Article
Thermodynamics
Ao Xu, Bo-Tao Li
Summary: This study evaluates the performance of a hybrid OpenACC and MPI approach for multi-GPUs accelerated thermal LB simulation. OpenACC is used to accelerate computation on a single GPU, while MPI synchronizes information between multiple GPUs. The results show promising performance improvement with single GPU achieving 1.93 billion GLUPS for 2D simulation and 1.04 GLUPS for 3D simulation. With 16 GPUs, the parallel efficiency remains high, reaching 30.42 GLUPS for 2D simulation and 14.52 GLUPS for 3D simulation.
INTERNATIONAL JOURNAL OF HEAT AND MASS TRANSFER
(2023)
Article
Computer Science, Software Engineering
Weicheng Xue, Christoper J. Roy
Summary: This article explores the multi-GPU performance of a 3D buoyancy driven cavity solver using MPI and OpenACC directives, revealing the significant impact of decomposing the problem in different dimensions on the strong scaling performance of the GPU. Various performance optimizations presented in the article can benefit performance using different decompositions, such as parallel message packing/unpacking and transferring different data based on stencil sizes. The optimizations aim to reduce communication costs and improve memory throughput between hosts and devices efficiently.
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE
(2021)
Article
Microbiology
Igor Sfiligoi, George Armstrong, Antonio Gonzalez, Daniel McDonald, Rob Knight
Summary: UniFrac is a widely used tool in microbiome research for comparing microbiome profiles. This study adapts UniFrac to be used on graphics processing units, resulting in a significant improvement in computational performance. The tool is successfully applied to the largest 16S rRNA V4 microbiome dataset analyzed to date.
Article
Computer Science, Theory & Methods
Marco Aldinucci, Valentina Cesare, Iacopo Colonnelli, Alberto Riccardo Martinelli, Gianluca Mittone, Barbara Cantalupo, Carlo Cavazzoni, Maurizio Drocco
Summary: The work presents a systematic methodology to modernize existing sequential scientific codes with little re-designing effort, turning them into parallel and robust code. The proposed semi-automatic methodology can parallelize scientific applications designed with purely sequential programming mindset, and it has been successfully demonstrated in shared memory, message passing, and GPU computing models. The method has been applied to parallelize real-world sequential codes in physics and material science domains.
JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING
(2021)
Article
Mechanics
Long Chen, Alban Potherat, Ming-Jiu Ni, Rene Moreau
Summary: High-resolution direct numerical simulations were conducted to study turbulent shear flow of liquid metal in a cylindrical container driven by an azimuthal Lorentz force. The simulations accurately reproduced experimental findings and revealed a transition to small scale turbulence at certain Reynolds numbers. The study also confirmed theoretical laws for cutoff scale and characteristics of sheared magnetohydrodynamics (MHD) turbulence.
JOURNAL OF FLUID MECHANICS
(2021)
Article
Meteorology & Atmospheric Sciences
Jian Sun, John M. Dennis, Sheri A. Mickelson, Brian Vanderwende, Andrew Gettelman, Katherine Thayer-Calder
Summary: In this study, the cloud microphysics parameterization in a climate model was ported from CPU to GPU for faster computation. The directive-based methods were found to be the best fit for the development practices and showed improved portability and maintainability. The experiments demonstrated that using GPU in climate simulations can significantly reduce computational costs.
JOURNAL OF ADVANCES IN MODELING EARTH SYSTEMS
(2023)
Article
Computer Science, Theory & Methods
Weicheng Xue, Charles W. Jackson, Christoper J. Roy
Summary: This study focuses on optimizing the performance of a CFD code on structured grids with multiple GPUs. By applying various optimizations, the performance is improved, and using 16 P100 GPUs and 16 V100 GPUs can be up to 30x and 90x faster than 16 Xeon CPU E5-2680v4 cores.
JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING
(2021)
Article
Computer Science, Hardware & Architecture
Matthew R. Norman, David A. Bader, Christopher Eldred, Walter M. Hannah, Benjamin R. Hillman, Christopher R. Jones, Jungmin M. Lee, L. R. Leung, Isaac Lyngaas, Kyle G. Pressel, Sarat Sreepathi, Mark A. Taylor, Xingqiu Yuan
Summary: This study utilizes the Multi-scale Modeling Framework (MMF) to enhance cloud effects representation in Global Climate Models (GCM), embedding high-resolution Cloud Resolving Models (CRMs). Porting code to GPUs and running on Summit supercomputer, the performance of E3SM-MMF code in large-scale climate simulation is demonstrated.
INTERNATIONAL JOURNAL OF HIGH PERFORMANCE COMPUTING APPLICATIONS
(2022)
Article
Astronomy & Astrophysics
Geoffroy R. J. Lesur
Summary: This study examines the parameter space of wind-driven protoplanetary discs and presents scaling laws. It found that mass accretion rates and ejection efficiency are related to field strength, with magnetic lever arms typically lower than 2. Simplified models like self-similar solutions can capture most of the physics seen in full 3D simulations.
ASTRONOMY & ASTROPHYSICS
(2021)
Article
Astronomy & Astrophysics
L. Adhikari, G. P. Zank, L-L Zhao, M. Nakanotani, S. Tasnim
Summary: The study developed a theoretical model combining nearly incompressible magnetohydrodynamic (NI MHD) turbulence with a solar wind model that includes electron pressure and heat flux. The results show that the model is consistent with the measurements of fast solar wind by PSP and Helios 2.
ASTRONOMY & ASTROPHYSICS
(2021)
Article
Astronomy & Astrophysics
G. Leidi, C. Birke, R. Andrassy, J. Higl, P. V. F. Edelmann, G. Wiest, C. Klingenberg, F. K. Roepke
Summary: Fully compressible magnetohydrodynamic (MHD) simulations are essential for studying magnetic field generation in deep convective layers of stars. However, traditional methods become inefficient due to the strict time step constraints. This study presents a new method that efficiently simulates low Mach number compressible MHD flows in strongly stratified setups.
ASTRONOMY & ASTROPHYSICS
(2022)
Article
Astronomy & Astrophysics
N. Andres, F. Sahraoui, S. Huang, L. Z. Hadid, S. Galtier
Summary: This study investigates the isotropy and anisotropy energy cascade rates in solar wind turbulence using more than two years of Parker Solar Probe (PSP) observations. The results show a dominance of perpendicular cascades over parallel cascades, and a dominant 2D cascade and/or geometry over the slab component in the largest MHD scales.
ASTRONOMY & ASTROPHYSICS
(2022)
Article
Computer Science, Interdisciplinary Applications
Francesco De Vanna, Filippo Avanzi, Michele Cogo, Simone Sandrin, Matt Bettencourt, Francesco Picano, Ernesto Benini
Summary: This paper presents URANOS, a massively parallel GPU-accelerated compressible flow solver for high-fidelity modeling of compressible wall flows. The solver employs modern high-fidelity and high-resolution discretization strategies, features multiple convective scheme implementations, innovative turbulence modeling frameworks, and utilizes high-order finite-difference approach. The computational performance and scalability of URANOS on different architectures are evaluated, and the solver is validated against various benchmark cases for wall flows.
COMPUTER PHYSICS COMMUNICATIONS
(2023)
Article
Computer Science, Interdisciplinary Applications
Jae Youp Kim, Ji-Sun Kang, Minsu Joh
Summary: This study accelerated a microphysics scheme within MPAS using OpenACC directives and focused on parallelizing WRF's microphysics scheme onto a GPU. By optimizing the performance and minimizing data transfer, significant speed-ups were achieved compared to MPI processes. A precise verification method successfully distinguished GPU-induced differences from nonlinear error growth.
COMPUTERS & GEOSCIENCES
(2021)
Article
Astronomy & Astrophysics
Hai Yang, Feng Yuan, Tom Kwan, Lixin Dai
Summary: This paper investigates the properties of wind and jet launched from a magnetized super-Eddington accretion flow around a supermassive black hole. The simulation data is analyzed using the 'virtual particle trajectory' approach, and the mass flux, velocity, momentum and energy fluxes of wind and jet are obtained. The results show that the mass flux is higher than previous methods suggest, and the momentum flux of wind is larger than that of jet, indicating that wind likely plays a more important role in AGN feedback. The acceleration mechanism of wind and jet is found to be dominated by Lorentz force rather than radiation force.
MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY
(2023)
Article
Remote Sensing
Kepeng Feng, Yang Hong, Juncang Tian, Xiangyu Luo, Guoqiang Tang, Guangyuan Kan
Summary: This study evaluated the accuracy of precipitation products in runoff simulation in small watersheds in the United States, finding that different datasets have varying levels of matching in different regions. In the runoff simulation effectiveness evaluation, NOAA-CPC-US and StageIV performed well, while TRMM-3B42V7 is not suitable for runoff simulation in small watersheds.
