Dear all,
I am solving an eigenvalue problem similar than step-36. After solving for
the eigenpairs, I evaluate the eigenfunctions in the standard way:
VectorTools::point_value ( mapping, dof_handler, efun[m],
q_points[j], Uq );
where: efun[m] is the m-th eigenfunction from step-36,
q_points[j] are selected quadrature points,
Uq is where I store the FE evaluation.
As expected, this operation is awfully slow! it takes seconds for a single
point evaluation with a decent discretization and having m eigenfunctions
makes it worse!
1) I wonder if there is a way to evaluate the whole vector q_points with a
single (and clever) call instead of looping on j and calling
VectorTools::point_value(... , q_points[j], ...).
I remember when coding basic FEM in matlab, I had loops over several points
in order to reduce the overhead of evaluating the FE
for c in cells
for x in q_points
if x is in cell
evaluate FE: evaluate FE: loop over shape functions with
support in c,
U(x)=sum ...
end
end
end
2) My issue today goes beyond evaluating in several points ... I also
require to evaluate several FE vectors (eigenfunctions). Intuitively, one
would evaluate the FE like:
for c in cells
for x in q_points
if x is in cell
for m ... in eigenfunctions
evaluate FE: loop over shape functions with support in c,
Um(x)=sum ...
end
end
end
end
Any ideas how to achieve this?
Thanks in advance.
Juan Carlos Araújo
Umeå Universitet
--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see
https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
