-
Notifications
You must be signed in to change notification settings - Fork 29
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
FV->FV bilin and related algorithms dont work with MPAS oEC60to30v3 source grid #107
Comments
for ocn->atm maps, bilin, intbilin and intbilinbg all fail:
|
Regarding the first issue, it appears to have been an issue with the tolerance for the iterative method used to solve for the weights int "intbilin." If you go to line 933 of "NewtonQuadrilateral" FiniteVolumeTools.cpp, and change the stopping criterion to 1e-30, then it works fine. Regarding the second issue, it is related to the ConstructEdgeMap() function. The methods "bilin," "intbilin," "bilingb," and "intbilingb" all rely on the dual of the source mesh and its edge map. There appears to be no issue constructing the dual of the ocean mesh, but the error "FacePair already has a full set of Faces" occurs with ConstructEdgeMap(), i.e. when the edge map is being constructed. I'm not sure why there would be a problem, however. Do you notice any problems with the dual of the ocean grid? |
Thanks. The 1e-30 tol for atm->ocn maps fixed the issue. For ocn->atm maps, when you wrote"Did you notice any problems with the dual of the ocean grid?" - what did you mean? My understanding is that this is computed internally in TR that I dont have access to? Note that these are MPAS ocean grids, which only mesh the ocean so they are not global. |
The first two steps of the schemes that are currently failing are to construct the dual of the source mesh and its edgemap. Something isn't working in one of these steps. Is there any reason why a mesh that is not global would cause either the function "Dual" (GridElements.cpp line 3128) or "ConstructEdgeMap" to fail? I've only tested these schemes on global meshes. There is nothing wrong with the underlying algorithms here; they will work as long as the dual of the source mesh and its edgemap are constructed properly. |
There's a new pull request that should fix the problems for FV->FV bilin for MPAS oEC60to30v3. Also, it should work for the function LinearRemapGeneralizedBarycentric, but there doesn't seem to be a command line parameter to use this function. I haven't implemented the changes for intbilin or intbilinbg yet. |
Is this fixed by PR #112 ? |
Yes, this is fixed by PR #112 (at least for bilin and bilingb. I'm working on the intbilin and intbilingb). |
error with intbilin. But bilin and intbilinbg both work.
The text was updated successfully, but these errors were encountered: