• Themes
• Newsmakers

# Fusion

07.06.2021
08:32 Heterogeneous Noisy Short Signal Camouflage in Multi-Domain Environment Decision-Making. (arXiv:2106.02044v1 [cs.LG])

Data transmission between two or more digital devices in industry and government demands secure and agile technology. Digital information distribution often requires deployment of Internet of Things (IoT) devices and Data Fusion techniques which have also gained popularity in both, civilian and military environments, such as, emergence of Smart Cities and Internet of Battlefield Things (IoBT). This usually requires capturing and consolidating data from multiple sources. Because datasets do not necessarily originate from identical sensors, fused data typically results in a complex Big Data problem. Due to potentially sensitive nature of IoT datasets, Blockchain technology is used to facilitate secure sharing of IoT datasets, which allows digital information to be distributed, but not copied. However, blockchain has several limitations related to complexity, scalability, and excessive energy consumption. We propose an approach to hide information (sensor signal) by transforming it to an

08:32 Heterogeneous Noisy Short Signal Camouflage in Multi-Domain Environment Decision-Making. (arXiv:2106.02044v1 [cs.LG])

Data transmission between two or more digital devices in industry and government demands secure and agile technology. Digital information distribution often requires deployment of Internet of Things (IoT) devices and Data Fusion techniques which have also gained popularity in both, civilian and military environments, such as, emergence of Smart Cities and Internet of Battlefield Things (IoBT). This usually requires capturing and consolidating data from multiple sources. Because datasets do not necessarily originate from identical sensors, fused data typically results in a complex Big Data problem. Due to potentially sensitive nature of IoT datasets, Blockchain technology is used to facilitate secure sharing of IoT datasets, which allows digital information to be distributed, but not copied. However, blockchain has several limitations related to complexity, scalability, and excessive energy consumption. We propose an approach to hide information (sensor signal) by transforming it to an

08:32 Heterogeneous Noisy Short Signal Camouflage in Multi-Domain Environment Decision-Making. (arXiv:2106.02044v1 [cs.LG])

Data transmission between two or more digital devices in industry and government demands secure and agile technology. Digital information distribution often requires deployment of Internet of Things (IoT) devices and Data Fusion techniques which have also gained popularity in both, civilian and military environments, such as, emergence of Smart Cities and Internet of Battlefield Things (IoBT). This usually requires capturing and consolidating data from multiple sources. Because datasets do not necessarily originate from identical sensors, fused data typically results in a complex Big Data problem. Due to potentially sensitive nature of IoT datasets, Blockchain technology is used to facilitate secure sharing of IoT datasets, which allows digital information to be distributed, but not copied. However, blockchain has several limitations related to complexity, scalability, and excessive energy consumption. We propose an approach to hide information (sensor signal) by transforming it to an

04.06.2021
19:07 It's Time For Congress to Support Fusion Energy

Fusion devices for clean, safe, and affordable electricity and industrial heat are making advances and need a push -- Read more on ScientificAmerican.com

09:45 A Normative Model of Classifier Fusion. (arXiv:2106.01770v1 [cs.LG])

Combining the outputs of multiple classifiers or experts into a single probabilistic classification is a fundamental task in machine learning with broad applications from classifier fusion to expert opinion pooling. Here we present a hierarchical Bayesian model of probabilistic classifier fusion based on a new correlated Dirichlet distribution. This distribution explicitly models positive correlations between marginally Dirichlet-distributed random vectors thereby allowing normative modeling of correlations between base classifiers or experts. The proposed model naturally accommodates the classic Independent Opinion Pool and other independent fusion algorithms as special cases. It is evaluated by uncertainty reduction and correctness of fusion on synthetic and real-world data sets. We show that a change in performance of the fused classifier due to uncertainty reduction can be Bayes optimal even for highly correlated base classifiers.

09:45 Fast simulations for large aspect ratio stellarators with the neoclassical code KNOSOS. (arXiv:2106.01727v1 [physics.plasm-ph])

In this work, a new version of KNOSOS is presented. KNOSOS is a low-collisionality radially-local, bounce-averaged neoclassical code that is extremely fast, and at the same time, includes physical effects often neglected by more standard codes: the component of the magnetic drift that is tangent to the flux-surface and the variation of the electrostatic potential on the flux-surface. An earlier version of the code could only describe configurations that were sufficiently optimized with respect to neoclassical transport. KNOSOS can now be applied to any large aspect ratio stellarator, and its performance is demonstrated by means of detailed simulations in the configuration space of Wendelstein 7-X.

09:45 A Normative Model of Classifier Fusion. (arXiv:2106.01770v1 [cs.LG])

Combining the outputs of multiple classifiers or experts into a single probabilistic classification is a fundamental task in machine learning with broad applications from classifier fusion to expert opinion pooling. Here we present a hierarchical Bayesian model of probabilistic classifier fusion based on a new correlated Dirichlet distribution. This distribution explicitly models positive correlations between marginally Dirichlet-distributed random vectors thereby allowing normative modeling of correlations between base classifiers or experts. The proposed model naturally accommodates the classic Independent Opinion Pool and other independent fusion algorithms as special cases. It is evaluated by uncertainty reduction and correctness of fusion on synthetic and real-world data sets. We show that a change in performance of the fused classifier due to uncertainty reduction can be Bayes optimal even for highly correlated base classifiers.

03.06.2021
05:30 Energetic particle transport in optimized stellarators. (arXiv:2106.00716v1 [physics.plasm-ph])

Nine stellarator configurations, three quasiaxisymmetric, three quasihelically symmetric and three non-quasisymmetric are scaled to ARIES-CS size and analyzed for energetic particle content. The best performing configurations with regard to energetic particle confinement also perform the best on the neoclassical {\Gamma}c metric, which attempts to align contours of the second adiabatic invariant with flux surfaces. Quasisymmetric configurations that simultaneously perform well on {\Gamma}c and quasisymmetry have the best overall confinement, with collisional losses under 3%, approaching the performance of ITER with ferritic inserts.

02.06.2021
22:06 China's 'artificial sun' nuclear fusion reactor sets a new world record after running at 216MILLION°F for 100 seconds - as the nation inches closer to its goal of 'limitless clean power'

China's 'artificial sun' nuclear fusion reactor has set a new world record after running at 216million degrees Fahrenheit (120million°C) for 100 seconds, according to state media.

08:16 An adaptive scalable fully implicit algorithm based on stabilized finite element for reduced visco-resistive MHD. (arXiv:2106.00260v1 [physics.comp-ph])

The magnetohydrodynamics (MHD) equations are continuum models used in the study of a wide range of plasma physics systems, including the evolution of complex plasma dynamics in tokamak disruptions. However, efficient numerical solution methods for MHD are extremely challenging due to disparate time and length scales, strong hyperbolic phenomena, and nonlinearity. Therefore the development of scalable, implicit MHD algorithms and high-resolution adaptive mesh refinement strategies is of considerable importance. In this work, we develop a high-order stabilized finite-element algorithm for the reduced visco-resistive MHD equations based on the MFEM finite element library (mfem.org). The scheme is fully implicit, solved with the Jacobian-free Newton-Krylov (JFNK) method with a physics-based preconditioning strategy. Our preconditioning strategy is a generalization of the physics-based preconditioning methods in [Chacon, et al, JCP 2002] to adaptive, stabilized finite elements. Algebraic

08:16 An adaptive scalable fully implicit algorithm based on stabilized finite element for reduced visco-resistive MHD. (arXiv:2106.00260v1 [physics.comp-ph])

The magnetohydrodynamics (MHD) equations are continuum models used in the study of a wide range of plasma physics systems, including the evolution of complex plasma dynamics in tokamak disruptions. However, efficient numerical solution methods for MHD are extremely challenging due to disparate time and length scales, strong hyperbolic phenomena, and nonlinearity. Therefore the development of scalable, implicit MHD algorithms and high-resolution adaptive mesh refinement strategies is of considerable importance. In this work, we develop a high-order stabilized finite-element algorithm for the reduced visco-resistive MHD equations based on the MFEM finite element library (mfem.org). The scheme is fully implicit, solved with the Jacobian-free Newton-Krylov (JFNK) method with a physics-based preconditioning strategy. Our preconditioning strategy is a generalization of the physics-based preconditioning methods in [Chacon, et al, JCP 2002] to adaptive, stabilized finite elements. Algebraic

08:16 An adaptive scalable fully implicit algorithm based on stabilized finite element for reduced visco-resistive MHD. (arXiv:2106.00260v1 [physics.comp-ph])

The magnetohydrodynamics (MHD) equations are continuum models used in the study of a wide range of plasma physics systems, including the evolution of complex plasma dynamics in tokamak disruptions. However, efficient numerical solution methods for MHD are extremely challenging due to disparate time and length scales, strong hyperbolic phenomena, and nonlinearity. Therefore the development of scalable, implicit MHD algorithms and high-resolution adaptive mesh refinement strategies is of considerable importance. In this work, we develop a high-order stabilized finite-element algorithm for the reduced visco-resistive MHD equations based on the MFEM finite element library (mfem.org). The scheme is fully implicit, solved with the Jacobian-free Newton-Krylov (JFNK) method with a physics-based preconditioning strategy. Our preconditioning strategy is a generalization of the physics-based preconditioning methods in [Chacon, et al, JCP 2002] to adaptive, stabilized finite elements. Algebraic

01.06.2021
12:52 China fires up 'artificial sun' at 120 MILLION DEGREES Celsius in quest for nuclear fusion – media

Chinese media have reported that researchers working on a nuclear fusion project have succeeded in holding plasma of 120 million degrees Celsius for close to two minutes. Read Full Article at RT.com

10:58 BAAI-VANJEE Roadside Dataset: Towards the Connected Automated Vehicle Highway technologies in Challenging Environments of China. (arXiv:2105.14370v1 [cs.CV])

As the roadside perception plays an increasingly significant role in the Connected Automated Vehicle Highway(CAVH) technologies, there are immediate needs of challenging real-world roadside datasets for bench marking and training various computer vision tasks such as 2D/3D object detection and multi-sensor fusion. In this paper, we firstly introduce a challenging BAAI-VANJEE roadside dataset which consist of LiDAR data and RGB images collected by VANJEE smart base station placed on the roadside about 4.5m high. This dataset contains 2500 frames of LiDAR data, 5000 frames of RGB images, including 20% collected at the same time. It also contains 12 classes of objects, 74K 3D object annotations and 105K 2D object annotations. By providing a real complex urban intersections and highway scenes, we expect the BAAI-VANJEE roadside dataset will actively assist the academic and industrial circles to accelerate the innovation research and achievement transformation in the field of intelligent

31.05.2021
06:29 ECG Heart-beat Classification Using Multimodal Image Fusion. (arXiv:2105.13536v1 [eess.SP])

In this paper, we present a novel Image Fusion Model (IFM) for ECG heart-beat classification to overcome the weaknesses of existing machine learning techniques that rely either on manual feature extraction or direct utilization of 1D raw ECG signal. At the input of IFM, we first convert the heart beats of ECG into three different images using Gramian Angular Field (GAF), Recurrence Plot (RP) and Markov Transition Field (MTF) and then fuse these images to create a single imaging modality. We use AlexNet for feature extraction and classification and thus employ end to end deep learning. We perform experiments on PhysioNet MIT-BIH dataset for five different arrhythmias in accordance with the AAMI EC57 standard and on PTB diagnostics dataset for myocardial infarction (MI) classification. We achieved an state of an art results in terms of prediction accuracy, precision and recall.

