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
They claim it significantly reduces the setup time compared to the existing AMG solver, since it assumes logically-Cartesian mesh connectivity (possibly with AMR) using the SStruct interface, rather than the IJ interface, while being as robust as AMG. I don't know if there's a GPU version yet.
A comparison paper between implicit FLD with this method and our method in Quokka would be potentially interesting.
The text was updated successfully, but these errors were encountered:
There is a new "semi-structured" AMG solver (SSAMG) in Hypre: https://arxiv.org/abs/2205.14273.
They claim it significantly reduces the setup time compared to the existing AMG solver, since it assumes logically-Cartesian mesh connectivity (possibly with AMR) using the SStruct interface, rather than the IJ interface, while being as robust as AMG. I don't know if there's a GPU version yet.
A comparison paper between implicit FLD with this method and our method in Quokka would be potentially interesting.
The text was updated successfully, but these errors were encountered: