devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

Commit Graph

  • 6b3f81b414 precision improvement. Wtf were we using sqrt(precision) for the cutoff? Replaced by precision*biggest. Benoit Jacob 2010-02-23 16:26:39 -05:00
  • 60325b8330 actually, this is not even meant to be a termination criterion. so the proper fix is this. Benoit Jacob 2010-02-23 16:10:26 -05:00
  • 3d066f4bc7 LDLT: * fix bug thanks to Ben Goodrich: we were terminating at the wrong place, leaving some matrix coefficients with wrong values. * don't use Higham's formula here: we're not trying to be rank-revealing. Benoit Jacob 2010-02-23 16:05:37 -05:00
  • d92df336ad Further LU test improvements. I'm not aware of any test failures anymore, not even with huge numbers of repetitions. Finally the createRandomMatrixOfRank() function is renamed to createRandomPIMatrixOfRank, where PI stands for 'partial isometry', that is, a matrix whose singular values are 0 or 1. Benoit Jacob 2010-02-23 15:40:24 -05:00
  • a1e1103328 add a 2D parallelizer Gael Guennebaud 2010-02-23 21:40:15 +01:00
  • 022e2f5ef4 fix typo Gael Guennebaud 2010-02-23 18:24:15 +01:00
  • 68eaefa5d4 update BTL (better timer, eigen2 => eigen3, etc) Gael Guennebaud 2010-02-23 18:23:12 +01:00
  • 7dc75380c1 * FullPivLU: replace "remaining==0" termination condition (from Golub) by a fuzzy compare (fixes lu test failures when testing solve()) * LU test: set appropriate threshold and limit the number of times that a specially tricky test is run. (fixes lu test failures when testing rank()). * Tests: rename createRandomMatrixOfRank to createRandomProjectionOfRank Benoit Jacob 2010-02-23 09:04:59 -05:00
  • 4a0d41c5fb merge Gael Guennebaud 2010-02-23 14:34:55 +01:00
  • 1fd8d7b96a Attempt to fix PGI compilation issue. Hauke Heibel 2010-02-23 11:35:51 +01:00
  • b528b747c1 merge Mark Borgerding 2010-02-22 21:44:30 -05:00
  • 5d530e0373 enable caller to supply FFT length for Eigen Matrix interface functions to effect zero pad or source shrink at Nyquist bin Mark Borgerding 2010-02-22 21:43:30 -05:00
  • 3beedba244 merge Gael Guennebaud 2010-02-22 21:32:29 +01:00
  • d3b314569b provide default values for CXX, remove duplicate define Thomas Capricelli 2010-02-22 15:39:17 +01:00
  • e6f0cd7121 Merge. Jitse Niesen 2010-02-22 16:17:43 +00:00
  • d2b0eadf52 fully adapt the gebp kernel and optimize it for CPU with only 8 registers (transplanted from 2ed88ebbf1995be90b8d0c25ff10248c8f56d023) Gael Guennebaud 2010-02-22 16:35:05 +01:00
  • 51a4b929a1 implement an even lower level version of the gebp kernel for MSVC (it seems to be faster with gcc as well) (transplanted from 9a5643551fe068497f84a81cd8986febf1918382) Gael Guennebaud 2010-02-22 15:18:29 +01:00
  • 3e6ab8f93b ups Hauke Heibel 2010-02-22 11:34:25 +01:00
  • 801440c519 fix BTL's eigen interface (transplanted from 437f40acc1 ) Gael Guennebaud 2010-02-22 09:32:16 +01:00
  • eb905500b6 significant speedup in the matrix-matrix products Gael Guennebaud 2010-02-23 13:06:49 +01:00
  • d579d4cc37 oops Gael Guennebaud 2010-02-22 17:57:15 +01:00
  • fc4a85ecd5 fully adapt the gebp kernel and optimize it for CPU with only 8 registers Gael Guennebaud 2010-02-22 16:35:05 +01:00
  • dd569c7d0f merge Gael Guennebaud 2010-02-22 15:19:19 +01:00