06:29 Inertial Sensor Data To Image Encoding For Human Action Recognition. (arXiv:2105.13533v1 [cs.CV])

Convolutional Neural Networks (CNNs) are successful deep learning models in the field of computer vision. To get the maximum advantage of CNN model for Human Action Recognition (HAR) using inertial sensor data, in this paper, we use 4 types of spatial domain methods for transforming inertial sensor data to activity images, which are then utilized in a novel fusion framework. These four types of activity images are Signal Images (SI), Gramian Angular Field (GAF) Images, Markov Transition Field (MTF) Images and Recurrence Plot (RP) Images. Furthermore, for creating a multimodal fusion framework and to exploit activity image, we made each type of activity images multimodal by convolving with two spatial domain filters : Prewitt filter and High-boost filter. Resnet-18, a CNN model, is used to learn deep features from multi-modalities. Learned features are extracted from the last pooling layer of each ReNet and then fused by canonical correlation based fusion (CCF) for improving the

28.05.2021
23:28 Chinese ‘Artificial Sun’ Experimental Fusion Reactor Sets World Record for Superheated Plasma Time

The Belt and Road infrastructure megaproject doesn’t just include rails and highways, it’s aimed at helping nations achieve energy independence, too. China is exporting nuclear power plants of a newer, safer design to partner nations like Pakistan, as well as building them at home to help transition away from fossil fuel use. China’s futuristic nuclear fusion reactor just set a new world record for the longest duration of time in sustaining the sun-like temperature needed for fusion to occur. While China is still a long way from a fusion power plant, the achievement is an important step towards clean, sustainable electricity generation. The achievement was announced on Friday by Gong Xianzu, a researcher at the Experimental Advanced Superconducting Tokamak (EAST) at the Hefei Institutes of Physical Science of the Chinese Academy of Sciences in China’s Anhui Province. The device, which replicates the atom-building process that occurs at the center of stars and gives them

27.05.2021
10:00 Debye source representation for type-I superconductors, I. (arXiv:2105.12246v1 [math.NA])

In this note, we analyze the classical magneto-static approach to the theory of type I superconductors, and a Debye source representation that can be used numerically to solve the resultant equations. We also prove that one of the fields, $\boldsymbol{J}^-$, found within the superconductor via the London equations, is the physical current in that the outgoing part of the magnetic field is given as the Biot-Savart integral of $\boldsymbol{J}^{-}$. Finally, we compute the static currents for moderate values of London penetration depth, $\lambda_L,$ for a sphere, a stellarator-like geometry and a two-holed torus.

10:00 Surrogate Approximation of the Grad-Shafranov Free Boundary Problem via Stochastic Collocation on Sparse Grids. (arXiv:2105.12217v1 [math.NA])

In magnetic confinement fusion devices, the equilibrium configuration of a plasma is determined by the balance between the hydrostatic pressure in the fluid and the magnetic forces generated by an array of external coils and the plasma itself. The location of the plasma is not known a priori and must be obtained as the solution to a free boundary problem. The partial differential equation that determines the behavior of the combined magnetic field depends on a set of physical parameters (location of the coils, intensity of the electric currents going through them, magnetic permeability, etc.) that are subject to uncertainty and variability. The confinement region is in turn a function of these stochastic parameters as well. In this work, we consider variations on the current intensities running through the external coils as the dominant source of uncertainty. This leads to a parameter space of dimension equal to the number of coils in the reactor. With the aid of a surrogate function

10:00 Surrogate Approximation of the Grad-Shafranov Free Boundary Problem via Stochastic Collocation on Sparse Grids. (arXiv:2105.12217v1 [math.NA])

In magnetic confinement fusion devices, the equilibrium configuration of a plasma is determined by the balance between the hydrostatic pressure in the fluid and the magnetic forces generated by an array of external coils and the plasma itself. The location of the plasma is not known a priori and must be obtained as the solution to a free boundary problem. The partial differential equation that determines the behavior of the combined magnetic field depends on a set of physical parameters (location of the coils, intensity of the electric currents going through them, magnetic permeability, etc.) that are subject to uncertainty and variability. The confinement region is in turn a function of these stochastic parameters as well. In this work, we consider variations on the current intensities running through the external coils as the dominant source of uncertainty. This leads to a parameter space of dimension equal to the number of coils in the reactor. With the aid of a surrogate function

10:00 Debye source representation for type-I superconductors, I. (arXiv:2105.12246v1 [math.NA])

In this note, we analyze the classical magneto-static approach to the theory of type I superconductors, and a Debye source representation that can be used numerically to solve the resultant equations. We also prove that one of the fields, $\boldsymbol{J}^-$, found within the superconductor via the London equations, is the physical current in that the outgoing part of the magnetic field is given as the Biot-Savart integral of $\boldsymbol{J}^{-}$. Finally, we compute the static currents for moderate values of London penetration depth, $\lambda_L,$ for a sphere, a stellarator-like geometry and a two-holed torus.

10:00 Surrogate Approximation of the Grad-Shafranov Free Boundary Problem via Stochastic Collocation on Sparse Grids. (arXiv:2105.12217v1 [math.NA])

In magnetic confinement fusion devices, the equilibrium configuration of a plasma is determined by the balance between the hydrostatic pressure in the fluid and the magnetic forces generated by an array of external coils and the plasma itself. The location of the plasma is not known a priori and must be obtained as the solution to a free boundary problem. The partial differential equation that determines the behavior of the combined magnetic field depends on a set of physical parameters (location of the coils, intensity of the electric currents going through them, magnetic permeability, etc.) that are subject to uncertainty and variability. The confinement region is in turn a function of these stochastic parameters as well. In this work, we consider variations on the current intensities running through the external coils as the dominant source of uncertainty. This leads to a parameter space of dimension equal to the number of coils in the reactor. With the aid of a surrogate function

26.05.2021
05:48 BORAY: An Axisymmetric Ray Tracing Code Supports Both Closed and Open Field Lines Plasmas. (arXiv:2105.12014v1 [physics.plasm-ph])

Ray tracing codes are useful to study the electromagnetic wave propagation and absorption in the geometrical optics approximation. In magnetized fusion plasma community, most ray tracing codes assume the plasma density and temperature be functions of the magnetic flux and study waves only inside the last closed flux surface, which are sufficient for the present day tokamak. However, they are difficult to be used for configurations with open magnetic field line plasmas, such as mirror machine and field-reversed-configuration (FRC). We develop a ray tracing code in cylindrical coordinates $(r,\phi,z)$ to support arbitrary axisymmetric configurations with both closed and open field lines plasmas. For wave propagation, the cold plasma dispersion relation is usually sufficient, and we require the magnetic field ${\bf B}(r,z)$ and species densities $n_{s0}(r,z)$ profiles as input. For wave absorption, we require a further temperature $T_{s0}(r,z)$ profile to solve a hot kinetic plasma

25.05.2021
08:28 Destabilizing effects of edge infernal components on n = 1 resistive wall modes in CFETR 1GW steady-state operating scenario. (arXiv:2105.11149v1 [physics.plasm-ph])

The stability of the $n=1$ resistive wall modes (RWMs) is investigated using the AEGIS code for the newly designed China Fusion Engineering Test Reactor (CFETR) 1GW steady-state operating (SSO) scenario. Here, $n$ is the toroidal mode number. Due to the large fraction of bootstrap current contribution, the profile of safety factor q is deeply reversed in magnetic shear in the central core region and locally flattened within the edge pedestal. Consequently the pressure-driven infernal components develop in the corresponding q-flattened regions of both core and edge. However, the edge infernal components dominate the $n=1$ RWM structure and lead to lower $\beta_N$ limits than the designed target $\beta_N$ for the CFETR 1GW SSO scenario. The edge rotation is found the most critical to the stabilization due to the dominant influence of the edge infernal components, which should be maintained above $1.5\%\Omega_{A0}$ in magnitude in order for the rotation alone to fully suppress the $n=1$

08:28 Progress toward Fusion Energy Breakeven and Gain as Measured against the Lawson Criterion. (arXiv:2105.10954v1 [physics.plasm-ph])

The Lawson criterion is a key concept in the pursuit of fusion energy, relating the fuel density $n$, (energy) confinement time $\tau$, and fuel temperature $T$ to the energy gain $Q$ of a fusion plasma. The purpose of this paper is to explain and review the Lawson criterion and to provide a compilation of achieved parameters for a broad range of historical and contemporary fusion experiments. Although this paper focuses on the Lawson criterion, it is only one of many equally important factors in assessing the progress and ultimate likelihood of any fusion concept becoming a commercially viable fusion energy system. Only experimentally measured or inferred values of $n$, $\tau$, and $T$ that have been published in the peer-reviewed literature are included in this paper. For extracting these parameters, we discuss methodologies that are necessarily specific to different fusion approaches (including magnetic, inertial, and magneto-inertial fusion). This paper is intended to serve as a

08:28 MCR-Net: A Multi-Step Co-Interactive Relation Network for Unanswerable Questions on Machine Reading Comprehension. (arXiv:2103.04567v2 [cs.CL] UPDATED)

Question answering systems usually use keyword searches to retrieve potential passages related to a question, and then extract the answer from passages with the machine reading comprehension methods. However, many questions tend to be unanswerable in the real world. In this case, it is significant and challenging how the model determines when no answer is supported by the passage and abstains from answering. Most of the existing systems design a simple classifier to determine answerability implicitly without explicitly modeling mutual interaction and relation between the question and passage, leading to the poor performance for determining the unanswerable questions. To tackle this problem, we propose a Multi-Step Co-Interactive Relation Network (MCR-Net) to explicitly model the mutual interaction and locate key clues from coarse to fine by introducing a co-interactive relation module. The co-interactive relation module contains a stack of interaction and fusion blocks to continuously

08:28 High-level camera-LiDAR fusion for 3D object detection with machine learning. (arXiv:2105.11060v1 [cs.CV])

This paper tackles the 3D object detection problem, which is of vital importance for applications such as autonomous driving. Our framework uses a Machine Learning (ML) pipeline on a combination of monocular camera and LiDAR data to detect vehicles in the surrounding 3D space of a moving platform. It uses frustum region proposals generated by State-Of-The-Art (SOTA) 2D object detectors to segment LiDAR point clouds into point clusters which represent potentially individual objects. We evaluate the performance of classical ML algorithms as part of an holistic pipeline for estimating the parameters of 3D bounding boxes which surround the vehicles around the moving platform. Our results demonstrate an efficient and accurate inference on a validation set, achieving an overall accuracy of 87.1%.

