return. However, under an IR&D project, an FPGA-based, reconfigurable computing data system has been developed that has been demonstrated to perform real-time signal extraction under realistic operating constraints. This reconfigurable data system is based on the commercially available Firebird Board from Annapolis Microsystems. This PCI board consists of a Xilinx Virtex 2000E FPGA along with 36 MB of SRAM arranged in five separately addressable banks. This board is housed in a rackmount PC with dual 850MHz Pentium processors running the Windows 2000 operating system. This data system performs all signal extraction in hardware on the Firebird, but also runs the existing "software based" signal extraction in tandem for comparison purposes. Using a relatively small amount of the Virtex XCV2000E resources, the reconfigurable data system has demonstrated to improve performance improvement over the existing software based data system by an order of magnitude. Performance could be further improved by employing parallelism. Ground testing and a preliminary engineering test flight aboard the NASA P3 has been performed, during which the reconfigurable data system has been demonstrated to match the results of the existing data system.

Powell, Wesley↗

Rarefied solids

One important limit to creating low density materials is the objects' own weight. As a solid or colloidal matrix becomes more rarefied, gravity acts destructively to compress its suporting skeleton. We describe experimental results and propose a model which matches the low gravity behavior of rarefied or fractal solids. On parabolic airplane flights, we sought to demonstrate a key component of producing higher surface area fractals. Flight paths were selected to give a range of gravity levels: 0.01 g/g(sub 0) (low), 0.16 g(sub 0) (Lunar), 0.33 g/g(sub 0) (Martian), 1 g/g(sub 0) (Earth) and 1.8 g/g(sub 0) (high) (where g(sub 0) = 980 cm/sq s). Results using the model material of hydrophobic silica indicated that stable agglomeration of such tenuous objects can increase markedly in reduced gravity. Optical characterization revealed that fractal dimension changed directly with varying gravity. As measured by fractal dimension, effective surface area and roughness increased by 40% in low gravity. This finding supports the conclusion that relieving internal weight stresses on delicate aggregates can enhance their overall size (by two orders of magnitude) and internal surface area. We conclude that gravitational restructuring limits the overall size and void content of low-density solids. These sparse colloidal regimes may present new and technologically attractive physics, ranging from improved insulators, liquid-like tension in a 'solid' matrix, and characteristically low conductivities for sound and (8 to 14 micrometers wavelength) infrared radiation.

Noever, D. A.↗

A preliminary assessment of the accuracy of selected meteorological parameters determined from Nimbus 6 satellite profile data

Published rms errors in rawinsonde data and discrepancies between satellite and rawinsonde profile data for temperature, dewpoint temperature, mixing ratio, and wind speed. Satellite rms errors were found to be 2 to 3 times as large as those for rawinsonde data. Gradients of the preceding parameters were computed for both rawinsonde and satellite data and compared with means and near extreme values computed from the AVE 2 and AVE 4 experiments. In all cases, it was found that satellite data can be used to determine with relatively good accuracy the near extreme gradients but not those whose value does not exceed the average. Synoptic charts were prepared to show that patterns of temperature could be determined with relatively good accuracy, while those of dew point were not as good as those for temperature. Winds represented by cloud motion vectors (satellite winds) were compared with rawinsonde winds, and it was found that large gaps exist in satellite values for a given pressure level and that errors in the satellite determined concluded that satellite profile data are very useful in synoptic analysis, particularly in data sparse regions as well as regions where near extreme gradients exist in the measured parameters.

