General question about domain decomposition in parallel

Hi all,

I am using an STL file input to generate my MultiBlockLattice.

I was wondering how Palabos decomposes the domain across nodes in a parallel run, and if there is any function to determine the region (say a Box3D) that is contained by the PE/node.



the only part that is not parallel is the stl file which is shared accross all nodes. Beware that this can cause issues if you are working with too fine stl files. This is on our “todo” list for some time now…

The domain decomposition methods can be found in the mutliBlock directory mutliBlockManagement*, sparseBlockStructure*, … What are you exactly looking for?