Search NASA⌕ Search

SEARCH · Search NASA

Results for “core optimization”

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 1,441 records · Page 80

ON-OFF neuromorphic ISING machines using Fowler-Nordheim annealers

We introduce NeuroSA, a neuromorphic architecture specifically designed to ensure asymptotic convergence to the ground state of an Ising problem using a Fowler-Nordheim quantum mechanical tunneling based threshold-annealing process. The core component of NeuroSA consists of a pair of asynchronous ON-OFF neurons, which effectively map classical simulated annealing dynamics onto a network of integrate-and-fire neurons. The threshold of each ON-OFF neuron pair is adaptively adjusted by an FN annealer and the resulting spiking dynamics replicates the optimal escape mechanism and convergence of SA, particularly at low-temperatures. To validate the effectiveness of our neuromorphic Ising machine, we systematically solved benchmark combinatorial optimization problems such as MAX-CUT and Max Independent Set. Across multiple runs, NeuroSA consistently generates distribution of solutions that are concentrated around the state-of-the-art results (within 99%) or surpass the current state-of-the-art solutions for Max Independent Set benchmarks. Furthermore, NeuroSA is able to achieve these superior distributions without any graph-specific hyperparameter tuning. For practical illustration, we present results from an implementation of NeuroSA on the SpiNNaker2 platform, highlighting the feasibility of mapping our proposed architecture onto a standard neuromorphic accelerator platform.

42 ENGINEERING↗

Integrated fusion simulation with self-consistent core-pedestal coupling

In this study, accurate prediction of fusion performance in present and future tokamaks requires taking into account the strong interplay between core transport, pedestal structure, current profile and plasma equilibrium. An integrated modeling workflow capable of calculating the steady-state self- consistent solution to this strongly-coupled problem has been developed. The workflow leverages state-of-the-art components for collisional and turbulent core transport, equilibrium and pedestal stability. Validation against DIII-D discharges shows that the workflow is capable of robustly pre- dicting the kinetic profiles (electron and ion temperature and electron density) from the axis to the separatrix in good agreement with the experiments. An example application is presented, showing self-consistent optimization for the fusion performance of the 15 MA D-T ITER baseline scenario as functions of the pedestal density and ion effective charge Z eff .

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Designing a parallel Feel-the-Way clustering algorithm on HPC systems

This paper introduces a new parallel clustering algorithm, named Feel-the-Way clustering algorithm, that provides better or equivalent convergence rate than the traditional clustering methods by optimizing the synchronization and communication costs. Our algorithm design centers on how to optimize three factors simultaneously: reduced synchronizations, improved convergence rate, and retained same or comparable optimization cost. To compare the optimization cost, we use the Sum of Square Error (SSE) cost as the metric, which is the sum of the square distance between each data point and its assigned clusters. Compared with the traditional MPI k-means algorithm, the new Feel-the-Way algorithm requires less communications among participating processes. As for the convergence rate, the new algorithm requires fewer number of iterations to converge. As for the optimization cost, it obtains the SSE costs that are close to the k-means algorithm. In the paper, we first design the full-step Feel-the-Way k-means clustering algorithm that can significantly reduce the number of iterations that are required by the original k-means clustering method. Next, we improve the performance of the full-step algorithm by adopting an optimized sampling-based approach, named reassignment-history-aware sampling. Our experimental results show that the optimized sampling-based Feel-the-Way method is significantly faster than the widely used k-means clustering method, and can provide comparable optimization costs. More extensive experiments with several synthetic datasets and real-world datasets (e.g., MNIST, CIFAR-10, ENRON, and PLACES-2) show that the new parallel algorithm can outperform the open source MPI k-means library by up to 110% on a high-performance computing system using 4,096 CPU cores. In addition, the new algorithm can take up to 51% fewer iterations to converge than the k-means clustering algorithm.

