Search NASA⌕ Search

SEARCH · Search NASA

Results for “scalable architecture”

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 55 records · Page 3
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Analog-to-digital converter based on voltage-controlled superconducting devices

The increasing demand for cryogenic electronics in superconducting and quantum computing systems calls for ultra-energy-efficient data conversion architectures that remain functional at deep cryogenic temperatures. Here, in this work, we present the first design of a voltage-controlled superconducting flash analog-to-digital converter (ADC) based on a voltage-controlled quantum-enhanced Josephson junction field-effect transistor (JJFET). Exploiting its strong gate tunability and transistor-like behavior, the JJFET offers a scalable alternative to conventional current-controlled superconducting devices while aligning naturally with CMOS-style design methodologies. Building on our previously developed Verilog-A compact model calibrated to experimental data, we design and simulate a three-bit JJFET-based flash ADC targeted for integration within cryogenic control and readout circuitry in quantum computing. The core comparator block is realized through careful bias current selection and augmented with a three-terminal nanocryotron to precisely define reference voltages. Cascaded JJFET comparators ensure robust voltage gain, cascadability, and logic-level restoration across stages. Simulation results demonstrate accurate quantization behavior with ultra-low power dissipation, underscoring the feasibility of voltage-driven superconducting mixed-signal circuits. This work establishes a critical step toward unifying superconducting logic and data conversion, paving the way for scalable cryogenic architectures in quantum–classical co-processors, low-power artificial intelligence accelerators, and next-generation energy-constrained computing platforms.

Analog-to-digital converter↗
DOE OSTI2025

Proton Selective Nanoporous Atomically Thin Graphene Membranes for Vanadium Redox Flow Batteries

Angstrom-scale proton-selective pores in atomically thin 2D materials present fundamentally new opportunities for advancing proton exchange membranes (PEMs). Vanadium Redox Flow Batteries (VRFBs) for grid-scale energy storage require PEMs with high areal proton conductance (>1 S cm −2 ) and minimal vanadium ion (VO 2+ ) crossover. However, state-of-the-art Nafion 212 membranes (N212 ≈50 µm thick), suffer from persistent VO 2+ crossover reducing performance and efficiency. Here, a layered PEM is demonstrated, comprising monolayer CVD graphene with Angstrom-scale proton-selective pores introduced via Ar plasma, integrated with an ultra-thin ≈300 nm polybenzimidazole (PBI) layer and sandwiched between two Nafion 211 (25 µm thick) layers. The layered architecture facilitates scalable membrane fabrication by mitigating defects while processing and facile stacking of graphene layers allows stochastic non-selective defect isolation enabling exceptionally low VO 2+ crossover (selectivity (H + areal conductance / VO 2+ permeability) ≈6709 × 10 6 S min cm −4 ), with proton conductance >8 S cm −2 . Systematic transport experiments supported by resistance-based transport modelling elucidate the role of defect size, defect isolation, and sealing, as well as layering/stacking, to enable orders of magnitude (>671× over N212) improvements in selectivity, along with areal proton conductance >8 S cm −2 . This work highlights the potential of atomic-scale proton-selective defect engineering in 2D materials, in conjunction with facile stacking and layering of materials as strategies for scalable, high-performance advances in PEMs for energy, electrochemical, and separation applications beyond VRFBs.

Chaturvedi, Pavan [Vanderbilt Univ., Nashville, TN↗
DOE OSTI2020

Fouling modeling and prediction approach for heat exchangers using deep learning

In this article, we develop a generalized and scalable statistical model for accurate prediction of fouling resistance using commonly measured parameters of industrial heat exchangers. This prediction model is based on deep learning where a scalable algorithmic architecture learns non-linear functional relationships between a set of target and predictor variables from large number of training samples. Here, the efficacy of this modeling approach is demonstrated for predicting fouling in an analytically modeled cross-flow heat exchanger, designed for waste heat recovery from flue-gas using room temperature water. The performance results of the trained models demonstrate that the mean absolute prediction errors are under 10 –4 KW –1 for flue-gas side, water side and overall fouling resistances. The coefficients of determination (R 2 ), which characterize the goodness of fit between the predictions and observed data, are over 99%. Even under varying levels of measurement noise in the inputs, we demonstrate that predictions over an ensemble of multiple neural networks achieves better accuracy and robustness to noise. We find that the proposed deep-learning fouling prediction framework learns to follow heat exchanger flow and heat transfer physics, which we confirm using locally interpretable model agnostic explanations around randomly selected operating points. Overall, we provide a robust algorithmic framework for fouling prediction that can be generalized and scaled to various types of industrial heat exchangers.

