Regularized focusing inversion for large-scale gravity data based on GPU parallel computing

Processing large-scale 3-D gravity data is an important topic in geophysics field. Many existing inversion methods lack the competence of processing massive data and practical application capacity. This study proposes the application of GPU parallel processing technology to the focusing inversion method, aiming at improving the inversion accuracy while speeding up calculation and reducing the memory consumption, thus obtaining the fast and reliable inversion results for large complex model. In this paper, equivalent storage of geometric trellis is used to calculate the sensitivity matrix, and the inversion is based on GPU parallel computing technology. The parallel computing program that is optimized by reducing data transfer, access restrictions and instruction restrictions as well as latency hiding greatly reduces the memory usage, speeds up the calculation, and makes the fast inversion of large models possible. By comparing and analyzing the computing speed of traditional single thread CPU method and CUDA-based GPU parallel technology, the excellent acceleration performance of GPU parallel computing is verified, which provides ideas for practical application of some theoretical inversion methods restricted by computing speed and computer memory. The model test verifies that the focusing inversion method can overcome the problem of severe skin effect and ambiguity of geological body boundary. Moreover, the increase of the model cells and inversion data can more clearly depict the boundary position of the abnormal body and delineate its specific shape.

Parallel numerical simulations for quantized vortices in Bose-Einstein condensates

We employ the parallel computing technology to study numerically the three-dimensional structure of quantized vortices of Bose-Einstein condensates, For anisotropic cases, the bending process of vortices is described in detail by the decrease of Gross-Pitaevskii energy. A completely straight vortex and the steady and symmetrical multiple-vortex configurations are obtained. We analyse the effect of initial conditions and angular velocity on the number and shape of vortices.

Effect of Parallel Computing Environment on the Solution Consistency of CFD Simulations—Focused on IC Engines

For CFD results to be useful in IC engine analysis, simulation results should be accurate and consistent. However, with wide spread use of parallel computing nowadays, it has been reported that a model would not give the same results against the same input when the parallel computing environment is changed. The effect of parallel environment on simulation results needs to be carefully investigated and understood. In this paper, the solution inconsistency of parallel CFD simulations is investigated. First, the concept of solution inconsistency on parallel computing is reviewed, followed by a systematic CFD simulations specific to IC engine applications. The solution inconsistency against the number of CPU cores was examined using a commercial CFD code CONVERGE. A test matrix was specifically designed to examine the core number effect on engine flow, spray and combustion submodels performance. It was found that the flow field simulation during the gas exchange process is the most sensitive to the number of cores among all submodels examined. An engineering solution was developed where local upwind scheme was used to control the variability, which showed good performance. The implication of the observed inconsistency was also discussed.

Combination Method for Parallel Computation in ODEs

In this paper, a 3rd order combination method with three processes and a 4th order combination method with five processes for solving ODEs are discussed. These methods are the Runge-Kutta method combined with a linear multistep method, which overcomes the defect of the 3rd order parallel Runge-Kutta method discussed in [1].

Development of Operational Technology for Meteorological High Performance Computing

As an important branch of information technology, high-performance computing has expanded its application field and its influence has been expanding. High-performance computing is always a key area of application in meteorology. We used field research and literature review methods to study the application of high performance computing in China’s meteorological department, and obtained the following results: 1) China Meteorological Department gradually established the first high-performance computer system since 1978. High-performance computing services can support operational numerical weather prediction models. 2) The Chinese meteorological department has always used the relatively advanced high-performance computing technology, and the business system capability has been continuously improved. The computing power has become an important symbol of the level of meteorological modernization. 3) High-performance computing technology and meteorological numerical forecasting applications are increasingly integrated, and continue to innovate and develop. 4) In the future, high-performance computing resource management will gradually transit from the current local pre-allocation mode to the local remote unified scheduling and shared use. In summary, we have come to the conclusion that the performance calculation business of the meteorological department will usher in a better tomorrow.

A Class of Semi-Implicit Parallel Difference Method for Time Fractional Diffusion Equations

In this paper, we construct a class of semi-implicit difference method for time fractional diffusion equations—the group explicit (GE) difference scheme, which is a difference scheme with good parallelism constructed using Saul’yev asymmetric scheme. The stability and convergence of the GE scheme of time fractional diffusion equation are analyzed by mathematical induction. Then, the theoretical analysis is verified by numerical experiments, which shows that the GE scheme is effective for solving the time fractional diffusion equation.

