mirror/eigen
2
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-01-06 14:14:46 +08:00
Code Issues Packages Projects Releases Wiki Activity
1,466 Commits 11 Branches 89 Tags 168 MiB
11c3762068
Commit Graph

1002 Commits

Author SHA1 Message Date
Gael Guennebaud
ec5c608aa3 Set of fixes and workaround to make sun studio more happy. 
Still remains the problem of alignment and vectorization.
 2009-07-10 16:10:03 +02:00
Gael Guennebaud
b47dea8b7a add a meta unroller for the triangular solver (only for vectors as rhs) 2009-07-10 11:30:46 +02:00
Gael Guennebaud
1a1b2e9f27 finally directly calling the low-level products is faster 2009-07-10 10:41:26 +02:00
Gael Guennebaud
1c52985aa7 merge 2009-07-10 08:21:57 +02:00
Gael Guennebaud
629e083d81 slight change in the comparison to -1 2009-07-10 08:21:20 +02:00
Gael Guennebaud
8885d56928 commit woking versions of triangular solvers naturally 
handling conjuagted expression. still have to bench whether it
is faster (runtime and compile time) to directly call the
cache friendly functions, whence all the commented piece of code...
 2009-07-09 23:59:18 +02:00
Gael Guennebaud
fa60c72398 started to simplify the triangular solvers 2009-07-09 17:11:03 +02:00
Gael Guennebaud
96e7d9f896 ok now all the complex mat-mat and mat-vec products involving conjugate, 
adjoint, -, and scalar multiple seems to be well handled. It only remains
the simpler case: C = alpha*(A*B) ... for the next commit
 2009-07-08 18:24:37 +02:00
Gael Guennebaud
13b2dafb50 conjugate expressions are now properly caught by Product 
=> significant speedup in expr. like a.adjoint() * b,
   for complex scalar type (~ x3)
 2009-07-07 21:30:20 +02:00
Gael Guennebaud
5ed6ce90d3 started to catch scalar multiple and conjugate xpr in Product 2009-07-07 16:55:51 +02:00
Gael Guennebaud
ea23f36c78 * change the nesting order of adjoint_return_type to 
1 - make it easier to catch conjugate expressions
 2 - make sure there is no unecessary copy (we had NestByValue<Derived> which seems to be very bad)
* update eigensolver wrt recent changes
 2009-07-07 15:56:13 +02:00
Gael Guennebaud
79877a9917 * take advantage of new possibilies in LLT (mat -= product) 
* fix Block::operator+= product which was not optimized
* fix some compilation issues
 2009-07-07 15:32:21 +02:00
Gael Guennebaud
92a35c93b2 * extended the cache friendly products to support C = alpha * A * M and C += alpha * A * B 
* this allows to optimize xpr like C -= lazy_product, still have to catch "scalar_product_of_lazy_product"
* started to support conjugate in cache friendly products (very useful to evaluate A * B.adjoint() without
  evaluating B.adjoint() into a temporary
* compilation fix
 2009-07-07 11:39:19 +02:00
Gael Guennebaud
544888e342 add a generic mechanism to copy a special matrix to a dense matrix so that 
we don't need to add other specialization of MatrixBase::operator=, Matrix::=,
and Matrix::Matrix(...)
 2009-07-07 09:05:20 +02:00
Gael Guennebaud
1aea45335f * bybye Part, welcome TriangularView and SelfAdjointView. 
* move solveTriangular*() to TriangularView::solve*()
* move .llt() to SelfAdjointView
* add a high level wrapper to the efficient selfadjoint * vector product
* improve LLT so that we can specify which triangular part is meaningless
=> there are still many things to do (doc, cleaning, improve the matrix products, etc.)
 2009-07-06 23:43:20 +02:00
Benoit Jacob
889726bf7c add matrixQR() method exposing the storage. that's where the householder thing impacts the API. 2009-07-06 17:24:11 +02:00
Benoit Jacob
e057beee4e fix some search-and-replace damage 2009-07-06 17:20:07 +02:00
Benoit Jacob
e093b43b2c * rename QR to HouseholderQR because really that impacts the API, not just the impl. 
* rename qr() to householderQr(), for same reason.
* clarify that it's non-pivoting, non-rank-revealing, so remove all the rank API, make solve() be void instead of bool, update the docs/test, etc.
