Uniform approximation of singularly perturbed reaction-diffusion problems by the finite element method on a Shishkin mesh. (English) Zbl 1026.65106

The authors consider the two-dimensional singularly perturbed problem \[ -\varepsilon^2\Delta u+ u= f\quad\text{in }\Omega\subset \mathbb{R}^2,\quad u=0\quad\text{on }\partial\Omega, \] where the parameter \(\varepsilon\) can approach zero, \(\partial\Omega\) is smooth and \(f\in H^{4M+2}(\Omega)\) for some \(M\in \{0,1,2,\dots\}\). Let \(u_\varepsilon\in H^1_0(\Omega)\) denote the unique solution to the following variational formulation of the above problem \[ B(u_\varepsilon, v)= F(v),\;\forall v\in H^1_0(\Omega), \] where \[ B(u,v):=\int_\Omega(\varepsilon^2\nabla u\cdot\nabla v+ uv) dx dy,\;F(v):= \int_\Omega fvdx dy \] and let \[ \Omega_0:= \{z- \rho\overline n_z; z\in\partial\Omega, 0<\rho\rho_0\}, \] where \(\rho_0\) is a positive constant less than the minimum radius of curvature of \(\partial\Omega\) and \(\overline n_z\) is the outward unit normal at a point \(z\in\partial\Omega\).
Introducing the so-called Shishkin mesh \(\Delta_0\) in \(\Omega_0\) and a quasi-uniform mesh \(\Delta_1\) in \(\Omega_1:= \Omega\setminus\Omega_0\) compatible with \(\Delta_0\) the authors obtain the following approximation of \(u_\varepsilon\) by its finite element counterpart \(v_N\in S_N\), uniformly with respect to \(\varepsilon\), namely \[ \|u_\varepsilon- v_N\|_{\varepsilon,\Omega}\leq CN^{-p/2}[\ln(n+1)]^p, \] where \(C\in\mathbb{R}\) does not depend of \(\varepsilon\) and \(N\). Here \(S_N\) is the finite element space on \(\Delta_0\cup\Delta_1\) of piecewise polynomials of degree \(\leq p\), \(N= \dim S_N\), \(n\) is the number of subdivisions within \(\Omega_0\) in the direction normal to the boundary and the energy norm \(\|\cdot\|_{\varepsilon,\Omega}\) is defined by \[ \|\cdot\|^2_{\varepsilon,\Omega}:= \varepsilon^2|\cdot|^2_{1,\Omega}+\|\cdot \|^2_{0,\Omega}, \] \(\|\cdot\|_{k,\Omega}\) and \(|\cdot|_{k,\Omega}\) being the norm and seminorm in \(H^k(\Omega)\), respectively.
Numerical examples are presented giving high agreement of numerics with theory.
Reviewer: S.Burys (Kraków)


65N30 Finite element, Rayleigh-Ritz and Galerkin methods for boundary value problems involving PDEs
65N15 Error bounds for boundary value problems involving PDEs
35J25 Boundary value problems for second-order elliptic equations
35B25 Singular perturbations in context of PDEs
65N50 Mesh generation, refinement, and adaptive methods for boundary value problems involving PDEs
Full Text: DOI


[1] and Robust computational techniques for boundary layers, Chapman & Hall/CRC Press, Boca Raton, FL, 2000.
[2] and Fitted numerical methods for singular perturbation problems, World Scientific, Singapore, 1995.
[3] Numerical solution of convection-diffusion problems, Chapman and Hall, London, 1996.
[4] and Numerical methods for singularly perturbed differential equations, Springer-Verlag, Berlin, 1996.
[5] Blatov, Differ Eq 28-7 pp 931– (1993)
[6] Blatov, Differ Eq 28-10 pp 1469– (1993)
[7] Xenophontos, Int J Numer Meth Eng 53 pp 929– (2002)
[8] Shishkin, Soviet J Numer Anal Math Modeling 4 pp 397– (1989)
[9] Li, Comp Meth Appl Mech Eng 171 pp 1– (1999)
[10] Sun, IMA J Numer Anal 15 pp 117– (1995)
[11] Zhang, Numer Methods PDE 18 pp 374– (2002)
[12] Schwab, SIAM J Numer Anal 35 pp 1520– (1998)
[13] Schwab, Math Comp 65 pp 1403– (1996)
[14] Schwab, Comp Meth Appl Mech Eng 157 pp 311– (1996)
[15] The hp finite element method for singularly perturbed problems, Ph.D. Dissertation, University of Maryland, Baltimore, 1996.
[16] Xenophontos, Math Model Methods Appl Sci 2 pp 299– (1998)
[17] Xenophontos, Numer Meth PDE 15 pp 63– (1998)
[18] Eckhaus, SIAM Rev 72 pp 225– (1973)
[19] The finite element method for elliptic problems, North-Holland, Amsterdam, 1978.
[20] Treatise on the theory of Bessel functions, Cambridge University Press, New York, 1966.
[21] Kellogg, SIAM J Numer Anal 34-5 pp 1808– (1997)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.