08:28 Insect-Computer Hybrid System Capable of Autonomous Navigation and Human Detection in Unknown Environments. (arXiv:2105.10869v1 [cs.RO])

There is still a long way to go before artificial mini robots are really used for search and rescue missions in disaster-hit areas due to hindrance in power consumption, computation load of the locomotion, and obstacle-avoidance system. Insect-computer hybrid system, which is the fusion of living insect platform and microcontroller, emerges as an alternative solution. This study demonstrates the first-ever insect-computer hybrid system conceived for search and rescue missions, which is capable of autonomous navigation and human presence detection in an unstructured environment. Customized navigation control algorithm utilizing the insect's intrinsic navigation capability achieved exploration and negotiation of complex terrains. On-board high-accuracy human presence detection using infrared camera was achieved with a custom machine learning model. Low power consumption suggests system suitability for hour-long operations and its potential for realization in real-life missions.

24.05.2021
04:06 Trimming Feature Extraction and Inference for MCU-based Edge NILM: a Systematic Approach. (arXiv:2105.10302v1 [cs.LG])

Non-Intrusive Load Monitoring (NILM) enables the disaggregation of the global power consumption of multiple loads, taken from a single smart electrical meter, into appliance-level details. State-of-the-Art approaches are based on Machine Learning methods and exploit the fusion of time- and frequency-domain features from current and voltage sensors. Unfortunately, these methods are compute-demanding and memory-intensive. Therefore, running low-latency NILM on low-cost, resource-constrained MCU-based meters is currently an open challenge. This paper addresses the optimization of the feature spaces as well as the computational and storage cost reduction needed for executing State-of-the-Art (SoA) NILM algorithms on memory- and compute-limited MCUs. We compare four supervised learning techniques on different classification scenarios and characterize the overall NILM pipeline's implementation on a MCU-based Smart Measurement Node. Experimental results demonstrate that optimizing the feature

21.05.2021
05:18 Ray-tracing Analysis for Cross-polarization Scattering Diagnostic on MAST-Upgrade Spherical Tokamak. (arXiv:2105.09818v1 [physics.plasm-ph])

A combined Doppler backscattering/cross-polarization scattering (DBS/CPS) system is being deployed on MAST-U, for simultaneous measurements of local density turbulence, turbulence flows, and magnetic turbulence. In this design, CPS shares the probing beam with the DBS and uses a separate parallel-viewing receiver system. In this study, we utilize a modified GENRAY 3D ray-tracing code, to simulate the propagation of the probing and scattered beams. The contributions of different scattering locations along the entire beam trajectories are considered, and the corresponding local $\tilde{\mathbf{B}}$ wavenumbers are estimated using the wave-vector matching criterion. The wavenumber ranges of the local $\tilde{\mathbf{B}}$ that is detectable to the CPS system are explored for simulated L- and H-mode plasmas.

20.05.2021
08:06 High performance and energy efficient inference for deep learning on ARM processors. (arXiv:2105.09187v1 [cs.DC])

We evolve PyDTNN, a framework for distributed parallel training of Deep Neural Networks (DNNs), into an efficient inference tool for convolutional neural networks. Our optimization process on multicore ARM processors involves several high-level transformations of the original framework, such as the development and integration of Cython routines to exploit thread-level parallelism; the design and development of micro-kernels for the matrix multiplication, vectorized with ARMs NEON intrinsics, that can accommodate layer fusion; and the appropriate selection of several cache configuration parameters tailored to the memory hierarchy of the target ARM processors. Our experiments evaluate both inference throughput (measured in processed images/s) and inference latency (i.e., time-to-response) as well as energy consumption per image when varying the level of thread parallelism and the processor power modes. The experiments with the new inference engine are reported for the ResNet50 v1.5 model

08:06 Fusion-DHL: WiFi, IMU, and Floorplan Fusion for Dense History of Locations in Indoor Environments. (arXiv:2105.08837v1 [cs.RO])

The paper proposes a multi-modal sensor fusion algorithm that fuses WiFi, IMU, and floorplan information to infer an accurate and dense location history in indoor environments. The algorithm uses 1) an inertial navigation algorithm to estimate a relative motion trajectory from IMU sensor data; 2) a WiFi-based localization API in industry to obtain positional constraints and geo-localize the trajectory; and 3) a convolutional neural network to refine the location history to be consistent with the floorplan. We have developed a data acquisition app to build a new dataset with WiFi, IMU, and floorplan data with ground-truth positions at 4 university buildings and 3 shopping malls. Our qualitative and quantitative evaluations demonstrate that the proposed system is able to produce twice as accurate and a few orders of magnitude denser location history than the current standard, while requiring minimal additional energy consumption. We will publicly share our code, data and models.

19.05.2021
10:58 Proof of concept of a fast surrogate model of the VMEC code via neural networks in Wendelstein 7-X scenarios. (arXiv:2105.08467v1 [physics.plasm-ph])

In magnetic confinement fusion research, the achievement of high plasma pressure is key to reaching the goal of net energy production. The magnetohydrodynamic (MHD) model is used to self-consistently calculate the effects the plasma pressure induces on the magnetic field used to confine the plasma. Such MHD calculations serve as input for the assessment of a number of important physics questions. The VMEC code is the most widely used to evaluate 3D ideal-MHD equilibria, as prominently present in stellarators. However, considering the computational cost, it is rarely used in large-scale or online applications. Access to fast MHD equilbria is a challenging problem in fusion research, one which machine learning could effectively address. In this paper, we present artificial neural network (NN) models able to quickly compute the equilibrium magnetic field of W7-X. Magnetic configurations that extensively cover the device operational space, and plasma profiles with volume averaged

10:58 UncertaintyFuseNet: Robust Uncertainty-aware Hierarchical Feature Fusion with Ensemble Monte Carlo Dropout for COVID-19 Detection. (arXiv:2105.08590v1 [eess.IV])

The COVID-19 (Coronavirus disease 2019) has infected more than 151 million people and caused approximately 3.17 million deaths around the world up to the present. The rapid spread of COVID-19 is continuing to threaten human's life and health. Therefore, the development of computer-aided detection (CAD) systems based on machine and deep learning methods which are able to accurately differentiate COVID-19 from other diseases using chest computed tomography (CT) and X-Ray datasets is essential and of immediate priority. Different from most of the previous studies which used either one of CT or X-ray images, we employed both data types with sufficient samples in implementation. On the other hand, due to the extreme sensitivity of this pervasive virus, model uncertainty should be considered, while most previous studies have overlooked it. Therefore, we propose a novel powerful fusion model named $UncertaintyFuseNet$ that consists of an uncertainty module: Ensemble Monte Carlo (EMC) dropout.

17.05.2021
09:00 How Bayesian methods can improve $R$-matrix analyses of data: the example of the $dt$ Reaction. (arXiv:2105.06541v1 [nucl-th])

The $^3{\rm H}(d,n)^4{\rm He}$ reaction is of significant interest in nuclear astrophysics and nuclear applications. It is an important, early step in big-bang nucleosynthesis and a key process in nuclear fusion reactors. We use one- and two-level $R$-matrix approximations to analyze data on the cross section for this reaction at center-of-mass energies below 215 keV. We critically examine the data sets using a Bayesian statistical model that allows for both common-mode and additional point-to-point uncertainties. We use Markov Chain Monte Carlo sampling to evaluate this $R$-matrix-plus-statistical model and find two-level $R$-matrix results that are stable with respect to variations in the channel radii. The $S$ factor at 40 keV evaluates to $25.36(19)$ MeV b (68% credibility interval). We discuss our Bayesian analysis in detail and provide guidance for future applications of Bayesian methods to $R$-matrix analyses. We also discuss possible paths to further reduction of the $S$-factor

13.05.2021
05:48 A Survey on Reinforcement Learning-Aided Caching in Mobile Edge Networks. (arXiv:2105.05564v1 [cs.NI])

Mobile networks are experiencing tremendous increase in data volume and user density. An efficient technique to alleviate this issue is to bring the data closer to the users by exploiting the caches of edge network nodes, such as fixed or mobile access points and even user devices. Meanwhile, the fusion of machine learning and wireless networks offers a viable way for network optimization as opposed to traditional optimization approaches which incur high complexity, or fail to provide optimal solutions. Among the various machine learning categories, reinforcement learning operates in an online and autonomous manner without relying on large sets of historical data for training. In this survey, reinforcement learning-aided mobile edge caching is presented, aiming at highlighting the achieved network gains over conventional caching approaches. Taking into account the heterogeneity of sixth generation (6G) networks in various wireless settings, such as fixed, vehicular and flying networks,

05:48 The ionization of carbon at 10-100 times the diamond density and in the 10$^6$ K temperature range. (arXiv:2105.05707v1 [physics.plasm-ph])