* fix warning in SVD
 2009-07-06 17:12:10 +02:00
Gael Guennebaud
f84bd3e7b1 include the fixes of the third edition 2009-07-06 15:01:30 +02:00
Gael Guennebaud
0c2232e5d9 quick reimplementation of SVD from the numeral recipes book: 
this is still not Eigen style code but at least it works for
n>m and it is more accurate than the JAMA based version. (I needed
it now, this is why I did that)
 2009-07-06 13:47:41 +02:00
Gael Guennebaud
0cd158820c switch from eigensolver to SVD which seems to be more accurate with float 2009-07-06 11:15:38 +02:00
Gael Guennebaud
90f1e24579 significantly improve the accuracy of setFromTwoVectors (fixes #21) 2009-07-06 10:35:20 +02:00
Gael Guennebaud
ecc4f07af5 fix doc of Quaternion::setFromTwoVectors 2009-07-06 09:17:25 +02:00
Gael Guennebaud
c6f610093b add a VectorBlock expr as a specialization of Block 2009-07-05 11:33:55 +02:00
Gael Guennebaud
eec334c604 fixes a segfault 2009-07-05 10:48:57 +02:00
Gael Guennebaud
08e419dcb1 * update sparse module wrt new diagonal matrix impl 
* fix a bug is SparseMatrix
 2009-07-04 11:16:27 +02:00
Gael Guennebaud
d457ec5810 fix #20: SVD::solve() now resize the result 2009-07-04 09:28:11 +02:00
Benoit Jacob
7b750182f2 * polish computeInverseWithCheck to share more code, fix documentation, fix coding style 
* add snippet for computeInverseWithCheck documentation
* expand unit-tests to cover computeInverseWithCheck
 2009-06-29 22:07:37 +02:00
Manuel Yguel
126a031a39 computeInverseWithCheck method added to matrix base (specialization for 1D to 4D) 2009-06-29 20:47:37 +02:00
Benoit Jacob
632cb7a4a1 patch by Myguel from the forum: fix documentation 2009-06-29 19:26:20 +02:00
Benoit Jacob
2de9b7f537 fully vectorize DiagonalProduct 
(it used to be partially vectorized and that had been lost in the big changes from the previous commit)
 2009-06-29 04:01:31 +02:00
Benoit Jacob
bf91003d37 FreeBSD: determine precisely when malloc is 16-byte aligned 2009-06-29 00:08:34 +02:00
Benoit Jacob
5eabf2b75d double precision() : change to 1e-12 instead of 1e-11 (as discussed several times on the list) 2009-06-29 00:00:29 +02:00
Benoit Jacob
6809f7b1cd new implementation of diagonal matrices and diagonal matrix expressions 2009-06-28 21:27:37 +02:00
Benoit Jacob
fc9000f23e only disable the inline ASM if we're NEITHER gcc nor icc. right ?? 2009-06-26 05:32:21 +02:00
Benoit Jacob
6ccb97620a patch by Patrick Mihelich: use empty struct + anonymous namespace for NoChange 2009-06-25 03:33:47 +02:00
Benoit Jacob
903acf0d5c add missing code snippets for newer Matrix methods and PartialLU::solve() 2009-06-25 00:57:51 +02:00
Benoit Jacob
03ad303d14 * add resize(int, NoChange) and resize(NoChange, int) 
* add missing assert in resize(int)
* add examples for all resize variants
* expand docs (part of which is from Tim Hutt's e-mail)
 2009-06-24 22:07:03 +02:00
Benoit Jacob
96dca681b0 use <...> for system headers 2009-06-24 16:35:02 +02:00
Gael Guennebaud
a44f7cf440 re-enable the fast unaligned loads for gcc and icc using inline assembly 
(this allows to avoid incompatible pointer casts and to specify the dependency to the data explicitely)
 2009-06-24 10:48:36 +02:00
Gael Guennebaud
aa17b5b514 use the slower unaligned load intrinsics in ei_ploadu because GCC mess up with my tricks 2009-06-23 23:28:34 +02:00
Benoit Jacob
7d48ed4be3 refine the check to disable alignment. now it's disabled on gcc3 (where we don't vectorize anyway) 2009-06-21 04:56:01 +02:00
Benoit Jacob
8be088bfb0 add Eigen/Eigen 2009-06-19 20:46:55 +02:00
Benoit Jacob
fe8ab0147b add "Dense" header 2009-06-19 19:09:35 +02:00
Benoit Jacob
032594cee2 forward port fix to #12 2009-06-19 18:51:15 +02:00
Benoit Jacob
a57325e971 fix #14: make llt::solve() and also ldlt::solve() work with uninitialized result 2009-06-19 17:01:32 +02:00
Moritz Lenz
c6e81869d0 fixed typo in SuperLUSupport.h 2009-06-17 11:55:57 +02:00
Gael Guennebaud
627595ad19 * rename PartialRedux to VectorwiseOp 
* add VectorwiseOp's +, -, +=, -= operators
 2009-06-10 11:20:30 +02:00
Gael Guennebaud
f3fd7fd22b fix #11: now the default Transform ctor set the last row in Affine mode. 2009-06-10 09:35:04 +02:00
Gael Guennebaud
d97d307fcf SparseMatrix::resize() always resets the matrix to an empty one 2009-06-08 14:12:11 +02:00
Gael Guennebaud
55de162cf6 fix #10: the reallocateSparse function was half coded 2009-06-08 14:05:23 +02:00
Hauke Heibel
f26c691678 Renamed internal helper functions from the Memory header. 2009-06-04 17:25:15 +02:00
Hauke Heibel
5f04f8eb6b Fixes #9. Thanks to the (unknown) bug contributor. 2009-06-04 09:11:35 +02:00
Hauke Heibel
6530c8f5b4 A much simplified version of the earlier commit introducing way fewer changes compared to changeset f292d2352e 
.
