Closed thehalfspace closed 5 years ago
You may find a better explanation of the operations here and (for irregular grids) in section 3.6 of my lecture notes.
Thanks. The lecture notes were super helpful. I am able to assemble the stiffness as sparse matrix now.
I am glad it helped. Please consider contributing your code to semlab, if you think others can benefit from it. If you make it available somewhere else, let me know.
I'll definitely contribute to SEMLAB once I am done testing my stuff. The multiplication (Ksparse*d) is much faster than assembling the local contributions using loop.
I have done the sparse matrix assembly here. I'll try to rewrite it in matlab and submit a pr for SEMLAB.
I am working on the earthquake cycle code which I will make available on Github once it is fully functional. Here is the prototype for now.
Great work in Julia. The most efficient versions we have achieved in SEMLAB so far are: sem2d_NewmarkA1_scec2_vectorized.m sem2d_NewmarkA1_scec2_vectorized_parfor.m We have used those optimizations to model cycles too. It will be interesting to see if a sparse matrix implementation performs better in matlab. Performance comparisons may depend on the language.
I am trying to assemble stiffness matrix as a sparse matrix in two dimensions, similar to what is shown in one dimension.
I know the current problem calculates K.u at a local level for efficiency as shown here, but in quasi-static problem the matrix solver is a very slow step.
I am trying to do something like this:
This doesn't give me correct answer. I am confused due to the symmetry of shape functions in x and y.