EUROPEAN JOURNAL OF REMOTE SENSING
(2021)
Article
Physics, Fluids & Plasmas
H. W. Zhang, X. Lin, Z. W. Ma, W. Zhang, T. E. Bagwell
Summary: The study shows that in a tokamak plasma, the ability of single-harmonic RMP to penetrate the mode-rational surface is limited by plasma screening effects; while in the multiple-harmonic RMP, non-resonant components can reduce plasma shielding effects, allowing the RMP to penetrate deeply into the central plasma region.
PLASMA PHYSICS AND CONTROLLED FUSION
(2021)
Article
Engineering, Civil
Depeng Zuo, Yuna Han, Zongxue Xu, Peijun Li, Chunguang Ban, Wenchao Sun, Bo Pang, Dingzhi Peng, Guangyuan Kan, Rui Zhang, Hong Yang
Summary: The study revealed that vegetation coverage generally increases from upstream to downstream in the Yarlung Zangbo River basin, with greening trend in the northwestern lower reaches and vegetation degradation in the southeastern region. Different vegetation types showed correlations and time-lag effects with precipitation, temperature and drought in the area.
JOURNAL OF HYDROLOGY
(2021)
Article
Environmental Sciences
Peijun Li, Depeng Zuo, Zongxue Xu, Rui Zhang, Yuna Han, Wenchao Sun, Bo Pang, Chunguang Ban, Guangyuan Kan, Hong Yang
Summary: This study explored the dynamic changes in land use/cover and landscape patterns in the Yarlung Tsangpo River basin of China from 1980 to 2015. Results showed a decrease in permanent glacier-snow area and an increase in built-up areas. Landscape fragmentation and diversity gradually increased from upstream to downstream.
LAND DEGRADATION & DEVELOPMENT
(2021)
Article
Computer Science, Interdisciplinary Applications
W. Zhang, S. C. Jardin, Z. W. Ma, A. Kleiner, H. W. Zhang
Summary: By comparing the simulation results of CLT code and M3D-C1 code, it was found that they were almost identical in terms of resistive kink mode and tearing mode, including linear and nonlinear growth rates. This indicates that the simulation results for the 1/1 resistive-kink mode and 2/1 tearing mode are accurate and reliable.
COMPUTER PHYSICS COMMUNICATIONS
(2021)
Article
Physics, Fluids & Plasmas
H. W. Zhang, Z. W. Ma, J. Zhu, W. Zhang, Z. Y. Qiu
Summary: The generation of the n = 0 zonal flow and the excitation of the n = 1 toroidal Alfven eigenmode (TAE) due to the redistribution of energetic particles (EPs) by the m/n = 2/1 tearing mode (TM) were systematically studied using hybrid drift-kinetic magnetohydrodynamic (MHD) simulations. The study showed that the amplitude of the n = 1 TAE decays slower after saturation due to wave-particle nonlinearity and the nonlinear generation of sidebands. Additionally, a strong n = 0 zonal flow component is generated when both TAE and TM reach high amplitudes. The redistribution of EPs by the m/n = 2/1 magnetic island drives the background plasma and leads to the generation of the zonal flow through MHD nonlinearity. Furthermore, the large m/n = 2/1 magnetic island is responsible for the formation of a strong spatial gradient of the EP distribution, which can result in the burst of unstable TAE and destabilization of originally stable TAE.
Article
Meteorology & Atmospheric Sciences
Mengjie Zhang, Qing Cao, Feilin Zhu, Upmanu Lall, Peng Hu, Yunzhong Jiang, Guangyuan Kan
Summary: This study evaluated the rainy season features in the Yellow River basin, China, and the influence of the El Nino-Southern Oscillation (ENSO) on precipitation. The findings showed that the rainy season began and ended earliest in the southern region of the Yellow River basin, with increasing precipitation from northwest to southeast. Rainy-season precipitation also exhibited strong correlations with sea surface temperature in the Nino regions, and different ENSO phases had varying impacts on precipitation.
THEORETICAL AND APPLIED CLIMATOLOGY
(2022)
Article
Physics, Fluids & Plasmas
L. Guo, H-W Zhang, H-C Wu
Summary: This article proposes an alternative mechanism for lightning to produce very-high-frequency radio signals through the transition radiation of runaway electrons. The calculated results show that the energy spectra and radiation powers of the transition radiation are consistent with the detection results. The proposed coherent transition radiation technique could potentially reconstruct the actual properties of electron beams in the lightning leaders.