The behaviour of partially ionized hot compressed matter is critical to the study of planetary interiors as well as for nuclear fusion studies. A recent quantum study of carbon in the 10-70 Gbar range and at a temperature of 100 eV used $N$-atom density functional theory (DFT) with $N\sim 32$-64 and molecular dynamics (MD). This involves band-structure type electronic calculations and averaging over many MD generated ion configurations. The calculated average number of free electrons per ion, viz., $\bar{Z}$, was systematically higher than from a standard average atom (AA) quantum calculation. To clarify this offset, we examine (a) the effect of the self-interaction (SI) error in such estimates (b) the possibility of carbon being a granular plasma containing Coulomb crystals. The possibility of magic-number' bound states is considered. The electrical conductivity, pressure, and the compressibility of the carbon system are examined. The very low conductivity and the high $\bar{Z}$

05:48 CoCoNet: Co-Optimizing Computation and Communication for Distributed Machine Learning. (arXiv:2105.05720v1 [cs.DC])

Modern deep learning workloads run on distributed hardware and are difficult to optimize -- data, model, and pipeline parallelism require a developer to thoughtfully restructure their workload around optimized computation and communication kernels in libraries such as cuBLAS and NCCL. The logical separation between computation and communication leaves performance on the table with missed optimization opportunities across abstraction boundaries. To explore these opportunities, this paper presents CoCoNet, which consists of a compute language to express programs with both computation and communication, a scheduling language to apply transformations on such programs, and a compiler to generate high performance kernels. Providing both computation and communication as first class constructs enables new optimizations, such as overlapping or fusion of communication with computation. CoCoNet allowed us to optimize several data, model and pipeline parallel workloads in existing deep learning

05:48 A Survey on Reinforcement Learning-Aided Caching in Mobile Edge Networks. (arXiv:2105.05564v1 [cs.NI])

Mobile networks are experiencing tremendous increase in data volume and user density. An efficient technique to alleviate this issue is to bring the data closer to the users by exploiting the caches of edge network nodes, such as fixed or mobile access points and even user devices. Meanwhile, the fusion of machine learning and wireless networks offers a viable way for network optimization as opposed to traditional optimization approaches which incur high complexity, or fail to provide optimal solutions. Among the various machine learning categories, reinforcement learning operates in an online and autonomous manner without relying on large sets of historical data for training. In this survey, reinforcement learning-aided mobile edge caching is presented, aiming at highlighting the achieved network gains over conventional caching approaches. Taking into account the heterogeneity of sixth generation (6G) networks in various wireless settings, such as fixed, vehicular and flying networks,

12.05.2021
08:31 An experimental characterization of core turbulence regimes in Wendelstein 7-X. (arXiv:2105.05107v1 [physics.plasm-ph])

First results from the optimized helias Wendelstein 7-X stellarator (W7-X) have shown that core transport is no longer mostly neoclassical, as is the case in previous kinds of stellarators. Instead, turbulent transport poses a serious limitation to the global performance of the machine. Several studies have found this particularly relevant for ion transport, with core ion temperatures becoming clamped at relatively low values of $T_{i} \simeq 1.7$ keV, except in the few scenarios in which turbulence can be suppressed. In order to understand turbulent mechanisms at play, it is important to have a clear understanding of the parametric dependencies of turbulent fluctuations, and the relation between them and turbulent transport. In this work we use Doppler reflectometry measurements carried out during a number of relevant operational scenarios to provide a systematic characterization of ion-scale ($k_\perp\rho_i\simeq 1$) density fluctuations in the core of W7-X. Then, we study the

08:31 Latest results on quiescent and post-disruption runaway electron mitigation experiments at Frascati Tokamak Upgrade. (arXiv:2105.04706v1 [physics.plasm-ph])

Analysis of experimental data collected in the last FTU campaigns provide interesting results on the deuterium large (wrt FTU volume) pellet capability of REs suppression, mainly due to the induced burst MHD activity expelling REs seed, on discharges with 0.5 MA and 5.3T. Clear signs of avalanche multiplication of REs after single pellet injection on 0.36 MA flat-top discharges is shown as well as quantitative indications of dissipative effects in terms of critical electrical field increase due to fan-like instabilities. Analysis of large fan-like instabilities on post-disruption RE beams, that seem to be correlated with current inversion in the central solenoid current as well as with background density drops, revealed the strong capability they have to suppress RE energy indicating a new possible strategy for RE energy suppression controlling large fan instabilities. We show how such useful density drops can be induced using modulated ECRH power.

08:31 Stability of pulsatile quasi-two-dimensional duct flows under a transverse magnetic field. (arXiv:2105.04686v1 [physics.flu-dyn])

This manuscript has been accepted for publication in Physical Review Fluids, see https://journals.aps.org/prfluids/accepted/53075Se8O0b1b109b1cc0061b280aaa122f0f92dc. The stability of a pulsatile quasi-two-dimensional duct flow was numerically investigated. Flow was driven, in concert, by a constant pressure gradient and by the synchronous oscillation of the lateral walls. This prototypical setup serves to aid understanding of unsteady magnetohydrodynamic flows in liquid metal coolant ducts subjected to transverse magnetic fields, motivated by the conditions expected in magnetic confinement fusion reactors. A wide range of wall oscillation frequencies and amplitudes were simulated. Focus was placed on the driving pulsation optimized for the greatest reduction in the critical Reynolds number, for a range of friction parameters $H$ (proportional to magnetic field strength). An almost $70$% reduction in the critical Reynolds number, relative to that for the steady base flow, was obtained

11.05.2021
11:18 Nonlinear MHD simulation of core plasma collapse events in stellarators. (arXiv:2105.04119v1 [physics.plasm-ph])

The core collapse events observed in a stellarator experiment are studied by a three-dimensional nonlinear MHD simulations. In the low magnetic shear configuration like the Wendelstein 7-X, the rotational transform profile is very sensitive to the toroidal current density. The 3D equilibrium with localized toroidal current density is studied. If the toroidal current density follows locally in the middle of the plasma minor radius, the rotational transform is also changed locally. Sometimes, the magnetic topology changes due to appearing the magnetic island. The nonlinear behaviors of the MHD instability are studied by a full three-dimensional nonlinear MHD code. It was found that a following sequence. At first, the high-n ballooning-type mode structure appears in the plasma core, and then the mode linearly grows. The high-n ballooning modes nonlinearly couple and saturate. The mode structure changes to the low-n mode. In that phase, the magnetic field structure becomes strongly

11:18 Experimental confirmation of efficient island divertor operation and successful neoclassical transport optimization in Wendelstein 7-X. (arXiv:2105.04002v1 [physics.plasm-ph])

We present recent highlights from the most recent operation phases of Wendelstein 7-X, the most advanced stellarator in the world. Stable detachment with good particle exhaust, low impurity content, and energy confinement times exceeding 100 ms, have been maintained for tens of seconds. Pellet fueling allows for plasma phases with reduced ITG turbulence, and during such phases, the overall confinement is so good (energy confinement times often exceeding 200 ms) that the attained density and temperature profiles would not have been possible in less optimized devices, since they would have had neoclassical transport losses exceeding the heating applied in W7-X. This provides proof that the reduction of neoclassical transport through magnetic field optimization is successful. W7-X plasmas generally show good impurity screening and high plasma purity, but there is evidence of longer impurity confinement times during turbulence-suppressed phases.

11:18 Real time prediction of probabilistic crack growth in complex structures with reduced-order models: Towards a Digital Twin of a helicopter component. (arXiv:2105.03668v1 [physics.app-ph])

To deploy the airframe Digital Twin or to conduct probabilistic evaluations of the remaining life of a structural component, a (near) real-time crack growth simulation method is critical. In this paper, a reduced-order simulation approach is developed to achieve this goal by leveraging two methods. On one hand, the SGBEM super element - FEM coupling method is combined with parametric modeling to generate the database of computed Stress Intensity Factors for cracks with various sizes/shapes in a complex structural component, by which hundreds of samples are automatically simulated within a day. On the other hand, machine learning methods are applied to establish the relation between crack sizes/shapes and crack front SIFs. By combining the reduced-order computational model with load inputs and fatigue growth laws, a real time prediction of probabilistic crack growth in complex structures with minimum computational burden is realized. In an example of a round-robin helicopter component,

10.05.2021
06:59 Elastically-isotropic open-cell minimal surface shell-lattices with superior stiffness via variable thickness design. (arXiv:2105.03046v1 [math.NA])

Triply periodic minimal surface (TPMS) shell-lattices are attracting increasingly attention due to their unique combination of geometric and mechanical properties, and their open-cell topology. However, uniform thickness TPMS shell-lattices are usually anisotropic in stiffness, namely having different Young's moduli along different lattice directions. To reduce the anisotropy, we propose a family of variable thickness TPMS shell-lattices with isotropic stiffness designed by a strain energy-based optimization algorithm. The optimization results show that all the five selected types of TPMS lattices can be made to achieve isotropic stiffness by varying the shell thickness, among which N14 and OCTO can maintain over 90% of the Hashin-Shtrikman upper bound of bulk modulus. All the optimized shell-lattices exhibit superior stiffness properties and significantly outperform elastically-isotropic truss-lattices. Both uniform and optimized types of N14 shell-lattices along [100], [110] and

06:59 Elastically-isotropic open-cell minimal surface shell-lattices with superior stiffness via variable thickness design. (arXiv:2105.03046v1 [math.NA])

Triply periodic minimal surface (TPMS) shell-lattices are attracting increasingly attention due to their unique combination of geometric and mechanical properties, and their open-cell topology. However, uniform thickness TPMS shell-lattices are usually anisotropic in stiffness, namely having different Young's moduli along different lattice directions. To reduce the anisotropy, we propose a family of variable thickness TPMS shell-lattices with isotropic stiffness designed by a strain energy-based optimization algorithm. The optimization results show that all the five selected types of TPMS lattices can be made to achieve isotropic stiffness by varying the shell thickness, among which N14 and OCTO can maintain over 90% of the Hashin-Shtrikman upper bound of bulk modulus. All the optimized shell-lattices exhibit superior stiffness properties and significantly outperform elastically-isotropic truss-lattices. Both uniform and optimized types of N14 shell-lattices along [100], [110] and

07.05.2021
08:01 A probabilistic model for missing traffic volume reconstruction based on data fusion. (arXiv:2105.02777v1 [physics.soc-ph])