Scoggins, J. R.↗

Biodosimetry of ionizing radiation by selective painting of prematurely condensed chromosomes in human lymphocytes

Painting of interphase chromosomes can be useful for biodosimetric purposes in particular cases such as radiation therapy, accidental exposure to very high radiation doses and exposure to densely ionizing radiation, for example during space missions. Biodosimetry of charged-particle radiation is analyzed in the present paper. Target cells were human peripheral blood lymphocytes irradiated in vitro with gamma rays, protons and iron ions. After exposure, lymphocytes were incubated for different times to allow repair of radiation-induced damage and then fused to mitotic hamster cells to promote premature condensation in the interphase chromosomes. Chromosome spreads were then hybridized with whole-chromosome DNA probes labeled with fluorescent stains. Dose-response curves for the induction of chromatin fragments shortly after exposure, as well as the kinetics of rejoining and misrejoining, were not markedly dependent on linear energy transfer. However, after exposure to heavy ions, more aberrations were scored in the interphase cells after incubation for repair than in metaphase samples harvested at the first postirradiation mitosis. On the other hand, no significant differences were observed in the two samples after exposure to sparsely ionizing radiation. These results suggest that interphase chromosome painting can be a useful tool for biodosimetry of particle radiation.

NASA Center JSC↗

The terminal area simulation system. Volume 2: Verification cases

The numerical simulation of five case studies are presented and are compared with available data in order to verify the three-dimensional version of the Terminal Area Simulation System (TASS). A spectrum of convective storm types are selected for the case studies. Included are: a High-Plains supercell hailstorm, a small and relatively short-lived High-Plains cumulonimbus, a convective storm which produced the 2 August 1985 DFW microburst, a South Florida convective complex, and a tornadic Oklahoma thunderstorm. For each of the cases the model results compared reasonably well with observed data. In the simulations of the supercell storms many of their characteristic features were modeled, such as the hook echo, BWER, mesocyclone, gust fronts, giant persistent updraft, wall cloud, flanking-line towers, anvil and radar reflectivity overhang, and rightward veering in the storm propagation. In the simulation of the tornadic storm a horseshoe-shaped updraft configuration and cyclic changes in storm intensity and structure were noted. The simulation of the DFW microburst agreed remarkably well with sparse observed data. The simulated outflow rapidly expanded in a nearly symmetrical pattern and was associated with a ringvortex. A South Florida convective complex was simulated and contained updrafts and downdrafts in the form of discrete bubbles. The numerical simulations, in all cases, always remained stable and bounded with no anomalous trends.

Proctor, F. H.↗

Optical Characterisation of Suspended Particles in the Mackenzie River Plume (Canadian Arctic Ocean) and Implications for Ocean Colour Remote Sensing

Climate change significantly impacts Arctic shelf regions in terms of air temperature, ultraviolet radiation, melting of sea ice, precipitation, thawing of permafrost and coastal erosion. Direct consequences have been observed on the increasing Arctic river flow and a large amount of organic carbon sequestered in soils at high latitudes since the last glacial maximum can be expected to be delivered to the Arctic Ocean during the coming decade. Monitoring the fluxes and fate of this terrigenous organic carbon is problematic in such sparsely populated regions unless remote sensing techniques can be developed and proved to be operational. The main objective of this study is to develop an ocean colour algorithm to operationally monitor dynamics of suspended particulate matter (SPM) on the Mackenzie River continental shelf (Canadian Arctic Ocean) using satellite imagery. The water optical properties are documented across the study area and related to concentrations of SPM and particulate organic carbon (POC). Robust SPM and POC : SPM proxies are identified, such as the light backscattering and attenuation coefficients, and relationships are established between these optical and biogeochemical parameters. Following a semi-analytical approach, a regional SPM quantification relationship is obtained for the inversion of the water reflectance signal into SPM concentration. This relationship is reproduced based on independent field optical measurements. It is successfully applied to a selection of MODIS satellite data which allow estimating fluxes at the river mouth and monitoring the extension and dynamics of the Mackenzie River surface plume in 2009, 2010 and 2011. Good agreement is obtained with field observations representative of the whole water column in the river delta zone where terrigenous SPM is mainly constrained (out of short periods of maximum river outflow). Most of the seaward export of SPM is observed to occur within the west side of the river mouth. Future work will require the validation of the developed SPM regional algorithm based on match-ups with field measurements, then the routine application to ocean colour satellite data in order to better estimate the fluxes and fate of SPM and POC delivered by the Mackenzie River to the Arctic Ocean.

