Abstract
This paper discusses the design of efficient algorithms for solving linear nonsymmetric systems and nonlinear systems associated with FEM approximation of elliptic PDEs. The novel feature of the designed linear solvers like GMRES, BICGSTAB(\(\ell \)), TFQMR, and nonlinear solvers like Newton and Picard, is the incorporation of error control in the ‘natural norm’ in combination with an effective a posteriori estimator for the PDE approximation error. This leads to robust black-box stopping criteria in the sense that the iteration is terminated as soon as the algebraic error is insignificant compared to the approximation error. Such a solver is called ‘balanced’ in this paper since the stopping criteria are obtained by balancing (comparing) the algebraic error and the approximation error.
Similar content being viewed by others
Notes
Galerkin approximation for (18) is inaccurate if the mesh is not fine enough to resolve the layers in the solution and these inaccuracies may also propagate and pollute the approximated solution in regions where the exact solution is well behaved. An alternative way to handle boundary layers is by using Shishkin grids; see [19].
Note that for BICGSTAB2, \(k_1\), \(k_2\) are the iteration counts for the built in stopping criterion in the IFISS implementation of BICGSTAB2 to tolerances of 1e−6, 1e−9 respectively. The iteration counts \(k^*\) for BICGSTAB2 is obtained by incorporating the weak stopping criterion in this IFISS implementation of BICGSTAB2.
The average cputimes were computed for each h by averaging over cputimes generated from 10 independent runs (of eigs and the IFISS script for computing \(\eta _h\)) for each h.
This observation is clear from \(\Lambda _h\) values in Table 5, however a rigorous mathematical proof is still under research.
\(\mathbf{X}^h_{{E}}\) is not a vector space unless its elements (which are functions) are zero on the boundary.
This is not a rigorous mathematical statement. A proof for this statement is an ongoing research.
This \(k^*\) will in general be different for different l.
This a posteriori approximation error estimate is for the linearized part \((\delta \overrightarrow{u}_h^{(l_k)}, \delta {p}_h^{(l_k)})\).
References
Arioli, M., Loghin, D., Wathen, A.J.: Stopping criteria for iterations in finite element methods. Numer. Math. 99(3), 381–410 (2005). https://doi.org/10.1007/s00211-004-0568-z
Brenner, S.C., Scott, L.R.: The Mathematical Theory of Finite Element Methods, Third edn. Springer, Berlin (2008)
Elman, H., Silvester, D., Wathen, A.: Finite Elements and Fast Iterative Solvers: With Applications in Incompressible Fluid Dynamics, Second edn. Oxford University Press, Oxford (2014)
Elman, H.C., Ramage, A., Silvester, D.J.: IFISS: a computational laboratory for investigating incompressible flow problems. SIAM Review 56(2), 261–273 (2014). https://doi.org/10.1137/120891393
Eriksson, K., Estep, D., Hansbo, P., Johnson, C.: Computational Differential Equations, First edn. Cambridge University Press, Cambridge (1996)
Freund, R.W.: A transpose-free quasi-minimal residual algorithm for non-hermitian linear systems. SIAM J. Sci. Comput. 14(2), 470–482 (1993). https://doi.org/10.1137/0914029
Freund, R.W., Gutknecht, M.H., Nachtigal, N.M.: An implementation of the look-ahead Lanczos algorithm for non-Hermitian matrices. SIAM J. Sci. Comput. 14(1), 137–158 (1993). https://doi.org/10.1137/0914009
Golub, G.H., Van Loan, C.F.: Matrix Computations, Fourth edn. The John Hopkins University Press, Baltimore (2013)
Gresho, P.M., Gartling, D.K., Torczynski, J.R., Cliffe, K.A., Winters, K.H., Garratt, T.J., Spence, A., Goodrich, J.W.: Is the steady viscous incompressible two-dimensional flow over a backward-facing step at Re = 800 stable? Int. J. Numer. Methods Fluids 17(6), 501–541 (1993). https://doi.org/10.1002/fld.1650170605