Parallel Iterative FEM Solver with Initial Guess for Frequency Domain Electromagnetic Analysis

The finite element method is a key player in computational electromag-netics for designing RF(Radio Frequency)components such as waveguides.The frequency-domain analysis is fundamental to identify the characteristics of the components.For the conventional frequency-domain electromagnetic analysis using FEM(Finite Element Method),the system matrix is complex-numbered as well as indefinite.The iterative solvers can be faster than the direct solver when the solver convergence is guaranteed and done in a few steps.However,such complex-numbered and indefinite systems are hard to exploit the merit of the iterative solver.It is also hard to benefit from matrix factorization techniques due to varying system matrix parts according to frequency.Overall,it is hard to adopt conventional iterative solvers even though the system matrix is sparse.A new parallel iterative FEM solver for frequency domain analysis is implemented for inhomogeneous waveguide structures in this paper.In this implementation,the previous solution of the iterative solver of Matlab(Matrix Laboratory)employ-ing the preconditioner is used for the initial guess for the next step’s solution process.The overlapped parallel stage using Matlab’s Parallel Computing Toolbox is also proposed to alleviate the cold starting,which ruins the convergence of early steps in each parallel stage.Numerical experiments based on waveguide structures have demonstrated the accuracy and efficiency of the proposed scheme.

Global SH-wavefield calculation for a two-dimensional whole-Earth model with the parallel hybrid PSM/FDM algorithm

We present a parallel hybrid algorithm based on pseudospectral method （PSM） and finite difference method （FDM） for two-dimensional （2-D） global SH- wavefield simulation. The whole-Earth model is taken as a cross section of spherical Earth, and corresponding wave equations are defined in 2-D cylindrical coordinates. Spatial derivatives in the wave equations are approximated with efficient and high accuracy PSM in the lateral and high-order FDM in the radial direction on staggered grids. This algorithm allows us to divide the whole-Earth into sub-domains in radial direction and implement efficient parallel computing on PC cluster, while retains high accuracy and efficiency of PSM in lateral direction. A transformation of moment tensor between 3-D spherical Earth and our 2-D model was proposed to give corre- sponding moment tensor components used in 2-D modeling. Comparison of modeling results with those obtained by direct solution method shows very good accuracy of our algorithm. We also demonstrate its feasibility with a lateral heterogeneous whole-Earth model with localized velocity perturbation.

Optimization Parameter of Transformation of Parallel Translation for Ship Pitch Simulation

The transformation of parallel translation can improve the smoothness of discrete series sometimes. In this paper, for ship pitch, a method to modify the system error is proposed via the transformation of parallel translation, which can give the optimize parameters using the Method of Minimum Squares. The series in the method can fit white exponential law better, and then be applied in GM (1,1) very well. The numerical experiments imply that the method is practical, which make the ship pitch system model more accurate than GM ( 1,1 ).

基于并行BICGSTAB的不可压流场计算方法

不可压缩流动广泛存在于日常生活及工程应用之中,随着数值模拟规模不断扩大,CFD算法越来越重视并行计算。该文采用SIMPLE算法对不可压流场进行求解,为求解算法中的大型稀疏矩阵,该文发展一套基于MPI框架的并行BICGSTAB矩阵求解方法。最后采用定常顶盖驱动流算例及非定常圆柱绕流算例验证并行算法的可行性及准确性。

提升PD效率的粒子对方法及信息传递接口并行方法

为了避免近场动力方法中粒子间物理信息重复求解消耗的计算成本和占据的额外内存,克服近场动力学低效率计算过程对其工程应用的制约,本文提出一种提升PD计算效率的粒子对方法并建立了该方法的信息传递接口并行方法。利用粒子对方法的信息传递接口并行策略并分析其计算效率的优越性,通过大型的工程应用实例给出了本文方法的工程实用性。研究结果表明:粒子对方法的应用不仅提升了计算效率,也优化了邻域粒子信息储存数组占据的内存和计算时间,粒子对方法可与信息传递接口并行编程相结合,能够进一步提升其计算效率,并且在较多线程并行时,计算效率高于OpenMP并行方法,而在实际计算中,加速效果也会受到不同线程间的通信任务影响。

