Innovative Methods and Theories in Numerical Algebra
Zhong‐Zhi Bai,Iain Duff,Lothar Reichel,Zuoqiang Shi
DOI: https://doi.org/10.1002/nla.1857
2012-01-01
Numerical Linear Algebra with Applications
Abstract:The Third International Conference on Numerical Algebra and Scientific Computing (NASC 2010) was held in Beijing, October 23–27, 2010. More than 150 participants from Australia, China, France, Germany, Greece, Japan, Korea, Russia, Sweden, Switzerland, UK, USA and other countries attended the conference. This issue of the journal ‘Numerical Linear Algebra with Applications’ contains carefully selected and refereed papers by participants of the conference. The conference was organized by the Chinese Academy of Sciences and was chaired by Martin H. Gutknecht, ETH Zurich, Switzerland, and by Zhong-Ci Shi, Chinese Academy of Sciences, China. The organizing committee consisted of Zhong-Zhi Bai, Xue-Bin Chi and Xue-Jun Xu, all from the Chinese Academy of Sciences, Beijing; Feng-Shan Bai from Tsinghua University; Xing-Ping Liu from Beijing Institute of Applied Physics and Computational Mathematics; and Shu-Fang Xu from Peking University. It was the purpose of the conference to gather experts in numerical algebra and scientific computing from all over the world to present state-of-the-art research results, exchange new ideas and discuss future developments, as well as to enhance academic communication and connections among them. The talks demonstrate the present activity in these areas and illustrate the diversity of the ongoing research. The topics discussed include direct and iterative methods for large systems of linear equations, numerical methods for systems of nonlinear equations, preconditioning methods for least-squares problems, sinc discretizations for linear ordinary differential equations, waveform relaxation methods for differential algebraic equations, spectral methods for time-dependent partial differential equations, robust methods for integral equations, splitting iteration methods for linear complementarity problems, iterative methods for ill-posed problems, multigrid and multilevel methods, preconditioning, homotopy methods, spectral deflation in Krylov solvers, linear and nonlinear eigenvalue problems, sparse and structured matrix computations, perturbation theory, algebraic Riccati equations, moments of linear operators and parallel computing. Applications to computational fluid dynamics, image restoration, control theory, model updating, computation of meshless elastic polyharmonic splines and thermal conduction were presented. We are happy to have been able to secure manuscripts at the leading edge of research on numerical algebra and scientific computing from seven of the speakers at the conference, and we hope that readers of this issue will enjoy the contributions as much as we have. Donatelli, Neuman and Reichel 1 discuss the design of square regularization matrices that can be used in iterative methods based on the Arnoldi process for large-scale Tikhonov regularization problems. Bai 2 establishes a class of block alternating splitting implicit iteration methods and demonstrates its unconditional convergence for the saddle-point problems arising from the finite element discretizations of the hybrid formulations of the time-harmonic eddy current problems. Experimental results are given to show the feasibility and effectiveness of this class of iterative methods when they are employed as preconditioners for Krylov subspace methods such as GMRES and BiCGSTAB. Brezinski, Fika and Mitrouli 3 give estimations of the trace of the inverse of an operator in a Hilbert space and of the norm of the error when solving a system of linear equations, which are obtained by extrapolation of the moments of the operator. By making use of the linearization method, Liu and Zhao 4 present new and computable estimates for the backward errors of approximate solutions of scaled total least-squares, linear equality constrained least-squares and least-squares over a sphere. Lambers 5 describes time-stepping methods for linear PDE that address stiffness through a componentwise approach. Krylov subspace spectral methods, which use techniques from ‘matrices, moments and quadrature’, are made far more efficient through asymptotic analysis of block Lanczos iteration used to compute each Fourier coefficient of the solution. Xia 6 studies a structured inner–outer preconditioner that uses a robust hierarchically semi-separable (HSS) inner factorization with low accuracy and a scalable HSS outer factorization. The method enhances both the robustness and the scalability of existing HSS methods for preconditioning and is shown to be very effective for ill-conditioned problems. Wang, Liang, Bai and Huo 7 find that the time spent for computing the permanent of a sparse matrix by a hybrid algorithm is strongly correlated to its permanent value. A strategy is introduced to improve a parallel algorithm for sparse permanents; methods for approximating permanents are used to approximate the permanent values of submatrices to decide the processing order of jobs, and numerical results are given to show that the parallel efficiency is improved remarkably for the permanents of fullerene graphs. Finally, we would like to thank Professors Maya G. Neytcheva and Panayot S. Vassilevski, the editors of the journal, for agreeing to and helping with the publication of this special issue. We would also like to take this opportunity to thank the sponsors of the conference: National Natural Science Foundation of China; The Hundred Talent Project of the Chinese Academy of Sciences; State Key Laboratory of Scientific/Engineering Computing; Academy of Mathematics and Systems Science, Chinese Academy of Sciences; Laboratory of Computational Physics, Beijing Institute of Applied Physics and Computational Mathematics; and Department of Mathematical Sciences, Tsinghua University. Our special thanks go to Dr. Zhi-Ru Ren and Drs. Jun-Liang Dong, Shuai Jian, Jian-Yu Pan, Yu-Hong Ran, Xi Yang, Jun-Feng Yin and Li-Li Zhang, as well as Ms. Ru-Juan Ding and others, for their enthusiasm and help with the organization of the conference.