the code for weak galerkin

[festung]

the code for weak galerkin

[reuterbal]

Hi @yoczhang,

at the moment there is no adaptivity support in FESTUNG. We are thinking of adding this at some point but it requires some major changes to data structures and thus won't happen anytime soon.

Balthasar

[yoczhang]

Dear Balthasar
Thanks very much for your kind reply!

best regards.
yongchao

[yoczhang]

Dear @reuterbal ,
I found a relatively simple way to implement the adaptive mesh, hoping to provide some help. We can refer to the following this document, as well as the program code:
Efficient Implementation of Adaptive P1-FEM in Matlab,
http://www.asc.tuwien.ac.at/~praetorius/matlab/
I think using this method we need not to change the data structures, in every circle we just to update the coordV and V0T, and then we can do g=generateGridData(coordV, V0T).

best regards,
yongchao

[reuterbal]

Dear @yoczhang ,
thank you for that reference. Regenerating grid data structures is an obvious choice but not the core of the problem.

Most work would go into implementing recursive edge bisection or any other refinement algorithm to actually create coordV and V0T. Furthermore, you would first need some error indicator to decide which elements to refine, and afterwards a method to project your solution from the coarse grid to the fine grid.

Other choices might include hierarchical meshes, which are better suited for hyperbolic problems, where the area of interest travels through the domain.

You can probably implement an easy mesh adaptivity algorithm for your problem using the first method and without projecting the solution but recomputing from the beginning. This is a common choice and should still save you some computation time compared to starting out with a uniformly fine grid.

Best,
Balthasar

[yoczhang]

Dear @reuterbal ,
One limitation of this document is that it only applies to P1 situations and I will continue to improve the program. Thank you very much for your suggestion.

Best,
yongchao.

[reuterbal]

Closing this due to inactivity.