Commit Graph

5120 Commits

Author SHA1 Message Date
Gael Guennebaud
7fbc9d8409 Introduce a ReplicateReturnType as a possible workaround of a compilation issue with MSVC+ICC 2014-12-10 14:26:25 +01:00
Gael Guennebaud
79c3cfabe3 Fix nomalloc_3 and binding reference to temporary issue 2014-12-09 19:01:25 +01:00
Gael Guennebaud
e0f390793c Fix dynamic allocation in JacobiSVD (regression)
(grafted from 30c849669d
)
2014-12-08 14:45:04 +01:00
Gael Guennebaud
97812ad0d3 UmfPack support: fix redundant evaluation/copies when calling compute() and support generic expressions as input 2014-12-02 17:30:57 +01:00
Gael Guennebaud
d66b5a1d91 Fix MSVC compilation issue
(grafted from a819fa148d
)
2014-12-02 14:35:31 +01:00
Gael Guennebaud
b0152fdb1d Fix bicgstab example 2014-12-02 14:32:55 +01:00
Gael Guennebaud
e9c5418249 bug #897: fix UmfPack usage with mapped sparse matrices
(grafted from 1a8dc85142
)
2014-12-02 13:57:13 +01:00
Gael Guennebaud
b25b517817 Fix bug #911: m_extractedDataAreDirty was not initialized in UmfPackLU
(grafted from 4974d1d2b4
)
2014-12-02 13:54:06 +01:00
Gael Guennebaud
ce0fb1bca1 Simplify return type of diagonal(Index) (and ease compiler job) 2014-11-28 14:39:47 +01:00
Christoph Hertzberg
92fce631ed added std:: scope to abs function call 2014-11-28 02:24:51 +00:00
Christoph Hertzberg
238308e0f7 bug #909: Removed unreachable return statement 2014-11-26 15:45:11 +01:00
Gael Guennebaud
719ac0d6b0 Fix Hyperplane::Through(a,b,c) when points are aligned or identical. We use the stratgey as in Quaternion::setFromTwoVectors.
(grafted from 8518ba0bbc
)
2014-11-26 15:01:53 +01:00
Gael Guennebaud
8e61a7aab6 Fix a case where 0-1 leads to Dynamic instead of 0. 2014-11-26 15:03:22 +01:00
Gael Guennebaud
09e992ce9f Add missing specialization of Block<const SparseMatrix> 2014-11-24 18:40:44 +01:00
Gael Guennebaud
cdd401f743 Enable Mx0 * 0xN matrix product. 2014-11-24 18:07:50 +01:00
Gael Guennebaud
59b7615d31 Fix memory pre-allocation when permuting inner vectors of a sparse matrix.
(grafted from da584912b6
)
2014-11-24 17:31:59 +01:00
Gael Guennebaud
a8cb0dfcf5 re-enable usage of ProductBase::m_result and workaround a compilation failure when m_result is too large but unused 2014-11-14 13:38:12 +01:00
Christoph Hertzberg
0e7a26c19f bug #898: add inline hint to const_cast_ptr 2014-10-28 14:51:05 +01:00
Christoph Hertzberg
13c636d864 Addendum to bug #859: pexp(NaN) for double did not return NaN, also, plog(NaN) did not return NaN.
psqrt(NaN) and psqrt(-1) shall return NaN if EIGEN_FAST_MATH==0
2014-10-20 13:35:03 +02:00
Gael Guennebaud
00ec1629ca Fix bug #859: pexp(NaN) returned Inf instead of NaN 2014-10-20 11:38:51 +02:00
Gael Guennebaud
a72eabec9b Fix bug #894: the sign of LDLT was not re-initialized at each call of compute()
(grafted from d04f23260d
)
2014-10-20 10:48:40 +02:00
Gael Guennebaud
235c97ba92 Fix SparseQR::rank for a completely empty matrix.
(grafted from 8838b0a1ff
)
2014-10-19 22:42:20 +02:00
Gael Guennebaud
4126cb6369 Fix SparseLU::absDeterminant and add respective unit test
(grafted from a370b1f2e2
)
2014-10-17 16:52:56 +02:00
Gael Guennebaud
8ea2ab4829 Fix JacobiSVD wrt undeR/overflow by doing scaling prior to QR preconditioning
(grafted from feacfa5f83
)
2014-10-17 15:32:06 +02:00
Christoph Hertzberg
9b79607579 bug #891: Determine sizeof(void*) via CMAKE variable instead of test program
(transplanted from 0ec1fc9e11
)
2014-10-14 14:14:25 +02:00
Gael Guennebaud
aadbfe78c2 bug #890: extract_data might returns 0x0 thus breaking aliasing detection 2014-10-10 16:42:32 +02:00
Gael Guennebaud
7d5e16c733 Add missing default ctor in Rotation2D 2014-09-30 16:59:28 +02:00
Christoph Hertzberg
e395a8042a Fix bug #884: No malloc for zero-sized matrices or for Ref without temporaries
manually ported from 4ba8aa1482
2014-09-25 16:25:31 +02:00
Gael Guennebaud
91f1a161ca bug #879: tri1 = mat * tri2 was compiling and running incorrectly if tri2 was not numerically triangular. Workaround the issue by evaluating mat*tri2 into a temporary. 2014-09-22 17:20:42 +02:00
Gael Guennebaud
16bca3bfe2 Fix SparseQR for row-major inputs.
