Article Contents
Article Contents

# Optimal number of Schur subdomains: Application to semi-implicit finite volume discretization of semilinear reaction diffusion problem

• * Corresponding author: N. Nagid
• The purpose of this paper is to establish a new numerical approach to solve, in two dimensions, a semilinear reaction diffusion equation combining finite volume method and Schur complement method. We applied our method for q = 2 non-overlapping subdomains and then we generalized in the case of several subdomains (q≥2). A large number of numerical test cases shows the efficiency and the good accuracy of the proposed approach in terms of the CPU time and the order of the error, when increasing the number of subdomains, without using the parallel computing. After several variations of the number of subdomains and the mesh grid, we remark two significant results. On the one hand, the increase related to the number of subdomains does not affect the order of the error, on the other hand, for each mesh grid when we augment the number of subdomains, the CPU time reaches the minimum for a specific number of subdomains. In order to have the minimum CPU time, we resorted to a statistical study between the optimal number of subdomains and the mesh grid.

Mathematics Subject Classification: Primary: 65M08, 65M55; Secondary: 62J05.

 Citation:

• Figure 1.  FV structured mesh of domain Ω

Figure 2.  Non-overlapping strip decomposition

Figure 3.  Domain decomposition and structured conforming mesh of domain Ω

Figure 4.  FE solution (left) and FV-1D (right), for h = 0:0138 and t = 0:5 in 3D

Figure 5.  FE solution (left) and FV-1D (right), for h = 0:0138 and t = 0:5 in 2D

Figure 6.  Point-wise errors between the solution by FV-1Dom and by FE method (with the basis function in P1 (left) and P2 (right)), for $h=0.0138$ and $t=0.5$

Figure 7.  Point-wise errors between the solution by FV-SC (2SD) and by FE method (with the basis function in P1 (left) and P2 (right)), for $h=0.0138$ and $t=0.5$

Figure 8.  $L^{\infty}(\Omega)$ error history between FE and FV-1Dom (left) (resp. FV-SC (right)), with the basis functions in P1 (FE-P1) and P2 (FE-P2), for $h=0.0138$ and $t=1$

Figure 9.  FV-1Dom solutions for different values of $h$, $y=0.0833$ (left) and $y=1.9167$ (right) at $t=0.5$

Figure 10.  FV-SC (2 subdomains) solutions for different values of $h$, $y=0.0833$ (left) and $y=1.9167$ (right) at $t=0.5$

Figure 11.  CPU time in seconds for FV-1Dom and FV-SC (h = 0:0138 and t = 1)

Figure 12.  Graphic representation of the adjustment to the power model

Table 1.  $L^{2}(\Omega)$ errors between FV and FE-P2 ($h_{FE}=0.0069$) for different values of $h$ and of the number of subdomains, at $t=0.5$

 Number of Subdomains 0.1666 h 0.0138 0.0069 1 Domain 0.0243 1.7708E-4 9.8594e-05 2 Subdomains 0.0243 1.7221E-4 8.9848e-05 3 Subdomains 0.0243 1.7430E-4 6.9495e-05 4 Subdomains 0.0243 1.7600E-4 7.0203e-05 6 Subdomains 0.0243 1.7485E-4 7.1316e-05 8 Subdomains - 1.7375E-4 6.922e-05 9 Subdomains - 1.7357E-4 7.0606e-05

Table 2.  CPU time calculation for FV-1Dom and FV-SC at t = 1

 Number of Subdomains 0.1666 h 0.0138 0.0069 1 Domain 0.1139 (second) 0.6285 (hour) 1.9884 (hour) 2 Subdomains 0.4891 (second) 0.4084 (hour) 1.4347 (hour) 3 Subdomains 0.4602 (second) 0.3264 (hour) 1.3845 (hour) 4 Subdomains 0.4823 (second) 0.3198 (hour) 1.4279 (hour) 6 Subdomains 0.5081 (second) 0.3299 (hour) 1.6049 (hour) 8 Subdomains - 0.3555 (hour) 1.1068 (hour) 9 Subdomains - 0.3700 (hour) 1.1577 (hour)