97 MATHEMATICS AND COMPUTING↗

A graphics processing unit accelerated sparse direct solver and preconditioner with block low rank compression

We present the GPU implementation efforts and challenges of the sparse solver package STRUMPACK. The code is made publicly available on github with a permissive BSD license. STRUMPACK implements an approximate multifrontal solver, a sparse LU factorization which makes use of compression methods to accelerate time to solution and reduce memory usage. Multiple compression schemes based on rank-structured and hierarchical matrix approximations are supported, including hierarchically semi-separable, hierarchically off-diagonal butterfly, and block low rank. Here, in this paper, we present the GPU implementation of the block low rank (BLR) compression method within a multifrontal solver. Our GPU implementation relies on highly optimized vendor libraries such as cuBLAS and cuSOLVER for NVIDIA GPUs, rocBLAS and rocSOLVER for AMD GPUs and the Intel oneAPI Math Kernel Library (oneMKL) for Intel GPUs. Additionally, we rely on external open source libraries such as SLATE (Software for Linear Algebra Targeting Exascale), MAGMA (Matrix Algebra on GPU and Multi-core Architectures), and KBLAS (KAUST BLAS). SLATE is used as a GPU-capable ScaLAPACK replacement. From MAGMA we use variable sized batched dense linear algebra operations such as GEMM, TRSM and LU with partial pivoting. KBLAS provides efficient (batched) low rank matrix compression for NVIDIA GPUs using an adaptive randomized sampling scheme. The resulting sparse solver and preconditioner runs on NVIDIA, AMD and Intel GPUs. Interfaces are available from PETSc, Trilinos and MFEM, or the solver can be used directly in user code. We report results for a range of benchmark applications, using the Perlmutter system from NERSC, Frontier from ORNL, and Aurora from ALCF. For a high frequency wave equation on a regular mesh, using 32 Perlmutter compute nodes, the factorization phase of the exact GPU solver is about 6.5× faster compared to the CPU-only solver. The BLR-enabled GPU solver is about 13.8× faster than the CPU exact solver. For a collection of SuiteSparse matrices, the STRUMPACK exact factorization on a single GPU is on average 1.9× faster than NVIDIA’s cuDSS solver.

97 MATHEMATICS AND COMPUTING↗

Influence of carbon chain length on the synthesis and yield of fatty amine-coated iron-platinum nanoparticles

Iron oxide nanoparticles are among the most widely used and characterized magnetic nanoparticles. However, metal alloys such as superparamagnetic iron-platinum particles (SIPPs), which have better magnetic properties, are receiving increased attention. Scalable techniques to routinely synthesize SIPPs in bulk need further study. Here, we focus on the role played by the fatty amine ligand in the formation of the bimetallic FePt nanocrystal. More specifically, we compare the effect of varying lengths of fatty amine ligands on the shape, structure, uniformity, composition, and magnetic properties of the SIPPs. We synthesized SIPPs by employing a ‘green’ thermal decomposition reaction using fatty amine ligands containing 12 to 18 carbons in length. Greater fatty amine chain length increased the polydispersity, particle concentration, iron concentration, and the stability of the SIPPs. Additionally, longer reflux times increased the diameter of the particles, but decreased the iron concentration, suggesting that shorter reaction times are preferable. Fourier transform infrared spectroscopy of the SIPPs indicates that the ligands are successfully bound to the FePt cores through the amine group. Superconducting quantum interference device magnetometry measurements suggest that all of the SIPPs were superparamagnetic at room temperature and that SIPPs synthesized using tetradecylamine had the highest saturation magnetization. Our findings indicate that the octadecylamine ligand, which is currently used for the routine synthesis of SIPPs, may not be optimal. Overall, we found that using tetradecylamine and a 30-min reflux reaction resulted in optimal particles with the highest degree of monodispersity, iron content, stability, and saturation magnetization.

36 MATERIALS SCIENCE↗

Rendering and Compositing Infrastructure Improvements to VisIt for Insitu Rendering

Compared to posthoc rendering, insitu rendering often generates larger numbers of images, as a result rendering performance and scalability are critical in the insitu setting. In this work we present improvements to VisIt's rendering and compositing infrastructure that deliver increased performance and scalability in both posthoc and insitu settings. We added the capability for alpha blend compositing and use it with ordered compositing when datasets have disjoint block domain decomposition to optimize the rendering of transparent geometry. We also made improvements that increase overall efficiency by reducing communication and data movement and have addressed a number of performance issues. We structured our code to take advantage of SIMD parallelization and use threads to overlap communication and compositing. We tested our improvements on a 20 core workstation using 8 cores to render geometry generated from a $256^3$ cosmology dataset and on a Cray XC31 using 512 cores to render geometry generated from a $2000^2 \times 800$ plasma dataset. Our results show that ordered compositing provides a speed up of up to $4 \times$ over the current sort first strategy. The other improvements resulted in modest speed up with one notable exception where we achieve up to $40 \times$ speed up of rendering and compositing of opaque geometry when both opaque and transparent geometry are rendered together. We also investigated the use of depth peeling, but found that the implementation provided by VTK is substantially slower,both with and without GPU acceleration, than a local camera order sort.