(grafted from 755e77266f
)
2014-09-19 09:58:56 +02:00
Gael Guennebaud
e0ab58d815 Fix bug #791: infinite loop in JacobiSVD in the presence of NaN.
(grafted from d6236d3b26
)
2014-09-10 11:54:20 +02:00
Gael Guennebaud
c67a7148c4 ArrayWrapper and MatrixWrapper classes should not be nested by reference.
(grafted from 921a645481
)
2014-09-10 10:33:19 +02:00
Gael Guennebaud
38dc683901 Fix bug #822: outer products needed linear access, and add respective unit tests
(grafted from 51b3f558bb
)
2014-09-08 10:21:22 +02:00
Jitse Niesen
cad0fa5d77 Replace asm by __asm__ (bug #873).
Thanks to Markus Eisenmann for report and initial patch.
2014-09-06 11:54:47 +01:00
Gael Guennebaud
5daebe0a27 bug #871: fix compilation on ARM/Neon regarding __has_builtin usage (backport) 2014-09-01 10:58:07 +02:00
Georg Drenkhahn
05fb735d1d Added missing STL include of <list> in main.h
Removed duplicated include of <sstream>
Added comments on the background of min/max macro definitions and STL header includes
(grafted from e49e84d979
)
2014-08-29 10:41:05 +02:00
Gael Guennebaud
7443d8b4e9 bug #867: forward the cmake generator when testing support for fortran. (was already fixed in the default branch) 2014-08-28 09:15:33 +02:00
Georg Drenkhahn
36506511a1 Fixed CMakeLists.txt files to prevent CMake 3.0.0 warnings about deprecated LOCATION target property.
Small whitespace cleanup in CMakelLists.txt.
2014-08-22 12:13:07 +02:00
Gael Guennebaud
3afdc6d95a In SparseQR, calling factorize() without analyzePattern() was broken. 2014-08-26 23:32:32 +02:00
Gael Guennebaud
c14c03490f merge 2014-08-26 13:00:11 +02:00
Gael Guennebaud
c880590d27 bug #861: enable posix_memalign with PGI
(grafted from 2e50289ba3
)
2014-08-26 12:54:19 +02:00
Gael Guennebaud
54294e2293 bug #857: workaround MSVC compilation issue. 2014-08-26 12:52:29 +02:00
Gael Guennebaud
c7331ebb06 Do not apply the preconditioner before starting the iterations as this might destroy a very good initial guess.
(grafted from b49ef99617
)
2014-08-21 22:14:25 +02:00
Gael Guennebaud
0321449944 bug #854: fix numerical issue in SelfAdjointEigenSolver::computeDirect for 3x3 matrices. The tolerance to detect stable cross products was too optimistic.
Add respective unit tests.
(grafted from 9c0aa81fbf
)
2014-08-21 10:49:09 +02:00
Gael Guennebaud
44c390a370 Added tag 3.2.2 for changeset bbaf01712c 2014-08-04 12:52:31 +02:00
Gael Guennebaud
bbaf01712c bump to 3.2.2 2014-08-04 12:51:54 +02:00
Gael Guennebaud
8e875d3c38 Memory allocated on the stack is freed at the function exit, so reduce iteration count to avoid stack overflow
(grafted from e51da9c3a8
)
2014-08-04 12:46:00 +02:00
Gael Guennebaud
8d69b87c53 Make the ordering method of SimplicialL[D]LT user configurable.
(grafted from d4cc1bdc7f
)
2014-07-20 14:22:58 +02:00
Christoph Hertzberg
49cbaf3856 Add note to EIGEN_DONT_PARALLELIZE into preprocessor documentation page (requested in IRC)
(transplanted from 68eafc10b1
)
2014-07-18 15:42:12 +02:00
Gael Guennebaud
9b00035438 bug #843: fix jacobisvd for complexes and extend respective unit test to chack with random tricky matrices,
and backport other JacobiSVD fixes
2014-07-17 17:09:15 +02:00