Hi there I am presently accelerating my flosolver (solving 2d euler equations) using multigrid. I get a fairly good early acceleration for the first classic v-cycles; however afterwards, when shifting to the fine grid (prolongation using simple averaging), I start getting oscillations that become progressively large until the residue fall saturates before reaching convergence. Can anyone tell me how to avoid the problem and how to prolong the data from coarse to fine grid inorder to avoid those oscillations. I am using a finite difference based code. thanks Jim

Do you have shock wave?, if have, the prolongation should have some damping ability. If have not, maybe there is soemthing wrong with your code, your order of interpolation, your treatment of the boundary condictions on the coarse grid.

hi there, thanks a lot versi, i observed that by using a direct solution on the coarse grid (that one which has run for about 700 iterations say), the oscillations disappear completely and i am also able to get some acceleration on the fine grid. can this be explained? in fact i'm using only two grid levels (with a v cycle). jim