PHYSICS OF PLASMAS
(2022)
Article
Geosciences, Multidisciplinary
Houfa Wu, Jianyun Zhang, Zhenxin Bao, Guoqing Wang, Wensheng Wang, Yanqing Yang, Jie Wang, Guangyuan Kan
Summary: The Yellow-Huai-Hai River Basin is an important ecological barrier in China, and the vegetation coverage in this region shows spatiotemporal differences and is influenced by various driving factors. The study found that vegetation coverage increased from 1982 to 2015, with a positive correlation between vegetation growth and temperature. Natural forcing had the greatest impact on vegetation change, while anthropogenic forcing and other driving factors also played significant roles, with spatial variations. Climate change and human activities both had positive effects on vegetation restoration, while land use change was the main factor causing vegetation degradation.
FRONTIERS IN EARTH SCIENCE
(2022)
Article
Physics, Fluids & Plasmas
Haowei Zhang, Zhiwei Ma
Summary: Both current and pressure coupling schemes have been verified to be numerically equivalent in the hybrid kinetic-magnetohydrodynamic code CLT-K. The stability of the m/n=2/1 tearing mode is influenced by the presence of co-/counter-passing and trapped energetic particles (EPs), with co-passing and trapped EPs stabilizing the mode and counter-passing EPs destabilizing it. The excitation of the m/n=2/1 energetic particle mode occurs after exceeding critical EP betas.
PLASMA SCIENCE & TECHNOLOGY
(2023)
Article
Environmental Sciences
Sijie Chen, Haiwen Yan, Wei Shao, Wenjun Yu, Lingna Wei, Zongji Yang, Ye Su, Guangyuan Kan, Shaohui Luo
Summary: This study conducted an inverse analysis of in situ measured soil moisture in an earthquake-induced landslide deposit and optimized the soil hydraulic parameters using the Differential Evolution Markov chain Monte Carlo method (DE-MC). The results showed that the DE-MC approach can significantly reduce uncertainties in specified prior soil hydraulic parameters and have a high level of accuracy in matching in situ measured soil moisture content.
Article
Physics, Fluids & Plasmas
X. Q. Lu, W. Guo, Z. W. Ma, H. W. Zhang, W. F. Guo
Summary: The effects of plasma resistivity and viscosity on the dynamic evolution of the m/n=3/1 double tearing mode (DTM) are quantitatively studied using the Ci-Liu-Ti (CLT) code. The study focuses on the change in the radial positions and oscillatory dynamics of magnetic islands on rational surfaces. The investigation reveals that the time required to enter the explosive phase decreases with decreasing viscosity, and high viscosity suppresses the oscillation amplitude of kinetic energy.
Article
Environmental Sciences
Xiaodi Fu, Guangyuan Kan, Ronghua Liu, Ke Liang, Xiaoyan He, Liuqian Ding
Summary: To improve the scientific nature, reliability, and accuracy of flood forecasting, constructing a flood forecasting scheme and considering different rain patterns is effective. Using machine learning methods, this study classified rain patterns and compared the performances of different models. LightGBM achieved the highest accuracy and fastest training speed, while LSTM and SVM showed higher accuracy but lower efficiency, and DT had fast classification speed but lower accuracy. Increasing the sampling size improved classification accuracy and training efficiency.
Article
Physics, Fluids & Plasmas
X. Wang, H. W. Zhang, W. J. Chen, T. F. Sun, W. Zhang, Z. W. Ma
Summary: Any arbitrary perturbation on a magnetic field separatrix can cause a structure named homoclinic tangle in tokamaks. Both an edge localized mode (ELM) and a resonant magnetic perturbation (RMP) can lead to a perturbation of the magnetic field on the separatrix. Under the appropriate circumstances, RMP could alleviate or even completely suppress a rapid collapse process of an ELM.
Article
Geochemistry & Geophysics
Zhihao Wei, Xunjian Long, Yingying Gai, Zekun Yang, Xinxin Sui, Xi Chen, Guangyuan Kan, Wenjie Fan, Yaokui Cui
Summary: Error estimation of precipitation products using the double instrumental variable (DIV) method was conducted in regions with complex terrain, and validated with ground-based precipitation data CMPA. The results showed DIV-based errors for IMERG and SM2RAIN ranging from 0 to 25 mm per day and from 0 to 15 mm per day, respectively, with root-mean-square errors compared with CMPA ranging from 0 to 23 mm and 0 to 22 mm. The spatial distribution of DIV-based errors exhibited consistency with CMPA-based errors, indicating the potential of using the DIV method for precipitation products fusion.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)