You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Reported by hywel on 7 Oct 2011 16:58 UTC
Before the collision operator had to go through and set the values of f_neq on the !HydrodynamicVars object. Now the kernels do this, and the collisions can only read the vars. Pointer look-ups avoided by using a struct to hold the 15 f-values.
Regression test passes (modulo images; see ticket:4), unit tests pass.
Assigning to Rupert for review.
The text was updated successfully, but these errors were encountered:
Comment by rupert on 13 Oct 2011 13:00 UTC
This is a small but definite improvement as described above. I do have reservations about the larger design which I will log another ticket about for the next milestone.
One point: In Code/lb/kernels/LBGK.h, I don't see why the !HydroVars template is being specialized on LBGK to be the same as the default. If this is not an accidental code duplication, please explain why.
Comment by hywel on 13 Oct 2011 14:54 UTC
Fixed some stuff to do with related changes in ticket 8. Mainly applying the same changes to LBGKNN that were necessary for LBGK and Entropic kernels.
Good spot on the unnecessary LBGK !HydroVars. Now removed.
Reported by hywel on 7 Oct 2011 16:58 UTC
Before the collision operator had to go through and set the values of f_neq on the !HydrodynamicVars object. Now the kernels do this, and the collisions can only read the vars. Pointer look-ups avoided by using a struct to hold the 15 f-values.
Regression test passes (modulo images; see ticket:4), unit tests pass.
Assigning to Rupert for review.
The text was updated successfully, but these errors were encountered: