mirror/eigen
2
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-03-07 18:27:40 +08:00
Code Issues Packages Projects Releases Wiki Activity
9,694 Commits 11 Branches 89 Tags 183 MiB
6d451cf2b6
Commit Graph

5574 Commits

Author SHA1 Message Date
Michael Figurnov
4bd158fa37 Derivative of the incomplete Gamma function and the sample of a Gamma random variable. 
In addition to igamma(a, x), this code implements:
* igamma_der_a(a, x) = d igamma(a, x) / da -- derivative of igamma with respect to the parameter
* gamma_sample_der_alpha(alpha, sample) -- reparameterization derivative of a Gamma(alpha, 1) random variable sample with respect to the alpha parameter

The derivatives are computed by forward mode differentiation of the igamma(a, x) code. Although gamma_sample_der_alpha can be implemented via igamma_der_a, a separate function is more accurate and efficient due to analytical cancellation of some terms. All three functions are implemented by a method parameterized with "mode" that always computes the derivatives, but does not return them unless required by the mode. The compiler is expected to (and, based on benchmarks, does) skip the unnecessary computations depending on the mode.
 2018-06-06 18:49:26 +01:00
Deven Desai
8fbd47052b Adding support for using Eigen in HIP kernels. 
This commit enables the use of Eigen on HIP kernels / AMD GPUs. Support has been added along the same lines as what already exists for using Eigen in CUDA kernels / NVidia GPUs.

Application code needs to explicitly define EIGEN_USE_HIP when using Eigen in HIP kernels. This is because some of the CUDA headers get picked up by default during Eigen compile (irrespective of whether or not the underlying compiler is CUDACC/NVCC, for e.g. Eigen/src/Core/arch/CUDA/Half.h). In order to maintain this behavior, the EIGEN_USE_HIP macro is used to switch to using the HIP version of those header files (see Eigen/Core and unsupported/Eigen/CXX11/Tensor)


Use the "-DEIGEN_TEST_HIP" cmake option to enable the HIP specific unit tests.
 2018-06-06 10:12:58 -04:00
Michael Figurnov
f216854453 Exponentially scaled modified Bessel functions of order zero and one. 
The functions are conventionally called i0e and i1e. The exponentially scaled version is more numerically stable. The standard Bessel functions can be obtained as i0(x) = exp(|x|) i0e(x)

The code is ported from Cephes and tested against SciPy.
 2018-05-31 15:34:53 +01:00
Gael Guennebaud
647b724a36 Define pcast<> for SSE types even when AVX is enabled. (otherwise float are silently reinterpreted as int instead of being converted) 2018-05-29 20:46:46 +02:00
Gael Guennebaud
49262dfee6 Fix compilation and SSE support with PGI compiler 2018-05-29 15:09:31 +02:00
Gael Guennebaud
f0862b062f Fix internal::is_integral<size_t/ptrdiff_t> with MSVC 2013 and older. 2018-05-22 19:29:51 +02:00
Gael Guennebaud
36e413a534 Workaround a MSVC 2013 compilation issue with MatrixBase(Index,int) 2018-05-22 18:51:35 +02:00
Gael Guennebaud
725bd92903 fix stupid typo 2018-05-18 17:46:43 +02:00
Gael Guennebaud
a382bc9364 is_convertible<T,Index> does not seems to work well with MSVC 2013, so let's rather use __is_enum(T) for old MSVC versions 2018-05-18 17:02:27 +02:00
Gael Guennebaud
4dd767f455 add some internal checks 2018-05-18 13:59:55 +02:00
Mark D Ryan
405859f18d Set EIGEN_IDEAL_MAX_ALIGN_BYTES correctly for AVX512 builds 
bug #1548

The macro EIGEN_IDEAL_MAX_ALIGN_BYTES is being incorrectly set to 32
on AVX512 builds.  It should be set to 64.  In the current code it is
only set to 64 if the macro EIGEN_VECTORIZE_AVX512 is defined.  This
macro does get defined in AVX512 builds in Core, but only after Macros.h,
the file that defines EIGEN_IDEAL_MAX_ALIGN_BYTES, has been included.
This commit fixes the issue by setting EIGEN_IDEAL_MAX_ALIGN_BYTES to
64 if __AVX512F__ is defined.
 2018-05-17 17:04:00 +01:00
Gael Guennebaud
7134fa7a2e Fix compilation with MSVC by reverting to char* for _mm_prefetch except for PGI (the later being the one that has the wrong prototype). 2018-06-07 09:33:10 +02:00
Robert Lukierski
b2053990d0 Adding EIGEN_DEVICE_FUNC to Products, especially Dense2Dense Assignment 
specializations. Otherwise causes problems with small fixed size matrix multiplication (call to
0x00 in call_assignment_no_alias in debug mode or trap in release with CUDA 9.1).
 2018-03-14 16:19:43 +00:00
Jeff Trull
9f0c5c3669 Make sparse QR result sizes consistent with dense QR, with the following rules: 
1) Q is always square
2) Q*R*P' is valid and recovers the original matrix

This implies that the size of Q is the number of rows in the original matrix, square,
and that the size of R is the size of the original matrix.
 2018-02-15 15:00:31 -08:00
Christoph Hertzberg
d655900953 bug #1544: Generate correct Q matrix in complex case. Original patch was by Jeff Trull in PR-386. 2018-05-17 19:17:01 +02:00
Christoph Hertzberg
0272f2451a Fix "suggest parentheses around comparison" warning 2018-05-15 19:35:53 +02:00
Gael Guennebaud
6e7118265d Fix compilation with NEON+MSVC 2018-04-26 10:50:41 +02:00
Gael Guennebaud
8810baaed4 Add multi-threading for sparse-row-major * dense-row-major 2018-04-25 10:14:48 +02:00
Gael Guennebaud
e8ca5166a9 bug #1428: atempt to make NEON vectorization compilable by MSVC. 
The workaround is to wrap NEON packet types to make them different c++ types.
 2018-04-24 11:19:49 +02:00
Benoit Steiner
6f5935421a fix AVX512 plog 2018-04-23 15:49:26 +00:00
Gael Guennebaud
e9da464e20 Add specializations of is_arithmetic for long long in c++11 2018-04-23 16:26:29 +02:00
Gael Guennebaud
a57e6e5f0f workaround MSVC 2013 compilation issue (ambiguous call) 2018-04-23 15:31:51 +02:00
Gael Guennebaud
11123175db typo in doc 2018-04-23 15:30:35 +02:00
Gael Guennebaud
5679e439e0 bug #1543: fix linear indexing in generic block evaluation (this completes the fix in commit 12efc7d41b 
)
 2018-04-23 14:40:16 +02:00
Christoph Hertzberg
34e499ad36 Disable -Wshadow when compiling with g++ 2018-04-21 22:08:26 +02:00
Jayaram Bobba
b7b868d1c4 fix AVX512 plog 2018-04-20 13:39:18 -07:00
Gael Guennebaud
686fb57233 fix const cast in NEON 2018-04-18 18:46:34 +02:00
Dmitriy Korchemkin
02d2f1cb4a Cast zeros to Scalar in RealSchur 2018-04-18 13:52:46 +03:00
Christoph Hertzberg
50633d1a83 Renamed .trans() et al. to .reverseFlag() et at. Adapted documentation of .setReverseFlag() 2018-04-17 11:30:27 +02:00
nicolov
39c2cba810 Add a specialization of Eigen::numext::conj for std::complex<T> to be used when compiling a cuda kernel. This fixes the compilation of TensorFlow 1.4 with clang 6.0 used as CUDA compiler with libc++. 
This follows the previous change in 2a69290ddb
, which mentions OSX (I guess because it uses libc++ too).
 2018-04-13 22:29:10 +00:00
Christoph Hertzberg
42715533f1 bug #1493: Make representation of HouseholderSequence consistent and working for complex numbers. Made corresponding unit test actually test that. Also simplify implementation of QR decompositions 2018-04-15 10:15:28 +02:00
Christoph Hertzberg
4d392d93aa Make hypot_impl compile again for types with expression-templates (e.g., boost::multiprecision) 2018-04-13 19:01:37 +02:00
Christoph Hertzberg
072e111ec0 SelfAdjointView<...,Mode> causes a static assert since commit d820ab9edc 2018-04-13 19:00:34 +02:00
Gael Guennebaud
7a9089c33c fix linking issue 2018-04-13 08:51:47 +02:00
Gael Guennebaud
e43ca0320d bug #1520: workaround some -Wfloat-equal warnings by calling std::equal_to 2018-04-11 15:24:13 +02:00
Gael Guennebaud
c91906b065 Umfpack: UF_long has been removed in recent versions of suitesparse, and fix a few long-to-int conversions issues. 2018-04-11 09:59:59 +02:00
Gael Guennebaud
0050709ea7 Merged in v_huber/eigen (pull request PR-378) 
Add interface to umfpack_*l_* functions
 2018-04-11 07:43:04 +00:00
Guillaume Jacob
8c1652055a Fix code sample output in block(int, int, int, int) doxygen 2018-04-09 17:23:59 +02:00
Gael Guennebaud
add15924ac Fix MKL backend for symmetric eigenvalues on row-major matrices. 2018-04-09 13:29:26 +02:00
Gael Guennebaud
04b1628e55 Add missing empty line. 2018-04-09 13:28:31 +02:00
Gael Guennebaud
2f833b1c64 bug #1509: fix computeInverseWithCheck for complexes 2018-04-04 15:47:46 +02:00
Gael Guennebaud
b903fa74fd Extend list of MSVC versions 2018-04-04 15:14:09 +02:00
Gael Guennebaud
403f09ccef Make stableNorm and blueNorm compatible with 2D matrices. 2018-04-04 15:13:31 +02:00
Gael Guennebaud
4213b63f5c Factories code between numext::hypot and scalar_hyot_op functor. 2018-04-04 15:12:43 +02:00
Gael Guennebaud
368dd4cd9d Make innerVector() and innerVectors() methods available to all expressions supported by Block. 
Before, only SparseBase exposed such methods.
 2018-04-04 15:09:21 +02:00
Gael Guennebaud
e116f6847e bug #1521: avoid signalling NaN in hypot and make it std::complex<> friendly. 2018-04-04 13:47:23 +02:00
Gael Guennebaud
13f5df9f67 Add a note on vec_min vs asm 2018-04-04 13:10:38 +02:00
Gael Guennebaud
e91e314347 bug #1494: makes pmin/pmax behave on Altivec/VSX as on x86 regading NaNs 2018-04-04 11:39:19 +02:00
Gael Guennebaud
112c899304 comment unreachable code 2018-04-03 23:16:43 +02:00
Gael Guennebaud
a1292395d6 Fix compilation of product with inverse transpositions (e.g., mat * Transpositions().inverse()) 2018-04-03 23:06:44 +02:00
Gael Guennebaud
8c7b5158a1 commit 45e9c9996da790b55ed9c4b0dfeae49492ac5c46 (HEAD -> memory_fix) 
Author: George Burgess IV <gbiv@google.com>
Date:   Thu Mar 1 11:20:24 2018 -0800

    Prefer `::operator new` to `new`

    The C++ standard allows compilers much flexibility with `new`
    expressions, including eliding them entirely
    (https://godbolt.org/g/yS6i91). However, calls to `operator new` are
    required to be treated like opaque function calls.

    Since we're calling `new` for side-effects other than allocating heap
    memory, we should prefer the less flexible version.