42 ENGINEERING↗
DOE OSTI2020

A Robust, Scalable Platform for the Electrochemical Conversion of CO 2 to Formate: Identifying Pathways to Higher Energy Efficiencies

This work demonstrated a robust, scalable cell architecture for electroreduction of CO 2 (CO 2 R). An up to 90% faradaic efficiency for the conversion of CO 2 R to formate at 500 mA/cm 2 was realized at a 25 cm2 gas diffusion electrode (GDE) with a carbon-supported SnO 2 electrocatalyst. Furthermore, a 1.27 mm thick catholyte was used between the bipolar membrane and cathode GDE, which could be further reduced to tens of micrometers upon refinement. The deconvolution of the potential drop from each individual component/process guides the pathways to higher energy efficiencies of CO 2 R at this platform. Significant changes in the agglomerate size and aspect ratio on the electrode before and after an 11 h test were revealed by nano-CT, suggesting reduced CO 2 accessibility from electrode degradation. The versatility of this CO 2 R testing platform enables the ability to assess materials, components, and interactions at scales more in line with future devices.

30 DIRECT ENERGY CONVERSION↗
DOE OSTI2023

Vertical organic electrochemical transistors for complementary circuits

Abstract Organic electrochemical transistors (OECTs) and OECT-based circuitry offer great potential in bioelectronics, wearable electronics and artificial neuromorphic electronics because of their exceptionally low driving voltages (<1 V), low power consumption (<1 µW), high transconductances (>10 mS) and biocompatibility 1–5 . However, the successful realization of critical complementary logic OECTs is currently limited by temporal and/or operational instability, slow redox processes and/or switching, incompatibility with high-density monolithic integration and inferior n-type OECT performance 6–8 . Here we demonstrate p- and n-type vertical OECTs with balanced and ultra-high performance by blending redox-active semiconducting polymers with a redox-inactive photocurable and/or photopatternable polymer to form an ion-permeable semiconducting channel, implemented in a simple, scalable vertical architecture that has a dense, impermeable top contact. Footprint current densities exceeding 1 kA cm −2 at less than ±0.7 V, transconductances of 0.2–0.4 S, short transient times of less than 1 ms and ultra-stable switching (>50,000 cycles) are achieved in, to our knowledge, the first vertically stacked complementary vertical OECT logic circuits. This architecture opens many possibilities for fundamental studies of organic semiconductor redox chemistry and physics in nanoscopically confined spaces, without macroscopic electrolyte contact, as well as wearable and implantable device applications.

Science & Technology - Other Topics↗
DOE OSTI2023

Theory of overparametrization in quantum neural networks

The prospect of achieving quantum advantage with quantum neural networks (QNNs) is exciting. Understanding how QNN properties (for example, the number of parameters $M$) affect the loss landscape is crucial to designing scalable QNN architectures. Here we rigorously analyze the overparametrization phenomenon in QNNs, defining overparametrization as the regime where the QNN has more than a critical number of parameters $M_c$ allowing it to explore all relevant directions in state space. In this study, our main results show that the dimension of the Lie algebra obtained from the generators of the QNN is an upper bound for $M_c$, and for the maximal rank that the quantum Fisher information and Hessian matrices can reach. Underparametrized QNNs have spurious local minima in the loss landscape that start disappearing when $M$ ≥ $M_c$. Thus, the overparametrization onset corresponds to a computational phase transition where the QNN trainability is greatly improved. We then connect the notion of overparametrization to the QNN capacity, so that when a QNN is overparametrized, its capacity achieves its maximum possible value.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Colloidal quantum dot based infrared detectors: extending to the mid-infrared and moving from the lab to the field