Mackenzie River plume↗

HARMONY: Large-Scale Architecture Search for Efficient Hybrid Language Models

As large language models scale to trillions of parameters, their computational and memory requirements present critical challenges for efficient training and deployment. While Mixture of Experts (MoE) architectures enable efficient scaling through sparse parameter activation, and state-space models like Mamba offer linear-time complexity, principled methods for combining these paradigms remain undeveloped. We introduce HARMONY (Hybrid Architecture Research for Mamba, Optimized with Neural efficiencY), a multi-objective evolutionary neural architecture search framework for discovering efficient hybrid language models that integrate Transformer attention mechanisms, Mixture-of-Experts routing, and Mamba state-space components. Through large-scale distributed search using 16,384 MI250X GPUs on the Frontier supercomputer, HARMONY explores a comprehensive design space encompassing six attention variants (MHA, MQA, GQA, MLA, SWA, and Mamba-2), variable MoE configurations with both routed and shared experts, and extensive Mamba hyperparameters. Our framework discovers heterogeneous architectures that balance training performance with computational efficiency through multi-objective optimization incorporating latency penalties and fitness-based selection. Analysis of discovered architectures reveals that optimal hybrid designs favor heterogeneous component mixing rather than homogeneous patterns, with Mamba-2 and Multi-Head Latent Attention (MLA) emerging as preferred mechanisms. Discovered architectures demonstrate superior training efficiency: our best configuration achieves a final perplexity of 1.0874 with 2.38B parameters while processing 4,320 tokens/second, outperforming significantly larger manually designed models. Full-scale evaluation shows HARMONY's top architectures achieve better loss trajectories than equivalently-sized models using state-of-the-art configurations including Mixtral, Jamba, and Samba. Additionally, we demonstrate 91% weak scaling efficiency when training discovered 36B-parameter models across 1,024 GPUs. HARMONY is released as an open framework with comprehensive tools for building and training hybrid models using expert-data-pipeline parallelism, democratizing access to automated architecture design for next-generation language models.

Herron, Emily [ORNL] (ORCID:0000000273008172)↗

Complex Sources and Impacts on Remediation and Monitoring Approaches

Remedy selection, implementation, and coordination with tank farm activities for adjacent waste sites need to be facilitated by a technically defensible description of sources and their relative strength and temporal contributions to groundwater contamination. Detailed data describing contaminant source releases are typically not available, so source locations are highly uncertain. Furthermore, data on sources and model parameters are not readily available through direct measurements. Instead, source data are often derived by extrapolation from a few direct and/or indirect measurements, while model parameters are commonly estimated using inverse modeling or calibration methods. Because the conceptual model of sources will influence the mass and timing of the flux of contaminants to groundwater, uncertainty in the source terms should be considered. Even though a best-fit model can often be identified, it may not be the best representation of the sources given the sparse, uncertain nature of the data used for estimating source terms. To support future feasibility studies, a systematic approach is needed for identifying remediation and monitoring activities appropriate under conditions of uncertainty. In this study, an integrated assessment of alternative conceptual model elements has been examined using the B-Complex site in the Hanford Central Plateau to demonstrate how multiple sources in a waste management complex can be described to identify their relative contributions to groundwater. The heterogeneous structure of the system is modeled based on available field information on the lithological data for both the vadose zone and aquifer. The approach includes a simulation assessment and linkages to site data, with an uncertainty analysis applied to evaluate the effects of source term uncertainty on contaminant mass flux to groundwater. An integrated assessment of alternative conceptual model elements was evaluated to identify how parameter uncertainty impacts both temporal and spatial uncertainties associated with vadose zone source terms. Modeling scenarios identify important characteristics of remediation target zones to provide context for how surface barriers, in situ remedies, and water-table treatment zones can be designed to account for source dynamics. Ramifications for remediation and monitoring are discussed within the context of the source term uncertainty. The described approach facilitates systematically addressing alternative conceptual model questions in relation to remediation and monitoring and supports the remedial investigation/ feasibility study process and the remediation strategy development for the Hanford Central Plateau.

Rockhold, Mark L.↗