Table 3.  Optimal number of subdomains for different values of h

 Optimal Number of subdomains h CPU time 1 Domain 0.1666 0.2018 (second) 2 Subdomains 0.083333 1.2547 (second) 3 Subdomains 0.01388 0.1344 (hour) 4 Subdomains 0.012820 0.4736 (hour) 5 Subdomains 0.00952 1.0142 (hour) 6 Subdomains 0.00925 0.9637 (hour) 7 Subdomains 0.00793 0.9124 (hour) 8 Subdomains 0.00694 1.0741 (hour) 9 Subdomains 0.00347 1.1570 (hour) 10 Subdomains 0.002222 1.4655 (hour) 11 Subdomains 0.002164 1.5632 (hour) 12 Subdomains 0.0021367 1.5932 (hour) 13 Subdomains 0.002051 2.1130 (hour) 35 Subdomains 0.00029304 5.4254 (hour)
•  [1] F. Caetano, M. J. Gander, L. Halpern and J. Szeftel, Schwarz Waveform Relaxation Algorithms for Semilinear Reaction-Diffusion Equations, Networks And Heterog. Media, 5 (2010), 487-505.  doi: 10.3934/nhm.2010.5.487. [2] L. M. Carvalho, L. Giraud and P. Le Tallec, Algebraic two-level preconditioners for the Schur complement method, SIAM Journal on Scientific Computing, 22 (2000), 1987-2005.  doi: 10.1137/S1064827598340809. [3] T. Cazenave and A. Haraux, An Introduction to Semilinear Evolution Equations Oxford University, Clarendon Press, 1998. [4] C. Chen and W. Liu, Two-grid finite volume element methods for semilinear parabolic problems, Applied Numerical Mathematics, 60 (2010), 10-18.  doi: 10.1016/j.apnum.2009.08.004. [5] M. Dryja and O. B. Widlund, Domain decomposition algorithms with small overlap, SIAM Journal on Scientific Computing, 15 (1994), 604-620.  doi: 10.1137/0915040. [6] S. Duminil, H. Sadok and D. B. Szyld, Nonlinear Schwarz iterations with reduced rank extrapolation, Applied Numerical Mathematics, 94 (2015), 209-221.  doi: 10.1016/j.apnum.2015.04.001. [7] R. I. Fernandes and G. Fairweather, An ADI extrapolated Crank-Nicolson orthogonal spline collocation method for nonlinear reaction-diffusion systems, Journal of Computational Physics, 231 (2012), 6248-6267.  doi: 10.1016/j.jcp.2012.04.001. [8] C. Hay-Jahans, An R Companion to Linear Statistical Models CRC Press, 2011. doi: 10.1201/b11157. [9] F. N. Hochbruck and X. C. Cai, A class of parallel two-level nonlinear Schwarz preconditioned inexact Newton algorithms, Computer methods in applied mechanics and engineering, 196 (2007), 1603-1611.  doi: 10.1016/j.cma.2006.03.019. [10] S. Khallouq and H. Belhadj, Schur complement technique for advection-diffusion equation using matching structured finite volumes, Advances in Dynamical Systems and Applications, 8 (2013), 51-62. [11] R. Klajn, M. Fialkowski, I. T. Bensemann, A. Bitner, C. J. Campbell, K. Bishop, S. Smoukov and B. A. Grzybowski, Multicolour Micropatterning of Thin Films of Dry Gels, Nature materials, 3 (2004), 729-735.  doi: 10.1038/nmat1231. [12] K. J. Lee, W. D. McCormick, J. E. Pearson and H. L. Swinney, Experimental Observation of Self-Replicating Spots in a Reaction-Diffusion System, Nature, 369 (1994), 215-218.  doi: 10.1038/369215a0. [13] J. W. Lottes and F. Fischer Paul, Hybrid multigrid/Schwarz algorithms for the spectral element method, Journal of Scientific Computing, 24 (2005), 45-78.  doi: 10.1007/s10915-004-4787-3. [14] A. Quarteroni and A. Valli, Domain Decomposition Methods for Partial Differential Equations Oxford University Press, 1999. [15] A. Quarteroni and A. Valli, Theory and application of stecklov-poincaré operators for boundary value problems, In Applied and Industrial Mathematics. Springer Netherlands, 56 (1991), 179-203. [16] L. Roques, Equations de Réaction-Diffusion non-Linéaires et Modélisation en Ecologie Thesis of Univ. Pierre et Marie Curie, Paris 6,2004. [17] N. Shigesada and K. Kawasaki, Biological Invasions -Theory and Practice Oxford Series in Ecology and Evolution, Oxford University Press, 1997. [18] J. H. Stapleton, Linear Statistical Models John Wiley and Sons, 2009. [19] T. P. A. Mathew, Domain Decomposition Methods for the Numerical Solution of Partial Differential Equations Lecture Notes in Computational Science and Engineering, 61, Springer-Verlag, Berlin, 2008. doi: 10.1007/978-3-540-77209-5.

Figures(12)

Tables(3)