RBF motion solver does not work well in parallel
Happy New Year!
I found that the RBF motion solver does ot work very well in parallel.
1. Since the solver needs movingIDs for interpolation, but this is not guaranteed after decomposition, some of the processors may not have the moving patches. Using manual decompose is a solution, but not a good solution.
2. Even if the moving patches' cells are decomposed into all the processors, isn't it a better idea to remain all the processors have the same controlPoints as based points for interpolation?
3. After forcing all the processors to have the same controlPoints, I still found another issue, don;t know how to solve it. That is after several deltaT runtime, some of the cell faces do not match with each other for different processors. Compared with differential based motion solver, I think it is bcoz the RBF motion solver is based on solving the pointsField, not volVectorField.
So what would be the best solution to correct the face mismatch. Or should the code be rewritten based on solving the volVectorField?
help of RBF example
I am looking for RBF example in OF 1.6-ext. But so far I have not found the RBF example. I will be appreciated If you can give me a tutorial about RBF.
My mail is 'firstname.lastname@example.org'
Thank you in advance.
are you sure that you tried to search in openfoam-extend?
there are two versions of OpenFoam one by SGI/OpenCFD and one by Extend-community.
to get the "Extend-version"
git clone git://openfoam-extend.git.sourceforge.net/gitroot/openfoam-extend/OpenFOAM-1.6-ext
Has anyone found a solution for the third problem? I'm have the same issue.
|All times are GMT -4. The time now is 17:50.|