  • e00f1fd125 implement an even lower level version of the gebp kernel for MSVC (it seems to be faster with gcc as well) Gael Guennebaud 2010-02-22 15:18:29 +01:00
  • 6730fd9f3f Port BenchTimer fix. Hauke Heibel 2010-02-22 11:42:58 +01:00
  • 4ba25a8d5c merge Gael Guennebaud 2010-02-22 11:30:36 +01:00
  • d5af5ab92b Added getRealTime() for windows. Hauke Heibel 2010-02-22 11:23:27 +01:00
  • aaaf855a88 add a small benchmark to quickly bench/compare SMP support Gael Guennebaud 2010-02-22 11:09:57 +01:00
  • 1c6bdb4060 merge Gael Guennebaud 2010-02-22 11:09:05 +01:00
  • 3e62fafce8 clean a bit the parallelizer Gael Guennebaud 2010-02-22 11:08:37 +01:00
  • f797ba0abe extend the bench timer to allow benchmarking of parallel code, improvements are welcome Gael Guennebaud 2010-02-22 11:04:35 +01:00
  • b20935be9b add initial openmp support for matrix-matrix products => x1.9 speedup on my core2 duo Gael Guennebaud 2010-02-22 09:40:34 +01:00
  • 437f40acc1 fix BTL's eigen interface Gael Guennebaud 2010-02-22 09:32:16 +01:00
  • 1a70f3b48d fix compilation Thomas Capricelli 2010-02-21 19:30:11 +01:00
  • a901bed33a Added IsRowMajor enum to DenseBase. Hauke Heibel 2010-02-21 18:26:14 +01:00
  • e2a059863e Added missing precision/eps functions to AutoDiffScalar. Hauke Heibel 2010-02-21 15:44:34 +01:00
  • f079f52b58 merge Hauke Heibel 2010-02-21 15:25:28 +01:00
  • f75e6773b0 Added ei_traits<Quaternion>::PlainObject. Hauke Heibel 2010-02-21 15:24:10 +01:00
  • ac8ff44278 Tried to get rid of MSVC warning D9025. Hauke Heibel 2010-02-21 15:23:51 +01:00
  • a7d085eb4e NonLinearOptimization : clean 'mode' handling from the old minpack code : * this is actually a boolean, not an int * use a better name * can be set at initialization time instead of bloating all methods signatures Thomas Capricelli 2010-02-21 12:41:37 +01:00
  • 1f4b8e6a36 compilation fix in ldlt() for non matrix types (transplanted from 608959aa6f ) Gael Guennebaud 2010-02-21 10:29:19 +01:00
  • 608959aa6f compilation fix in ldlt() for non matrix types Gael Guennebaud 2010-02-21 10:29:19 +01:00
  • 71fecd2371 add missing return value Gael Guennebaud 2010-02-20 18:19:34 +01:00
  • 6b4cecc1c6 CMake cleanup. Hauke Heibel 2010-02-20 17:39:58 +01:00
  • 4e389b195d Change MatrixFunction::separation() parameter from 0.01 to 0.1 . The latter is actually the value used in the literature. Jitse Niesen 2010-02-20 16:27:04 +00:00
  • abc8c01080 Renamed PlainMatrixType to PlainObject (Array != Matrix). Renamed ReturnByValue::ReturnMatrixType ReturnByValue::ReturnType (again, Array != Matrix). Hauke Heibel 2010-02-20 15:53:57 +01:00
  • 67ce07ea83 matrix_function test: replace expm(A).inverse() by expm(-A) The latter is more stable. This fixes one of the issues with the test. Also, make typedef's in MatrixFunctionReturnValue public; this is necessary to get the test to compile. Jitse Niesen 2010-02-20 14:45:50 +00:00
  • f0c8dcf1e2 Renamed AnyMatrixBase to EigenBase. Hauke Heibel 2010-02-20 15:26:02 +01:00
  • 4f8773c23a fix stupid enum values Gael Guennebaud 2010-02-19 17:46:36 +01:00
  • 5491531a81 add Stride.h Benoit Jacob 2010-02-18 20:44:17 -05:00
  • b73e22905d miserable half-working state, commiting to a fork just in case, just to perfect my day, my hard disk would die. Will write a more detailed commit message once it's working. Benoit Jacob 2010-02-18 20:42:38 -05:00
