Article
Physics, Fluids & Plasmas
Abraham Chien, Lan Gao, Shu Zhang, Hantao Ji, Eric Blackman, Hui Chen, Gennady Fiksel, Kenneth Hill, Philip Nilson
Summary: In the experiment, Megagauss magnetic fields were generated by currents passing through a U-shaped coil connecting parallel copper foils. The study showed that the magnetic field strength decays with increasing laser pulse width, and a lumped-circuit model demonstrated an ion shorting effect on the coil current.
PHYSICS OF PLASMAS
(2021)
Article
Geochemistry & Geophysics
Alfonso Lopez, Carlos J. Ogayar, Juan M. Jurado, Francisco R. Feito
Summary: In this work, a GPU-based LiDAR scanner simulator is presented for efficient generation of large labeled datasets. The simulator can emulate a wide range of LiDAR sensor models and use compute shaders to generate beams and solve their collision with the environment. It is mainly intended for rapid dataset generation for neural networks and has shown superior performance compared to sequential LiDAR scanning.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Review
Engineering, Aerospace
Xi Chen, Lei Dong, Hong-Chang Li, Xin-Peng Yao, Peng Wang, Shuang Yao
Summary: Defects and errors in code pose a potential risk to software operation and require a proper code review process, especially for safety-critical software. The traditional manual review method is no longer sufficient due to the increasing size and variety of code. Deep Reviewer is a flexible framework that automatically detects code defects and correlates review comments. It achieves high precision and F1 scores and outperforms other methods in multi-classification tasks.
Article
Multidisciplinary Sciences
Adam V. Dvorak, Dushyant Kumar, Jing Zhang, Guillaume Gilbert, Sharada Balaji, Neale Wiley, Cornelia Laule, G. R. Wayne Moore, Alex L. Mackay, Shannon H. Kolind
Summary: The study introduces the CALIPR framework to improve MRI techniques, allowing high-quality images to be obtained in a shorter period of time. This framework has significantly improved the precision and sensitivity of myelin water imaging, and is of great significance for research on the brain and spinal cord.
Article
Astronomy & Astrophysics
Kengo Tomida, James M. Stone
Summary: We describe the implementation of multigrid solvers in the Athena++ AMR framework for solving the Poisson equation for self-gravity. The new solvers are built on top of the AMR hierarchy and TaskList framework for efficient parallelization, and various boundary conditions are implemented. The method significantly outperforms fast Fourier transform-based methods on a uniform grid and requires only a small fraction of the computing time of the magnetohydrodynamic solver in Athena++.
ASTROPHYSICAL JOURNAL SUPPLEMENT SERIES
(2023)
Article
Chemistry, Physical
Stefan Doerr, Maciej Majewski, Adria Perez, Andreas Kramer, Cecilia Clementi, Frank Noe, Toni Giorgino, Gianni De Fabritiis
Summary: TorchMD is a molecular simulation framework that utilizes both classical and machine learning potentials. It allows for various force computations and supports learning and simulating neural network potentials. Through validation, it has been proven to be a useful toolkit for molecular simulations with machine learning potentials.
JOURNAL OF CHEMICAL THEORY AND COMPUTATION
(2021)
Article
Biotechnology & Applied Microbiology
Xiaogai Li
Summary: The study presents a personalization framework for generating subject-specific models across the lifespan and for pathological brains with significant anatomical changes. The framework includes hierarchical multiple feature and multimodality imaging registrations, mesh morphing, and mesh grouping, shown to be efficient with a heterogeneous dataset. The generated models demonstrate competitive personalization accuracy, allowing for age-dependent and groupwise brain injury mechanisms to be studied.
FRONTIERS IN BIOENGINEERING AND BIOTECHNOLOGY
(2021)
Article
Engineering, Civil
Ramesh Ramasamy Pandi, Song Guang Ho, Sarat Chandra Nagavarapu, Justin Dauwels
Summary: This paper introduces a GPU-based solution methodology for the dial-a-ride problem to generate good solutions in a short time. By accelerating time-critical neighborhood exploration and employing device-oriented optimization strategies, the utilization of a current-generation GPU architecture (Tesla P100) is enhanced. Experimental results show that the proposed GPU methodology can generate better solutions in a short time compared to existing sequential approaches.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2021)
Article
Astronomy & Astrophysics
Lena Murchikova, Christopher J. White, Sean M. Ressler
Summary: By comparing the 230 GHz near-horizon emission from Sagittarius A* with simulations, this study finds that the models based on feeding by stellar winds match the observations very well, while the models based on torus-fed accretion disks show noticeable discrepancies.