Quantum dots (QDs) that absorb in the mid-wave infrared (MWIR) regime (3–5 μm) have recently generated significant interest as possible detector materials for MWIR cameras, with promises to reduce materials and device fabrication costs and potentially increase device operating temperatures. However, these materials have been primarily explored in single pixel devices – not multi-pixel cameras – and their current performance lags behind that of commercially available MWIR cameras based on single crystalline materials. To realize the potential of QD MWIR photodetectors with imaging capabilities, improvements are needed in the quality of the MWIR-absorbing QD materials, ligands dictating inter-QD charge transport, ordering of QD films, device architectures, and scalability of these methods to larger areas. With three families of QDs being researched – lead-, silver-, and mercury-based chalcogenides – as well as a multitude of possible capping ligands and film deposition techniques, the experimental phase space for MWIR QDs is vast. Here, we provide a roadmap, considering the pros and cons of various film deposition and ligand exchange techniques, as well as reintroduce lessons learned over years of research on QD film formation. We also extracted and created a database of reported quantum dot photodetectors’ performance and fabrication methods and have developed an interactive data visualization dashboard for this database, which provides researchers in the field a quick snapshot of the existing state of the art. Finally, we outline figures of merit and information that should be presented in papers moving forward which would help to clarify the reported results, as well as offer ideas for future steps. We provide a database visualization of the QD photodetector literature at https://public.tableau.com/app/profile/tom.nakotte/viz/IRQDphotodetectors_16384709473480/Dashboard1#1.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Unraveling spin entanglement using quantum gates with scanning tunneling microscopy-driven electron spin resonance

Quantum entanglement is a fundamental resource for quantum information processing, and its controlled generation and detection remain key challenges in scalable quantum architectures. Here, we numerically demonstrate the deterministic generation of entangled spin states in a solid-state platform by implementing quantum gates via electron spin resonance combined with scanning tunneling microscopy (ESR-STM). Using two titanium atoms on a MgO/Ag(100) substrate as a model, we construct a two-qubit system whose dynamics are coherently manipulated through tailored microwave pulse sequences. We generate Bell states by implementing a Hadamard gate followed by a controlled-NOT gate, and evaluate its fidelity and concurrence using the quantum-master equation-based code TimeESR. Our results demonstrate that ESR-STM can create entangled states with significant fidelity. This study paves the way for the realization of atom-based quantum circuits and highlights ESR-STM as a powerful tool for probing and engineering entangled states on surfaces.

Switzer, Eric D. [Donostia International Physics C↗
DOE OSTI2023

Densified vertically lamellar electrode architectures for compact energy storage

As one of the most compact electrochemical energy storage systems, lithium-ion batteries (LIBs) are playing an indispensable role in the process of vehicle electrification to accelerate the shift to sustainable mobility. Making battery electrodes thicker is a promising strategy for improving the energy density of LIBs which is essential for applications with weight or volume constraints, such as electric-powered transportation; however, their power densities are often significantly restricted due to elongated and tortuous charge traveling distances. Here, we propose an effective methodology that couples bidirectional freeze-casting and compression-induced densification to create densified vertically lamellar electrode architectures for compact energy storage. The vertically lamellar architectures not only overcome the critical thickness limit for conventional electrodes but also facilitate and redistribute the lithium-ion flux enabling both high rate capability and stable cyclability. Furthermore, this proposed methodology is universal as demonstrated in various electrochemical active material systems. This study offers a facile approach that realizes simultaneous high energy and high power in high-loading battery electrodes and provides useful rationales in designing electrode architectures for scalable energy storage systems.

42 ENGINEERING↗
DOE OSTI2025

HydroForecast Long-term: Improving hydropower’s resilience to climate change through accurate climate-scale

With hydrologic patterns and water availability across the globe shifting due to climate change, advancements in hydrologic prediction systems can help significantly reduce the uncertainties that utilities and water supply entities have in their decision making. Understanding and estimating hydrology at the climate scale is critical for managing water resources under changing climate scenarios. This project focuses on integrating state-of-the-art neural network modeling with downscaled climate projections to deliver the reliable water supply projections decades into the future to meet an urgent need from hydropower operators and water utilities. In this Phase 1 DOE SBIR proposal, we developed and validated a theory-guided neural network model, HydroForecast Long-term, for climate-scale hydrology and implemented the model within existing HydroForecast infrastructure. HydroForecast Long-term combines the most accurate streamflow modeling system with a flexible and scalable data architecture to generate water supply projections out to the year 2100. This report illustrates that we have achieved our four objectives: 1) create a prototype of HydroForecast Long-term, building the neural network prediction model, 2) build an automated data input pipeline that processes large amounts of data from the latest global temperature and precipitation climate models; 3) benchmark the accuracy of the hydrologic model over the recent two decades over a large set of diverse basins, and 4) create a set of output visuals and summary metrics informed by customer feedback that connect the data to critical decision points. This work empowers water users to make data-informed decisions supporting a resilient, renewable-powered grid and water system. The results advance the Department of Energy’s mission by addressing critical gaps in water supply planning under climate change.

13 HYDRO ENERGY↗
DOE OSTI2025

Ion Traps and Packaging for Heterogenous Integration - Chimera

Microfabricated surface ion traps and silicon-based photonics are critical technologies for scaling quantum systems. Current ion trap architectures face scalability and integration challenges due to limitations in optical access, fabrication techniques, and material compatibility. State-of-the-art quantum computers and atomic clocks are investigating monolithic integration, which necessitates custom traps for each ion species and has not overcome the integration hurdles presented by merging these technologies. The Chimera (Ion Traps and Packaging for Heterogeneous Integration) project proposes a novel approach utilizing heterogeneous integration (HI) of ion traps and photonic circuits. This separation of components allows for flexibility in ion trap design and reduces fabrication compromises. The Chimera project specifically designed an ion trap to interface vertically with a separately fabricated waveguide chip and demonstrates the first steps to integrating them at the packaging level. The ion trap features a large area of removed silicon, allowing the photonics chip outputs closer to the ion trap, improving alignment and packaging processes. The alignment must be accurate to < 1 µm to ensure that the light from the waveguide can overlap with the trapping region. This fine alignment must also be maintained through an ultra-high vacuum bake, a critical step in preparing an ion trap experiment. By combining separate chips, we demonstrate a new path for scaling trapped ion technology that is less reliant on monolithic integration. We successfully fabricated a trap with a large area of oxide removed, resulting in a region thinned to about 40 µm, a key milestone toward successful integration.

42 ENGINEERING↗
DOE OSTI2026

Organic Rankine Cycle Integration and Optimization for High Efficiency CHP Genset Systems (Final Technical Report)

This project successfully advanced the integration of Organic Rankine Cycle (ORC) technology with reciprocating engine–based combined heat and power (CHP) systems to improve electrical efficiency, total CHP efficiency, and grid-responsive operation. Over three budget periods, the work progressed from high-temperature ORC component development and thermodynamic model validation to next-generation system design, working fluid transition, and techno-economic analysis. Key technical accomplishments include development and validation of a thermodynamic model capable of accurately predicting ORC performance across an expanded temperature and pressure envelope; successful identification and validation of low-global-warming-potential (GWP) working fluids—most notably R1233zd(E)—as viable replacements for R245fa; and demonstration of scalable ORC architectures suitable for integration with 1–20 MW class reciprocating engines. These advances enable flexible CHP configurations that can increase electrical output while maintaining high overall utilization of available thermal energy. The project also produced a clean-sheet design for a next-generation ORC system targeting substantially higher power output per unit, supported by detailed component selection, heat exchanger evaluation, and system-level modeling. Techno-economic analyses indicate that ORC-enabled flexible CHP systems can meet or exceed Department of Energy (DOE) efficiency targets while providing value to both facility operators and the electric grid.. Late-stage testing of the largest next-generation ORC prototype identified limitations related to pump net positive suction head (NPSH) requirements and condenser flooding under certain operating conditions. Although these issues constrained full validation of that configuration within the project timeframe, they provided clear and actionable design guidance for future system refinements. Importantly, validated modeling, smaller-scale testing, and working fluid evaluations confirmed the technical viability of the overall approach. In aggregate, this project met its core objectives by establishing validated design tools, de-risking key ORC technologies for CHP applications, and defining a credible pathway toward commercialization of flexible, high-efficiency CHP systems. The results form a strong foundation for continued development and deployment beyond the conclusion of the DOE-funded effort.

20 FOSSIL-FUELED POWER PLANTS↗
DOE OSTI2022

Scalable Nano-Scaffold SOFC Anode Architecture Enabling Direct Hydrocarbon Utilization

