A uniform numerical method for semilinear reaction-diffusion problems with a boundary turning point

From MaRDI portal





The paper deals with the semilinear reaction-diffusion boundary-turning-point problem: \[ -\varepsilon^2 u + a(x,u)[1-e_\varepsilon(x)] = f(x)\text{ in }(0,1) \] with Dirichlet boundary conditions, where \( 0 < \varepsilon \ll 1\), \(a \in C^4\), \(a_u(x,u) > \alpha^2 > 0\), \(e_\varepsilon(x) = \exp(-\beta x^2/\varepsilon^2)\), and \(\beta > 0\). The authors prove an a priory estimate \(|u^{(k)}(x)| \leq C[1 + \varepsilon^{-k}\exp(-\alpha x/\varepsilon) + \varepsilon^{-k}\exp(-\alpha (x-1)/\varepsilon)]\) for \(1 \leq k \leq 4\) and a quadratic convergence rate of the finite-difference approximations on Bakhvalov-type meshes in \(L^\infty\)-norm. This convergence is uniform with respect to \(\varepsilon\). The concluding numerical experiments verify this uniform quadratic convergence by computing the experimental orders of convergence for a linear and a semilinear problem. The results and methods are motivated by \textit{C. De Falco} and \textit{E. O'Riordan} [Hegarty, Alan F. (ed.) et al., BAIL 2008 -- Boundary and interior layers. Proceedings of the international conference on boundary and interior layers -- computational and asymptotic methods, Limerick, Ireland, July 28--August 1, 2008. Berlin: Springer. Lecture Notes in Computational Science and Engineering 69, 129-139 (2009; Zbl 1180.35048)].











This page was built for publication: A uniform numerical method for semilinear reaction-diffusion problems with a boundary turning point

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5961875)