XZXunwang Zhao

Xunwang Zhao

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Xunwang Zhao has a total of 95 co-authors. Here are the top co-authors by number of shared publications:

  • Zhongchao Lin

    Shaanxi Key Laboratory of Large Scale Electromagnetic Computing, Shaanxi Innovation Center for High Performance CAE Software, Xidian University, Xian, China

    28 shared publications
  • Yu Zhang

    School of Electronic Engineering Xidian University Xi'an Shaanxi China

    26 shared publications
  • Sheng Zuo

    Shaanxi Key Laboratory of Large Scale Electromagnetic Computing, Shaanxi Innovation Center for High Performance CAE Software, Xidian University, Xian, China

    13 shared publications
  • Yu Zhang

    Shaanxi Key Laboratory of Large Scale Electromagnetic Computing, Shaanxi Innovation Center for High Performance CAE Software, Xidian University, Xian, China

    13 shared publications
  • Yu Zhang

    Shaanxi Key Laboratory of Large Scale Electromagnetic Computing, Shaanxi Innovation Center for High Performance CAE Software, Xidian University, Xian, China

    7 shared publications
  • Yan Chen

    School of Electronic Engineering, Xidian University, Xi’an, China

    5 shared publications

Latest Publications

  • An Efficient DFG-Optimized Parallelization Scheme for DGTD-FDTD Method

    by Qingkai Wu, Kunyi Wang, Sheng Zuo, Zhongchao Lin, Yu Zhang, Xunwang Zhao, Yi-Xuan Zhang

    An efficient data-flow-graph (DFG) optimized parallelization scheme is proposed for the discontinuous Galerkin time-domain and finite-difference time-domain (DGTD-FDTD) hybrid method. In the parallel hybrid method, load imbalance in conventional FDTD partitioning strategy and communication overhead between methods limit the parallel efficiency. To address these issues, a novel grid partitioning strategy based on graph node merging is introduced for the irregular FDTD region, which effectively ensures load balance while reducing inter-method communication. Additionally, a collective communication strategy is designed to reduce the waiting time in multi-process communication. The presented numerical examples demonstrate the efficiency and capability of the proposed method.

  • Efficient Evaluation of Weakly Singular Integrals Arising From Electromagnetic Volume Integral Equations

    by Yong Wang, Ming-Da Zhu, Xun-Wang Zhao, Zhong-Chao Lin, Yu Zhang, Yi-Xuan Zhang

    The Duffy method for the volume integral equations (VIE) is affected by the shape of the integration domain, which limits its accuracy, particularly for nearly singular integrals in the tetrahedral cells. This paper addresses the challenges posed by singularities in the use of tetrahedral elements, especially in regions with sharp geometric or material property variations. To overcome the shape-dependent issues, the paper proposes the application of the 3D Duffy transformation with the enhanced smoothing strategies for a good computational efficiency and stability. Some numerical examples are given to demonstrate the accuracy and efficiency of the proposed transformations in solving VIE with the tetrahedral elements.

  • Open Access

    An Adaptive Partitioning Parallel MLFMA for Simulating Radome‐Enclosed Antenna Arrays

    by Lei Yin, Peng Hou, Xunwang Zhao, Zhongchao Lin, Yu Zhang, Xifeng Pu, Yong‐Chang Jiao

    Accurate and efficient simulation of large‐scale radome‐enclosed antenna arrays is challenging in the fields such as radar and communication. The parallel multilevel fast multipole algorithm (MLFMA) is suitable for solving radiation problems of large‐scale radome‐enclosed antenna arrays. To enhance the efficiency and balance the workloads for the MLFMA, this paper presents a parallel MLFMA framework. The proposed algorithm creates multiple MLFMA octrees for different medium regions and waveport excitations. In the parallel computation of near‐field interactions, an auxiliary octree is constructed to avoid the duplication of triangular integrals and employ the K‐way approach to partition the workloads. In the computation of far‐field interactions through evaluating the workloads related to each tree and dynamically partitioning the process groups, an adaptive partitioning strategy is designed to balance the workloads of multiple octrees, which reduces the communication traffic and achieves high parallel efficiency. Numerical results show that efficiencies of the parallel MLFMA are higher than 80% when the number of processes increases from 36 to 720.