  • 39d9f0275b Update matrix_exponential test after API change in ei_matrix_function Apologies for forgetting this yesterday and not testing properly. Jitse Niesen 2010-02-17 09:50:11 +00:00
  • 2cd1ad2ded typo in merge Mark Borgerding 2010-02-16 22:06:23 -05:00
  • f200c84d9f merge Mark Borgerding 2010-02-16 21:41:04 -05:00
  • 8f51a4ac90 found out about little-documented FFTW_PRESERVE_INPUT which has effect on c2r transforms Mark Borgerding 2010-02-16 20:44:48 -05:00
  • 319bf3130b Use ReturnByValue to return result of ei_matrix_function(), ... Jitse Niesen 2010-02-16 16:43:11 +00:00
  • 25019f0836 Use ReturnByValue to return result of ei_matrix_exponential() . Jitse Niesen 2010-02-15 19:17:25 +00:00
  • a9096b626b not all versions of gcc support -Wno-variadic-macros Gael Guennebaud 2010-02-15 11:39:57 +01:00
  • 016943f870 avoid 2 redundant calls to resize Gael Guennebaud 2010-02-15 11:31:36 +01:00
  • dcb395c6f5 explicitly disable the use of evalTo for dense object Gael Guennebaud 2010-02-15 11:09:33 +01:00
  • 21d0eb3f11 the default implementation should really call evalTo Gael Guennebaud 2010-02-15 11:01:55 +01:00
  • d00bff91ad workaround weird gcc 4.0.1 compilation error Gael Guennebaud 2010-02-15 11:00:30 +01:00
  • 8519558d11 Workaround for compounds affected by #94. Hauke Heibel 2010-02-15 10:11:10 +01:00
  • b18f737aa1 Test matrix functions with matrices with clustered imaginary eivals. The idea is that these test MatrixFunction::swapEntriesInSchur(), which is not covered by existing tests. This did not work out as expected, but nevertheless it is a good test so I left it in. Jitse Niesen 2010-02-13 22:49:01 +00:00
  • a4a2671fd0 Refactor matrix_function test in preparation of next commit. Jitse Niesen 2010-02-13 22:38:27 +00:00
  • 9251cfed9b this had to be done here, not at the end. Benoit Jacob 2010-02-12 09:03:16 -05:00
  • e1f61b40c8 oops, this had to be done here, not at the end. Benoit Jacob 2010-02-12 09:02:25 -05:00
  • 37ca4200b2 Piotr's patch was missing many occurences of size_t. So, using std::size_t; This is the only way that we can ensure QCC support in the long term without having to think about it everytime. Benoit Jacob 2010-02-12 08:58:29 -05:00
  • f369dc873e Piotr's patch was missing many occurences of size_t. So, using std::size_t; This is the only way that we can ensure QCC support in the long term without having to think about it everytime. Benoit Jacob 2010-02-12 08:57:04 -05:00
  • a76950bdab fix a couple of ICE with gcc 4.0.1 Gael Guennebaud 2010-02-12 09:41:56 +01:00
  • c03bca21c4 Added tag 2.0.12 for changeset ed6eb5a625 Benoit Jacob 2010-02-11 21:39:46 -05:00
  • ed6eb5a625 bump 2.0.12 Benoit Jacob 2010-02-11 21:39:41 -05:00
  • 9488a12125 work around brain dead ICC Benoit Jacob 2010-02-11 19:32:56 -05:00
  • 7b44957c4b std:: namespace fixup for more restricive compilers such as QNX's QCC Piotr Trojanek 2010-02-10 22:27:35 +01:00
  • 1701a5d1f8 std:: namespace fixup for more restricive compilers such as QNX's QCC Piotr Trojanek 2010-02-10 13:24:47 +01:00
  • ae0a17d30b Here is the proper fix. Hauke Heibel 2010-02-11 11:39:02 +01:00
  • 13ce92f5ba Fixed notemporary unit test. Hauke Heibel 2010-02-11 11:31:36 +01:00
