diff options
Diffstat (limited to 'extern/libmv/third_party/ceres/ChangeLog')
| -rw-r--r-- | extern/libmv/third_party/ceres/ChangeLog | 884 |
1 files changed, 445 insertions, 439 deletions
diff --git a/extern/libmv/third_party/ceres/ChangeLog b/extern/libmv/third_party/ceres/ChangeLog index c9f5b3426bc..cd168a44b35 100644 --- a/extern/libmv/third_party/ceres/ChangeLog +++ b/extern/libmv/third_party/ceres/ChangeLog @@ -1,646 +1,652 @@ -commit 8c62487e437b91d3d354cd1ae8957e43fe540732 +commit d3ecd18625ba260e0d00912a305a448b566acc59 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri May 16 09:44:18 2014 -0700 +Date: Tue Sep 23 10:12:42 2014 -0700 - Preparations for 1.9.0 release. + Add an explicit include for local_parameterization.h - Version bump. - minor docs update. + Thanks to cooordz for reporting this. - Change-Id: I2fbe20ba4af6b2e186fe244c96ce6d6464fe0469 + Change-Id: I7d345404e362a94ff1eb433ad6b9dcc4960ba76d -commit 0831275a78ab65e4c95979598cb35c54d03d3185 -Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri May 16 08:17:54 2014 -0700 +commit 5dd76869cf45122c79579423f09e0de08cf04092 +Author: Alex Stewart <alexs.mac@gmail.com> +Date: Fri Sep 19 16:08:25 2014 +0100 - Documentation update. + Fix unused-function warning with Eigen < 3.2.2. - 1. Update iOS build instructions. - 2. Update version history. + - CreateBlockJacobian() is only ever used when Eigen >= 3.2.2 is + detected, but was previously defined whenever CERES_USE_EIGEN_SPARSE + was defined with no check on the Eigen version. + - This resulted in an unused-function compile warning that became an + error due to -Werror, preventing compilation when using Eigen < 3.2.2. - Change-Id: I49d62e86ecff39190b50c050cb12eef4e2773357 + Change-Id: I24628ff329f14b087ece66bf2626bdc0de4ba224 -commit c7c7458625996a20203f1366d11bd701e5fb621b -Author: Jack Feng <jackfengji@gmail.com> -Date: Mon May 12 10:23:56 2014 +0800 +commit 820cb7b14831aa03eca1e8186000cebfdf0a42f3 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Wed Sep 17 09:46:08 2014 -0700 - add support for building for ios + Add solver_utils.cc to Android.mk - use ios-cmake to build for ios - - Change-Id: I6b17c33339f3121322a4004d79629b22a62f7a94 + Change-Id: I358522971711280f4362a1fa39b1568160e21e63 -commit 36c2ce87d13b9b7123bd0473b8b45fb3b6ae4271 -Author: Sergey Sharybin <sergey.vfx@gmail.com> -Date: Mon Jan 13 21:18:08 2014 +0600 +commit 092b94970a073f8b47179d96160226fc19095898 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Fri Sep 5 11:56:29 2014 -0700 - Libmv 2D homography estimation example application + Add GradientProblem and GradientProblemSolver. - Add an example application of homography matrix estimation - from a 2D euclidean correspondences which is done in two - steps: + The line search minimizer in Ceres does not require that the + problems that is solving is a sum of squares. Over the past + year there have been multiple requests to expose this algorithm + on its own so that it can be used to solve unconstrained + non-linear minimization problems on its own. - - Coarse algebraic estimation - - Fine refinement using Ceres minimizer + With this change, a new optimization problem called + GradientProblem is introduced which is basically a thin + wrapper around a user defined functor that evaluates cost + and gradients (FirstOrderFunction) and an optional LocalParameterization. - Nothing terribly exciting apart from an example of how to - use user callbacks. + Corresponding to it, a GradientProblemSolver and its associated + options and summary structs are introduced too. - User callback is used here to stop minimizer when average - of symmetric geometric distance becomes good enough. - This might be arguable whether it's the best way to go - (in some cases you would want to stop minimizer when - maximal symmetric distance is lower than a threshold) but - for a