- changed status to open
- removed comment
tensorspecial for ML_BSSN::phi is not set
if I am not mistaken then the default parameters (conformalMethod == 0) make phi = ln(gamma)/12 a tensor density of type tensorspecial == "log" and weight 1/6. The weight is in the code, the special is missing. This of course breaks once one sets conformalMethod == 1 and actually re-interprets phi as W.
Fix in the Mathematica script would seem to add
SetTensorAttribute[phi, TensorSpecial, "log"];
and to define a new grid function W with the proper type for the coformal case. Only should affect multipatch runs.
Keyword:
Comments (4)
-
reporter -
- removed comment
This attribute is only relevant if the BSSN variables are stored in patch-local coordinates. This is typically not the case, so this change would have no impact.
Adding a new variable would increase memory requirements. To make this feasible, there should also be a way to tell Kranc to not allocate storage for phi/W if unused.
-
reporter - changed status to resolved
- removed comment
Indeed (Christian Reisswig also kindly pointed this out to me). I'll close the ticket.
-
reporter - changed status to closed
- edited description
- Log in to comment