    Signed-off-by: George Burgess IV <gbiv@google.com>
 2018-04-03 17:15:38 +02:00
Gael Guennebaud
dd4cc6bd9e bug #1527: fix support for MKL's VML (destination was not properly resized) 2018-04-03 17:11:15 +02:00
Gael Guennebaud
c5b56f1fb2 bug #1528: better use numeric_limits::min() instead of 1/highest() that with underflow. 2018-04-03 16:49:35 +02:00
Gael Guennebaud
8d0ffe3655 bug #1516: add assertion for out-of-range diagonal index in MatrixBase::diagonal(i) 2018-04-03 16:15:43 +02:00
Gael Guennebaud
407e3e2621 bug #1532: disable stl::*_negate in C++17 (they are deprecated) 2018-04-03 15:59:30 +02:00
Gael Guennebaud
40b4bf3d32 AVX512: _mm512_rsqrt28_ps is available for AVX512ER only 2018-04-03 14:36:27 +02:00
Gael Guennebaud
584951ca4d Rename predux_downto4 to be more accurate on its semantic. 2018-04-03 14:28:38 +02:00
Gael Guennebaud
7b0630315f AVX512: fix psqrt and prsqrt 2018-04-03 14:12:50 +02:00
Gael Guennebaud
6719409cd9 AVX512: add missing pinsertfirst and pinsertlast, implement pblend for Packet8d, fix compilation without AVX512DQ 2018-04-03 14:11:56 +02:00
vhuber
267a144da5 Remove unnecessary define 2018-03-30 23:04:53 +02:00
vhuber
baf9a5a776 Add interface to umfpack_*l_* functions 2018-03-30 18:53:34 +02:00
luz.paz
e3912f5e63 MIsc. source and comment typos 
Found using `codespell` and `grep` from downstream FreeCAD
 2018-03-11 10:01:44 -04:00
Basil Fierz
624df50945 Adds missing EIGEN_STRONG_INLINE to support MSVC properly inlining small vector calculations 
When working with MSVC often small vector operations are not properly inlined. This behaviour is observed even on the most recent compiler versions.
 2017-10-26 22:44:28 +02:00
Benoit Steiner
d2631ef61d Merged in facaiy/eigen/ENH/exp_support_complex_for_gpu (pull request PR-359) 
ENH: exp supports complex type for cuda
 2018-03-23 00:59:15 +00:00
Benoit Steiner
8fcbd6d4c9 Merged in dtrebbien/eigen (pull request PR-369) 
Move up the specialization of std::numeric_limits
 2018-03-23 00:54:58 +00:00
Gael Guennebaud
f7d17689a5 Add static assertion for fixed sizes Ref<> 2018-03-09 10:11:13 +01:00
Gael Guennebaud
f6be7289d7 Implement better static assertion checking to make sure that the first assertion is a static one and not a runtime one. 2018-03-09 10:00:51 +01:00
Gael Guennebaud
d820ab9edc Add static assertion on selfadjoint-view's UpLo parameter. 2018-03-09 09:33:43 +01:00
Daniel Trebbien
0c57be407d Move up the specialization of std::numeric_limits 
This fixes a compilation error seen when building TensorFlow on macOS:
https://github.com/tensorflow/tensorflow/issues/17067
 2018-02-18 15:35:45 -08:00
Gael Guennebaud
adb134d47e Fix implicit conversion from 0.0 to scalar 2018-02-16 22:26:01 +04:00
Gael Guennebaud
5deeb19e7b bug #1517: fix triangular product with unit diagonal and nested scaling factor: (s*A).triangularView<UpperUnit>()*B 2018-02-09 16:52:35 +01:00
Gael Guennebaud
12efc7d41b Fix linear indexing in generic block evaluation. 2018-02-09 16:45:49 +01:00
Gael Guennebaud
f4a6863c75 Fix typo 2018-02-09 16:43:49 +01:00
