Chen-Song Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Chen-Song Zhang has a total of 68 co-authors. Here are the top co-authors by number of shared publications:
- Xuefeng Xu
School of Mathematics, Southeast University, Nanjing, China
7 shared publications - Jinchao Xu
Department of Mathematics Pennsylvania State University 16802 University Park PA USA
6 shared publications - Xiaowen Xu
Laboratory of Computational Physics, Institute of Applied Physics and Computational Mathematics, Beijing, China
4 shared publications - Chunsheng Feng
Xiangtan University
3 shared publications Li Zhao
Academy of Mathematics and System Sciences, Chinese Academy of Sciences and University of Chinese Academy of Sciences
3 shared publications- Shi Shu
Hunan Key Laboratory for Computation and Simulation in Science and Engineering, Key Laboratory of Intelligent Computing and Information Processing of Ministry of Education, School of Mathematics and Computational Science, Xiangtan University, Xiangtan 411105, China
3 shared publications
Latest Publications
- 22 September 2026
- Published by Elsevier in Computers & Mathematics with Applications
- Vol. 223, 74-98
Physics-informed neural networks for solving two-phase flow problems with moving interfaces
- 22 September 2026
- Published by Elsevier in Computers & Mathematics with Applications
- Vol. 223, 74-98
Physics-informed neural networks for solving two-phase flow problems with moving interfaces
- Open Access
Massively parallel Schwarz methods for the high frequency Helmholtz equation
We investigate the parallel one-level overlapping Schwarz method for solving finite element discretization of high-frequency Helmholtz equations. The resulting linear systems are large, indefinite, ill-conditioned, and complex-valued. We present a practical variant of the restricted additive Schwarz method with Perfectly Matched Layer transmission conditions (RAS-PML), which was originally analyzed in a theoretical setting in {\tt arXiv:2404.02156}, with some numerical experiments given in {\tt arXiv:2408.16580}. In our algorithm, the width of the overlap and the additional PML layer on each subdomain is allowed to decrease with $\mathcal{O}(k^{-1} \log(k))$, as the frequency $k \rightarrow \infty$, and this is observed to ensure good convergence while avoiding excessive communication. In experiments, the proposed method achieves $\mathcal{O}(k^d)$ parallel scalability under Cartesian domain decomposition and exhibits $\mathcal{O}(k)$ iteration counts and convergence time for $d$-dimensional Helmholtz problems ($d = 2,3$) as $k$ increases. In this preliminary note we restrict to experiments on 2D problems with constant wave speed. Details, analysis and extensions to variable wavespeed and 3D will be given in future work.Open AccessMassively parallel Schwarz methods for the high frequency Helmholtz equation
We investigate the parallel one-level overlapping Schwarz method for solving finite element discretization of high-frequency Helmholtz equations. The resulting linear systems are large, indefinite, ill-conditioned, and complex-valued. We present a practical variant of the restricted additive Schwarz method with Perfectly Matched Layer transmission conditions (RAS-PML), which was originally analyzed in a theoretical setting in {\tt arXiv:2404.02156}, with some numerical experiments given in {\tt arXiv:2408.16580}. In our algorithm, the width of the overlap and the additional PML layer on each subdomain is allowed to decrease with $\mathcal{O}(k^{-1} \log(k))$, as the frequency $k \rightarrow \infty$, and this is observed to ensure good convergence while avoiding excessive communication. In experiments, the proposed method achieves $\mathcal{O}(k^d)$ parallel scalability under Cartesian domain decomposition and exhibits $\mathcal{O}(k)$ iteration counts and convergence time for $d$-dimensional Helmholtz problems ($d = 2,3$) as $k$ increases. In this preliminary note we restrict to experiments on 2D problems with constant wave speed. Details, analysis and extensions to variable wavespeed and 3D will be given in future work. - 14 January 2026
- Published by Elsevier in Journal of Computational Physics
- Vol. 551
R-PINN: Recovery-type a-posteriori estimator enhanced adaptive PINN
- 14 January 2026
- Published by Elsevier in Journal of Computational Physics
- Vol. 551
R-PINN: Recovery-type a-posteriori estimator enhanced adaptive PINN
- Open Access
Recurrent Neural Operators: Stable Long-Term PDE Prediction
Neural operators have emerged as powerful tools for learning solution operators of partial differential equations. However, in time-dependent problems, standard training strategies such as teacher forcing introduce a mismatch between training and inference, leading to compounding errors in long-term autoregressive predictions. To address this issue, we propose Recurrent Neural Operators (RNOs)-a novel framework that integrates recurrent training into neural operator architectures. Instead of conditioning each training step on ground-truth inputs, RNOs recursively apply the operator to their own predictions over a temporal window, effectively simulating inference-time dynamics during training. This alignment mitigates exposure bias and enhances robustness to error accumulation. Theoretically, we show that recurrent training can reduce the worst-case exponential error growth typical of teacher forcing to linear growth. Empirically, we demonstrate that recurrently trained Multigrid Neural Operators significantly outperform their teacher-forced counterparts in long-term accuracy and stability on standard benchmarks. Our results underscore the importance of aligning training with inference dynamics for robust temporal generalization in neural operator learning.
Open AccessRecurrent Neural Operators: Stable Long-Term PDE Prediction
Neural operators have emerged as powerful tools for learning solution operators of partial differential equations. However, in time-dependent problems, standard training strategies such as teacher forcing introduce a mismatch between training and inference, leading to compounding errors in long-term autoregressive predictions. To address this issue, we propose Recurrent Neural Operators (RNOs)-a novel framework that integrates recurrent training into neural operator architectures. Instead of conditioning each training step on ground-truth inputs, RNOs recursively apply the operator to their own predictions over a temporal window, effectively simulating inference-time dynamics during training. This alignment mitigates exposure bias and enhances robustness to error accumulation. Theoretically, we show that recurrent training can reduce the worst-case exponential error growth typical of teacher forcing to linear growth. Empirically, we demonstrate that recurrently trained Multigrid Neural Operators significantly outperform their teacher-forced counterparts in long-term accuracy and stability on standard benchmarks. Our results underscore the importance of aligning training with inference dynamics for robust temporal generalization in neural operator learning.
- 10 March 2025
- Published by Elsevier in Applied Mathematics Letters
- Vol. 166
A super-localized finite element method for inhomogeneous fourth-order singular perturbation problem
- 10 March 2025
- Published by Elsevier in Applied Mathematics Letters
- Vol. 166
A super-localized finite element method for inhomogeneous fourth-order singular perturbation problem
- Open Access
- 28 February 2025
- Published by Association for Computing Machinery (ACM)
- p. 496-511
Semi-StructMG: A Fast and Scalable Semi-Structured Algebraic Multigrid
Parallel multigrid methods are widely used as preconditioned for solving large sparse linear systems. Most multigrids rely on general sparse matrix formats, which prevent them from achieving optimal performance. There is an emerging trend towards semi-structured multigrids that balance flexibility with performance. However, existing libraries often fall short in terms of speed and scalability for semi-structured problems. To address these limitations, we have designed and implemented Semi-StructMG. It employs multi-dimensional coarsening to reduce complexity and simplify communication patterns. It also considers the special role of inter-block connections in smoothers and triple-matrix products to improve convergence under large-scale parallelism. We evaluated Semi-StructMG using two benchmark problems and four real-world applications from petroleum reservoir simulation, ship manufacturing, numerical weather prediction, and ocean modeling. Compared to hypre's multigrids, Semi-StructMG achieves the fastest time-to-solution across all cases, with average speedups of 5.97x, 15.2x, and 3.85x over SSAMG, Split, and BoomerAMG, respectively. Additionally, Semi-StructMG significantly improves both strong and weak scaling efficiencies in all tests. These results suggest that it can serve as an effective alternative to SSAMG and Split.
