1. Add CUDADenseQR & tests.
CUDADenseQR uses the cuSolverDN LAPACK implementation
of QR factorization. A key limitation, however, is that
this solver does not perform singularity checking --
this is because cuSolverDN does not have a trtrs
implementation; we instead use cuBLAS' trsv for
backsubstitution.
2. All CPU -> GPU memory transfers are now async, and both
CUDADenseQR and CUDADenseCholesky explicitly manage their
own streams for async operations.
3. Simplified CUDADenseCholesky to only use the legacy 32-bit
cuSolverDN API.
Change-Id: I2a9b7b65469658ddfe33b5b2a3892c8744d6e437
Currently, the logic for exporting symbols is rather complicated: when
tests are enabled internal symbols are exported in addition to the
public symbols. Such logic causes several problems. (1) Test binaries
link against a Ceres build that is different from the final release
since fewer optimizations are applied if more symbols are exported. (2)
Also, some toolchains hide symbols by default breaking the existing
logic eventually causing linker errors.
Since internal symbols are not intended to be used outside of the
project, we can compile them into object files and use exactly the same
binary code both for the final build and the tests without relying on
conditionals.
By default, all symbols are now hidden unless annotated as public.
Internal symbols are explicitly marked as not being exported in case
users chose not to hide symbols by default.
Change-Id: I589dd10be2f6f438508783cf99d141af0120057b
This change is needed because compiler attributes and c++
annotations like [[deprecated]] do not mix well and we need that
for our public API.
https://github.com/ceres-solver/ceres-solver/issues/749
Change-Id: I79eda795f2912f9af9ab36ee24b8428b47104743
Enabling the AVX2 instruction set causes a segmentation fault in mocked
manifold tests. This is due to Eigen vectors stored in a std::shared_ptr
for which the memory allocated by gmock is not aligned even though it is
expected to be by Eigen for correct use of packet math.
The problem does not occur if Ceres is compiled with C++17 (or later)
enabled due to the support for aligned new allocations.
Change-Id: I711abe9439cc411bd7a8b4936f3b93af07b7fbd6
Add [[deprecate]] notices to everything LocalParameterization
related.
Make sure that Ceres can be compiled without triggering
deprecation warnings.
Update the documentation:
a. Add deprecation notices.
b. Document interaction between LocalParameterization and Manifold
coexisting in the Problem.
c. Add documentation for Manifold(s)
Change-Id: Ie4ad48963c83fded86e533c8c60561af402fbaff
This MR ports the LineParameterization of manifolds. The unit test are
rewritten to use the manifold test facilities.
The LineManifold is extended so that it can also handle dynamic size
ambient space dimensions.
Change-Id: I1fe3cd34b56f74b72ca028c34f5368e9df9fe4d7
Do not define trivial constructors or destructors unless necessary
(e.g., for implementing pimpl) following the rule of zero. Define
virtual base class destructors out-of-line to avoid emitting vtables in
every translation unit.
Change-Id: Iea2d8978e62a8ee5a97b86cbb4e858d56e0fb274
Applied changes correspond to clang-tidy fixes
stemming from the modernize-use-equals-default check.
Change-Id: I254b0908a76d464131564b637cd0e42a6b03fb5a
virtual can be ambiguous. Applied changes correspond to clang-tidy fixes
stemming from the modernize-use-override check.
Change-Id: I973afd4680a5df587419777504aeb94467196b89
1. Add CUDADenseCholesky64Bit, CUDADenseCholesky32Bit, & tests.
CUDADenseCholesky32Bit uses the legacy versions of potrf/potrs
in cuSolverDN, while CUDADenseCholesky64Bit uses the new 64-bit
versions available since Cuda 11.1. The legacy versions are
provided since some platforms such as the Nvidia Jetsons only
support Cuda 10.2.
2. Expose CUDA as a new option under DenseLinearAlgebraLibraryType.
The relevant option to string and string to option helper functions
are modified accordingly.
3. Add cuda as a dense_linear_algebra_library option in bundle_adjuster
to demonstrate the use of the new CUDA option.
Change-Id: I23615e1d301df5185ed646b3e33ee802508dae86
This MR adds SphereManifold ported from
HomogeneousVectorParameterization. Additionally the minus operator
and jacobian evaluation was implemented.
The unit tests were almost completly reimplemented and uses the
test facilities provided for manifolds.
Change-Id: Iccf72a2333bc921ff24c4d831db35020c653ee86
Complete support for all floating-point classification functions
(fpclassify, signbit) and consistently apply all overloads recursively
to the scalar part of a Jet only. This is now inline with how comparison
operators work. Sanity checks of derivatives should be performed
explicitly on the dual part of a Jet due an ambiguity on reducing the
classification results of multiple values.
Provide an fdim overload (in addition to fmin and fmax) and support
quiet versions of comparison operators also applied recursively to the
scalar part of a Jet but without type promotion.
Additionally, deprecate Ceres legacy classification functions. New code
should use C++11 function names for consistency.
Finally, simplify expressions using introduced scalar classification and
comparison.
Change-Id: I397e37425760717b991eb7ae5da0892f20c5a365
1. Add EigenDenseQR & tests.
This implementation now uses an in place decomposition,
which means that we are not allocating, deallocating
memory every call.