callback usage example it's good enough to stick - to current logic. + An example that uses the new API to find the minimum of Rosenbrock's + function is also added. - Change-Id: I60c8559cb10b001a0eb64ab71920c08bd68455b8 + Change-Id: I42bf687540da25de991e9bdb00e321239244e8b4 -commit d99a3a961e4a6ff7218d0ab749da57cf1a1677bd -Author: Björn Piltz <bjornpiltz@gmail.com> -Date: Wed May 7 14:59:12 2014 +0200 +commit 6c45d6b891aac01489b478a021f99081c61792cb +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Thu Sep 11 07:48:30 2014 -0700 - Separate PUBLIC and PRIVATE library dependencies - Do not propagate 3d party libs through - IMPORTED_LINK_INTERFACE_LIBRARIES_[DEBUG/RELEASE] mechanism - when building shared libraries. SuiteSparse, lapack & co - are considered private. Glog still gets propagated since - it is part of the public interface. See documentation of - TARGET_LINK_LIBRARIES(). + Add more inspection methods to Problem. + + Problem::GetCostFunctionForResidualBlock + Problem::GetLossFunctionForResidualBlock + + are added, so that users do not have to maintain this mapping + outside the Problem. - Change-Id: If0563b0c705b102876f5190e9a86694d10f79283 + Change-Id: I38356dfa094b2c7eec90651dafeaf3a33c5f5f56 -commit 1c089e8453583876f417b214f76a5863d7694986 +commit 6ad9b8e2ae66c9009441d0f9304486ec8dfa9a6a Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Tue May 13 16:58:25 2014 -0700 +Date: Tue Sep 9 14:29:28 2014 -0700 - Notational fix in modeling.rst by William Rucklidge. + Ignore row/column blocks structure when using dynamic sparsity - Change-Id: Iffa127541380fcc32da13fe4ac474692e1e3d0ec + The row/column blocks can be huge when using dynamic sparsity. This + can result in very large memory usage when augmenting the jacobian + with the LM diagonal. + + Thanks to Mingsong Dou for reporting this. + + Change-Id: I6aa140ceefa98389ae17958f89ca76e0c76f95b8 -commit 082d9e2a1b43b26a81157a6c711de0ff34c40ba4 -Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Mon May 12 20:41:27 2014 -0700 +commit 7e43460d42e20be1ba13121655dbbfd0d1c751ae +Author: Martin Baeuml <baeuml@gmail.com> +Date: Mon Sep 8 16:49:06 2014 +0200 - Add iOS.cmake from the ios-cmake project. + Fix a few typos in the documentation. - URL: https://github.com/cristeab/ios-cmake - Commit: 86dc085f0d5ed955cd58e2657cc3efc7c1aabbc8 - - Change-Id: I8fe6023d4cb6655b5a724e8b695fdae87ce3b685 + Change-Id: I541db56b2b81ae758e233ce850d78c3cbb4b6fa3 -commit a97056c9752fe7223c8560da58862ecb1fd241ad +commit 1aef66eeae7042902655a11b0d6a1a32900abb7b Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Tue May 13 16:45:36 2014 -0700 +Date: Sun Sep 7 21:18:44 2014 -0700 - Various documentation fixes from William Rucklidge. + Remove errant space. - Change-Id: I102e98f41f4b5fe2a84d1224d5ed7517fdfdb022 + Change-Id: Iedc06960417a9b938d57f623b4beb87a98e3d081 -commit 2f8fb218f0a08102231ace07ef02b34b4aad7336 -Author: Alex Stewart <alexs.mac@gmail.com> -Date: Tue May 13 20:57:39 2014 +0100 +commit 89080ab153a33008782759187fa8e9af7d2f83f1 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Sat Sep 6 21:13:48 2014 -0700 - Adding function to update CMake cache variables and preserve help. + Add LocalParameterization::MultiplyByJacobian. - - Previously we were replicating the same two lines to update a cache - variable whilst preserving its help string. - - This commit adds a function which wraps up this common operation into - a single line. + This is needed to efficiently support LocalParameterization objects + in GradientProblemSolver. - Change-Id: Ic78a5adf5d59262bbbcec1e353ded7620391e862 + Change-Id: Ic7b715b8be694b099dc95d6707a67474297533e6 -commit 8f4dcb25f1be74a8c12c0f9eeb67b6b0755563f5 +commit d76da16f49d419ae3664ca1bdc2286c1ea78ebed Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Tue Apr 29 21:40:57 2014 -0700 +Date: Sun Sep 7 18:42:49 2014 -0700 - Documentation