This has a measurable impact on interpolation performance.
Also remove an accidentally named enum with an anonymous enum.
Change-Id: Ied6a4b2b06bb27a7f004bd0e01353742e1f84034
The key change is that there is a new layer of abstract,
a Array object that the interpolator depends on.
The Array provides a one dimension or two dimensional
array like interface independent of the underlying representation
of the data.
Also included here is support for vector valued functions.
Change-Id: Ica68f03778cf0d84192db00cd55653f8b4124d51
Make the definition of CERES_NO_THREADS conditional on the presence
of the CERES_HAVE_PTHREADS define.
This allows the user to enable threading in ceres using a combination
of CERES_EXTRA_DEFINES and linker flags which they need to do in
their app.
Change-Id: I43b99fc829001166e2f6dfb0a770768e61470591
Delete code needed by old versions of the NDK. We do not build
with these versions of the NDK and do not use STLPort anymore.
Change-Id: I61092db0aa3980cfae6ff57f3f318482027e627f
Due to floating point and conditioning issues, a system
matrix which is guaranteed to be PSD in exact arithmetic
can appear indefinite to the ConjugateGradientsSolver.
Previously, x'Ax <= 0, the solver returned with numerical
failure. Which the trust region solver will treat as a failed
solve.
But, more general truncated Newton when they encounter indefiniteness
use the step computed till that point instead of declaring failure.
This changes does this and adds a bit more logging.
Change-Id: I0e0cc56ef7d856f1c54ac6d638327b8353039f70
The example code for DynamicNumericDiffCostFunction and
DynamicAutoDiffCostFunction in the documentation was
allocating the cost function objects on the stack rather
than the heap.
Thanks to Rodney Hoskinson for reporting this.
Change-Id: I217aee02d1e2c1e9c25b8197451e9f9e0482915d
Example code demonstrates how a sampled function can be
minimized. Also, in the process uncovered some deficiencies
in the CubicInterpolator and BicubicInterpolator interfaces and
fixed them.
Change-Id: I18c8f670fbee076bf1e94d1f45c7477fd71640e8
This bi-cubic interpolation implementation is based
on the cubic convolution algorithm of keys, which allows
us to implement a bi-cubic spline like interpolation scheme
using five one dimensional cubic spline operations.
Change-Id: I116aa8036191c3e654af788323fc8298ae8252a6
Add a cubic interpolator based on the Catmull-Rom spline,
with support for automatic differentiation.
Change-Id: I02ae4c4ea37805ff1f717b05ea805989b474bd59
The previous commit broke levenberg_marquardt_strategy_test due
a change in the logging string being used.
Change-Id: I6ea69c0bba195c3e7d457d509b9a6605458d69b4
When a user requests an operation on a parameter block which has
not been added to the problem, we now display a more helpful
error message with more context.
Change-Id: Ia7e54815111eb6375b31ed0d1cddca5917138ae1
- Results of find_program() are by default visible in the CMake GUI
(cache), but the user has no reason to care in general about the
location of the Homebrew executable detected.
- Note it is still available in the GUI, but only via toggling to the
advanced display.
Change-Id: Ifb0e54c00d350cbfb3b2717d889570cab61cce54
- As part of adding support for detection of non-standard Homebrew
install on OSX, an Eigen specific search directory was added to the
search path in the main CMakeLists. This commit moves that search
into FindEigen.cmake.
Change-Id: Ia6142211961a0d1fdddeea77b83438e4ef94434f
- Call through to Homebrew on OSX to determine it's install root
in case it is not /usr/local and add the result to the CMake
prefix path that is searched for all find_path/library calls.
- This should allow Ceres to compile even when Homebrew is installed
in a non-standard location.
Change-Id: I230a5e12aef54617567bdfd20c4fd45c5a04a8bf
For historical reasons we had a "using namespace std;" in port.h. This
is generally a bad idea. So removing it and along the way doing a bunch
of cpplint cleanup.
Change-Id: Ia125601a55ae62695e247fb0250df4c6f86c46c6
- Previously we had no defined default value for
sparse_linear_algebra_library_type in Solver::Options if Ceres
was compiled with no sparse library available. Thus in that case,
the default value (dependent upon the compiler) would indicate that
one was available.
- Now we have an explicit option that means no sparse library is
available, which is now the default value in Solver::Options in this
case.
- Add a warning in CMake when the user disables all sparse libraries.
- Fix typos in trust_region_preprocessor_test:
(SUITE/CX)_SPARSE -> (SUITE/CX)SPARSE that induced failures when
no sparse libraries were available.
Change-Id: I869c399a12d42bfc44220cbb25ce6d6dd80236bd
The line search used by the trust region minimizer when enforcing
the bounds constraints starts by using the trust region step
as the line search direction and if that fails, uses the gradient
as the fallback.
The problem with this logic is that the calling code only sees
whether one of the line searches succeeds or not. It does not see
that the fallback happened. So if the fallback line search suceeeds
it still thinks that the line search direction was the trust region
step. This is clearly wrong.
This change, removes the broken fallback logic. This has no effect
on current solution quality as it stands.
Change-Id: Ibc8edd98f77c782ec4708d1e66eaa76d6867b990
- If compiling without glog (but with gflags) on OSX, unistd.h is
required for close() et al, when using glog this was pulled in
indirectly.
Change-Id: I8f0807d98479e386921fb48da30683d027d4bc61
- At version 2.1, gflags changed from using the google namespace, to
using gflags by default. However, it can be configured at build time
to be something else (which would be google for legacy compatibility
unless you were evil).
- Ceres previously assumed that gflags was in the google namespace.
- Now, FindGFlags.cmake extracts the namespace when gflags.h is found
and saves it in GFLAGS_NAMESPACE.
- When building the tests and examples that require gflags,
CERES_GFLAGS_NAMESPACE is defined to be the detected namespace, and
all tests/examples now use CERES_GFLAGS_NAMESPACE:: instead of
google:: when calling gflags functions.
Change-Id: Ia333df7a7e2f08ba9f26bbd339c3a785b88f04c4
- We now compute & report the cumulative time spent performing the
following tasks as part of a line search:
- Evaluation of the univariate cost function value & gradient.
- Minimization of the interpolating polynomial.
- Total time spent performing line searches.
- This information is now reported for all minimizers, although only in
the case of a constrained problem for the TR minimizer.
- Remove LineSearch::Function abstraction in place of using
LineSearchFunction implementation directly, and remove virtual
functions from LineSearchFunction.
-- LineSearch::Function added an unnecessary level of abstraction since
the user always had to create a LineSearchFunction anyway to use a
Ceres Evaluator, and it added an unncessary virtual function call.
Change-Id: Ia4e1921d78f351ae119875aa97a3ea5e8b5d9877
Logging was happening unconditionally and that is rather jarring
in certain circumstances, e.g., when ceres is being called
inside a RANSAC loop.
Thanks to Bryan Klingner for reporting this.
Change-Id: I97a0b6c48df51f5f7362eb9d46654712ef044a00
Since the trust region minimizer can use the line search
minimizer when it is solving a box constrained problem,
ensure that the line search options are valid.
Also some minor spacing fixes in the line search code.
Change-Id: Ife04204855cfac389cf980f0a79155d4accc8662
Sameer Agarwal