Evolutionary Neural Architecture Search and Its Applications in Healthcare

Most of the neural network architectures are based on human experience,which requires a long and tedious trial-and-error process.Neural architecture search(NAS)attempts to detect effective architectures without human intervention.Evolutionary algorithms(EAs)for NAS can find better solutions than human-designed architectures by exploring a large search space for possible architectures.Using multiobjective EAs for NAS,optimal neural architectures that meet various performance criteria can be explored and discovered efficiently.Furthermore,hardware-accelerated NAS methods can improve the efficiency of the NAS.While existing reviews have mainly focused on different strategies to complete NAS,a few studies have explored the use of EAs for NAS.In this paper,we summarize and explore the use of EAs for NAS,as well as large-scale multiobjective optimization strategies and hardware-accelerated NAS methods.NAS performs well in healthcare applications,such as medical image analysis,classification of disease diagnosis,and health monitoring.EAs for NAS can automate the search process and optimize multiple objectives simultaneously in a given healthcare task.Deep neural network has been successfully used in healthcare,but it lacks interpretability.Medical data is highly sensitive,and privacy leaks are frequently reported in the healthcare industry.To solve these problems,in healthcare,we propose an interpretable neuroevolution framework based on federated learning to address search efficiency and privacy protection.Moreover,we also point out future research directions for evolutionary NAS.Overall,for researchers who want to use EAs to optimize NNs in healthcare,we analyze the advantages and disadvantages of doing so to provide detailed guidance,and propose an interpretable privacy-preserving framework for healthcare applications.

基于GLC多项式谱元法的大地电磁场二维正演模拟

为了提高大地电磁场数值模拟的精度和效率,提出基于Gauss-Lobatto-Chebyshev(GLC)基函数谱元法的大地电磁场正演模拟方法。该方法首先从麦克斯韦方程组出发,推导了二维大地电磁场边值问题;然后基于Galerkin加权余量法将微分形式的边值问题转换成积分弱形式;最后采用GLC正交多项式插值基函数对全局问题进行离散化,利用Pardiso求解器求解大型稀疏线性方程组得到大地电磁场,实现了二维大地电磁场数值模拟。为了提高数值模拟效率,文中算法采用变密度规则网格剖分技术,即在电性复杂区域使用细网格,在电性均匀区域使用粗网格,采用OpenMP编程模式实现了多个频点的并行计算,以此达到缩短计算时间的目的。通过一维层状介质模型数值模拟结果验证了文中算法的正确性和精度,相比于Gauss-Lobatto-Legendre(GLL)多项式谱元法的数值模拟结果,GLC多项式谱元法的模拟结果精度更高。针对国际标准模型COMMEMI 2D-1和带地形模型,进行了基于GLC多项式谱元法、有限差分法和基于三角网格有限元法的正演模拟,三者数值结果对比表明,GLC多项式谱元法计算精度更高、网格依赖性更低。

面向SW26010间断有限元算法的多级并行计算

间断有限元算法(Discontinuous Galerkin Finite Element Method,DGM)是一种高精度的数值求解算法,针对电磁工程应用中DGM并行计算效率低、计算量较大的问题,提出了基于SW26010平台的并行DGM算法。通过区域分解、数据结构重构、热点函数从核并行计算、计算与通信重叠及从核缓冲优化技术完成了DGM算法的并行优化。实现结果表明,与基于MPI进程级的DGM并行算法相比,可以获得46.8的平均加速比。

A physics-based hydro-geomorphologic simulation utilizing cluster parallel computing