update. + Move some routines to solver_utils.h/cc - Update modeling.rst and solving.rst to reflect - changes to the API. + This moves a couple of routines from solver.cc into solver_utils.h/cc + so that they can also be used by the upcoming GradientProblemSolver. - Change-Id: Id1a8adfed1486f08e5fd67c5af2d29708a26490c + Change-Id: I627b32ad3dc639422aacde78a8e391459d947e99 -commit d48e7050225730f61eaef851def5b43bc439e991 -Author: Alex Stewart <alexs.mac@gmail.com> -Date: Sat May 10 08:58:58 2014 +0100 +commit cbf03ac292a0c0e9e6b7fcc1b08b67e95965922f +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Sat Sep 6 21:07:08 2014 -0700 - Configure config.h and include it from the build directory. - - - Previously we overwrote the default (empty) config.h in the source - tree with a configured config.h, generated using the current compile - options. - - This was undesirable as it could lead to inadvertant commits of the - generated config.h. + Make LineSearchMinimizer consistent with TrustRegionMinimizer. - - This patch moves the default config.h to <src>/config/ceres/internal, - separate from the other headers, thus if Ceres is compiled without - CMake this directory will now also have to be included. This - directory is _not_ added to the CMake include directories for Ceres - (thus the default config.h is never used when compiling with CMake). - - When using CMake, the generated config.h is now placed in - <build>/config/ceres/internal, which is in turn added to the include - directories for Ceres when it is compiled, and the resulting config.h - is copied to ceres/internal when installed. + Change the logic for how IterationSummary objects are added to + Summary::iterations to match the one in TrustRegionMinimizer. - Change-Id: Ib1ba45e66e383ade2ebb08603af9165c1df616f2 + Change-Id: I57851ad8294e58f83b9115cca9c24695d86ee92a -commit 11c496164ffe9809306945c2b81276efcd51533d +commit f04c32319751e1efd610acd3699bca0a6dd6c6d1 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri May 9 16:27:03 2014 -0700 +Date: Sat Sep 6 21:05:41 2014 -0700 - Fix 80cols violation in local_parameterization.h + Fix some obsolete documentation in CostFunction::Evaluate - Change-Id: I07f59baa9e4aba7c5ae028f0c144ea9ad153d49a + Change-Id: I1d7ee5c596fbf6a4d886dce5b989c8eb18af2dce -commit af3154422b63b7792ecd23b00ca1a0c003764dae -Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri May 9 05:45:03 2014 -0700 +commit 9263547c02a1807532b159c217e7acd124d3db10 +Author: Johannes Schönberger <hannesschoenberger@gmail.com> +Date: Sat Sep 6 17:26:15 2014 -0400 - Add Alex Stewart as a maintainer. + Fix CG solver options for ITERATIVE_SCHUR, which did not copy min_num_iterations - Update contributing.rst to mention Alex - as one of the people who can be added - as a reviewer. + Change-Id: If31bc53b49ec20426fd438b79b8fa1f69d11e861 + +commit b41f048256d1a8184cbe874b5a96dffa7fa4630d +Author: Martin Baeuml <baeuml@gmail.com> +Date: Fri Sep 5 15:03:32 2014 +0200 + + Remove obsolete include of numeric_diff_functor.h. - Change-Id: I30ff3e635e8c419e11e8f20394aaea5f284a10d5 + numeric_diff_functor.h was removed and does not exist anymore. + + Change-Id: I07bf04bf81142551e867b95b83a0653e11cad54c -commit ea765850685f1ff0431da5212656378fc20d3673 -Author: Alex Stewart <alexs.mac@gmail.com> -Date: Wed May 7 20:46:17 2014 +0100 - - Adding autogenerated Ceres config.h to #define Ceres compile options. - - - Previously we passed all compile options to Ceres via add_definitions - in CMake. This was fine for private definitions (used only by Ceres) - but required additional work for public definitions to ensure they - were correctly propagated to clients via CMake using - target_compile_definitions() (>= 2.8.11) or add_definitions(). - - A drawback to these approaches is that they did not work for chained - dependencies on Ceres, as in if in the users