During the initializing stage, there are errors when mapping the composition to the 3d grid. This causes some unwanted density perturbations to occur. Switching the QuadInterp function so that limit = false by default fixed the issue. Additionally, using linear interpolation by setting s0_interp_type = 2 were able to resolve the issue.
Small example to reproduce:
run make -j USE_MPI=FALSE in the urca directory. Then run using attached inputs
inputs.zip. This is a 1 level 256^3 resolution run, but have seen the issue persist up to 2048^3 resolution
Looking at plt_InitData/ and particularly rhopert should show the density perturbations. There is a lack of urca products (na23 and ne23) at these points. c12 and o16 make up for this lack and so you should be able to see that as well by looking at slices of X(c12) and X(o16)
biboyd
commented
9 months ago
During the initializing stage, there are errors when mapping the composition to the 3d grid. This causes some unwanted density perturbations to occur. Switching the
QuadInterpfunction so thatlimit = falseby default fixed the issue. Additionally, using linear interpolation by settings0_interp_type = 2were able to resolve the issue.Small example to reproduce:
run
make -j USE_MPI=FALSEin the urca directory. Then run using attached inputs inputs.zip. This is a 1 level 256^3 resolution run, but have seen the issue persist up to 2048^3 resolutionLooking at
plt_InitData/and particularlyrhopertshould show the density perturbations. There is a lack of urca products (na23 and ne23) at these points. c12 and o16 make up for this lack and so you should be able to see that as well by looking at slices ofX(c12)andX(o16)