This project is based on WVU’s pending patents, technology and aims to design and modify the internal surfaces of the Ni/YSZ anode from currently commercially viable Solid Oxide Fuel Cells (SOFCs) using the additive manufacturing process of Atomic Layer Deposition (ALD). The surface architecture/scaffold added onto the internal surface of the anode possesses an engineered nanostructure but it features only commonly-used oxide conductors and electro-catalyst materials. The surface layer possesses a minimum thickness of ~2-40 nm and is solely designed to control the surface reforming reactions and to increase catalytic activity. Three-dimensional (3D) nano scaffold architectures with the noble metal nano-catalyst, low-cost bimetallic catalytic alloys, and nano-scale ionic conducting oxide fully compatible with the state-of-the-art Ni/YSZ anode, were applied to the internal surface of the entire porous SOFC anode using ALD. In the present work, the surface scaffold architecture is essentially multi-functional at the nano-scale, facilitated by the multiple heterostructured interfaces. It will significantly enhance the power density and cell durability for direct hydrocarbon utilization by (1) increasing the number of electrochemical reaction sites to enhance the hydrogen/hydrocarbon oxidation reactions; (2) reducing carbon formation; (3) mitigating the coarsening of backbone Ni phase and the oxidation attack of Ni from oxidants (e.g., H2O, CO2); and (4) promoting the internal reforming capabilities, especially for natural gas applications. ALD is employed to generate stable anode surface architectures that are uniform, precisely controllable at the atomic scale, and accurately repeatable for processing. The engineered anode surface nano-scaffold architecture was cataloged and analyzed using High-resolution Transmission Electron Microscopy (TEM), and cell power/durability performance assessed via comprehensive electrochemical performance testing with commercial specimens and relevant environments using hydrocarbon fuels. To the best of our knowledge, this project is the First Report on ALD of Ni/YSZ. The actual achievement of this Project includes (1). Successful demonstration of 7 types of ALD layers on Ni/YSZ anode, including Co, Ni, Mn, Pt, Ru, ZrOx and multi-functional nano-composite. (1). Conformal coating and subsequently spontaneously pinning the discrete nano-catalyst, including the precious metal nano-catalyst and the Ni and Co catalysts, on the YSZ surface upon the electrochemical operation in the reduced atmosphere. Those nano-catalysts on the ionic-conducting YSZ provided excellent sites for promoting internal reforming; (2). Demonstrated ALD coating increased both catalytic activity and conductivity of Ni/YSZ. Conformal coating provided dopants and introduced additional electrical conducting pathways on the YSZ ionic conductor. The doped surface layer of YSZ with mixed conductivity thus further introduces the active triple phase boundaries adjacent to the ALD-coated nano-catalysts such as Pt, Co, and Ni that are pinned on the YSZ surface. The nano-composite ALD coating on Ni/YSZ anode has significantly increased cell durability; and (3). ALD coating of Ni/YSZ anode increased the power density of the entire cell by 300%. For a long time, the SOFC performance, such as the power density, was deemed hindered by the cathode. The sluggish oxygen reduction reaction (ORR) in the cathode was deemed as hindering the power density of the SOFCs. For the anode-supported commercial SOFCs, the cell performance is considered to be limited by the cathode's performance. For the first time in the field of SOFC, this project has demonstrated that (1). the performance of commercial SOFCs can be further increased by the ALD coating on Ni/YSZ anode backbone. (2). ALD coating on Ni/YSZ fuel electrodes results in the enhancement of power density, and increased reliability, robustness, and endurance of SOFCs, for their application using both hydrogen and hydrocarbon fuels over the entire operating temperature range of 650-800ºC for the inherently functional commercial cells. (3). ALD coating provides alternative approaches of exsolutions for introducing the stable catalyst onto the internal surface of the Ni/YSZ electrode. ALD coating could be much more versatile than exsolution in employing the catalysts with various chemistries onto the various backbones. (4). Due to the negligible amount of ALD materials coated onto the internal surface of the porous cathode of the as-fabricated cells, a peak power density increase up to 300 % induced by ALD coating was simultaneously achieved in terms of both power density and specific power. (5). The ALD coating developed through this project was applied to both the SOFC and Solid Oxide Electrolysis Cells (SOEC). SOEC’s face a similar but more demanding need to improve the fuel electrode's performance. It opens further research directions for electrocatalytic surface nanoionics with a wide range of chemistry. It will revolutionize our ability to render the formation of a nanostructured electrode that has been constantly pursued yet barely achieved for practical SOFC/SOEC applications. The research is also immediately transformative since both the preliminary data and the proposed work are on the direct implantation of nanoionics into the state-of-the-art inherently functional SOCs. It represents an immediate impact on the commercial sectors in SOC technology since the applied ALD processing is computer-controlled ALD coating using the commercial ALD systems, and it is scalable to both the single cells and SOC stacks.