To conduct a large-scale hydrologic-response and landform evolution simulation at high resolution,a complex physics-based numerical model,the Integrated Hydrology Model(InHM),was revised utilizing cluster parallel computing.The parallelized InHM(ParInHM) divides the simulated area into multiple catchments based on geomorphologic features,and generates boundary-value problems for each catchment to construct simulation tasks,which are then dispatched to different computers to start the simulation.Landform evolution is considered during simulating and implemention in one framework.The dynamical Longest-Processing-Time(LPT) first scheduling algorithm is applied to job management.In addition,a pause-integratedivide-resume routine method is used to ensure the hydrologic validity during the simulation period.The routine repeats until the entire simulation period is finished.ParInHM has been tested in a computer cluster that uses 16 processors for the calculation,to simulate 100 years' hydrologic-response and soil erosion for the 117-km2 Kaho'olawe Island in the Hawaiian Islands under two different mesh resolutions.The efficiency of ParInHM was evaluated by comparing the performance of the cluster system utilizing different numbers of processors,as well as the performance of non-parallelized system without domain decomposition.The results of this study show that it is feasible to conduct a regional-scale hydrologic-response and sediment transport simulation at high resolution without demanding significant computing resources.

Three-dimensional finite element simulation and reconstruction of jointed rock models using CT scanning and photogrammetry

The geometry of joints has a significant influence on the mechanical properties of rocks.To simplify the curved joint shapes in rocks,the joint shape is usually treated as straight lines or planes in most laboratory experiments and numerical simulations.In this study,the computerized tomography (CT) scanning and photogrammetry were employed to obtain the internal and surface joint structures of a limestone sample,respectively.To describe the joint geometry,the edge detection algorithms and a three-dimensional (3D) matrix mapping method were applied to reconstruct CT-based and photogrammetry-based jointed rock models.For comparison tests,the numerical uniaxial compression tests were conducted on an intact rock sample and a sample with a joint simplified to a plane using the parallel computing method.The results indicate that the mechanical characteristics and failure process of jointed rocks are significantly affected by the geometry of joints.The presence of joints reduces the uniaxial compressive strength (UCS),elastic modulus,and released acoustic emission (AE) energy of rocks by 37%–67%,21%–24%,and 52%–90%,respectively.Compared to the simplified joint sample,the proposed photogrammetry-based numerical model makes the most of the limited geometry information of joints.The UCS,accumulative released AE energy,and elastic modulus of the photogrammetry-based sample were found to be very close to those of the CT-based sample.The UCS value of the simplified joint sample (i.e.38.5 MPa) is much lower than that of the CT-based sample (i.e.72.3 MPa).Additionally,the accumulative released AE energy observed in the simplified joint sample is 3.899 times lower than that observed in the CT-based sample.CT scanning provides a reliable means to visualize the joints in rocks,which can be used to verify the reliability of photogrammetry techniques.The application of the photogrammetry-based sample enables detailed analysis for estimating the mechanical properties of jointed rocks.

Extreme-scale parallel computing:bottlenecks and strategies

Extreme-scale numerical simulations seriously demand extreme parallel computing capabilities. To address the challenges of these capabilities toward exascale, we systematically analyze the major bottlenecks of parallel computing research from three perspectives： computational scale, computing efficiency, and programming productivity. For these bottlenecks, we propose a series of urgent key issues and coping strategies. This study will be useful in synchronizing development between the numerical computing capability and supercomputer peak performance.

基于并行自适应遗传算法的水文模型率定研究

【目的】参数率定是影响水文模型预报精度的重要因素,采用人工智能算法可以有效提高水文模型参数的率定效果。【方法】采用基于种群离散程度的自适应算子,对GA算法的交叉、变异和迁移过程进行自适应优化,并利用粗粒度并行计算模型提高种群进化效率,综合以上手段研究了一种基于自适应策略的并行遗传算法。将传统遗传算法(GA),串行自适应遗传算法(AGA)和并行自适应遗传算法(PAGA),应用于屯溪流域新安江模型的参数率定,从率定效率、率定收敛性、率定稳定性和率定效果四个方面,验证PAGA算法的综合性能。【结果】结果表明:PAGA算法的计算加速效果显著,在10核环境下相对于AGA算法计算时间减少了87.9%;在进化后期,PAGA算法能够更加稳定的收敛于最优解,收敛后的目标函数值具有更好的稳定性;在验证期的场次洪水模拟中,采用PAGA算法率定的模型模拟效果最优,总体洪水合格率大于90%,确定性系数均值为0.85。【结论】PAGA算法能够明显降低模型参数寻优耗时,改善模型率定效果和收敛性能,为水文模型参数的率定提供了新思路。