project B <- A <- Ceres, - then although the required Ceres public compile definitions would - be used when compiling A, they would not be propagated to B. - - - This patch replaces the addition of compile definitions via - add_definitions() with an autogenerated config.h header which - is installed with Ceres and defines all of the enabled Ceres compile - options. - - This removes the need for the user to propagate any compile - definitions in their projects, and additionally allows post-install - inspect of the options with which Ceres was compiled. - - Change-Id: Idbdb6abdad0eb31e7540370e301afe87a07f2260 - -commit cbf955474acf8f275b272da6ff5acd3a629cc806 -Author: Björn Piltz <bjornpiltz@gmail.com> -Date: Wed May 7 17:10:15 2014 +0200 +commit b7fb6056a717cc3c372cfb7115c527ee8bc05ddb +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Wed Sep 3 11:19:02 2014 -0700 - Fixes swapped verboselevel and condition. + Remove NumericDiffFunctor. + + Its API was broken, and its implementation was an unnecessary + layer of abstraction over CostFunctionToFunctor. - Change-Id: I296d86e6bbf415be4bfd19d6a0fe0963e3d36d74 + Change-Id: I18fc261fc6a3620b51a9eeb4dde0af03d753af69 -commit 3209b045744ea31f38d74bd9e9c8f88e605e7f76 -Author: Björn Piltz <bjornpiltz@gmail.com> -Date: Wed May 7 17:02:27 2014 +0200 +commit 175fa8ff09049110a8509409f60cee5fd52cdbe6 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Tue Sep 2 06:49:08 2014 -0700 - Fixed warning : 'va_copy' : macro redefinition - MSVC 2013 has got va_copy - Compare - http://msdn.microsoft.com/en-us/library/kb57fad8(v=vs.110).aspx - and - http://msdn.microsoft.com/en-us/library/kb57fad8.aspx. + CostFunctionToFunctor allows dynamic number of residuals. + + The code itself was perfectly capable of handling residuals, but there + was an overly strict runtime check that had to be removed. - Change-Id: If0937c76e8d250cde4b343844f3d35c980bf0921 + Thanks to Domink Reitzle for reporting this. + + Change-Id: I6a6d000a7c5203dd5945a61b4caeda1b8aeb09c9 -commit 1df2f0f5d704f0cc458cf707e2602d495979e3c6 -Author: Björn Piltz <bjornpiltz@gmail.com> -Date: Wed May 7 11:10:30 2014 +0200 +commit 70ace0d5a5601901288974fcf27919754260cf0e +Author: Johannes Schönberger <hannesschoenberger@gmail.com> +Date: Sat Aug 30 15:52:34 2014 -0400 - Removed MSVC warnings - These are warnings which show up when using Ceres. + Fix max. linear solver iterations in ConjugateGradientsSolver - Change-Id: Id1f382f46b8a60743f0b12535b5b3cdf46f988e0 + Change-Id: Ice0cef46441dbc1c121eeb42113667a46c96936f -commit eca7e1c635581834c858794e09c1e876323b7775 +commit c5d8d0680250f5eb554577d30d28fc805b03fab9 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Tue May 6 10:16:19 2014 -0700 +Date: Fri Aug 29 20:31:19 2014 -0700 - Remove BlockRandomAccessCRSMatrix. + Fix a unused function error with Eigen version 3.2.1 or less. - It is not used anywhere. + Thanks to Johannes Schoenberger for reporting this. - Change-Id: I2a8ebbdacf788582f21266825ead3f76646da29e + Change-Id: Ie17d28f2a68734a978a8c95007724bc4055de43a -commit 7088a08f5d9e04e75a5a4c3823ef7927e13ff0e4 +commit 0e1cc2a55488e4cf381833baaa3531c02ce9d69e Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Mon May 5 09:02:05 2014 -0700 +Date: Fri Aug 29 09:16:56 2014 -0700 - Fix some 80col violations and reflow the comments in cmake.in file. + Fix the build on Eigen version 3.2.1 and older. - Change-Id: I4c65c89b794845aeef69159a03350c727e2ee812 + Change-Id: I18f5cb5d42113737d7b8f78a67acee28bd5b3e08 -commit 95cce0834d5a2d72568e6d2be968a51c244c2787 +commit 5f96c62b56222f27e606f2246a8a16b6942af8d1 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Mon May 5 08:54:50 2014 -0700 +Date: Thu Aug 28 23:06:17 2014 -0700 - Remove some errant tabs. + Add Block AMD ordering for SPARSE_SCHUR + EIGEN_SPARSE. + + Ordering routines for the Schur complement when using