Runge-Kutta Methods for Partial Differential Equations and Fractional Orders of Convergence (Q4031571)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 149258
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Runge-Kutta Methods for Partial Differential Equations and Fractional Orders of Convergence |
scientific article; zbMATH DE number 149258 |
Statements
1 April 1993
0 references
fractional orders of convergence
0 references
method of lines
0 references
\(A\)-stability
0 references
\(A\)-stable methods
0 references
numerical examples
0 references
global error estimate
0 references
implicit Runge-Kutta methods
0 references
order of consistency
0 references
heat equation
0 references
Runge-Kutta Methods for Partial Differential Equations and Fractional Orders of Convergence (English)
0 references
0 references
0 references
0 references
0.8537716269493103
0 references
0.8260170221328735
0 references
0.8149394392967224
0 references
0.8112508654594421
0 references
This elegant paper studies the (possibly fractional) order behaviour of implicit Runge-Kutta methods when applied to linear partial differential equations \(u_ t(x,t) = Lu(x,t) + f(x,t)\), \({u(0) = u_ 0}\), where \(L\) is a differential operator with a pure point spectrum of nonpositive real part. An \(A\)-stable Runge-Kutta method with stage order \(q\) and order of consistency \(p\) is shown to have global order given by \(\alpha_ r = \min\{p,q + 2 + \nu_ r\}\), where \(\nu_ r\) depends on the \(L_ r\) norm used to estimate the global error. In the case of the one- dimensional heat equation \(u_ t = u_{xx} + a(x)g(t)\) on \(\Omega = (0,1)\) with \(a(x)\) differentiable and nonvanishing on \(\partial\Omega\), \(\nu_ r = {1\over 2r}\). A similar analysis for nonhomogeneous boundary conditions is given -- in which case \(q+2\) is replaced by \(q+1\) or (in the case of even stage Gauss methods) \(q\). The results are numerically verified for a simple semi- discretized problem.
0 references