ASTROPHYSICAL JOURNAL LETTERS
(2022)
Article
Multidisciplinary Sciences
T. Kurz, T. Heinemann, M. F. Gilljohann, Y. Y. Chang, J. P. Couperus Cabadag, A. Debus, O. Kononenko, R. Pausch, S. Schoebel, R. W. Assmann, M. Bussmann, H. Ding, J. Goetzfried, A. Koehler, G. Raj, S. Schindler, K. Steiniger, O. Zarini, S. Corde, A. Doepp, B. Hidding, S. Karsch, U. Schramm, A. Martinez de la Ossa, A. Irman
Summary: Plasma wakefield accelerators can achieve higher accelerating fields than traditional accelerator modules, with beam-driven wakefields being particularly suitable for high-quality beam generation and acceleration. The demonstration of a millimeter-scale plasma accelerator powered by laser-accelerated electron beams shows promise for compact and high-gradient acceleration. By combining controlled drive and witness electron bunches, fundamental studies of beam-driven plasma accelerators can be conducted at high-power laser facilities, potentially providing compact sources of energetic high-brightness electron beams.
NATURE COMMUNICATIONS
(2021)
Article
Multidisciplinary Sciences
A. Lawrence Gould, Robert K. Campbell, John W. Loewy, Robert A. Beckman, Jyotirmoy Dey, Anja Schiel, Carl-Fredrik Burman, Joey Zhou, Zoran Antonijevic, Eva R. Miller, Rui Tang
Summary: The FDA's Accelerated Approval program aims to expedite the availability of evidence-based products for treating serious diseases. However, there is debate over the sufficiency of supporting evidence and meeting the needs of stakeholders. Therefore, it is important to provide an approach for evaluating the impact of Accelerated Approval that considers the views of different stakeholders.
Article
Biochemical Research Methods
Xia-An Bi, Yuhua Mao, Sheng Luo, Hao Wu, Lixia Zhang, Xun Luo, Luyun Xu
Summary: Imaging genetics integrates multi-level medical data to provide unique insights into the pathological studies of complex brain diseases. This paper proposes a deep learning method called feature information aggregation and diffusion generative adversarial network (FIAD-GAN) to efficiently classify samples and select features.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Astronomy & Astrophysics
Christopher J. White, Patrick D. Mullen, Yan-Fei Jiang, Shane W. Davis, James M. Stone, Viktoriya Morozova, Lizhong Zhang
Summary: We enhance the capabilities of Athena++ in general-relativistic magnetohydrodynamics (GRMHD) to incorporate radiation simulations. The numerical procedure is described in detail and its correctness is verified using a series of tests. The method is applied to the problem of black hole accretion in the high-accretion-rate, thin-disk regime, and the port of the algorithm to GPU-enabled AthenaK code enables the simulation of previously challenging radiation-GRMHD systems.
ASTROPHYSICAL JOURNAL
(2023)
Article
Chemistry, Analytical
Whai-En Chen, Yi-Bing Lin, Tai-Hsiang Yen, Syuan-Ru Peng, Yun-Wei Lin
Summary: This paper discusses the challenges of developing distributed intelligent systems in both the network and device domains. It proposes a low-code or no-code approach to automate code generation and introduces DeviceTalk, an environment that automatically generates code for IoT devices to speed up software development.
Article
Multidisciplinary Sciences
Mitsugu Araki, Shigeyuki Matsumoto, Gert-Jan Bekker, Yuta Isaka, Yukari Sagae, Narutoshi Kamiya, Yasushi Okuno
Summary: Capturing the dynamic processes of biomolecular systems in atomistic detail remains difficult, but researchers have developed a method using high-frequency ultrasound perturbation to accelerate MD simulations and detect binding events between proteins and inhibitors. This innovative approach has successfully accelerated slow binding rates and revealed microscopic kinetic features, offering deeper insights into the interactions controlling biomolecular processes.