Gael Guennebaud
09a16ba42f bug #1412: fix compilation with nvcc+MSVC 2018-01-17 23:13:16 +01:00
Yan Facai (颜发才)
42a8334668 ENH: exp supports complex type for cuda 2018-01-04 16:01:01 +08:00
Eugene Chereshnev
f558ad2955 Fix incorrect ldvt in LAPACKE call from JacobiSVD 2018-01-03 12:55:52 -08:00
Gael Guennebaud
73629f8b68 Fix gcc7 warning 2018-01-09 08:59:27 +01:00
nluehr
f9bdcea022 For cuda 9.1 replace math_functions.hpp with cuda_runtime.h 2017-12-18 16:51:15 -08:00
Gael Guennebaud
06bf1047f9 Fix compilation of stableNorm with some expressions as input 2017-12-15 15:15:37 +01:00
Gael Guennebaud
546ab97d76 Add possibility to overwrite EIGEN_STRONG_INLINE. 2017-12-14 14:47:38 +01:00
Gael Guennebaud
9c3aed9d48 Fix packet and alignment propagation logic of Block<Xpr> expressions. In particular, (A+B).col(j) lost vectorisation. 2017-12-14 14:24:33 +01:00
nluehr
aefd5fd5c4 Replace __float2half_rn with __float2half 
The latter provides a consistent definition for CUDA 8.0 and 9.0.
 2017-11-28 10:15:46 -08:00
Gael Guennebaud
d0b028e173 clarify Pastix requirements 2017-11-27 22:11:57 +01:00
Gael Guennebaud
3587e481fb silent MSVC warning 2017-11-27 21:53:02 +01:00
nluehr
dd6de618c3 Fix incorrect integer cast in predux<half2>(). 
Bug corrupts results on Maxwell and earlier GPU architectures.
 2017-11-21 10:47:00 -08:00
Gael Guennebaud
672bdc126b bug #1479: fix failure detection in LDLT 2017-11-16 17:55:24 +01:00
Gael Guennebaud
7cc503f9f5 bug #1485: fix linking issue of non template functions 2017-11-15 21:33:37 +01:00
Gael Guennebaud
00bc67c374 Move KLU support to official 2017-11-10 14:11:22 +01:00
Gael Guennebaud
1495b98a8e Merged in spraetor/eigen (pull request PR-305) 
Issue with mpreal and std::numeric_limits::digits
 2017-11-10 10:28:54 +00:00
Gael Guennebaud
d306b96fb7 Merged in carpent/eigen (pull request PR-342) 
Use col method for column-major matrix
 2017-11-10 10:09:53 +00:00
Gael Guennebaud
f86bb89d39 Add EIGEN_MKL_NO_DIRECT_CALL option 2017-11-09 11:07:45 +01:00
Gael Guennebaud
5fa79f96b8 Patch from Konstantin Arturov to enable MKL's direct call by default 2017-11-09 10:58:38 +01:00
Gael Guennebaud
4c03b3511e Fix issue with boost::multiprec in previous commit 2017-11-08 23:28:01 +01:00
Gael Guennebaud
e9d2888e74 Improve debugging tests and output in BDCSVD 2017-11-08 10:26:03 +01:00
Gael Guennebaud
e8468ea91b Fix overflow issues in BDCSVD 2017-11-08 10:24:28 +01:00
Christoph Hertzberg
11ddac57e5 Merged in guillaume_michel/eigen (pull request PR-334) 
- Add support for NEON plog PacketMath function
 2017-10-23 13:22:22 +00:00
Benoit Steiner
f16ba2a630 Merged in LaFeuille/eigen-1/LaFeuille/typo-fix-alignmeent-alignment-1505889397887 (pull request PR-335) 
Typo fix alignmeent ->alignment
 2017-10-21 01:59:55 +00:00
Henry Schreiner
9bb26eb8f1 Restore __device__ 2017-10-21 00:50:38 +00:00
Henry Schreiner
4245475d22 Fixing missing inlines on device functions for newer CUDA cards 2017-10-20 03:20:13 +00:00
Justin Carpentier
a020d9b134 Use col method for column-major matrix 2017-10-17 21:51:27 +02:00