Hughes, T.J.R., Brooks, A.: A multi-dimensional upwind scheme with no crosswind diffusion. In: T. Hughes (ed.) Finite Element Methods for Convection Dominated Flows, ASME Winter Annual Meeting, New York, USA, vol. 34, pp. 19–35. https://www.researchgate.net/publication/297681092 (1979)
Jiránek, P., Strakos, Z., Vohralík, M.: A posteriori error estimates including algebraic error and stopping criteria for iterative solvers. SIAM J. Sci. Comput. 32(3), 1567–1590 (2010). https://doi.org/10.1137/08073706X
Oden, J.T., Demkowicz, L.F.: Applied Functional Analysis, First edn. CRC Press, Boca Raton (1996)
Paige, C.C., Saunders, M.A.: Solution of sparse indefinite systems of linear equations. SIAM J. Numer. Anal. 12(4), 617–629 (1975). https://doi.org/10.1137/0712047
Pietro, D.A.D., Flauraud, E., Vohralík, M., Yousef, S.: A posteriori error estimates, stopping criteria, and adaptivity for multiphase compositional refinement for thermal multiphase compositional flows in porous media. J. Comput. Phys. 276, 163–187 (2014). https://doi.org/10.1016/j.jcp.2014.06.061
Pietro, D.A.D., Vohralík, M., Yousef, S.: An a posteriori-based, fully adaptive algorithm with adaptive stopping criteria and mesh refinement for thermal multiphase compositional flows in porous media. Comput. Math. Appl. 68(12 B), 2331–2347 (2014). https://doi.org/10.1016/j.camwa.2014.08.008
Powell, C.E., Silvester, D.J.: Preconditioning steady-state Navier–Stokes equations with random data. SIAM J. Sci. Comput. 34(5), A2482–A2506 (2012). https://doi.org/10.1137/120870578
Pranjal, P.: Optimal iterative solvers for linear systems with stochastic PDE origins: balanced black-box stopping tests. PhD thesis. University of Manchester, UK. http://eprints.maths.manchester.ac.uk/2596/ (2017)
Saad, Y., Schultz, M.: A generalized minimal residual algorithm for solving nonsymmetric linear systems. SIAM J. Sci. Comput. 7(3), 856–869 (1986). https://doi.org/10.1137/0907058
Shishkin, G.I.: Methods of constructing grid approximations for singularly perturbed boundary-value problems. Condensing grid methods. Rus. J. Numer. Anal. Math. Model. 7(6), 537–562 (1992). https://doi.org/10.1515/rnam.1992.7.6.537
Silvester, D., Pranjal, P.: An optimal solver for linear systems arising from stochastic FEM approximation of diffusion equations with random coefficients. SIAM/ASA J. Uncertain. Quantif. 4(1), 298–311 (2016). https://doi.org/10.1137/15M1017740
Sleijpen, G.L.G., Fokkema, D.R.: BICGSTAB(L) for linear equations involving unsymmetric matrices with complex spectrum. Electron. Trans. Numer. Anal. 1, 11–32 (1993)
Syamsudhuha, Silvester D.J.: Efficient solution of the steady-state Navier–Stokes equations using a multigrid preconditioned Newton–Krylov method. Int. J. Numer. Methods Fluids 43(12), 1407–1427 (2003). https://doi.org/10.1002/fld.627
Verfürth, R.: A Posteriori Error Estimation Techniques for Finite Element Methods, First edn. Oxford University Press, Oxford (2013)
Wathen, A.: Preconditioning and convergence in the right norm. Int. J. Comput. Math. 84(8), 1199–1209 (2007). https://doi.org/10.1080/00207160701355961
Wu, C.T.: On the implementation of an accurate and efficient solver for convection–diffusion equations. PhD thesis. University of Maryland, USA. https://drum.lib.umd.edu/handle/1903/32 (2003)
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Pranjal, Silvester, D. Balanced Iterative Solvers for Linear Nonsymmetric Systems and Nonlinear Systems with PDE Origins: Efficient Black-Box Stopping Criteria. J Sci Comput 81, 271–290 (2019). https://doi.org/10.1007/s10915-019-01018-w
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10915-019-01018-w