Element evaluation and Galerkin projections

(from [a1358143a00a238c9e8d98c20591ac3c8c35e991]) Fixed memory leaks, Hex evaluation seems to work. [#1 Element evaluation and Galerkin projections status:open]

src/jacobi/tests/ex1.c: Hex evaluation tested, demonstrates spectral convergence for nontrivial synthetic input. Still need to implement for other topologies.

Signed-off-by: Jed Brown jed@59A2.org http://github.com/jedbrown/dohp/...

(from [7a47b22f503278c6b9f6a50eaa45ac0bf9ccd4f0]) Implemented dEFSApply for Line and Quad topologies. [#1 Element evaluation and Galerkin projections]

Spectral convergence is confirmed as for Hex. Anisotropic approximation order and quadrature seems to work correctly.

Still need to implement Discontinuous Galerkin projections.

Signed-off-by: Jed Brown jed@59A2.org http://github.com/jedbrown/dohp/...

(from [d022dba2e105ab10b58767087497466aa2c2f80b]) Solve projection problems with two elements. [#1 Element evaluation and Galerkin projections status:resolved]

Two main weaknesses:

• Uses -snes_mf for projection, we should also assemble a matrix and be able to apply the Jacobian matrix-free.

• Discretization error is not yet checked, but evaluation at quadrature points demonstrates spectral convergence so the projection should as well.

Signed-off-by: Jed Brown jed@59A2.org http://github.com/jedbrown/dohp/...