Closed AlexanderCicchino closed 7 months ago
Currently running ctest on Narval. Will post log file once it is done.
Fixed the euler split testcases from failing due to memory. output_3D_euler_split_TGV.log
Yes of course! Thanks for the catch! Fixed the alignment.
Thanks @PranshulThakur! To use strong DG with AD: for inviscid problems it will need the modifications from weak DG's assemble_volume_term_derivatives for the dRdW and dRdX. I don't think there needs to be any changed in assemble volume/face/boundary functions. For viscous problems, strong DG's DR/Du will need to incorporate the auxiliary variable q and its respective solution: DR/Du +DR/Dq*Dq/Du. Again, this can be computed outside of the assemble volume/face/boundary functions for the primary and auxiliary equations.
Thanks @PranshulThakur! To use strong DG with AD: for inviscid problems it will need the modifications from weak DG's assemble_volume_term_derivatives for the dRdW and dRdX. I don't think there needs to be any changed in assemble volume/face/boundary functions. For viscous problems, strong DG's DR/Du will need to incorporate the auxiliary variable q and its respective solution: DR/Du +DR/Dq*Dq/Du. Again, this can be computed outside of the assemble volume/face/boundary functions for the primary and auxiliary equations.
Makes sense. Thanks for the explanation.
Templated operators class with type real for input, output, and weight vectors. The basis is assumed to stay as double for the sum-factorization. This allows operators to be used with AD type.