97 MATHEMATICS AND COMPUTING↗

RAFT: Reconfigurable Array of High-Efficiency Ducted Turbines for Hydrokinetic Energy Harvesting

Diversifying the energy harvesting portfolio is crucial to achieving the ambitious goal of transitioning to clean energy by 2030. Marine hydrokinetic energy has garnered renewed interest due to its high harvesting potential in the U.S., and the resource's reliability and predictability—remaining relatively constant on a daily basis and available 24/7. However, there are currently few commercial devices capable of harnessing the energy from flowing water. This project aims to bridge that gap by designing and evaluating a novel hydrokinetic turbine concept that can efficiently harvest energy from both rivers and tidal streams. The RAFT (Reconfigurable Array of High-Efficiency Ducted Turbines) concept introduces a duct surrounding the turbine rotor and creates an array of small 5-kW units. The duct serves two primary purposes: (1) it enhances hydrodynamic efficiency by accelerating flow to the rotor, and (2) it functions as a structural component, facilitating the formation of modular arrays that lower costs. This project focuses on demonstrating this concept and validating these benefits through simulations and scaled prototype testing. The project team includes 8 faculty members and over 20 students from 3 universities, organized into three core areas: hydrodynamics, electrical systems, and structural analysis, with additional teams dedicated to system integration, environmental assessment and risk management, and tech-to-market strategy. The team successfully demonstrated the increased hydrodynamic efficiency of a ducted turbine compared to an unducted version using high-fidelity simulations and prototype tests. Moreover, design optimization efforts led to surpassing the SHARKS program's goal of 60% reduction of the levelized cost of energy with a significant margin.

13 HYDRO ENERGY↗

Synchrotron Radiation Research and Analysis of the Particulate Matter in Deep Ice Cores: An Overview of the Technical Challenges

Airborne dust extracted from deep ice core perforations can provide chemical and mineralogical insight into the history of the climate and atmospheric conditions, with unrivalled temporal resolution, time span and richness of information. The availability of material for research and the natural complexity of the particulate, however, pose significant challenges to analytical methods. We present the developments undertaken to optimize the experimental techniques, materials and protocols for synchrotron radiation-based analysis, in particular for the acquisition of combined Synchrotron Radiation X-Ray Fluorescence and X-ray Absorption Spectroscopy data.

Cibin, Giannantonio↗

A Methodology for Dynamic Security Risk Quantification and Optimal Resource Allocation of Security Assets

In this presentation we overview a methodology for dynamic security risk quantification and optimal resource allocation of security assets for high profile venues. This methodology is especially applicable to venues that require security screening operations such as mass transit (e.g., train or airport terminals), critical infrastructure protection (e.g., government buildings), and largescale public events (e.g., concerts or professional sports). The method starts by decomposing the three core components of risk -- threat, vulnerability, and consequence -- into their various subcomponents. For instance, vulnerability can be decomposed into availability, accessibility, organic security, and target hardness and each of these can be evaluated against the potential threats of interest for the given venue. Once evaluated, these subcomponents are rolled back up to compute the specific value for the vulnerability core risk component. Likewise, the same is done for consequence and threat, and then risk is computed as the product of these three components. A key aspect of our methodology is dynamically quantifying risk. That is, we incorporate the ability to uniquely allow the subcomponents and core components, and in turn, risk, to be quantified as a continuous function of time throughout the day, week, month, or year as appropriate.

Brigantic, Robert T.↗

Neutronics Analysis of TREAT Multi-SERTTA Calibration Test Vehicle (Multi-SERTTA CAL)

Transient testing is a necessary step in the performance evaluation of Accident Tolerant Fuels (ATF) developed as part of the Fuel Cycle Research and Development Advanced Fuels Campaign initiated and supported by the United States Department of Energy (DOE-NE). These advanced fuels are expected to be used in future reactors or replace the current fuel design used in existing power reactors for enhanced performance and safety. Upon completion of a process proscribed by the National Environmental Policy Act, the Transient Reactor Test (TREAT) facility, located at Idaho National Laboratory (INL), was chosen by DOE-NE to provide the desired specialized testing. TREAT has successfully operated from start-up in 1959 to 1994 when it was placed on operational stand-by. Careful preparations are being made for operational readiness and TREAT is projected to resume testing no later than 2018. Past TREAT testing primarily focused on sodium-cooled reactor fuels (SRF). Test vehicle designs varied to accommodate testing needs ranging from separate effects tests to integral tests approaching simulation of actual reactor environment conditions including forced convection cooling. The current focus on light water reactor (LWR) ATF requires design of new experiment vehicles to accommodate the water/steam environment and associated complexities regarding desired energy deposition into the specimen and proper containment. The first set of experiments in TREAT restart is expected to be conducted with the Multi-SERTTA (Static Environment Rodlet Transient Test Apparatuses). Difficult to quantify and calculate integral and time dependent effects inherent in transient testing limit predictive modeling capability; therefore, calibration testing in TREAT prior to the actual experiment is necessary to empirically determine the effectiveness of core-to-specimen power coupling in order to achieve the desired energy deposition into the test specimen during a nuclear power transient. This has been done in the past, in part, by evaluating the difference in energy deposition into dosimeter wires subjected to various TREAT time dependent power conditions. These calibration tests were typically conducted in a test vehicle neutronically similar to the actual test vehicle, but more simplistic in design in order to quickly retrieve and analyze dosimeter wires and minimize containment structure costs due to several trials. Prior calibration testing for SRF experiments could be conducted without coolant as sodium has no notable neutronic effect on the core-to-specimen power coupling factor (PCF). Such is not the case with LWR fuel where water serves as both moderator and coolant. This presents a design challenge as the calibration test vehicle must retain heated, high-pressure water to preserve neutronic equivalence representative of near prototypic reactor conditions while allowing for quick retrieval of dosimeters. The solution to this challenge comes in the form of a thin walled dry well positioned through the center of the basic Multi-SERTTA design. Test vehicle reactivity worth and PCF are two important factors with respect to ascertainment of neutronic equivalence. Similarity in test vehicle reactivity worth is desired to avoid the need to change operational core rod positions in TREAT which could significantly alter neutron flux distribution and consequently power distribution throughout the core and into the test specimen. The geometry and composition of various materials within the experiment containment structure can also significantly alter the quality of neutron signal delivery from the core to the test specimen. Introduction of an Inconel tube, albeit very thin, is expected to impact PCF. Simple modifications to the CAL design may compensate for the anticipated change. Evaluation of these factors and proposed modifications through neutronics analysis serves to inform experiment design and is a valuable tool towards achieving the optimal CAL design.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

National Spherical Torus Experiment Upgrade Collaborative Research on Configuration Optimization of Advanced Operating Scenarios and Control Including Macroscopy Stability EOG for Period 3/1/14 through 2/28/18

General Atomics has collaborated with PPPL since 1999 in support of NSTX and NSXT-U by leveraging DIII-D program expertise. John Ferron was the principal investigator (PI) for the first nine years (to 2008) and Rob La Haye since then. The research effort evolved over a number of three-year grants and originally included both the core (control, stability, resonant error fields) and the boundary (ELMs, divertor, error fields). In 2014 the current and last grant extension was stretched to four years and split into a core grant (advanced scenarios and control, macroscopic plasma stability and extending 3D magnetics diagnostics) and a boundary grant. Todd Evans is the PI for the new boundary grant which while also for four years expires later than the core grant; a separate end of grant (EOG) report for that is due in July 2018.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Risk-informed Physical Security Using Dynamic Force-on-Force Modelling Tools (ABSTRACT)