EIGEN_SPARSE. + Also integration into SchurComplementSolver. - Change-Id: Ie1f7051e99bcb15ad068711b68a9d8f317b12ed7 + Part of this CL is also a refactoring of the block jacobian matrix + construction. + + Change-Id: I11d665cc7d4867c64190e6fed1118f4d2e13d59b -commit a536ae76dfa2dbe2bc487900b98cf6c15276c649 +commit 7344626c04d19ca1dc4871c377c4422c744b1bca Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Sun May 4 21:18:42 2014 -0700 +Date: Thu Aug 28 22:03:09 2014 -0700 - Lazily initialize the bounds arrays in ParameterBlock. - - Problems that do not use bounds do not have to pay the - price of storing bounds constraints. + Let EIGEN_SPARSE + SPARSE_NORMAL_CHOLESKY use block AMD. - Also replace the raw pointer access to the upper and - lower bounds arrays with accessors which hides the - lazy initialization from the user. + Modify SparseNormalCholeskySolver to use a pre-ordered Jacobian + matrix. - Change-Id: I0325a35de9c29f853559f891e32e7c777686e537 + Change-Id: Ib4d725d7a2d7bb94ea76dbb3a9b172784dbc8ea0 -commit 633b50b7af9841607c07133f585e131fba7de177 +commit 9f7032369ea4e432f0fb507cb6d2209741ee6946 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri May 2 22:46:20 2014 -0700 +Date: Thu Aug 28 21:46:43 2014 -0700 - Add the (2,4,8) template specialization. + Block AMD for SparseNormalCholesky + EIGEN_SPARSE. - Change-Id: I058bcebdd1725031d573404133b184d6f27dc005 + This is just the reordering routine. The integration with + SparseNormalCholesky shall happen in a subsequent CL. + + Change-Id: I39ddc32aa66b11c368faf75404850fa0ae0d2b3a -commit 5ffe06019a6c741ee7edc940ffeeceaaeabfa05d -Author: Alex Stewart <alexs.mac@gmail.com> -Date: Thu May 1 12:06:46 2014 +0100 +commit b9331cd4077100d645be22a912d5743eeda72878 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Thu Aug 28 14:42:27 2014 -0700 - Export Ceres compile definitions to targets compiled against Ceres. + Cleanup reorder_program.cc - - Previously all Ceres compile definitions were private to Ceres, that - is they were not exported to users via the CMake export mechanism. - - Now that we use compile definitions in public (installed) Ceres - headers, we need to export the Ceres compile definitions. - - If we did not do this, then the client's code 'see's' a different - version of the Ceres headers to those which were in fact compiled, - or in the case of shared_ptr, may not find the required header. + Program::SetParameterOffsetsAndIndex() was being called willy nilly. + Now the invariant is that any function that actually reorders the + program, updates the offsets and indices. - - This patch makes use of the new, in CMake 2.8.11, function: - target_compile_definitions() to export all of the Ceres compile - definitions using CMake's export functionality. - - For CMake versions < 2.8.11, we have to use the blunter instrument of - calling add_definitions() in CeresConfig.cmake (invoked by a call to - find_package(Ceres)). This is messy because it ends up adding the - Ceres compile definitions to any target declared in the user's code - after the call to find_package(Ceres). Although this should do no - harm as all of our defines are prefaced with CERES_, so any - unintentional name clashes are unlikely. + Also the logic around handling EIGEN_SPARSE has been simplified in + anticipation of the block AMD code that is forthcoming. - Change-Id: I5dea80949190eaf4fb08ea4ac568ce28c32dd4e0 + Last but not the least, num_eliminate_blocks, which is a rather + cryptic name to begin with has been replaced by the more meaningful + size_of_first_elimination_group. + + Change-Id: I77e684f699a93b53e76aa406d64f40f8704df813 -commit 0e811b0881f1f21df0ae04fd745ae4ba5189cac1 +commit 79491a3f4a3939a3cce4644da7a998b7782b963a Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Thu May 1 07:54:12 2014 -0700 +Date: Thu Aug 28 13:57:50 2014 -0700 - Fix a bug in