A parallel boundary value technique for singularly perturbed two-point boundary value problems (Q704687): Difference between revisions
From MaRDI portal
Removed claims |
Set profile property. |
||
(One intermediate revision by one other user not shown) | |||
Property / author | |||
Property / author: Jesus Vigo Aguiar / rank | |||
Normal rank | |||
Property / author | |||
Property / author: Srinivasan Natesan / rank | |||
Normal rank | |||
Property / MaRDI profile type | |||
Property / MaRDI profile type: MaRDI publication profile / rank | |||
Normal rank |
Latest revision as of 02:01, 5 March 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | A parallel boundary value technique for singularly perturbed two-point boundary value problems |
scientific article |
Statements
A parallel boundary value technique for singularly perturbed two-point boundary value problems (English)
0 references
19 January 2005
0 references
The problem under consideration is the singularly perturbed boundary value problem (BVP) for the differential equation \[ \varepsilon u''(x)+a(x)u'(x)-b(x)u(x)=f(x), \quad 0 < x < 1, \] under the boundary conditions \[ a(0)u'(0)-b(0)u(0)=f(0),\quad u(1)=\beta, \] where \(\varepsilon\) is a small positive parameter, \(a(x)\geq\alpha>0\), \(b(x)\geq\gamma>0\). The BVP has a unique solution \(u(x)\) exhibiting a boundary layer at \(x=0\). In order to obtain the numerical solution by using an \(r\)-processor system the authors suggest an iterative non-overlapping domain decomposition method. The initial segment \([0, 1]\) is divided in one boundary layer domain \([0, k\varepsilon]\) and \((r-1)\) regular domains. The initial BVP is substituted by \(r\) separated BVPs on each domain. The boundary conditions in the decomposition knots are defined as values of the solution \(u^0(t)\) to the reduced initial problem created by the initial BVP with \(\varepsilon=0\) and the right hand value \(u^0(1)=\beta\). These BVPs could be resolved simultaneously, as the authors affirm. However, any proof of such a substitution is absent. Numerical examples are presented.
0 references
boundary layer
0 references
parallel computing
0 references
exponential fitting
0 references
adapted multistep methods
0 references
singular perturbation problems
0 references
domain decomposition
0 references
numerical examples
0 references