The reason of the previous commit was incorrect. The smart pointers issues were actually a result of issue 9.
 2009-06-03 22:22:15 +02:00
Hauke Heibel
71e5cbcbc4 Added specializations for DontAlign when using Dynamic matrices. 
This allows users to store Matrices in smart pointers without the
need for a specialized allocator/de-allocator.
 2009-06-03 16:47:38 +02:00
Hauke Heibel
f292d2352e Relaxed checks againts _MaxRows and _MaxCols in Matrix::_check_template_params(). 2009-05-29 09:09:48 +02:00
Benoit Jacob
ee92009fd8 make Umeyama, and its unit-test, work for me on gcc 4.3 2009-05-27 23:10:24 +02:00
Benoit Jacob
86be59124d fix the static assert checking the size template parameters. 2009-05-27 23:07:29 +02:00
Hauke Heibel
4d1e492c00 * Umeyama has now similar performance for RowMajor and ColMajor layouts. 
* Fixed a bug in umeyama for fixed size matrices.
* Fixed the umeyama unit test for fixed size matrices.
* Added XprHelper::ei_plain_matrix_type_row_major.
 2009-05-27 19:24:05 +02:00
Hauke Heibel
db5647abae Added Umeyama implementation. 2009-05-26 19:22:25 +02:00
Gael Guennebaud
9d5728c511 fix #4 
and also improve performance of Tridiag::diag/subDiag at the same time
 2009-05-23 13:31:20 +00:00
Benoit Jacob
42848498aa fixes #5 : freebsd really has aligned malloc 2009-05-22 23:54:52 +02:00
Benoit Jacob
7667a93cbe merge 2009-05-22 20:31:26 +02:00
Benoit Jacob
6347b1db5b remove sentence "Eigen itself is part of the KDE project." 
it never made very precise sense. but now does it still make any?
 2009-05-22 20:25:33 +02:00
Thomas Capricelli
0ed7c2f6d7 fix typo 2009-05-22 19:46:29 +02:00
Hauke Heibel
c7303a876f Oops, here the actual LLT and LDLT patch. 2009-05-22 15:58:20 +02:00
Hauke Heibel
0523b64fe9 Eigensolver decomposition interface unification. 
Added default ctor and public compute method as
well as safe-guards against uninitialized usage.
Added unit tests for the safe-guards.
 2009-05-22 14:27:58 +02:00
Hauke Heibel
2c247fc8a8 LU and PartialLU decomposition interface unification. 
Added default ctor and public compute method as well
as safe-guards against uninitialized usage.
Added unit tests for the safe-guards.
 2009-05-22 14:27:58 +02:00
Hauke Heibel
5c5789cf0f QR and SVD decomposition interface unification. 
Added default ctor and public compute method as
well as safe-guards against uninitialized usage.
Added unit tests for the safe-guards.
 2009-05-22 14:27:58 +02:00
Benoit Jacob
c7baddb132 add internal comment (mostly a pretext to test the eigen-commits list) 2009-05-20 16:54:40 +02:00
Gael Guennebaud
dd45c4805c * add a writable generic coeff wise expression (CwiseUnaryView) 
* add writable .real() and .imag() functions
 2009-05-20 15:41:23 +02:00
Benoit Jacob
a697cb4094 fix comments (old comments that were copied from LU) 2009-05-19 21:54:29 +02:00
Gael Guennebaud
510029f2bc * optimize sum() for sparse matrices and vectors 
* fix the row()/col() functions of some InnerVector
 2009-05-19 09:40:00 +02:00
Gael Guennebaud
b83d9b48fa fix compilation with ICC 2009-05-18 18:26:45 +02:00
Gael Guennebaud
c8629e12f4 fix #3, remove inline keywords in QtAlignedMalloc (MSVC fix) 2009-05-18 18:09:21 +02:00
Gael Guennebaud
e186728867 fix #1 : need to nest by value the affine part in homogeneous product 2009-05-18 17:55:50 +02:00
Gael Guennebaud
e0832d5d93 fix bug reported by Moritz Lenz about random setter 2009-05-18 17:26:01 +02:00
Benoit Jacob
934d6b4749 fix #2, bug in Diagonal::MaxRowsAtCompileTime when Index==Dynamic 2009-05-17 16:00:56 +02:00
Benoit Jacob
dd390470e1 simplification (no reason anymore to write that in that convoluted way) 2009-05-15 16:05:45 +00:00
Benoit Jacob
5ee9f1a705 argh, forgot to re-add the throw() 2009-05-15 15:54:52 +00:00
Benoit Jacob
126284d08b * fix bugs in EigenTesting.cmake: it didn't work with -DEIGEN_NO_ASSERTION_CHECKING=ON 
* only try...catch if exceptions are enabled
 2009-05-15 15:53:26 +00:00
Benoit Jacob
7c14e1eac4 add partial-pivoting LU decomposition 
the name 'PartialLU' is not meant to be definitive!
make inverse() and determinant() use it, so it's *almost* considered
well tested.
 2009-05-13 02:02:22 +00:00
Gael Guennebaud
f5b5571a5a compilation fixes 2009-05-12 07:32:34 +00:00
Gael Guennebaud
6a4e94f349 bugfix from Jens Mueller (s/RowMajor/IsRowMajor) 2009-05-11 10:59:27 +00:00
Benoit Jacob
9afd1324fd constant Diagonal ---> DiagonalBits 
introduce ei_is_diagonal to check for it
DiagonalCoeffs ---> Diagonal and allow Index to by Dynamic
-> add MatrixBase::diagonal(int) with unittest and doc
 2009-05-10 16:24:39 +00:00
Benoit Jacob
c500415e18 result of our experiments with LU tuning: implement very simple formula, that 
turns out to be similar to Higham's formula already in use in LDLt
 2009-05-07 20:35:26 +00:00
Gael Guennebaud
b6c76c30cd apply patch from Hauke Heibel cleaning overloaded operator new/detete 2009-05-07 20:33:48 +00:00
Benoit Jacob
47fc3858b9 oops, didn't want to commit that 2009-05-07 18:40:06 +00:00
Benoit Jacob
c8a22dbc08 CREDIT Hauke Heibel, more std::vector::insert fixes 2009-05-07 18:38:07 +00:00
Gael Guennebaud
6dffdca123 fix realloc when initial size was 0 (bug reported by Jens Mueller) 2009-05-07 13:13:42 +00:00
Benoit Jacob
4f0af00e51 *add missing overloads of setZero, etc... that were mentioned in the tutorial 
--->they go into Matrix as they resize.
*add isConstant() alias to isApproxToConstant()
*extend unit-test
*change an assert into a static assert
 2009-05-06 21:40:24 +00:00
Benoit Jacob
0c0d38272e add copyright on two public headers that are not so trivial... 2009-05-06 15:48:28 +00:00
Benoit Jacob
e1abffa513 add missing function, thanks to Hauke Heibel 2009-05-06 13:33:35 +00:00
Gael Guennebaud
3361ad4589 one more gcc 3.3 fix 2009-05-06 10:51:46 +00:00
Gael Guennebaud
db079f3a8c fix internal error with gcc 3.3 (all tests are now ok with 3.3 !) 2009-05-06 09:37:30 +00:00
Gael Guennebaud
1e286464ab * compilation fixes for gcc 3.3 
* test Part::swap
 2009-05-06 08:43:38 +00:00
Benoit Jacob
66f059b99d remove unused member remean that was just initialized 2009-05-05 17:29:44 +00:00
Benoit Jacob
46e1162a39 reimplement linearRegression on top of fitHyperplane which is much better 2009-05-05 17:15:35 +00:00
Benoit Jacob
2b2f0c0220 fix linearRegression, fix doc, add unit test (it was untested since the change 
making fitHyperplane no longer use it)
 2009-05-05 16:50:58 +00:00
Gael Guennebaud
5b364a68cb oops 2009-05-04 14:53:21 +00:00
Gael Guennebaud
2829314284 new simplified API to fill sparse matrices (the old functions are 
deprecated). Basically there are now only 2 functions to set a
coefficient:
1) mat.coeffRef(row,col) = value;
2) mat.insert(row,col) = value;
coeffRef has no limitation, insert assumes the coeff has not already
been set, and raises an assert otherwise.
In addition I added a much lower level, but more efficient filling
mechanism for
internal use only.
 2009-05-04 14:25:12 +00:00
Thomas Capricelli
ddb6e96d48 fix warnings with recent gcc 2009-05-04 13:55:21 +00:00
Benoit Jacob
dec09a618d fix warning, unused variable dummy 2009-05-04 13:01:23 +00:00
Benoit Jacob
b60571a193 fix warnings with unused static functions 2009-05-04 12:49:56 +00:00
Benoit Jacob
8aa8854bbf fix SSE2 detection on win64, reported by 'kajala' 2009-05-04 12:13:37 +00:00
Benoit Jacob
95bda5e6ab let the user disable alignment altogether by #defining EIGEN_DONT_ALIGN. 
Until now, the user had to edit the source code to do that.
Internally, add EIGEN_ALIGN that takes into account both EIGEN_DONT_ALIGN.and
EIGEN_ARCH_WANTS_ALIGNMENT. From now on, only EIGEN_ALIGN should be used to
test whether we want to align.
 2009-05-03 13:50:56 +00:00
Gael Guennebaud
facee57b8d add auto transposition for vectors 2009-04-30 12:09:45 +00:00
Gael Guennebaud
7029ed6b88 ok, this time cast should really work ; sorry for the noise 2009-04-29 17:39:39 +00:00
Gael Guennebaud
428a12902a fixed my mistake in cast 2009-04-29 17:05:19 +00:00
Gael Guennebaud
9e9e99a42e casting to the same type no longer generates a CwiseUnaryOp 2009-04-29 14:57:18 +00:00
Gael Guennebaud
e722f55382 fix the Matrix(int,int)/vector 2D ctors issue so that we really 
have a Matrix(Scalar,Scalar) ctor. (useful for std::complex, user
defined types, etc.
 2009-04-24 14:38:40 +00:00
Benoit Jacob
32e9801890 CREDIT Ross Smith: fix posix_memalign detection 2009-04-24 13:26:36 +00:00
Gael Guennebaud
acb32c69d4 * update BVH to explicitely use aligned_allocator 
* fix warning in StdVector
 2009-04-23 11:33:36 +00:00
Gael Guennebaud
c7bb7436f9 make the ei_p* math functions overloads instead of template 
specializations
 2009-04-22 21:35:50 +00:00
Gael Guennebaud
6f6b5ad016 add a generic version of std::vector::resize for other stl 
implementations
 2009-04-21 21:40:33 +00:00
Gael Guennebaud
2697877fac std::vector now explicitely requires the use of Eigen::aligned_allocator 
CREDIT Hauke Heibel
 2009-04-21 21:30:38 +00:00
Gael Guennebaud
c5c384cf06 add missing vector ctor reported by Markus Moll on the ML 2009-04-21 11:52:25 +00:00
Gael Guennebaud
5e5bac52d7 this should fix the linking issue with StdVector without any user 
changes... I cross my fingers...
 2009-04-21 11:44:58 +00:00
Gael Guennebaud
4efcc14b91 fix ctor issues with ei_workaround_msvc_std_vector 2009-04-21 08:12:07 +00:00
Benoit Jacob
6ee4eb94fb fix the 4x4 inverse -- unit test passes again 2009-04-14 13:42:23 +00:00
Gael Guennebaud
804a239d30 patch from Moritz Lenz to allow solving transposed problem with superlu 2009-04-10 19:54:43 +00:00
Gael Guennebaud
fb3078fb62 fix memory leak in superlu backend 2009-04-10 18:49:38 +00:00
Gael Guennebaud
7254201632 bugfix when the diagonal is not stored and assumed to be 1 2009-04-10 18:13:37 +00:00
Gael Guennebaud
d78eb02627 add aligned_allocator operator == and != as suggested by Hauke Heibel 2009-04-09 21:22:02 +00:00
Benoit Jacob
0c99de5a17 more patches from Hauke Heibel: compilation/warning fixes from VC++ 2009-04-09 17:19:17 +00:00
Benoit Jacob
e6332cba4b forward-port r951449: patch by Hauke Heibel: compile fix with VS 9 2009-04-09 12:06:13 +00:00
Gael Guennebaud
e8329f9f45 relicence Julien Pommier's SSE code to Eigen's licenses 2009-04-09 06:03:51 +00:00
Gael Guennebaud
aabca5a44f stupid typo 2009-04-06 13:35:48 +00:00
Benoit Jacob
502bf4a81d * fix the binary bloat issue, Rohit's idea was the good one 
* a few dox fixes (alloc routines do return 0 on error) and forgot to update version number in CMakeLists
 2009-04-06 13:33:42 +00:00
Gael Guennebaud
38f501a596 fix computation of aligned_bit (has been broken by the change of 
AutoAlign/DontAlign)
 2009-04-05 17:34:59 +00:00
Gael Guennebaud
5b1d0cebc5 sparse module: new API proposal for triangular solves and experimental 
solver support with a sparse matrix as the rhs.
 2009-04-05 16:30:10 +00:00
Kenneth Frank Riddile
9a8096dd1d * fixed truncation warnings caused by MatrixBase::CoeffReturnType under MSVC 2009-04-03 20:19:07 +00:00
Gael Guennebaud
ff3a3209ca add an assertion in sparse LLT for invalid input matrix 2009-04-03 08:30:15 +00:00
Gael Guennebaud
adf5104bae oops, bad copy paste in ei_psqrt, thanks to Jitse Niesen 2009-04-02 06:29:05 +00:00
Benoit Jacob
40432c84b9 CwiseUnaryOp -> SparseCwiseUnaryOp 2009-04-01 14:47:56 +00:00
Benoit Jacob
0f8e692b3f * Find SuperLU also when it is installed without a superlu/ prefix 
* Some more CoeffReturnType changes
 2009-04-01 14:07:38 +00:00
Benoit Jacob
113fc3a260 now if Derived has the DirectAccess flag, then MatrixBase<Derived>::coeff() const returns a const Scalar& and not a Scalar as before. 
useful for people doing direct access. + 1 bugfix thanks to Patrick Mihelich, forgot a & in MapBase::coeff(int).
 2009-04-01 00:39:56 +00:00
Benoit Jacob
2f45eeb0c6 More Cholesky fixes. 
* Cholesky decs are NOT rank revealing so remove all the rank/isPositiveDefinite etc stuff.
* fix bug in LLT: s/return/continue/
* introduce machine_epsilon constants, they are actually needed for Higman's formula determining
  the cutoff in Cholesky. Btw fix the page reference to his book (chat with Keir).
* solve methods always return true, since this isn't a rank revealing dec. Actually... they already did always return true!! Now it's explicit.
* updated dox and unit-test
 2009-04-01 00:21:16 +00:00
Benoit Jacob
1e6097a810 fix mistake in static assertion, patch by Markus Moll. 2009-03-31 16:07:12 +00:00
Benoit Jacob
bf596d0b3a add adjointInPlace() and add documentation warnings on adjoint() and transpose() about aliasing effects. 2009-03-31 13:55:40 +00:00
Benoit Jacob
a1ba995f05 Many improvements in LLT and LDLT: 
* in LDLT, support the negative semidefinite case
* fix bad floating-point comparisons, improves greatly the accuracy of methods like
  isPositiveDefinite() and rank()
* simplifications
* identify (but not resolve) bug: claim that only triangular part is used, is inaccurate
* expanded unit-tests
 2009-03-30 21:45:45 +00:00
Gael Guennebaud
df9dfa1455 fix superLU backend: missing operator= 2009-03-27 19:25:22 +00:00
Gael Guennebaud
e9f6167485 make special ei_p functions static to avoid linking issues (they are too 
complex to be inlined)
 2009-03-27 14:55:46 +00:00
Gael Guennebaud
49fc1e3e84 add vectorization of sqrt for float 2009-03-27 14:41:46 +00:00
Gael Guennebaud
3499f6eccd fix Taucs support (it appears Taucs does not return sorted matrices) 2009-03-26 17:11:43 +00:00
Benoit Jacob
1b7b538e05 The ABI break: 
* set AutoAlign=0, DontAlign!=0
* set Dynamic=33331
* add check on fixed sizes
* bump version to 2.0.52
 2009-03-26 16:30:54 +00:00
Gael Guennebaud
ce5669dbf9 * enable vectorization of sin, cos, etc. by default with an option to 
disable them (-DEIGEN_FAST_MATH=0)
* add a specialization of MatrixBase::operator*(RealScalar) for fast
  "matrix of complex" times scalar products (even more useful for
  autodiff scalar types)
 2009-03-26 12:50:24 +00:00
Gael Guennebaud
62de40f8bb oops forgot to include a file in previous commit (I had other local 
changes I did not want to commit yet...)
 2009-03-26 07:10:59 +00:00
Gael Guennebaud
a22ef7e1f3 for some reason passing the argument by const reference killed the perf 
(in the packet version of sin, cos, exp, lop), so let's pass them by
value. Also, improve the perf of ei_plog by reducing dependencies.
 2009-03-25 18:33:36 +00:00
Gael Guennebaud
17860e578c add SSE2 versions of sin, cos, log, exp using code from Julien 
Pommier. They are for float only, and they return exactly the same
result as the standard versions in about 90% of the cases. Otherwise the max error
is below 1e-7. However, for very large values (>1e3) the accuracy of sin and cos
slighlty decrease. They are about 3 or 4 times faster than 4 calls to their respective
standard versions. So, is it ok to enable them by default in their respective functors ?
 2009-03-25 12:26:13 +00:00
Gael Guennebaud
70c0174bf9 * allows fixed size matrix with size==0 (via a specialization of 
MatrixStorage returning a null pointer). For instance this is very
  useful to make Tridiagonalization compile for 1x1 matrices
* fix LLT and eigensolver for 1x1 matrix
 2009-03-23 14:44:44 +00:00
Konstantinos A. Margaritis
fe00e864a1 ei_pnegate implemented for AltiVec 2009-03-20 17:26:50 +00:00
Gael Guennebaud
fbf415c547 add vectorization of unary operator-() (the AltiVec version is probably 
broken)
 2009-03-20 10:03:24 +00:00
Gael Guennebaud
3d385ae968 angle-axis doc: make it more clear the axis must be normalized 2009-03-18 19:38:55 +00:00
Gael Guennebaud
dcf49e5a28 more MSVC fixes: restrict keywords (sorry for all these commits) 2009-03-17 13:32:26 +00:00
Gael Guennebaud
718af05517 more MSVC fixes (asm comments...) 2009-03-17 13:25:26 +00:00
Gael Guennebaud
497d77b08c fix compilation MSVC 2009-03-17 13:18:53 +00:00
Gael Guennebaud
c6264d9b7e add affine * homogeneous vector for backward compatibility (e.g., 
kgllib)
 2009-03-11 14:32:03 +00:00
Gael Guennebaud
02e97ac0ae minor changes in AlignedBox needed for BVH module 2009-03-11 14:21:55 +00:00
Gael Guennebaud
b8f46090ff add optimized cross3 function (code from Rohit Garg) 2009-03-11 14:20:36 +00:00
Gael Guennebaud
f697ea6d30 fix a few compilation errors and warnings (ICC) 2009-03-11 08:45:53 +00:00
Gael Guennebaud
14691d6836 fix compilation with old, and future gcc 2009-03-10 11:55:50 +00:00
Gael Guennebaud
3e4307d8a8 compilation fix in MapBase 2009-03-10 09:06:38 +00:00
Gael Guennebaud
8aa5aa269a add "slice vectorization" of redux (eg. m.block().minCoeff() is now 
vectorized)
 2009-03-09 23:16:39 +00:00
Gael Guennebaud
c087373968 because of a missing specialization, operator/(scalar) was not vectorized 2009-03-09 23:14:53 +00:00
Gael Guennebaud
db6c3d0197 fix MapBase's ForceAligned concept which was not working at all.... 2009-03-09 19:23:31 +00:00
Gael Guennebaud
3f80c68be5 add the vectorization of abs 2009-03-09 18:40:09 +00:00
Gael Guennebaud
314aff875e fix warning in Tridiag 2009-03-09 10:41:37 +00:00
Gael Guennebaud
44f218988c add a small note in Transform doc. 2009-03-08 11:38:45 +00:00
Gael Guennebaud
3ac42fed94 big rework of the Transform class: 
* add Projective and AffineCompact modes as an optional third template
  argument
* extend Transform::operator* to support more use cases
 2009-03-08 11:35:30 +00:00
Gael Guennebaud
7718a8ed83 slight optimization of SSE base integer mul (thanks to Rohit Garg) 2009-03-08 10:14:07 +00:00
Jure Repinc
f9790a649c Install the newly added eigen2/Eigen/src/Geometry/arch/Geometry_SSE.h 2009-03-08 06:21:21 +00:00
Gael Guennebaud
e4f64ce098 add optimized quaternion * quaternion product specialization for 
float/SSE using code from Rohit Garg
 2009-03-07 13:52:44 +00:00
Gael Guennebaud
6f95270ede significantly reduce the default stack allocation limit which was much 
too high
 2009-03-06 16:52:43 +00:00
Gael Guennebaud
2c78ca62a6 patch from Stjepan Rajko (MSVC fix in RotationBase) 2009-03-06 11:45:39 +00:00
Andrew Coles
093e3cd5b5 Commented out duplicate definition of TransformTraits - was causing 
compile-time errors.