NATURE COMMUNICATIONS
(2021)
Correction
Neurosciences
James Bennett, Thomas Nowotny
Article
Biochemical Research Methods
Ho Ka Chan, Fabian Hersperger, Emiliano Marachlian, Brian H. Smith, Fernando Locatelli, Paul Szyszka, Thomas Nowotny
PLOS COMPUTATIONAL BIOLOGY
(2018)
Article
Computer Science, Cybernetics
Alan Diamond, Michael Schmuker, Thomas Nowotny
BIOLOGICAL CYBERNETICS
(2019)
Article
Neurosciences
Leonie S. Brebner, Joseph J. Ziminski, Gabriella Margetts-Smith, Meike C. Sieburg, Hayley M. Reeve, Thomas Nowotny, Johannes Hirrlinger, Tristan G. Heintz, Leon Lagnado, Shigeki Kato, Kazuto Kobayashi, Leslie A. Ramsey, Catherine N. Hall, Hans S. Crombag, Eisuke Koya
JOURNAL OF NEUROSCIENCE
(2020)
Review
Physiology
Mario Pannunzi, Thomas Nowotny
FRONTIERS IN PHYSIOLOGY
(2019)
Article
Multidisciplinary Sciences
Marcel Stimberg, Dan F. M. Goodman, Thomas Nowotny
SCIENTIFIC REPORTS
(2020)
Article
Neurosciences
Naoki Kogo, Felix B. Kern, Thomas Nowotny, Raymond van Ee, Richard van Wezel, Takeshi Aihara
Summary: In this study, a hybrid system was developed to construct a mutual inhibition circuit between two real-life pyramidal neurons, revealing bistable behavior when the neurons are simultaneously activated. By adding modeled synaptic noise and varying activation strength, the dynamics of the circuit were found to closely resemble the properties of bistable visual perception.
JOURNAL OF NEUROSCIENCE
(2021)
Article
Multidisciplinary Sciences
James E. M. Bennett, Andrew Philippides, Thomas Nowotny
Summary: Effective decision making in a changing environment requires accurate predictions about decision outcomes, which in Drosophila is partially orchestrated by the mushroom body where dopamine neurons signal reinforcing stimuli. The authors propose a model where dopaminergic learning signals encode reinforcement prediction errors by utilizing feedback reinforcement predictions from mushroom body output neurons.
NATURE COMMUNICATIONS
(2021)
Article
Mathematical & Computational Biology
James Paul Turner, Thomas Nowotny
Summary: Researchers developed the Arpra library for investigating the reproducibility of spiking neural network simulations by implementing a mixed IA/AA method, minimizing error terms, and improving efficiency with novel affine term reduction strategies.
FRONTIERS IN NEUROINFORMATICS
(2021)
Article
Biochemistry & Molecular Biology
Yuri Ogawa, Sarah Nicholas, Malin Thyselius, Richard Leibbrandt, Thomas Nowotny, James C. Knight, Karin Nordstrom
Summary: Many animals, including predators and male hoverflies, use motion vision information to detect and pursue moving targets. This study investigated the response of target-selective descending neurons (TSDNs) in male hoverflies during pursuits of artificial targets. The results showed that individual TSDNs responded consistently at specific time points, but with variations between neurons. The overall response rate was low, suggesting that different neurons may control different parts of the behavioral output.
Proceedings Paper
Computer Science, Theory & Methods
James C. Knight, Thomas Nowotny
Summary: The mlGeNN interface provides an easy way to define, train, and test spiking neural networks on the GPU-based GeNN framework. By using the e-prop learning rule, mlGeNN allows for convenient and rapid prototyping of different network architectures. The performance of recurrent spiking neural networks trained to recognize hand gestures using mlGeNN is compared to other recent results on the DVS gesture dataset.
PROCEEDINGS OF THE 2023 ANNUAL NEURO-INSPIRED COMPUTATIONAL ELEMENTS CONFERENCE, NICE 2023
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
James P. Turner, Jens E. Pedersen, Joerg Conradt, Thomas Nowotny
Summary: This article presents a new data processing pipeline for automatically labelled stereo vision data. The method utilizes 3D positioning of 3D printed props with active LED markers to provide both RGB frame and event-based data, which are automatically annotated with segmentation masks and ground truth position and orientation data.
PROCEEDINGS OF THE 2022 ANNUAL NEURO-INSPIRED COMPUTATIONAL ELEMENTS CONFERENCE (NICE 2022)
(2022)
Article
Engineering, Electrical & Electronic
James Paul Turner, James C. Knight, Ajay Subramanian, Thomas Nowotny
Summary: This paper introduces mlGeNN, a Python library for converting artificial neural networks specified in Keras to spiking neural networks. The authors evaluate the converted SNNs on CIFAR-10 and ImageNet classification tasks and find that they perform better than the original ANNs and other SNN simulators.
NEUROMORPHIC COMPUTING AND ENGINEERING
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Ho Ka Chan, Thomas Nowotny
NEURAL INFORMATION PROCESSING (ICONIP 2017), PT IV
(2017)