FS#108 Investigate non-linear Multi-Grid methods
Metadata
Property |
Value |
Project |
dumux |
Category |
General |
Reported by |
Anonymous (Id=0) |
Reported at |
May 4, 2011 17:20 |
Type |
Bug Report |
Version |
Git |
Last edited by |
Bernd Flemisch (bernd@iws.uni-stuttgart.de) |
Last edited at |
Jul 9, 2012 09:13 |
Closed by |
Bernd Flemisch (bernd@iws.uni-stuttgart.de) |
Closed at |
Jul 9, 2012 09:13 |
Closed in version |
unknown (Id=0) |
Resolution |
Won't implement |
Description
Typically the maximum time step size which can be achieved to solve a non-linear problem is highly dependent on the discretization length of the spatial discretization. It could be worth investigating if a hierarchical approach improves the situation if using implicit models. The algorithm might look like this:
grid.globalRefine(maxLevel);
initialSolution(grid);
dt = dtInitial;
for (t = 0; t < maxTime; t += dt) {
coarsenFinestSolutionToLevel(0);
for (refinementLevel in 0..maxLevel) {
solveAt(refinementLevel);
transferSolutionFrom(refinementLevel to refinementLevel+1);
}
dt = estimateNextDt();
}