Minimizer::RunCallbacks. + Solver::FullReport now reports build config. + + The header of Summary::FullReport now looks like - Solver::Summary::message was not being updated when the solver - terminated because of a user's iteration callback indicating - success or failure. + Solver Summary (v 1.10.0-suitesparse-cxsparse-lapack-no_openmp) - Thanks to Sergey Sharybin for reporting this. + Original Reduced + Parameter blocks 22122 22122 + Parameters 66462 66462 + Residual blocks 83718 83718 + Residual 167436 167436 - Change-Id: I27e6e5eed086920ddf765461b0159417ac79d7b3 + Change-Id: Id1b81bbf90ba412d19e2dd3687eeb9d372b72c1b -commit 31b503792611d2119bb1acb3528fc8d58c5bd029 +commit 48068c753e91d77f6c96ef2d529a27ef8ee3947c Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Wed Apr 30 15:02:38 2014 -0700 +Date: Thu Aug 28 13:03:40 2014 -0700 Lint cleanup from William Rucklidge. - Change-Id: If545f114c1a2b07edd660a3c71ecfc16ffa25e43 + Change-Id: Ie0e0aa58440be7a4f67dcd633dbb6f1bb0c051a8 -commit 15c1210a8bdf3e936b4ef600d75f0fbb70878fb5 +commit 6a51b135e6298e8ba44a58cc2b54a170ab61a82f Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Tue Apr 29 08:12:19 2014 -0700 +Date: Thu Aug 28 10:48:29 2014 -0700 - Lint cleanup from Jim Roseborough. + Fix solver_test.cc + + When Eigen is not installed, Solver::IsValid was not detecting + it correctly. - Change-Id: I53f4e0d020602443b397387b8c5908f25649403d + Change-Id: Id285a84d829a9e20bc5de663adfca66ac31e08f3 -commit b1668067f1c97520d5d28eecf2c11d2afc1b01b3 +commit 62a8d64453ee41dae56710a4eead3fadf2fe1a4e Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Tue Apr 29 08:12:19 2014 -0700 +Date: Wed Aug 27 22:54:00 2014 -0700 - Variety of changes to documentation and example code. + Expand check for lack of a sparse linear algebra library. - 1. Update version history. - 2. Minor changes to the tutorial to reflect the bounds constrained - problem. - 3. Added static factory methods to the SnavelyReprojectionError. - 4. Removed relative gradient tolerance from types.h as it is - not true anymore. + The LinearSolver factory was creating a NULL linear solver + if only Eigen's sparse linear algebra backend was available. - Change-Id: I8de386e5278a008c84ef2d3290d2c4351417a9f1 + Thanks to Michael Samples and Domink Reitzle for reporting this. + + Change-Id: I35e3a6c0fd0da2a31934adb5dfe4cad29577cc73 -commit 658407dacc351a999206980fbb3265099e50e7a3 +commit 12eb389b4ec4113a2260c1a192a1d3f8d1b6a2d3 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Mon Apr 28 13:25:17 2014 -0700 +Date: Wed Aug 27 22:18:33 2014 -0700 - Add missing template specializations to the NDK build. + Fix Eigen Row/ColMajor bug in NumericDiffCostFunction. + + If the parameter block size is 1, asking Eigen to create + a row-major matrix triggers a compile time error. Previously + we were handling the case where the number of rows in the + jacobian block was known statically, but the problem is present + when the nummber of rows is dynamic. + + This CL fixes this problem. - Change-Id: I42bb6c3bd47648050298472af80333aa900e79bf + Thanks to Dominik Reitzle for reporting this. + + Change-Id: I99c3eec3558e66ebf4efa51c4dee8ce292ffe0c1 -commit 5d7eed87b47871bc882af765188fa4fbca976855 -Author: Björn Piltz <bjornpiltz@gmail.com> -Date: Wed Apr 23 22:13:37 2014 +0200 +commit 6c25185bb1643d8d0f3d8e1a7b82a058156aa869 +Author: Alex Stewart <alexs.mac@gmail.com> +Date: Thu Aug 28 16:07:51 2014 +0100 - Suppport for MSVC DLLs. + Fix crash in Covariance if # threads > 1 requested without OpenMP. + + - Previously if options.num_threads > 1 was given to Covariance compiled + without OpenMP, a CHECK() would be triggered in program_evalutor. - Change-Id: Ibbcc4ba4e59f5bbf1cb91fe81c7d3b9042d03493 + Change-Id: Iaade4f5ed5326b0c59a7014c750c41ee026e1124 -commit c830820a5c2be0d0cecb0822f2cff8b4ffe88f36 +commit 6f89d850fb4ace0104abccf467c4fe37ad378b79 