So far, integration order has been erroneously adjusted when the Gauss-Lobatto basis is used with square element, as below:
if (intRuleType == 1 && vfes->GetFE(i)->GetGeomType() == Geometry::SQUARE)
integrationOrder--; // when Gauss-Lobatto
This somehow passed the consistency test with the time integration. However, it does not pass the convergence test with utils/compute_rhs. After this fix, the errors of manufactured solutions significantly decreased (about a factor of 4).
Beside this, another manufactured solution test mms_euler_2d is added.
So far, integration order has been erroneously adjusted when the Gauss-Lobatto basis is used with square element, as below:
This somehow passed the consistency test with the time integration. However, it does not pass the convergence test with
utils/compute_rhs
. After this fix, the errors of manufactured solutions significantly decreased (about a factor of 4).Beside this, another manufactured solution test
mms_euler_2d
is added.