| Commit message (Collapse) | Author | Age |
... | |
| | |
|
| |
| |
| |
| | |
but with carefull handling of under/over-flow
|
| |
| |
| |
| | |
shift until the factorization succeed (with at most 10 attempts).
|
| |
| |
| |
| | |
Pritikin)
|
| | |
|
| | |
|
| | |
|
| |
| |
| |
| | |
evaluator_traits::AssumeAliasing to evaluator_assume_aliasing.
|
| | |
|
|/ |
|
|\
| |
| |
| | |
Add special functions to eigen: lgamma, erf, erfc.
|
| | |
|
| | |
|
| | |
|
| |
| |
| |
| | |
compliant with the C99 specification.
|
| | |
|
| | |
|
| |
| |
| |
| | |
Cleaned up the code a bit.
|
| |
| |
| |
| | |
Includes CUDA support and unit tests.
|
| |
| |
| |
| | |
the initialization of complex coeffs to 0 before we can check for alignedness
|
|/ |
|
|
|
|
| |
coefficient-wise inverse square root much more efficiently than calling sqrt().inverse().
|
| |
|
| |
|
|
|
|
| |
that solves A^T x = b or A^* x = b.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
uncompressed mode
|
| |
|
| |
|
|\
| |
| |
| | |
Add round, ceil and floor for SSE4.1/AVX (Bug #70)
|
| |
| |
| |
| | |
This also fixes a regression regarding (dense*sparse).diagonal()
|
| | |
|
|/ |
|
| |
|
| |
|
| |
|
|
|
|
| |
expensive when running on x87 fp unit
|
| |
|
| |
|
|
|
|
| |
gcc4.4 and older msvc with fast-math.
|
| |
|
| |
|