- 28 February 2025
- Published by Association for Computing Machinery (ACM)
- p. 496-511
Open AccessSemi-StructMG: A Fast and Scalable Semi-Structured Algebraic Multigrid
Parallel multigrid methods are widely used as preconditioned for solving large sparse linear systems. Most multigrids rely on general sparse matrix formats, which prevent them from achieving optimal performance. There is an emerging trend towards semi-structured multigrids that balance flexibility with performance. However, existing libraries often fall short in terms of speed and scalability for semi-structured problems. To address these limitations, we have designed and implemented Semi-StructMG. It employs multi-dimensional coarsening to reduce complexity and simplify communication patterns. It also considers the special role of inter-block connections in smoothers and triple-matrix products to improve convergence under large-scale parallelism. We evaluated Semi-StructMG using two benchmark problems and four real-world applications from petroleum reservoir simulation, ship manufacturing, numerical weather prediction, and ocean modeling. Compared to hypre's multigrids, Semi-StructMG achieves the fastest time-to-solution across all cases, with average speedups of 5.97x, 15.2x, and 3.85x over SSAMG, Split, and BoomerAMG, respectively. Additionally, Semi-StructMG significantly improves both strong and weak scaling efficiencies in all tests. These results suggest that it can serve as an effective alternative to SSAMG and Split.
- 3 January 2025
- Published by Springer Nature in Journal of Scientific Computing
- Vol. 102 (2) , 1-47
A Stokes–Dual–Porosity–Poroelasticity Model and Discontinuous Galerkin Method for the Coupled Free Flow and Dual Porosity Poroelastic Medium Problem
- 3 January 2025
- Published by Springer Nature in Journal of Scientific Computing
- Vol. 102 (2) , 1-47
A Stokes–Dual–Porosity–Poroelasticity Model and Discontinuous Galerkin Method for the Coupled Free Flow and Dual Porosity Poroelastic Medium Problem
- 16 May 2024
- Published by Elsevier in Applied Numerical Mathematics
- Vol. 203, 235-254
Lowest-degree robust finite element schemes for inhomogeneous bi-Laplace problems
- 16 May 2024
- Published by Elsevier in Applied Numerical Mathematics
- Vol. 203, 235-254
Lowest-degree robust finite element schemes for inhomogeneous bi-Laplace problems
- Open Access
- 20 February 2024
- Published by Association for Computing Machinery (ACM)
- p. 478-480
POSTER: StructMG: A Fast and Scalable Structured Multigrid
Parallel multigrid is widely used as preconditioners in solving large-scale sparse linear systems. However, the current multigrid library still needs more satisfactory performance for structured grid problems regarding speed and scalability. To this end, we design and implement StructMG, a fast and scalable multigrid that constructs hierarchical grids automatically based on the original matrix. As a preconditioner, StructMG can achieve both low cost per iteration and good convergence. Two idealized and five real-world problems from four application fields, including radiation hydrodynamics, petroleum reservoir simulation, numerical weather prediction, and solid mechanics, are evaluated on ARM and X86 platforms. In comparison to hypre's multigrid preconditioners, StructMG achieves the fastest time-to-solutions in all cases with average speedups of 17.6x, 5.7x, 4.6x, 8.5x over SMG, PFMG, SysPFMG, and BoomerAMG, respectively. Additionally, StructMG significantly improves strong and weak scaling efficiencies in most tests.
- 20 February 2024
- Published by Association for Computing Machinery (ACM)
- p. 478-480
Open AccessPOSTER: StructMG: A Fast and Scalable Structured Multigrid
Parallel multigrid is widely used as preconditioners in solving large-scale sparse linear systems. However, the current multigrid library still needs more satisfactory performance for structured grid problems regarding speed and scalability. To this end, we design and implement StructMG, a fast and scalable multigrid that constructs hierarchical grids automatically based on the original matrix. As a preconditioner, StructMG can achieve both low cost per iteration and good convergence. Two idealized and five real-world problems from four application fields, including radiation hydrodynamics, petroleum reservoir simulation, numerical weather prediction, and solid mechanics, are evaluated on ARM and X86 platforms. In comparison to hypre's multigrid preconditioners, StructMG achieves the fastest time-to-solutions in all cases with average speedups of 17.6x, 5.7x, 4.6x, 8.5x over SMG, PFMG, SysPFMG, and BoomerAMG, respectively. Additionally, StructMG significantly improves strong and weak scaling efficiencies in most tests.