  • An Anisotropic h -Adaptive Finite Element Method for Time-Harmonic Electromagnetic Simulation

    by Haoxiang Wu, Kejie Fu, Sheng Zuo, Zhongchao Lin, Xunwang Zhao, Yu Zhang

    In this communication, we investigate an anisotropic adaptive finite element method (FEM) suitable for efficient time-harmonic electromagnetic simulations on unstructured meshes. Anisotropic adaptive methods are particularly effective in improving convergence by reducing element sizes in specific directions. We propose an anisotropic metric field constructed from an adjoint-weighted error estimation method, which captures the directional sensitivity of local error contributions. By utilizing the metric field as a constraint for mesh optimization, we develop a novel anisotropic adaptive FEM for unstructured meshes. Numerical results demonstrate that the mesh modified by the anisotropic adaptive technique offers higher resolution than that modified by an isotropic method with fewer elements. The presented case studies highlight the superior convergence of this method compared to isotropic FEM, thus demonstrating the potential of the anisotropic method in engineering applications.


  • A Parallel Direct Finite Element Solver Empowered by Machine Learning for Solving Large-Scale Electromagnetic Problems

    by Junjie Wang, Sheng Zuo, Xunwang Zhao, Min Tian, Zhongchao Lin, Yu Zhang

    In this communication, a parallel direct finite element solver is proposed for solving complex electromagnetic problems. Our proposed solver features adaptive fill-reducing ordering and highly assembled lazy offloading (HALO) factorization. In our adaptive fill-reducing ordering approach, we propose a graph convolutional network (GCN) model that adaptively selects the optimal fill-reducing ordering method for each electromagnetic problem. Furthermore, we propose a graph sampling training algorithm to enhance the scalability of the GCN model. In our HALO factorization approach, we focus on small irregular level-1 BLAS (BLAS-1) and level-2 BLAS (BLAS-2) subproblems for the supernode factorization and panel update steps, aggregates unstructured L and U factors stored in vector format into structured dense blocks stored in matrix format, assembles these small irregular BLAS-1 and BLAS-2 subproblems into highly assembled level-3 BLAS (BLAS-3) problems that are well suited for the parallelism of various dedicated manycore accelerators, and proposes a lazy offloading technique to achieve high computational efficiency and low data transfer overhead. Our numerical results demonstrate that our proposed method exhibits high accuracy and significant performance improvement for a variety of complex electromagnetic problems and arbitrarily shaped radiating and coupling objects. Moreover, our proposed solver can achieve a 4.25× speedup in comparison to the state-of-the-art swSuperLU on Sunway manycore computing systems, which is an unprecedented performance gain for the direct finite element solver on dedicated manycore accelerators.


  • Conformal Anisotropic Periodic Boundary Condition for FDTD Method

    by Kunyi Wang, Qingkai Wu, Feilong Yu, Sheng Zuo, Zhongchao Lin, Xunwang Zhao, Yu Zhang

    A novel conformal periodic boundary condition (CPBC) is proposed for the finite-difference time-domain (FDTD) method to accurately model anisotropic periodic structures. The conformal FDTD algorithms based on the effective material parameters are attractive due to their simplicity of implementation, but they require a reduced time step to maintain the Courant-Friedrichs-Lewy condition. In this letter, we introduce a novel conformal FDTD scheme for anisotropic materials. This scheme integrates the simplified conformal (SC) finite integration technique with the effective material parameters, which is not only simple to implement, but also avoids the time step reduction due to the effectively small cells. Additionally, we extend this conformal algorithm to periodic boundary condition (PBC) to accurately simulate anisotropic periodic structures. Finally, to validate the proposed method's accuracy and efficiency, we present two illustrative examples.