Traffic volume information is critical for intelligent transportation systems. It serves as a key input to transportation planning, roadway design, and traffic signal control. However, the traffic volume data collected by fixed-location sensors, such as loop detectors, often suffer from the missing data problem and low coverage problem. The missing data problem could be caused by hardware malfunction. The low coverage problem is due to the limited coverage of fixed-location sensors in the transportation network, which restrains our understanding of the traffic at the network level. To tackle these problems, we propose a probabilistic model for traffic volume reconstruction by fusing fixed-location sensor data and probe vehicle data. We apply the probabilistic principal component analysis (PPCA) to capture the correlations in traffic volume data. An innovative contribution of this work is that we also integrate probe vehicle data into the framework, which allows the model to solve both

08:01 A probabilistic model for missing traffic volume reconstruction based on data fusion. (arXiv:2105.02777v1 [physics.soc-ph])

Traffic volume information is critical for intelligent transportation systems. It serves as a key input to transportation planning, roadway design, and traffic signal control. However, the traffic volume data collected by fixed-location sensors, such as loop detectors, often suffer from the missing data problem and low coverage problem. The missing data problem could be caused by hardware malfunction. The low coverage problem is due to the limited coverage of fixed-location sensors in the transportation network, which restrains our understanding of the traffic at the network level. To tackle these problems, we propose a probabilistic model for traffic volume reconstruction by fusing fixed-location sensor data and probe vehicle data. We apply the probabilistic principal component analysis (PPCA) to capture the correlations in traffic volume data. An innovative contribution of this work is that we also integrate probe vehicle data into the framework, which allows the model to solve both

08:01 Estimating Presentation Competence using Multimodal Nonverbal Behavioral Cues. (arXiv:2105.02636v1 [cs.CV])

Public speaking and presentation competence plays an essential role in many areas of social interaction in our educational, professional, and everyday life. Since our intention during a speech can differ from what is actually understood by the audience, the ability to appropriately convey our message requires a complex set of skills. Presentation competence is cultivated in the early school years and continuously developed over time. One approach that can promote efficient development of presentation competence is the automated analysis of human behavior during a speech based on visual and audio features and machine learning. Furthermore, this analysis can be used to suggest improvements and the development of skills related to presentation competence. In this work, we investigate the contribution of different nonverbal behavioral cues, namely, facial, body pose-based, and audio-related features, to estimate presentation competence. The analyses were performed on videos of 251 students

05.05.2021
06:20 An asymptotic-preserving 2D-2P relativistic Drift-Kinetic-Equation solver for runaway electron simulations in axisymmetric tokamaks. (arXiv:2105.01623v1 [physics.plasm-ph])

We propose an asymptotic-preserving (AP), uniformly convergent numerical scheme for the relativistic collisional Drift-Kinetic Equation (rDKE) to simulate runaway electrons in axisymmetric toroidal magnetic field geometries typical of tokamak devices. The approach is derived from an exact Green's function solution with numerical approximations of quantifiable impact, and results in a simple, two-step operator-split algorithm, consisting of a collisional Eulerian step, and a Lagrangian orbit-integration step with analytically prescribed kernels. The AP character of the approach is demonstrated by analysis of the dominant numerical errors, as well as by numerical experiments. We demonstrate the ability of the algorithm to provide accurate answers regardless of plasma collisionality on a circular axisymmetric tokamak geometry.

06:20 Approach to nonlinear magnetohydrodynamic simulations in stellarator geometry. (arXiv:2105.01186v1 [physics.plasm-ph])

The capability to model the nonlinear magnetohydrodynamic (MHD) evolution of stellarator plasmas is developed by extending the M3D-$C^1$ code to allow non-axisymmetric domain geometry. We introduce a set of logical coordinates, in which the computational domain is axisymmetric, to utilize the existing finite-element framework of M3D-$C^1$. A $C^1$ coordinate mapping connects the logical domain to the non-axisymmetric physical domain, where we use the M3D-$C^1$ extended MHD models essentially without modifications. We present several numerical verifications on the implementation of this approach, including simulations of the heating, destabilization, and equilibration of stellarator plasmas with strongly anisotropic thermal conductivity, and of the relaxation of stellarator equilibria to integrable and non-integrable magnetic field configurations in realistic geometries.

06:20 Application of Gaussian process regression to plasma turbulent transport model validation via integrated modelling. (arXiv:2105.01177v1 [physics.plasm-ph])

This paper outlines an approach towards improved rigour in tokamak turbulence transport model validation within integrated modelling. Gaussian process regression (GPR) techniques were applied for profile fitting during the preparation of integrated modelling simulations allowing for rigourous sensitivity tests of prescribed initial and boundary conditions as both fit and derivative uncertainties are provided. This was demonstrated by a JETTO integrated modelling simulation of the JET ITER-like-wall H-mode baseline discharge #92436 with the QuaLiKiz quasilinear turbulent transport model, which is the subject of extrapolation towards a deuterium-tritium plasma. The simulation simultaneously evaluates the time evolution of heat, particle, and momentum fluxes over $\sim10$ confinement times, with a simulation boundary condition at $\rho_{tor} = 0.85$. Routine inclusion of momentum transport prediction in multi-channel flux-driven transport modelling is not standard and is facilitated here

06:20 Neural network surrogate of QuaLiKiz using JET experimental data to populate training space. (arXiv:2105.01168v1 [physics.plasm-ph])

Within integrated tokamak plasma modelling, turbulent transport codes are typically the computational bottleneck limiting their routine use outside of post-discharge analysis. Neural network (NN) surrogates have been used to accelerate these calculations while retaining the desired accuracy of the physics-based models. This paper extends a previous NN model, known as QLKNN-hyper-10D, by incorporating the impact of impurities, plasma rotation and magnetic equilibrium effects. This is achieved by adding a light impurity fractional density ($n_{imp,light} / n_e$) and its normalized gradient, the normalized pressure gradient ($\alpha$), the toroidal Mach number ($M_{tor}$) and the normalized toroidal flow velocity gradient. The input space was sampled based on experimental data from the JET tokamak to avoid the curse of dimensionality. The resulting networks, named QLKNN-jetexp-15D, show good agreement with the original QuaLiKiz model, both by comparing individual transport quantity

06:20 Neural Weighted A*: Learning Graph Costs and Heuristics with Differentiable Anytime A*. (arXiv:2105.01480v1 [cs.LG])

Recently, the trend of incorporating differentiable algorithms into deep learning architectures arose in machine learning research, as the fusion of neural layers and algorithmic layers has been beneficial for handling combinatorial data, such as shortest paths on graphs. Recent works related to data-driven planning aim at learning either cost functions or heuristic functions, but not both. We propose Neural Weighted A*, a differentiable anytime planner able to produce improved representations of planar maps as graph costs and heuristics. Training occurs end-to-end on raw images with direct supervision on planning examples, thanks to a differentiable A* solver integrated into the architecture. More importantly, the user can trade off planning accuracy for efficiency at run-time, using a single, real-valued parameter. The solution suboptimality is constrained within a linear bound equal to the optimal path cost multiplied by the tradeoff parameter. We experimentally show the validity of

04.05.2021
09:04 Representing the boundary of stellarator plasmas. (arXiv:2105.00768v1 [physics.plasm-ph])

In stellarator optimization studies, the boundary of the plasma is usually described by Fourier series that are not unique: several sets of Fourier coefficients describe approximately the same boundary shape. A simple method for eliminating this arbitrariness is proposed and shown to work well in practice.

09:04 Code O-SUKI-N 3D: Upgraded Direct-Drive Fuel Target 3D Implosion Code in Heavy Ion Inertial Fusion. (arXiv:2105.00092v1 [physics.plasm-ph])

The Code O-SUKI-N 3D is an upgraded version of the 2D Code O-SUKI (Comput. Phys. Commun. 240, 83 (2019)). Code O-SUKI-N 3D is an integrated 3-dimensional (3D) simulation program system for fuel implosion, ignition and burning of a direct-drive nuclear-fusion pellet in heavy ion beam (HIB) inertial confinement fusion (HIF).The Code O-SUKI-N 3D consists of the three programs of Lagrangian fluid implosion program, data conversion program, and Euler fluid implosion, ignition and burning program. The Code O-SUKI-N 3D can also couple with the HIB illumination and energy deposition program of OK3 (Comput. Phys. Commun. 181, 1332 (2010)). The spherical target implosion 3D behavior is computed by the 3D Lagrangian fluid code until the time just before the void closure of the fuel implosion. After that, all the data by the Lagrangian implosion code are converted to the data for the 3D Eulerian code. In the 3D Euler code, the DT fuel compression at the stagnation, ignition and burning are

03.05.2021
04:57 RSSI-Based Location Classification Using a Particle Filter to Fuse Sensor Estimates. (arXiv:2104.14874v1 [cs.IT])

For Cyper-Physical Production Systems (CPPS), localization is becoming increasingly important as wireless and mobile devices are considered an integral part. While localizing targets in a wireless communication system based on the Received Signal Strength Indicators (RSSIs) is a usual solution, it is limited by sensor quality. We consider the scenario of a car moving in and out of a chamber and propose to use a particle filter for sensor fusion, allowing us to incorporate non-idealities in our model and achieve a high-quality position estimate. Then, we use Machine Learning (ML) to classify the vehicle position. Our results show that the location output of the particle filter is a better input to the classifiers than the raw RSSI data, and we achieve improved accuracy while simultaneously reducing the number of features that the ML has to consider. We also compare the performance of multiple ML algorithms and show that SVMs provide the overall best performance for the given task.

04:57 SRS-SBS competition and nonlinear laser energy absorption in a high temperature plasma. (arXiv:2104.15102v1 [physics.plasm-ph])

Stimulated Raman and Brillouin scattering of laser radiation in a plasma corona are outstanding issues for the inertial confinement fusion. Stimulated Raman scattering may produce absorption of a significant fraction of laser energy near the plasma quarter critical density associated with plasma cavitation and generation of hot electrons. By contrast, stimulated Brillouin scattering operates in a lower density plasma and prevents the laser light access to the absorption region. In the present paper, we report the results of analysis of competition of these two parametric instabilities with a series of one-dimensional kinetic simulations of laser-plasma interactions. By controlling the Brillouin backscattering through variation such plasma parameters as ion acoustic wave damping, divergence of the plasma expansion velocity or the laser bandwidth, we demonstrate the possibility of controlling the level of nonlinear laser absorption and scattering in a hot, weakly collisional plasma.

