markcmiller86
commented
3 days ago Ok, I discovered I was not correctly handling the multimesh-adjacency object's back pointers properly after removing periodic domain neighbors. I have a working solution now but need to clean it up a bit. The logic is a tad complicated and I introduced a number of functions to help confirm the data structures were being kept consistent, both what is being read off disk from data file and what is modified in memory in plugin.
Describe the bug
The new logic to support a mesh completely surrounded by periodic boundaries (PBCs) has a bug in it which is reproducible with the original dataset doing a contour plot of
denvariable.