  • An Efficient Attributed Scattering Center Extraction Method Accelerated by Dictionary Reuse

    by Guopeng Peng, Xiaokun Zhang, Xunwang Zhao, Zhongchao Lin, Yu Zhang, Sheng Zuo

    The attributed scattering center (ASC) model provides a concise and physically relevant description of the targets and plays a vital role in inverse scattering problems. However, ASC extraction is a high-dimensional problem, which involve complex computations and is memory-demanding. In this letter, an efficient attributed scattering center extraction method is proposed. The ASCs are extracted one by one, with alternative optimization integrated into it to reduce dictionary dimension. Two properties of the ASC model, namely superposition property and norm-invariant property, are revealed. The superposition property and norm-invariant property are utilized to reuse dictionaries and to reduce the computational amount of l2-norm, respectively, which accelerates the ASC extraction procedure. Numerical example shows that the average time to extract one ASC is less than 0.2 s, which verify the effectiveness of the proposed method.


  • A Fast RCS Analysis Method Based on Parallel Half-Space MLFMA for Formation Problems

    by Peng Hou, Lei Yin, Zhongchao Lin, Xunwang Zhao, Yu Zhang, Xi Chen

    A fast radar cross section (RCS) analysis method based on the half-space multilevel fast multipole algorithm (MLFMA) is proposed for formation scattering problems. In the proposed method, the computational redundancy can be effectively avoided by separating the free-space terms from the half-space impedance and reusing the octree constructed by a single member. The coupling effects among all members are strictly considered via translation matrix in free space and a real image, leading to reduced time complexity and memory. Combining data reuse and adaptive direction partitioning (ADP) strategy, a balanced parallel framework is designed to keep high parallel efficiency. The numerical results validate the high accuracy and efficiency of the proposed method and demonstrate that the proposed method can solve large-scale formation problems.


  • Solution of Volume Integral Equation Using the Edge-MB Basis Functions for Inhomogeneous Dielectric Objects

    by Yong Wang, Ming-Da Zhu, Xun-Wang Zhao, Zhong-Chao Lin, Yu Zhang

    This letter proposes a volume integral equation (VIE) scheme with reduced unknowns for analyzing electromagnetic scattering from complex inhomogeneous dielectric objects. This scheme utilizes a novel edge basis for multi-branch Schaubert-Wilton-Glisson function (EDGE-MB). The problem of over-mesh and increased unknowns is severe for high-contrast media and multi-domain structures. In the proposed scheme, the divergence-free condition is applied to the multi-branch Schaubert-Wilton-Glisson (MB-SWG) basis function, and we use an EDGE-MB basis function defined within boundary tetrahedral elements instead of SWG basis functions. At the same time, the calculation process of matrix elements with the EDGE-MB scheme is more efficient than traditional SWGVIE or MB-VIE for multi-boundary and multi-medium problems. Finally, several numerical results are presented to demonstrate the accuracy and efficiency of the proposed method.


  • An Efficient h-Adaptive MoM for Modeling Composite Metallic-Dielectric Structures

    by Ning Ding, Lei Yin, Xunwang Zhao, Zhongchao Lin, Yu Zhang, Shugang Jiang

    This letter introduces the surface magnetic current density into error estimations of the Rao-Wilton-Glisson (RWG) method of moments (MoM). Three combined error estimators are constructed and evaluated through correlation coefficients. In addition, an approximation scheme is incorporated with error estimators to accelerate the adaptive refinement procedure. Numerical examples including composite metallic-dielectric structures show that, two of the error estimators can accurately indicate the regions with significant errors in meshed models and guide the mesh generation method to adaptively refine the corresponding triangles. The proposed method improves the modeling and computing efficiency of complex models.