04:57 Solenoid-free current drive via ECRH in EXL-50 spherical torus plasmas. (arXiv:2104.14844v1 [physics.plasm-ph])

As a new spherical tokamak (ST) designed to simplify engineering requirements of a possible future fusion power source, the EXL-50 experiment features a low aspect ratio (A) vacuum vessel (VV), encircling a central post assembly containing the toroidal field coil conductors. Multiple electron cyclotron resonance heating (ECRH) resonances are located within the VV to possibly improve current drive effectiveness. The energetic electrons are observed via hard X-ray detectors, carry the bulk of the plasma current ranging from 50kA to 150kA, which is maintained for more than 1s duration. It is observed that over one Ampere current can be maintained per Watt of ECRH power issued from the 28-GHz gyrotrons. The plasma current with high line-density (approaching 1019m-2) has been achieved for plasma currents as high as 76kA. An analysis was carried out combining reconstructed multi-fluid equilibrium, guiding-center orbits, and resonant heating mechanisms. It is verified that in EXL-50 a broadly

04:57 RSSI-Based Location Classification Using a Particle Filter to Fuse Sensor Estimates. (arXiv:2104.14874v1 [cs.IT])

For Cyper-Physical Production Systems (CPPS), localization is becoming increasingly important as wireless and mobile devices are considered an integral part. While localizing targets in a wireless communication system based on the Received Signal Strength Indicators (RSSIs) is a usual solution, it is limited by sensor quality. We consider the scenario of a car moving in and out of a chamber and propose to use a particle filter for sensor fusion, allowing us to incorporate non-idealities in our model and achieve a high-quality position estimate. Then, we use Machine Learning (ML) to classify the vehicle position. Our results show that the location output of the particle filter is a better input to the classifiers than the raw RSSI data, and we achieve improved accuracy while simultaneously reducing the number of features that the ML has to consider. We also compare the performance of multiple ML algorithms and show that SVMs provide the overall best performance for the given task.

30.04.2021
17:47 Electron beam melting gets brittle metal into shape

Tungsten has the highest melting point of all metals, 3,422 degrees Celsius. This makes the material ideal for use at high temperatures in e.g. space rocket nozzles, heating elements of high-temperature furnaces, or the fusion reactor. However, the metal is highly brittle and, hence, difficult to process. Researchers of Karlsruhe Institute of Technology (KIT) have developed an innovative approach to making this brittle material soft. To process tungsten, they have determined new process parameters for electron beam melting.

05:43 Charge-state resolved laser acceleration of gold ions to beyond 7 MeV/u. (arXiv:2104.14520v1 [physics.plasm-ph])

In the past years, the interest in the laser-driven acceleration of heavy ions in the mass range of A ~ 200 has been increasing due to promising application ideas like the fission-fusion nuclear reaction mechanism, aiming at the production of neutron-rich isotopes relevant for the astrophysical r-process nucleosynthesis. In this paper, we report on the laser acceleration of gold ions to beyond 7 MeV/u, exceeding for the first time an important prerequisite for this nuclear reaction scheme. Moreover, the gold ion charge states have been detected with an unprecedented resolution, which enables the separation of individual charge states up to 4 MeV/u. The recorded charge-state distributions show a remarkable dependency on the target foil thickness and differ from simulations, lacking a straight-forward explanation by the established ionization models.

05:43 Kinetic simulation of electron cyclotron resonance assisted gas breakdown in split-biased waveguides for ITER collective Thomson scattering diagnostic. (arXiv:2104.14303v1 [physics.plasm-ph])

For the measurement of the dynamics of fusion-born alpha particles $E_\alpha \leq 3.5$ MeV in ITER using collective Thomson scattering (CTS), safe transmission of a gyrotron beam at mm-wavelength (1 MW, 60 GHz) passing the electron cyclotron resonance (ECR) in the in-vessel tokamak port plug' vacuum is a prerequisite. Depending on neutral gas pressure and composition, ECR-assisted gas breakdown may occur at the location of the resonance, which must be mitigated for diagnostic performance and safety reasons. The concept of a split electrically biased waveguide (SBWG) has been previously demonstrated in [C.P. Moeller, U.S. Patent 4,687,616 (1987)]. The waveguide is longitudinally split and a kV bias voltage applied between the two halves. Electrons are rapidly removed from the central region of high radio frequency electric field strength, mitigating breakdown. As a full scale experimental investigation of gas and electromagnetic field conditions inside the ITER equatorial port plugs is

27.04.2021
06:54 Effect of collisions on non-adiabatic electron dynamics in ITG-driven microturbulence. (arXiv:2104.12585v1 [physics.plasm-ph])

Non-adiabatic electron response leads to significant changes in Ion Temperature Gradient (ITG) eigenmodes. For instance, it can produce fine-structures that are significantly extended along the magnetic field lines at the Mode Rational Surface (MRS) corresponding to each eigenmode. These eigenmodes can then nonlinearly interact with themselves to produce zonal flows via a process called the self-interaction mechanism. In this paper, the effect of collisions on these processes are studied. In presence of non-adiabatic electrons, growth rate of ITG eigenmodes decreases with increasing collisionality. Through detailed velocity space analysis of the distribution function, we show that this can be explained with collisions leading to a more adiabatic-like response of electrons away from MRSs. Collisions broaden the radial width of the fine structures, which translates to narrower tails of the eigenmode in extended ballooning space. The characteristic parallel scale length associated to

00:22 Fooling fusion fuel: How to discipline unruly plasma

Scientists have developed a type of deception to calm unruly plasma and accelerate the harvesting on Earth of fusion energy.

26.04.2021
22:26 Fooling fusion fuel: How to discipline unruly plasma

The process designed to harvest on Earth the fusion energy that powers the sun and stars can sometimes be tricked. Researchers at the U.S. Department of Energy's (DOE) Princeton Plasma Physics laboratory have derived and demonstrated a bit of slight-of-hand called "quasi-symmetry" that could accelerate the development of fusion energy as a safe, clean and virtually limitless source of power for generating electricity.

09:45 A self-consistent model of the plasma staircase and nonlinear Schr\"odinger equation with subquadratic power nonlinearity. (arXiv:2104.11582v1 [physics.plasm-ph])

A new basis has been found for the theory of self-organization of transport avalanches and jet zonal flows in L-mode tokamak plasma, the so-called "plasma staircase." The jet zonal flows are considered as a wave packet of coupled nonlinear oscillators characterized by a complex time- and wave-number dependent wave function; in a mean-field approximation this function is argued to obey a discrete nonlinear Schr\"odinger equation with subquadratic power nonlinearity. It is shown that the subquadratic power leads directly to a white L\'evy noise, and to a L\'evy-fractional Fokker-Planck equation for radial transport of test particles (via wave-particle interactions). In a self-consistent description the avalanches, which are driven by the white L\'evy noise, interact with the jet zonal flows, which form a system of semi-permeable barriers to radial transport. We argue that the plasma staircase saturates at a state of marginal stability, in whose vicinity the avalanches undergo an

23.04.2021
04:05 Self-consistent simulation of resistive kink instabilities with runaway electrons. (arXiv:2104.10806v1 [physics.plasm-ph])

A new fluid model for runaway electron simulation based on fluid description is introduced and implemented in the magnetohydrodynamics code M3D-C1, which includes self-consistent interactions between plasma and runaway electrons. The model utilizes the method of characteristics to solve the continuity equation for the runaway electron density with large convection speed, and uses a modified Boris algorithm for pseudo particle pushing. The model was employed to simulate magnetohydrodynamics instabilities happening in a runaway electron final loss event in the DIII-D tokamak. Nonlinear simulation reveals that a large fraction of runaway electrons get lost to the wall when kink instabilities are excited and form stochastic field lines in the outer region of the plasma. Plasma current converts from runaway electron current to Ohmic current, and get pinched at the magnetic axis. Given the good agreement with experiment, the simulation model provides a reliable tool to study macroscopic

22.04.2021
15:33 Team improves polar direct drive fusion neutron sources for use in laser experiments

Scientists from Lawrence Livermore National Laboratory (LLNL) and the Laboratory for Laser Energetics (LLE) are working to improve polar direct drive (PDD) neutron sources on the National Ignition Facility (NIF), the world's most energetic laser.

07:39 Team improves polar direct drive fusion neutron sources for use in National Ignition Facility experiments

Scientists from Lawrence Livermore National Laboratory (LLNL) and the Laboratory for Laser Energetics (LLE) are working to improve polar direct drive (PDD)

20.04.2021
05:01 On the accuracy of the binary-collision algorithm in particle-in-cell simulations of magnetically confined fusion plasmas. (arXiv:2104.09100v1 [physics.plasm-ph])

Ideally, binary-collision algorithms conserve kinetic momentum and energy. In practice, the finite size of collision cells and the finite difference in the particle locations affect the conservation properties. In the present work, we investigate numerically how the accuracy of these algorithms is affected when the size of collision cells is large compared with gradient scale length of the background plasma, a parameter essential in full-f fusion plasma simulations. Additionally, we discuss implications for the conserved quantities in drift-kinetic formulations when fluctuating magnetic and electric fields are present: we suggest how the accuracy of the algorithms could potentially be improved with minor modifications.

19.04.2021
07:50 Study of the Alfven Eigenmodes stability in CFQS plasma using a Landau closure model. (arXiv:2104.07987v1 [physics.plasm-ph])

The aim of this study is to analyze the stability of the Alfven eigenmodes (AE) in the Chinese First Quasi-axisymmetric Stellarator (CFQS). The AE stability is calculated using the code FAR3d that solves the reduced MHD equations to describe the linear evolution of the poloidal flux and the toroidal component of the vorticity in a full 3D system, coupled with equations of density and parallel velocity moment for the energetic particles (EP) species including the effect of the helical couplings and acoustic modes. The Landau damping and resonant destabilization effects are added in the model by a given closure relation. The simulation results indicate the destabilization of n = 1 to 4 AEs by EP during the slowing down process, particularly n = 1 and n = 2 toroidal AEs (TAE), n = 3 elliptical AE (EAE) and n = 4 non circular AE (NAE). If the resonance is caused by EPs with an energy above 17 keV (weakly thermalized EP), n = 2 EAEs and n = 3 NAEs are unstable. On the other hand, EPs with

07:50 Cross-beam energy transfer saturation by ion trapping-induced detuning. (arXiv:2104.07725v1 [physics.plasm-ph])

The performance of direct-drive inertial confinement fusion implosions relies critically on the coupling of laser energy to the target plasma. Cross-beam energy transfer (CBET), the resonant exchange of energy between intersecting laser beams mediated by ponderomotively driven ion-acoustic waves (IAW), inhibits this coupling by scattering light into unwanted directions. The variety of beam intersection angles and varying plasma conditions in an implosion results in IAWs with a range of phase velocities. Here we show that CBET saturates through a resonance detuning that depends on the IAW phase velocity and that results from trapping-induced modifications to the ion distribution functions. For smaller phase velocities, the modifications to the distribution functions can rapidly thermalize in the presence of mid-Z ions, leading to a blueshift in the resonant frequency. For larger phase velocities, the modifications can persist, leading to a redshift in the resonant frequency. Ultimately,

07:50 Serial or Parallel? Plug-able Adapter for multilingual machine translation. (arXiv:2104.08154v1 [cs.CL])

Developing a unified multilingual translation model is a key topic in machine translation research. However, existing approaches suffer from performance degradation: multilingual models yield inferior performance compared to the ones trained separately on rich bilingual data. We attribute the performance degradation to two issues: multilingual embedding conflation and multilingual fusion effects. To address the two issues, we propose PAM, a Transformer model augmented with defusion adaptation for multilingual machine translation. Specifically, PAM consists of embedding and layer adapters to shift the word and intermediate representations towards language-specific ones. Extensive experiment results on IWSLT, OPUS-100, and WMT benchmarks show that \method outperforms several strong competitors, including series adapter and multilingual knowledge distillation.

18.04.2021
00:33 Tracking the progress of fusion power through 60 years of neutral particle analysis

A review paper examines the 6-decade history of neutral particle analysis (NPA), a vital diagnostic tool used in magnetic plasma confinement devices such as tokamaks that will house the nuclear fusion process and generate the clean energy of the future.

16.04.2021
15:50 Scientists reject restrictive heat flux models using directly driven gold spheres

A team of scientists has conducted an analysis of directly driven gold sphere experiments to test heat transport models used in inertial confinement fusion (ICF) and high energy density (HED) modeling. It was found that overly restricting the heat flux caused disagreement with measurement.

08:54 Benchmarking the Scrape-Off-Layer Fast Ion (SOLFI) particle tracer code in a collisionless magnetic mirror with electrostatic potential drop. (arXiv:2104.07547v1 [physics.plasm-ph])

Optimizing the confinement and transport of fast ions is an important consideration in the design of modern fusion reactors. For spherical tokamaks in particular, fast ions can significantly influence global plasma behavior because their large drift orbits often sample both core and scrape-off-layer (SOL) plasma conditions. Their Larmor radii are also comparable to the SOL width, rendering the commonly chosen guiding center approximations inappropriate. Accurately modeling the behavior of fast ions therefore requires retaining a complete description of the fast ion orbit including its Larmor motion. Here, we introduce the Scrape-Off-Layer Fast Ion (SOLFI) code, which is a new and versatile full-orbit Monte Carlo particle tracer being developed to follow fast ion orbits inside and outside the separatrix. We benchmark SOLFI in a simple straight mirror geometry and show that the code (i) conserves particle energy and magnetic moment, (ii) obtains the correct passing boundary for particles

08:54 Nanosecond machine learning event classification with boosted decision trees in FPGA for high energy physics. (arXiv:2104.03408v1 [hep-ex] CROSS LISTED)

We present a novel implementation of classification using the machine learning / artificial intelligence method called boosted decision trees (BDT) on field programmable gate arrays (FPGA). The firmware implementation of binary classification requiring 100 training trees with a maximum depth of 4 using four input variables gives a latency value of about 10 ns, which corresponds to 3 clock ticks at 320 MHz in our setup. The low timing values are achieved by restructuring the BDT layout and reconfiguring its parameters. The FPGA resource utilization is also kept low at a range from 0.01% to 0.2% in our setup. A software package called fwXmachina achieves this implementation. Our intended audience is a user of custom electronics-based trigger systems in high energy physics experiments or anyone that needs decisions at the lowest latency values for real-time event classification. Two problems from high energy physics are considered, in the separation of electrons vs. photons and in the

08:54 Serverless Federated Learning for UAV Networks: Architecture, Challenges, and Opportunities. (arXiv:2104.07557v1 [cs.NI])

Unmanned aerial vehicles (UAVs), or say drones, are envisioned to support extensive applications in next-generation wireless networks in both civil and military fields. Empowering UAVs networks intelligence by artificial intelligence (AI) especially machine learning (ML) techniques is inevitable and appealing to enable the aforementioned applications. To solve the problems of traditional cloud-centric ML for UAV networks such as privacy concern, unacceptable latency, and resource burden, a distributed ML technique, i.e., federated learning (FL), has been recently proposed to enable multiple UAVs to collaboratively train ML model without letting out raw data. However, almost all existing FL paradigms are server-based, i.e., a central entity is in charge of ML model aggregation and fusion over the whole network, which could result in the issue of a single point of failure and are inappropriate to UAV networks with both unreliable nodes and links. To address the above issue, in this

15.04.2021
20:25 Tracking the progress of fusion power through 60 years of neutral particle analysis

As the world's energy demands grow, so too does growing concern over the environmental impact of power production. The need for a safe, clean, and reliable energy source has never been clearer. Fusion power could fulfil such a need. A review paper published in The European Physical Journal H examines the 6-decade history of neutral particle analysis (NPA), developed in Ioffe Institute, Saint Petersburg, Russia, a vital diagnostic tool used in magnetic plasma confinement devices such as tokamaks that will house the nuclear fusion process and generate the clean energy of the future.

04:20 Lack of Fusion in Additive Manufacturing: Defect or Asset?. (arXiv:2104.07014v1 [cond-mat.mtrl-sci])

Rapid cooling rates and stochastic interactions between the heat source and feedstock in additive manufacturing (AM) result in strong anisotropy and process-induced defects deteriorating the tensile ductility and fatigue resistance of printed parts. We show that by deliberately introducing a high density of lack of fusion (LoF) defects, a processing regime that has been avoided so far, followed by pressure assisted heat treatment, we can print Ti-6Al-4V with reduced texture and exceptional properties surpassing that of wrought, cast, forged, annealed, and solution-treated and aged counterparts. Such improvement is achieved through the formation of low aspect ratio {\alpha}-grains around LoF defects upon healing, surrounded by {\alpha}-laths. This occurrence is attributed to surface energy reduction and recrystallization events taking place during healing of LoF defects. Our approach to design duplex microstructures is applicable to a wide range of AM processes and alloys and can be

04:20 Tungsten boride shields in a spherical tokamak. (arXiv:2104.06886v1 [physics.comp-ph])

The favourable properties of tungsten borides for shielding the central High Temperature Superconductor (HTS) core of a spherical tokamak fusion power plant are modelled using the MCNP code. The objectives are to minimize the power deposition into the cooled HTS core, and to keep HTS radiation damage to acceptable levels by limiting the neutron and gamma fluxes. The shield materials compared are W2B, WB, W2B5 and WB4 along with a reactively sintered boride B0.329C0.074Cr0.024Fe0.274W0.299, monolithic W and WC. Of all these W2B5 gave the most favourable results with a factor of ~10 or greater reduction in neutron flux and gamma energy deposition as compared to monolithic W. These results are compared with layered water-cooled shields, giving the result that the monolithic shields, with moderating boron, gave comparable neutron flux and power deposition, and (in the case of W2B5) even better performance. Good performance without water-coolant has advantages from a reactor safety

01:33 Dell, Nvidia power new cloud-native supercomputer in the UK (ZDNet - Between the Lines)

ZDNet - Between the LinesDell, Nvidia power new cloud-native supercomputer in the UK - The expanded system at the University of Cambridge will deliver multi-tenant high performance computing for research spanning astrophysics, nuclear fusion power generation and clinical medicine applications. ...

14.04.2021
13:31 The Electric Universe Heresy

In this groundbreaking paper Wal Thornhill introduces a new Theory of Everything: The Electric Universe. Set aside everything you think you know about all things great and small because the ideas presented here overturn it all. Was there a big bang? Not likely. Einstein’s Relativity? Doesn’t hold up. Is the Sun a thermonuclear fusion reactor which will eventually run out of fuel and burn out? Nope. Are there black holes? No such thing. What about dark matter and dark energy? Forget about that nonsense and start learning about the science of the 21st century. “. . .the Electric Universe is the only coherent cosmology that has correctly predicted and explained discoveries in the space age.” For example, Thornhill specifically predicted the unexpected results of the Deep Impact mission to comet Tempel 1 in October 2001, almost four years before the event. He was alone in successfully predicting what would be seen beneath the clouds of Saturn’s moon Titan.

07:00 Numerics and computation in gyrokinetic simulations of electromagnetic turbulence with global particle-in-cell codes. (arXiv:2104.06282v1 [physics.plasm-ph])

Electromagnetic turbulence is addressed in tokamak and stellarator plasmas with the global gyrokinetic particle-in-cell codes ORB5 [E. Lanti et al, Comp. Phys. Comm, vol. 251, 107072 (2020)] and EUTERPE. The large-aspect-ratio tokamak, down-scaled ITER, and Wendelstein 7-X geometries are considered. The main goal is to increase the plasma beta, the machine size, the ion-to-electron mass ratio, as well as to include realistic-geometry features in such simulations. The associated numerical requirements and the computational cost for the cases on computer systems with massive GPU deployments are investigated. These are necessary steps to enable electromagnetic turbulence simulations in future reactor plasmas.