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Mon Apr 28 10:28:24 2014 -0700 +Date: Wed Aug 27 11:51:50 2014 -0700 - Add missing files to Android.mk + Further build breakage fixes. + + 1. Allow the minimum number of linear solver iterations to be zero. + 2. Fix conjugate gradients solver's iteration loop to be sane again. - Change-Id: Ibdf577c592bcde0fe5c2ce343ed8e9028b82af8f + Change-Id: I8594815fec940c2b30e28eb58ec5d8baacf13dae -commit ceb7a3beaad140762b499f9a306fd7230715941a -Author: Sergey Sharybin <sergey.vfx@gmail.com> -Date: Mon Apr 28 13:50:09 2014 +0600 +commit dd596d0f0d6d08951efc2c11a639b546db2080c6 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Wed Aug 27 11:26:50 2014 -0700 - Fix compilation error when using G++ compiler + Fix the broken build. - This compiler defines shared_ptr in std::tr1 namespace, but - for this <tr1/memory> is to be included. Further, this compiler - also does have <memory> header which confused previous shared - pointer check. + Change-Id: I083cf1cca1bf4cca956193022d450364e73f833a + +commit d906afae22b05b9b9a9a2657924f4c0bf1a9b5ea +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Mon Aug 25 22:32:38 2014 -0700 + + A number of bug fixes. - Simplified logic around defines now, so currently we've got: - - CERES_TR1_MEMORY_HEADER defined if <tr1/memory> is to be - used for shared_ptr, otherwise <memory> is to be used. - - CERES_TR1_SHARED_PTR defined if shared_ptr is defined in - std::tr1 namespace, otherwise it's defined in std namespace. + 1. Fix a build breakage in graph_test. + 2. Respect Solver::Options::min_num_linear_solver_iterations in + conjugate_gradients_solver.cc - All the shared_ptr checks are now moved to own file FindSharedPtr - which simplifies main CMakeLists. + Thanks to Johannes Schönberger for reporting these. - Change-Id: I558a74793baaa0bd088801910a356be4ef17c31b + Change-Id: Ib32e3929bf5d92dd576ae5b53d4d88797095136e -commit 02db9414fb6739857a37e268500083a0546cd0a3 +commit dab955928c6d0942d6acc5b5f1c4c11260d0767d Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Mon Apr 28 08:32:51 2014 -0700 +Date: Sun Aug 17 13:14:50 2014 -0700 - Fix the documentation for RandNormal. + Add an unweighted graph. + + Rename Graph -> WeightedGraph. + Add a new Graph class, which is cheaper to construct and + work with if the weights are not needed. - As pointed out by Jim Roseborough, this is the Marsaglia Polar - method and not the Box-Muller method. + This cuts down the cost of building the Hessian graph + significantly. - Change-Id: Id5332bcd4b4c23a3885cc296729b44eaa5edd0a8 + Change-Id: Id0cfc81dd2c0bb5ff8f63a1b55aa133c53c0c869 -commit 32530788d08c53f8d2c8a5f9bd61aa60a23d6e03 -Author: Richard Stebbing <richie.stebbing@gmail.com> -Date: Sat Apr 26 07:42:23 2014 +0100 +commit a0c282adbd268c2ad82551fab31fe1cf8d0c4282 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Sun Aug 24 22:19:03 2014 -0700 - Add dynamic_sparsity option. + Add EIGEN_STRONG_INLINE annotation to jet.h - The standard sparse normal Cholesky solver assumes a fixed - sparsity pattern which is useful for a large number of problems - presented to Ceres. However, some problems are symbolically dense - but numerically sparse i.e. each residual is a function of a - large number of parameters but at any given state the residual - only depends on a sparse subset of them. For these class of - problems it is faster to re-analyse the sparsity pattern of the - jacobian at each iteration of the non-linear optimisation instead - of including all of the zero entries in the step computation. + This improves performance when using MSVC on Windows. On GCC + there will be no effect. - The proposed solution adds the dynamic_sparsity option which can - be used with SPARSE_NORMAL_CHOLESKY. A - DynamicCompressedRowSparseMatrix type (which extends - CompressedRowSparseMatrix) has been introduced