36 MATERIALS SCIENCE↗
DOE OSTI2019

Extreme-bandwidth scalable performance-per-watt GPU architecture

A technique for accessing memory in an accelerated processing device coupled to stacked memory dies is provided herein. The technique includes receiving a memory access request from an execution unit and identifying whether the memory access request corresponds to memory cells of the stacked dies that are considered local to the execution unit or non-local. For local accesses, the access is made “directly”, that is, without using a bus. A control die coordinates operations for such local accesses, activating particular through-silicon-vias associated with the memory cells that include the data for the access. Non-local accesses are made via a distributed cache fabric and an interconnect bus in the control die. Various other features and details are provided below.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Massively scalable Kerr comb-driven silicon photonic link

Abstract The growth of computing needs for artificial intelligence and machine learning is critically challenging data communications in today’s data-centre systems. Data movement, dominated by energy costs and limited ‘chip-escape’ bandwidth densities, is perhaps the singular factor determining the scalability of future systems. Using light to send information between compute nodes in such systems can dramatically increase the available bandwidth while simultaneously decreasing energy consumption. Through wavelength-division multiplexing with chip-based microresonator Kerr frequency combs, independent information channels can be encoded onto many distinct colours of light in the same optical fibre for massively parallel data transmission with low energy. Although previous high-bandwidth demonstrations have relied on benchtop equipment for filtering and modulating Kerr comb wavelength channels, data-centre interconnects require a compact on-chip form factor for these operations. Here we demonstrate a massively scalable chip-based silicon photonic data link using a Kerr comb source enabled by a new link architecture and experimentally show aggregate single-fibre data transmission of 512 Gb s −1 across 32 independent wavelength channels. The demonstrated architecture is fundamentally scalable to hundreds of wavelength channels, enabling massively parallel terabit-scale optical interconnects for future green hyperscale data centres.

Rizzo, Anthony (ORCID:000000034752797X)↗
DOE OSTI2021

Portage: A Modular Data Remap Library for Multiphysics Applications on Advanced Architectures

Portage is a scalable and extensible remap library for numerical simulations. It supports state-of-the-art remap schemes for meshes and particles in 2D and 3D up to a second-order accuracy. Portage ensures critical properties such as local/global conservation and bounds preservation for mesh remap. It enables multi-material field remap through a dedicated plugin, and leverages the hybrid parallelism exposed by advanced architectures using multi-processing and multi-threading.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Optimizing High Performance Markov Clustering for Pre-Exascale Architectures

HipMCL is a high-performance distributed memory implementation of the popular Markov Cluster Algorithm (MCL) and can cluster large-scale networks within hours using a few thousand CPU-equipped nodes. It relies on sparse matrix computations and heavily makes use of the sparse matrix-sparse matrix multiplication kernel (SpGEMM). The existing parallel algorithms in HipMCL are not scalable to Exascale architectures, both due to their communication costs dominating the runtime at large concurrencies and also due to their inability to take advantage of accelerators that are increasingly popular. In this work, we systematically remove scalability and performance bottlenecks of HipMCL. We enable GPUs by performing the expensive expansion phase of the MCL algorithm on GPU. Additionally, we propose a CPU-GPU joint distributed SpGEMM algorithm called pipelined Sparse SUMMA and integrate a probabilistic memory requirement estimator that is fast and accurate. Furthermore, we develop a new merging algorithm for the incremental processing of partial results produced by the GPUs, which improves the overlap efficiency and the peak memory usage. We also integrate a recent and faster algorithm for performing SpGEMM on CPUs. We validate our new algorithms and optimizations with extensive evaluations. With the enabling of the GPUs and integration of new algorithms, HipMCL is up to 12.4x faster, being able to cluster a network with 70 million proteins and 68 billion connections just under 15 minutes using 1024 nodes of ORNL's Summit supercomputer.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

EFIT-Prime: Probabilistic and physics-constrained reduced-order neural network model for equilibrium reconstruction in DIII-D

We introduce EFIT-Prime, a novel machine learning surrogate model for EFIT (Equilibrium FIT) that integrates probabilistic and physics-informed methodologies to overcome typical limitations associated with deterministic and ad hoc neural network architectures. EFIT-Prime utilizes a neural architecture search-based deep ensemble for robust uncertainty quantification, providing scalable and efficient neural architectures that comprehensively quantify both data and model uncertainties. Physically informed by the Grad–Shafranov equation, EFIT-Prime applies a constraint on the current density J tor and a smoothness constraint on the first derivative of the poloidal flux, ensuring physically plausible solutions. Furthermore, the spatial location of the diagnostics is explicitly incorporated in the inputs to account for their spatial correlation. Extensive evaluations demonstrate EFIT-Prime's accuracy and robustness across diverse scenarios, most notably showing good generalization on negative-triangularity discharges that were excluded from training. Timing studies indicate an ensemble inference time of 15 ms for predicting a new equilibrium, offering the possibility of plasma control in real-time, if the model is optimized for speed.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