[Nek5000-users] Problem with intp_rstd

[nek5000-users at lists.mcs.anl.gov]

Dear Paul,


I am confused about something. I realize now that for the components of B_{ij} to be exact, quadrature for


\int_\Omega \phi_i \phi_j dx = (\phi_i, \phi_j ).


Since in Nek5000 we have (N+1)^dim GLL points for the velocity grid, does that mean every time an inner product is being evaluated in the code this is not exact?


If I want B_{ij} to be exact, do I need to evaluate the basis functions on more GLL quadrature points (say, N+2)?


I'm just wondering if I am understanding the problem correctly. For now, multiplying "rhs" by "binvm1" like you suggested produced accurate results for my dummy problem (vx**2), I may have to do exact integration in the long run.


Thank you,


Juan Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20170411/04c3da33/attachment.html>