0
I Use This!
Inactive

Commits : Listings

Analyzed 16 days ago. based on code collected 17 days ago.
Jan 29, 2024 — Jan 29, 2025
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Start preconditioner for Stokes, still need to assemble matrices. More... over 15 years ago
Stokes Jacobian by blocks works More... over 15 years ago
Change meaning of InsertMode argument to dFSExpandedToGlobal. More... over 15 years ago
Matrix-free analytic Jacobians for Stokes. More... over 15 years ago
Use INSERT_VALUES to remove superfluous VecZeroEntries More... over 15 years ago
Correct Stokes exact solution, we do indeed converge now More... over 15 years ago
Fix StokesPointwiseFunction bug, work on checking errors More... over 15 years ago
Partial Stokes implementation, still need errors and Jacobians. More... over 15 years ago
Compress and uncompress symmetric tensors. Eventually we will offer dAPPLY_SYMGRAD. More... over 15 years ago
dFS depends on Mat so initialize so that MatOrderingList is non-null More... over 15 years ago
remove warning More... over 15 years ago
Added dMeshMorph and test of elliptic solver on deformed mesh. Results are nearly mesh independent. More... almost 16 years ago
Add elasticity test. This is a nonlinear vector problem and it is large enough that it takes a couple seconds to run. It uses the (now default) RCM ordering. More... almost 16 years ago
Compute low-bandwidth ordering. -dfs_ordering_type More... almost 16 years ago
Simplified elasticity exact solution. Add logging to ElastShellMatMult. -q1mat_type More... almost 16 years ago
Critical bug fix in dFSGetCoordinates() for bs>1. More... almost 16 years ago
Alternate SSE unrolling, looks to be marginally better, though very close to measurement error. More... almost 16 years ago
Reworking of TensorMult_Hex to provide unrolled implementations More... almost 16 years ago
Add ability to apply left- and right-scaling to the sparse Jacobian. The scaling was optimized in the 1D case to minimize the condition number of the preconditioned spectral operator. Unfortunately, it does not seem to be a clear win on real problems (it is usually worse). It can be activated by -djac_tensor_{mass,laplace}_scale TRUE if supported by the assembly function (see ellip.c for example). More... almost 16 years ago
More convenient logging of tests More... almost 16 years ago
Add elasticity example More... almost 16 years ago
Bug fixes for bs>1 More... almost 16 years ago
Clean up ellip exact solutions, only use sympy function values More... almost 16 years ago
Clean up testing More... almost 16 years ago
Inhomogeneous Dirichlet implemented for ellip. Closes gh-13. More... almost 16 years ago
Implement dFSGetCoordinates(), closes gh-14. More... almost 16 years ago
Start boundary condition unit test More... almost 16 years ago
Verify the correct memory More... almost 16 years ago
Write geometry associations so they can be inferred by iRel. More... almost 16 years ago
Use VecDohp to efficiently manage Dirichlet part of vectors. closes gh-4. More... almost 16 years ago