| Commit message (Collapse) | Author | Age |
| |
|
| |
|
| |
|
|
|
|
| |
matrices, from MatrixMarket for instance
|
| |
|
| |
|
|
|
|
| |
(no need to use .twistedBy(I) anymore)
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Provide a default constructed step size as opposed to an int when the size is 1.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
ReverseInnerIterator should probably be removed anyway since we currently don't have real use cases for them. The only one in TriangularSolver could be advantageously replaced by a binary search.
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
threshold (ILUT)
Modify the BiCGSTAB function to check the residual norm of the initial guess
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
alignment status.
There probably many other places in Transform.h where such optimizations could be done.
|
| |
|
| |
|
| |
|