Physical security requirement of nuclear power plants (NPPs) in the US is regulated in a mainly prescriptive manner. This “one size fits all” approach may consequently overlook plant-specific characteristics which are beneficial for the inherent safety and security of the plant. Although a performance-based security measure has been promulgated in 10 CFR 73.55(r), current evaluation methods generally emphasize static and conservative assumptions on physical security uncertainties. This conservatism contributes to the overall operational and maintenance costs, which account to 70% of total operating expenditures for an NPP. This paper describes an ongoing work at the Idaho National Laboratory within the Light Water Reactor Sustainability pathway, which aims to optimize physical protection system with regards to system effectiveness and dollar costs. To achieve this goal, a risk-informed approach to physical security is proposed. This approach leverages the Event Modeling Risk Assessment using Linked Diagrams (EMRALD) tool to model facility infiltration and Force-on-Force (FoF) engagement in a dynamic manner. The dynamic FoF analysis was conducted by coupling human performance uncertainties modeled in EMRALD to existing commercial 3-dimensional FoF software, i.e. SCRIBE3D, AVERT and Simajin. Meanwhile, risk analysis was accomplished by coupling the attack outcomes to existing plant Probabilistic Risk Assessment (PRA) model with FLEX strategy. This approach reduces the conservatisms in security modeling and further acknowledges the possibility for plant operators to mitigate the attack outcomes. A sabotage attack on a hypothetical NPP facility was devised to demonstrate the proposed methodology. Dynamic FoF rules were implemented in EMRALD coupled to SCRIBE3D. This dynamic evaluation revealed an additional 21% system effectiveness in protecting the target set relative to the static modelling approach. Furthermore, the conditional core damage probability could be reduced by up to 46% by crediting operator actions in mitigating the sabotage outcomes. We evaluated the relative importance of each component of the physical security posture included in the model and assessed tradeoffs between components to identify potential opportunities for cost reduction while maintaining a specific level of system effectiveness. Results of the cost analysis will be presented in the conference. This study is expected to help NPP operators in allocating resources optimally to satisfy the security requirements.

97 MATHEMATICS AND COMPUTING↗

Carbon fiber synthesis from pitch: Insights from ReaxFF based molecular dynamics simulations

Despite the potential to lowering fabrication costs of pitch-based carbon fiber (CF) compared to polyacrylonitrile-based CF, large scale utilization is currently limited by the complexity in manufacturing. This is mainly due to the lack of consistency between the pitch structure/chemistry and CF properties under different processing conditions. Here, we employed full atomistic simulations based on reactive force fields to investigate the conversion of pitch precursors to CFs. Additionally, it was found that at the carbonization stage, the consumption rates of precursors, gas formations, and solid yields (carbonized CFs) are all dependent on the composition and arrangement of molecular moieties in the precursor. Specifically, precursors with sp 3 carbons in the cata-condensed core area have limited decomposition pathways (i.e., limited reaction sites), which leads to the formation of carbonized CFs with similar compositions. Furthermore, carbonized CFs possess distinct structural characteristics, and finally lead to the formation of CFs with different moduli. By analyzing CF crystallinity and elemental composition, general rationales are proposed for the selection of optimal precursors and processing conditions, and thus shed lights on parameter screening during industrial CF synthesis. For instance, inclusion of sp 3 carbons in cata- condensed rings can be preferred over the presence of sp 3 carbons on side chains.

01 COAL, LIGNITE, AND PEAT↗

Structural basis for MTR4–ZCCHC8 interactions that stimulate the MTR4 helicase in the nuclear exosome-targeting complex

The nuclear exosome-targeting (NEXT) complex functions as an RNA exosome cofactor and is involved in surveillance and turnover of aberrant transcripts and noncoding RNAs. NEXT is a ternary complex composed of the RNA-binding protein RBM7, the scaffold zinc-knuckle protein ZCCHC8, and the helicase MTR4. While RNA interactions with RBM7 are known, it remains unclear how NEXT subunits collaborate to recognize and prepare substrates for degradation. Here, we show that MTR4 helicase activity is enhanced when associated with RBM7 and ZCCHC8. While uridine-rich substrates interact with RBM7 and are preferred, optimal activity is observed when substrates include a polyadenylated 3' end. We identify a bipartite interaction of ZCCHC8 with MTR4 and uncover a role for the conserved C-terminal domain of ZCCHC8 in stimulating MTR4 helicase and ATPase activities. A crystal structure reveals that the ZCCHC8 C-terminal domain binds the helicase core in a manner that is distinct from that observed for Saccharomyces cerevisiae exosome cofactors Trf4p and Air2p. Furthermore, our results are consistent with a model whereby effective targeting of substrates by NEXT entails recognition of elements within the substrate and activation of MTR4 helicase activity.