2. Add LAPACKDenseQR and tests.
The LAPACK implementation instead of using dgels which is a
routine which does the factorization and solve in one
call, now uses dgeqrf for factorization and then
dormqr and dtrtrs for solving. This allows us to
have a factorize and solve interface like DenseCholesky.
And opens the door to iterative refinement and mixed
precision solves.
3. The refactor also allows us to simplify the interface to
DenseSparseMatrix considerably. The internals of this
class were complicated because we had the AppendDiagonal
and RemoveDiagonal methods and we did not want to allocate
deallocate memory every call. But since we pay the cost
of the copy anyways, we can just hold that buffer
in DenseQRSolver.
4. Delete lapack.cc/h
5. The net result is that everything seems to be a bit faster.
For LAPACK we are not doing some of the scaling work that
dgels was doing. For Eigen I think it maybe the inplace
decomposition.
Benchmark Time CPU Time Old Time New CPU Old CPU New
----------------------------------------------------------------------------------------------------------------------------------------------------------
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/1/1 -0.1154 -0.1159 692 612 691 611
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/2/1 -0.1601 -0.1553 717 603 712 601
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/3/1 -0.1673 -0.1575 733 610 724 610
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/6/2 -0.1008 -0.1003 886 797 884 796
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/10/3 -0.1489 -0.1514 1283 1092 1281 1087
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/12/4 -0.1040 -0.1104 1556 1394 1553 1381
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/20/5 -0.0007 -0.0097 1911 1910 1908 1890
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/40/5 -0.1033 -0.1022 2981 2673 2957 2655
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/100/10 -0.0147 +0.0015 9275 9138 9026 9040
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/200/10 -0.1408 -0.1284 15093 12968 14778 12880
BM_DenseSolver<ceres::EIGEN, ceres::DENSE_QR>/200/20 -0.0310 -0.0355 38973 37765 38837 37460
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/1/1 -0.1228 -0.1256 736 646 731 640
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/2/1 -0.1401 -0.1396 740 636 735 633
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/3/1 -0.1731 -0.1695 744 615 738 613
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/6/2 -0.1399 -0.1408 1121 965 1113 956
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/10/3 -0.1110 -0.1145 1571 1397 1560 1382
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/12/4 -0.1411 -0.1417 2006 1722 1993 1710
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/20/5 -0.1740 -0.1729 2741 2264 2724 2253
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/40/5 -0.0966 -0.1123 3462 3128 3425 3040
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/100/10 -0.0387 -0.0998 10365 9964 10339 9307
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/200/10 -0.2044 -0.2049 16031 12754 15998 12720
BM_DenseSolver<ceres::LAPACK, ceres::DENSE_QR>/200/20 -0.2391 -0.2386 35777 27223 35716 27193
Change-Id: I782f0d7664efe1435eebda92ddf47a0fe66c9c72
This is needed to make the dense_linear_solver_benchmark.cc
compile with the currently stable versions of the benchmark
library available on various linux distributions.
Change-Id: I1e391d5c2d16250d213bcfa3d50f9560aad9a363
- The intended use-case for these accessors is in client code tests to
support verification of the configuration with which cost functions
were constructed.
Change-Id: Ib77afa6409804ba7f724138f579e0c51b154f5ad
When computing the MinusJacobian we were passing two different
values to AutoDifferentiate for the output dimension. The template
argument was correct which is why the method was working correctly
but the function argument was incorrect (cut and paste error).
This would be fine in release mode, but in debug mode it would
trigger a check failure.
Change-Id: I0327656d1a4d34c82e4d3a8c04f27c264bce80eb
Like SparseCholesky, the DenseCholesky interface abstracts
away the solution of dense linear systems using Cholesky factorization.
This allows the client code to not worry about the type of dense
linear algebra library being used.
DenseNormalCholeskySolver and DenseSchurComplementSolver code
is considerably simpler as a result.
Change-Id: Ie15f09ee376d5f9a64609e6a55ad83e99c76352a
AutoDiffManifold allows the user to define a templated
functor that implements the Plus and Minus operations
on the Manifold and will compute the Jacobians needed
to define the Manifold object using automatic differentiation.
Change-Id: Ibd073c25847389308ca1ab66e6f5fe78aae77205
Manifolds are now part of the public API and co-exist
with LocalParameterizations.
1. Add Manifolds to the Problem API.
a. AddParameterBlock(double*, int, Manifold*)
b. SetParameterization(double*, Manifold*)
b. GetManifold(const double*)
c. HasManifold(const double*)
2. Internally Ceres now only uses Manifolds. When the user uses
a LocalParameterization, it is wrapped in a ManifoldAdapter.
3. To preserve the API semantics while keeping the internals clean
we need a new map in ProblemImpl which stores the association
between parameter blocks and local parameterizations. This
is temporary, it will go away once this transition is complete.
4. There are NO algorithmic changes, as in we are not using
any of the expanded interface of the Manifold objects yet.
That will come later.
5. All tests that use LocalParameterization have been duplicated
to use Manifolds, and when this transition is complete the
LocalParameterization based tests will be deleted.
6. Public documentation for the API has been updated. Deprecation
notices to the documentation as well as C++ annotations will come
later.
7. Similar changes have been made to GradientProblem.
Change-Id: I8e03c8ced6e141876ef3eca5740c113afa788f0c
Joydeep Biswas