  • 93e86b0884 Fixed typos. Replace NumTraits<bool>::dummy_precision() (three locations) by false in order to suppress warnings. Hauke Heibel 2010-02-11 11:31:22 +01:00
  • 48f5980669 fix compilation (cwise and epsilon) Gael Guennebaud 2010-02-11 09:24:29 +01:00
  • 602778ea26 also fix tests for NumTraits<double>::epsilon() Thomas Capricelli 2010-02-11 01:49:27 +01:00
  • fd19dd57d8 fix usage of epsilon wrt to latest API change Thomas Capricelli 2010-02-11 01:43:36 +01:00
  • ccc58e935e fix usage of epsilon wrt to latest API change Thomas Capricelli 2010-02-11 01:43:03 +01:00
  • fc5fa77743 unsupported/Eigen/AlignedVector3: dummy_precision is now in NumTraits Jitse Niesen 2010-02-10 17:59:27 +00:00
  • d0e8342a04 Fixed warning. Hauke Heibel 2010-02-10 18:00:36 +01:00
  • 0ca67afe6a finally here is a simple solution making (a*b).diagonal() even faster than a.lazyProduct(b).diagonal() !! Gael Guennebaud 2010-02-10 14:08:47 +01:00
  • 743ad75595 BenchTimer backport (clock_gettime & QueryPerformanceCounter). Hauke Heibel 2010-02-03 21:55:01 +01:00
  • a9eabed421 Patch by 'Wolf' from the issue tracker: Fix bug #90, missing type cast in LU, allow to use LU with MPFR. Benoit Jacob 2010-02-02 07:06:15 -05:00
  • cd34a1d351 backport bug fix by Jitse. Benoit Jacob 2010-01-28 14:00:09 -05:00
  • 3e963ee69d EIGEN_ENUM_MIN ---> EIGEN_SIZE_MIN Benoit Jacob 2010-01-26 20:37:57 -05:00
  • 6cc9dc17f2 In LU / Inverse, decouple the output type from the input type. This has long been done in the default branch Benoit Jacob 2010-01-26 18:45:23 -05:00
  • 7852a48a2f fix matrix product with EIGEN_DEFAULT_TO_ROW_MAJOR Gael Guennebaud 2010-01-25 21:56:01 +01:00
  • d209120180 * Introduce EIGEN_DEFAULT_TO_ROW_MAJOR tests option ---> Now only product_large fails with EIGEN_DEFAULT_TO_ROW_MAJOR. * Fix EIGEN_NO_ASSERTION_CHECKING tests option * Fix a crash in Tridiagonalization on row-major matrices + SSE * Fix inverse test (numeric stability noise) * Extend map test (see previous fixes in MapBase) * Fix vectorization_logic test for row-major * Disable sparse tests with EIGEN_DEFAULT_TO_ROW_MAJOR Benoit Jacob 2010-01-25 14:00:02 -05:00
  • 1d342e135c changed destination argument to reference Mark Borgerding 2010-01-22 22:52:13 -05:00
  • 55c0707b1d fix the script again (definitely?) + cleaning Thomas Capricelli 2010-01-22 19:28:33 +01:00
  • 141c746fc7 if the src.stride() != 1, then the layout is not continuous -- need to copy to temporary Mark Borgerding 2010-01-22 01:17:36 -05:00
  • cd7912313d changed FFT function vector and Matrix args to pointer as Benoit suggested implemented 2D Complex FFT for FFTW impl Mark Borgerding 2010-01-22 00:35:03 -05:00
  • 72044ca925 fix a super nasty bug: on row-major expressions that are NOT vectors but that do have LinearAccess, the MapBase::coeff(int) and MapBase::coeffRef(int) methods were broken. Benoit Jacob 2010-01-21 23:33:20 -05:00
  • c2b8ca7493 if EIGEN_DONT_ALIGN then don't try to vectorize (was giving a #error later on). Benoit Jacob 2010-01-21 22:32:16 -05:00
  • a30d42354f updated comments and played around with Map Mark Borgerding 2010-01-21 21:10:16 -05:00
  • 7a6cb2a39c added benchmark for unscaled and half-spectrum FFTs Mark Borgerding 2010-01-21 21:09:26 -05:00