 2009-03-05 22:35:06 +00:00
Gael Guennebaud
fa9f7708d4 add efficient matrix product specializations for Homogeneous 2009-03-05 16:40:56 +00:00
Gael Guennebaud
31332fca0b remove bad #include of SelfadjointRank2Update.h 2009-03-05 10:29:20 +00:00
Gael Guennebaud
0be89a4796 big addons: 
* add Homogeneous expression for vector and set of vectors (aka matrix)
  => the next step will be to overload operator*
* add homogeneous normalization (again for vector and set of vectors)
* add a Replicate expression (with uni-directional replication
  facilities)
=> for all of them I'll add examples once we agree on the API
* fix gcc-4.4 warnings
* rename reverse.cpp array_reverse.cpp
 2009-03-05 10:25:22 +00:00
Gael Guennebaud
6a26506341 add ReturnByValue pseudo expression for in-place evaluation with a 
return-by-value API style (will soon use it for the transform products)
 2009-03-04 13:00:00 +00:00
Gael Guennebaud
3288e9e168 add much faster versions of unaligned stores (and slightly faster 
unaligned loads)
 2009-03-03 14:01:30 +00:00
Gael Guennebaud
c4601314be cleaned Tridiagonalization impl, with improved performance 2009-03-02 15:07:39 +00:00
Gael Guennebaud
ed134a0ce5 performance improvement: rewrite of the matrix-matrix product following 
Goto's paper => x1.4 speedup with more consistent perf results
 2009-03-02 13:15:15 +00:00
Gael Guennebaud
8ed186b9ab improve WIP new matrix product 2009-02-27 17:18:52 +00:00
Gael Guennebaud
170128770a compilation fix + test orho methods for complex 2009-02-26 10:09:23 +00:00
Gael Guennebaud
91af389a9a fix unitOrthogonal() for size > 3 2009-02-25 09:23:09 +00:00
Laurent Montel
2d6d14a3d3 Add COMPONENT Devel 2009-02-23 07:50:56 +00:00
Laurent Montel
8e7c4df0db Fix install header 2009-02-22 10:28:31 +00:00
Gael Guennebaud
de014efdaf * split CacheFriendlyProduct into multiple smaller files 
* add an efficient selfadjoint * vector implementation (= blas symv)
  perf are inbetween MKL and GOTO
  => the interface is still missing (have to be rethougth)
 2009-02-21 20:20:38 +00:00
Gael Guennebaud
3d86dcf473 oops, got confused by the preprocessor directives around 
posix_memalign...
 2009-02-21 16:35:57 +00:00
Gael Guennebaud
7c4f9ecf0c fix posix_memalign return value warning 2009-02-21 16:23:18 +00:00
Daniel Gomez Ferro
032880074e Added new product implementation. 
Just works for square, power of 2 matrices of floats.
 2009-02-20 22:23:25 +00:00
Gael Guennebaud
e2ee7a6a58 increase version number for step 2009-02-19 15:46:10 +00:00
Gael Guennebaud
752d95c20e eventually c++ does not provide any optimized pow(int,int) function, 
so here you go :) (should also fix Timothy's troubles)
 2009-02-18 18:24:31 +00:00
Gael Guennebaud
e6f1104b57 * fix Quaternion::setFromTwoVectors (thanks to "benv" from the forum) 
* extend PartialRedux::cross() to any matrix sizes with automatic
  vectorization when possible
* unit tests: add "geo_" prefix to all unit tests related to the
  geometry module and start splitting the big "geometry.cpp" tests to
  multiple smaller ones (also include new tests)
 2009-02-17 09:53:05 +00:00
Gael Guennebaud
67b4fab4e3 fix assertion issue in slice vectorization 2009-02-16 10:17:21 +00:00
Konstantinos A. Margaritis
349557db9a no reason for 3 vec_mins, 2 are enough apparently in ei_predux_min 2009-02-12 22:03:30 +00:00
Konstantinos A. Margaritis
ad2bf14dbb modified ei_predux_min/max to actually use altivec instructions 2009-02-12 21:58:44 +00:00
Gael Guennebaud
3b12f48aa3 compilation fix for SuperLU 3.1 2009-02-12 17:50:51 +00:00
Gael Guennebaud
20a8bb96eb fix m = m*m with m sparse (gug found by Frederik Heinz) 2009-02-12 15:57:13 +00:00