07:00 Energy Balance Within Thermonuclear Reactors. (arXiv:2104.06251v1 [physics.plasm-ph])

Thermonuclear reactors hold a great promise for the future of Humankind. Within Tokamak and Stellarator reactors, plasma is confined by twisted magnetic fields. Reactors which produce fusion energy have existed since Princeton Large Torus Tokamak in 1978, nevertheless in all reactors built up to now, energy loss from plasma vastly exceeded fusion energy production. In order for a thermonuclear power plant to run, generated fusion energy must significantly exceed energy loss by the plasma. There are four processes by which plasma looses energy -- neutron radiation, Bremsstrahlung radiation, synchrotron radiation, and heat conduction to the walls. For a deuterium -- tritium reactor, 80\% of energy produced by fusion is lost to neutron radiation, about 4\% to 6\% of fusion energy is lost to Bremsstrahlung and synchrotron radiation. For a deuterium -- $^3$He reactor, 5\% of energy produced by fusion is lost to neutron radiation, about 50\% to 75\% of fusion energy is lost to Bremsstrahlung

07:00 Relevance-Aware Anomalous Users Detection in Social Network. (arXiv:2104.06095v1 [cs.SI])

Anomalous users detection in social network is an imperative task for security problems. Graph Neural Networks (GNNs) have been widely applied to reveal the suspicious nodes or relevance by aggregating the neighborhood information. However, the increasing scale of users' behaviors and interactions as well as the manifold camouflage techniques have largely complicated the discrimination process, which result in the serious loss of detection accuracy. In this paper, we propose an innovate Relevance-aware Anomalous Users Detection model (RAUD), which extracts and learns both the explicit and implicit relevance from meta data based on heterogeneous information network (HIN) and GNNs. We employ two distinct exploring layers including a self-attention layer and a similarity calculation layer to fully consider the users' intimacy across different relationships, and a following fusion layer based on graph convolution network(GCN) to consolidate the similarity and enhance the node embedding. We

13.04.2021
09:39 High-Throughput Virtual Screening of Small Molecule Inhibitors for SARS-CoV-2 Protein Targets with Deep Fusion Models. (arXiv:2104.04547v1 [cs.LG])

Structure-based Deep Fusion models were recently shown to outperform several physics- and machine learning-based protein-ligand binding affinity prediction methods. As part of a multi-institutional COVID-19 pandemic response, over 500 million small molecules were computationally screened against four protein structures from the novel coronavirus (SARS-CoV-2), which causes COVID-19. Three enhancements to Deep Fusion were made in order to evaluate more than 5 billion docked poses on SARS-CoV-2 protein targets. First, the Deep Fusion concept was refined by formulating the architecture as one, coherently backpropagated model (Coherent Fusion) to improve binding-affinity prediction accuracy. Secondly, the model was trained using a distributed, genetic hyper-parameter optimization. Finally, a scalable, high-throughput screening capability was developed to maximize the number of ligands evaluated and expedite the path to experimental evaluation. In this work, we present both the methods

09:39 Stellarators as a Fast Path to Fusion. (arXiv:2104.04621v1 [physics.plasm-ph])

The enhancement of the atmospheric concentration of carbon dioxide is doubling every forty years. Options must be developed for energy production that do not drive carbon-dioxide emissions and can be fully deployed within a few doubling times. Unit size and cost of electricity are only relevant in comparison to alternative worldwide energy solutions. Intermittency, site specificity, waste management, and nuclear proliferation make fusion attractive as the basis for a carbon-free energy system compared to the alternatives. Nonetheless, fusion will not be an option for deployment until a power plant has successfully operated. A critical element in a minimal time and risk program to operate a fusion power plant is the use of computational design as opposed to just extrapolation. The importance of minimizing time and risk is so great that ideally more than one concept would be pursued. Unfortunately, only the stellarator has an empirical demonstration of the reliability of computational

09:39 Model fusion with physics-guided machine learning. (arXiv:2104.04574v1 [physics.comp-ph])

The unprecedented amount of data generated from experiments, field observations, and large-scale numerical simulations at a wide range of spatio-temporal scales have enabled the rapid advancement of data-driven and especially deep learning models in the field of fluid mechanics. Although these methods are proven successful for many applications, there is a grand challenge of improving their generalizability. This is particularly essential when data-driven models are employed within outer-loop applications like optimization. In this work, we put forth a physics-guided machine learning (PGML) framework that leverages the interpretable physics-based model with a deep learning model. The PGML framework is capable of enhancing the generalizability of data-driven models and effectively protect against or inform about the inaccurate predictions resulting from extrapolation. We apply the PGML framework as a novel model fusion approach combining the physics-based Galerkin projection model and

09:39 High-Throughput Virtual Screening of Small Molecule Inhibitors for SARS-CoV-2 Protein Targets with Deep Fusion Models. (arXiv:2104.04547v1 [cs.LG])

Structure-based Deep Fusion models were recently shown to outperform several physics- and machine learning-based protein-ligand binding affinity prediction methods. As part of a multi-institutional COVID-19 pandemic response, over 500 million small molecules were computationally screened against four protein structures from the novel coronavirus (SARS-CoV-2), which causes COVID-19. Three enhancements to Deep Fusion were made in order to evaluate more than 5 billion docked poses on SARS-CoV-2 protein targets. First, the Deep Fusion concept was refined by formulating the architecture as one, coherently backpropagated model (Coherent Fusion) to improve binding-affinity prediction accuracy. Secondly, the model was trained using a distributed, genetic hyper-parameter optimization. Finally, a scalable, high-throughput screening capability was developed to maximize the number of ligands evaluated and expedite the path to experimental evaluation. In this work, we present both the methods

12.04.2021
05:04 Evaluation of silicon carbide as a divertor armor material in DIII-D H-mode discharges. (arXiv:2104.04083v1 [physics.plasm-ph])

Silicon carbide (SiC) represents a promising but largely untested plasma-facing material (PFM) for next-step fusion devices. In this work, an analytic mixed-material erosion model is developed by calculating the physical (via SDTrimSP) and chemical (via empirical scalings) sputtering yield from SiC, Si, and C. The Si content in the near-surface SiC layer is predicted to increase during D plasma bombardment due to more efficient physical and chemical sputtering of C relative to Si. Silicon erosion from SiC thereby occurs primarily from sputtering of the enriched Si layer, rather than directly from the SiC itself. SiC coatings on ATJ graphite, manufactured via chemical vapor deposition, were exposed to repeated H-mode plasma discharges in the DIII-D tokamak to test this model. The qualitative trends from analytic modeling are reproduced by the experimental measurements, obtained via spectroscopic inference using the S/XB method. Quantitatively the model slightly under-predicts measured

05:04 Fast Regression of the Tritium Breeding Ratio in Fusion Reactors. (arXiv:2104.04026v1 [physics.comp-ph])

The tritium breeding ratio (TBR) is an essential quantity for the design of modern and next-generation D-T fueled nuclear fusion reactors. Representing the ratio between tritium fuel generated in breeding blankets and fuel consumed during reactor runtime, the TBR depends on reactor geometry and material properties in a complex manner. In this work, we explored the training of surrogate models to produce a cheap but high-quality approximation for a Monte Carlo TBR model in use at the UK Atomic Energy Authority. We investigated possibilities for dimensional reduction of its feature space, reviewed 9 families of surrogate models for potential applicability, and performed hyperparameter optimisation. Here we present the performance and scaling properties of these models, the fastest of which, an artificial neural network, demonstrated $R^2=0.985$ and a mean prediction time of $0.898\ \mu\mathrm{s}$, representing a relative speedup of $8\cdot 10^6$ with respect to the expensive MC model. We

05:04 Fast Regression of the Tritium Breeding Ratio in Fusion Reactors. (arXiv:2104.04026v1 [physics.comp-ph])

The tritium breeding ratio (TBR) is an essential quantity for the design of modern and next-generation D-T fueled nuclear fusion reactors. Representing the ratio between tritium fuel generated in breeding blankets and fuel consumed during reactor runtime, the TBR depends on reactor geometry and material properties in a complex manner. In this work, we explored the training of surrogate models to produce a cheap but high-quality approximation for a Monte Carlo TBR model in use at the UK Atomic Energy Authority. We investigated possibilities for dimensional reduction of its feature space, reviewed 9 families of surrogate models for potential applicability, and performed hyperparameter optimisation. Here we present the performance and scaling properties of these models, the fastest of which, an artificial neural network, demonstrated $R^2=0.985$ and a mean prediction time of $0.898\ \mu\mathrm{s}$, representing a relative speedup of $8\cdot 10^6$ with respect to the expensive MC model. We

10.04.2021
05:38 TAE Technologies Claims Landmark In Fusion Energy, Sees Commercialization By 2030 (Slashdot)

SlashdotTAE Technologies Claims Landmark In Fusion Energy, Sees Commercialization By 2030 - TAE Technologies, a 20-year-old fusion energy technology developer, is claiming to have hit a milestone in the development of a new technology for generation power from nuclear fusion. The company said its reactors could be operating at commercial scale by the end of the decade, thanks to its newfound ability ...

09.04.2021
17:03 This Nuclear Reactor Just Made Fusion Viable by 2030. Seriously.

TAE Technologies, the world’s largest private fusion company, has announced it will have a commercially viable nuclear fusion power plant by 2030, which puts it years—or even decades—ahead of other fusion technology companies. ? You love nuclear. So do we. Let’s nerd out over nuclear together. The California-based company has raised $880 million in funding for its hydrogen-boron reactor. This reactor isn’t a traditional tokamak or stellarator; instead, it uses a confined particle acceleration mechanism that produces and confines plasma. All fusion technology has plasma, which mimics the extreme reactions that power all the stars—it’s what we’re emulating when we make fusion energy experiments. “Plasma is an oozy substance; the challenge of containing it is akin to holding Jell-O together using rubber bands,” TAE says on its website. 00:39 CORRECTED-Google-backed TAE Technologies raises$280 mln from new, existing investors (Reuters)

ReutersCORRECTED-Google-backed TAE Technologies raises $280 mln from new, existing investors - TAE Technologies, a California-based firm building technology to generate power from nuclear fusion, said on Thursday it had raised$280 million from new and existing investors, including Google and New Enterprise Associates. ...

00:39 Google-backed TAE Technologies raises $280 million from new, existing investors (Reuters) ReutersGoogle-backed TAE Technologies raises$280 million from new, existing investors - (Corrects to remove inaccurate company descriptor from headline. Also corrects paragraph 3 to say that the company plans to use the funds for further work related to nuclear fusion) ...

08.04.2021
23:31 Google-backed nuclear energy firm TAE Technologies raises $280 million (Reuters) ReutersGoogle-backed nuclear energy firm TAE Technologies raises$280 million - TAE Technologies, a California-based firm building technology to generate power from nuclear fusion, said on Thursday it had raised \$280 million from new and existing investors, including Google and New Enterprise Associates. ...