59 BASIC BIOLOGICAL SCIENCES↗

Mechanism of electronic-excitation transfer in organic light-emitting devices based on semiconductor quantum dots

The results of an experimental study of organic light-emitting diodes (LEDs) with luminescent layers based on two types of CdSe/CdS semiconductor quantum dots (QDs) with an average CdSe core diameter of 3 and 5 nm and a characteristic CdS shell thickness of 0.5 nm are presented. The dependences of the LED efficiency on the QD concentration are determined. The experimental data are used to determine the mechanism of electronic-excitation transfer from the organic matrix to the semiconductor QDs. Ways of optimizing the design of the LEDs in order to improve their efficiency are suggested on this basis.

36 MATERIALS SCIENCE↗

P3DFFT: A Framework for Parallel Computations of Fourier Transforms in Three Dimensions

Fourier and related transforms are a family of algorithms widely employed in diverse areas of computational science, notoriously difficult to scale on high-performance parallel computers with a large number of processing elements (cores). This paper introduces a popular software package called P3DFFT which implements fast Fourier transforms (FFTs) in three dimensions in a highly efficient and scalable way. It overcomes a well-known scalability bottleneck of three-dimensional (3D) FFT implementations by using two-dimensional domain decomposition. Designed for portable performance, P3DFFT achieves excellent timings for a number of systems and problem sizes. On a Cray XT5 system P3DFFT attains 45% efficiency in weak scaling from 128 to 65,536 computational cores. Library features include Fourier and Chebyshev transforms, Fortran and C interfaces, in- and out-of-place transforms, uneven data grids, and single and double precision. P3DFFT is available as open source at http://code.google.com/p/p3dfft/. This paper discusses P3DFFT implementation and performance in a way that helps guide the user in making optimal choices for parameters of their runs. Read More: https://epubs.siam.org/doi/10.1137/11082748X

Mathematics↗

A.I. for nuclear physics

Nuclear science is concerned with the understanding of the nature of matter, its basic constituents and their interaction to form the elements and the properties we observe. This includes the forms of matter we see around us and also exotic forms such as those that existed in the first moments after the Big Bang and that exist today inside neutron stars. The techniques, tools, and expertise needed for nuclear physics (NP) research are therefore diverse in nature. State-of-the art accelerators are being developed to illuminate the dynamical basis of the core of the atom in terms of the fundamental constituents called quarks and gluons and to increase the number of isotopes with known properties. This scientific infrastructure is reaching scales and complexities that require computational methods for tasks such as anomaly detection in operational data. New methodologies are needed to detect anomalies and to optimize operating parameters, predict failures as well as to discover new optimization algorithms.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Divertor Plasma Detachment Control Neural Network

DivControlNN is a state-of-the-art software tool that leverages advanced machine learning techniques to predict and control divertor plasma behavior in fusion reactors. Plasma, a highly energetic and electrically charged gas, requires meticulous management to protect reactor components and maintain optimal energy production. Conventional simulation methods, although extremely detailed, typically demand extensive computational time-making them unsuitable for real-time control scenarios. DivControlNN addresses this challenge by learning from tens of thousands of high-fidelity simulations, thereby creating a rapid surrogate model that can deliver near-instantaneous predictions. At the core of its functionality is a sophisticated technique known as latent space mapping, which condenses complex, high-dimensional plasma data into a compact, lower-dimensional representation. This streamlined representation enables the system to quickly forecast essential plasma properties and determine the precise conditions required for effective detachment. Detachment is a crucial process in which the plasma is cooled before reaching the divertor plates, thereby reducing heat loads and mitigating material erosion. In recent experiments conducted on the KSTAR tokamak in South Korea, DivControlNN successfully guided the detachment process without any fine-tuning-even when applied to a new tungsten divertor configuration. By achieving a computational speed-up of over one hundred million times compared to traditional simulation methods while maintaining low prediction errors, DivControlNN stands to significantly enhance real-time control and diagnostic capabilities in future fusion reactors. This breakthrough paves the way for safer, more reliable reactor operation and represents a major advancement toward realizing fusion energy as a practical, sustainable, and clean power source.

Xu, Xueqiao [Lawrence Livermore National Laborator↗