which allows - dynamic addition and removal of elements. A Finalize method is - provided which then consolidates the matrix so that it can be - used in place of a regular CompressedRowSparseMatrix. An - associated jacobian writer has also been provided. + Change-Id: I555a81ff6823c2855d64773073f75af50c48d716 + +commit 20de0a7793c574e964350a623446136889f74632 +Author: Björn Piltz <bjornpiltz@gmail.com> +Date: Mon Aug 25 17:05:54 2014 +0200 + + Fixed Malformed regex - Changes that were required to make this extension were adding the - SetMaxNumNonZeros method to CompressedRowSparseMatrix and adding - a JacobianFinalizer template parameter to the ProgramEvaluator. + I got the following error with MSVC: + Syntax error at index 9 in simple regular expression "NumGroups()": '(' is unsupported. - Change-Id: Ia5a8a9523fdae8d5b027bc35e70b4611ec2a8d01 + Change-Id: Id1952831d81d3eb5d73bbed8c311914c4c8ab51f -commit 2569076ff0bf8ffb3938da8b5df7edc4883aa053 -Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri Apr 25 23:54:48 2014 -0700 +commit ccf8aea988269841d84d746e52164d5056c67a10 +Author: Björn Piltz <bjornpiltz@gmail.com> +Date: Mon Aug 25 16:16:01 2014 +0200 - More NDK fixes. + Fixed MSVC error C2124: divide or mod by zero - Fix variable names in port.h and fix fpclassify when - using gnustl. This was tested by switching to gnustl - in the JNI build. + Alternatively, if quiet_NaN is not available on all platforms a workaround would be: + volatile double zero = 0.0; + double x = 1.0/zero; + The 'volatile' is needed to shut up "warning C4723: potential divide by 0". - Thanks to Carlos Hernandez for suggesting the gnustl fixes. - - Change-Id: I690b73caf495ccc79061f45288e416da1604cc72 + Change-Id: If2bbdab8540595aa2e0079e1eb6b6fed6d4a6ef7 -commit e55596f8860a09b12b5e1f949237f15357c1ac59 +commit 8de27be218d42b282d7f15867733ad07058b0887 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri Apr 25 16:17:19 2014 -0700 +Date: Tue Aug 19 08:22:40 2014 -0700 - Change the defaults for shared_ptr. - - By default shared_ptr is now assumed to be - in the standard <memory> header and in the - std namespace. + Fix a bug in TrustRegionPreprocessor - Previously the way the ifdefs were structured if the appropriate - variable was not defined, it would default to <t1/memory>. + TrustRegionPreprocessor was not setting Minimizer::Options::is_constrained. + This meant that the line search for bounds constraints was not being + invoked for bounds constrained problems. - The new defaults are more future proof. + And some minor lint cleanup. - Change-Id: If457806191196be2b6425b8289ea7a3488a27445 + Change-Id: I18852cfaf1b33fd90b7d8c196f2063c128126658 -commit bb05be341b8436f611e4b69954a529edcca5b577 +commit 1745dd615b3897a3ef9896acfdba67eee1739bf4 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Sun Apr 13 14:22:19 2014 -0700 +Date: Thu Jun 5 21:30:13 2014 -0700 - Solver::Options uses shared_ptr to handle ownership. + Refactor SolverImpl. - Solver::Options::linear_solver_ordering and - Solver::Options::inner_iteration_ordering - were bare pointers even though Solver::Options took ownership of these - objects. + Replace SolverImpl with - This lead to buggy user code and the inability to copy Solver::Options - objects around. + a. A minimizer specific preprocessor class. + b. A generic Solve function inside solver.cc + c. Presummarize and Postsummarize functions to handle + updates to the summary object. - With this change, these naked pointers have been replaced by a - shared_ptr object which will managed the lifetime of these objects. This - also leads to simplification of the lifetime handling of these objects - inside the solver. + The existing SolverImpl class was a mixture of the above three + things and was increasingly complicated code to follow. This change, + breaks it into its three separate constituents, with the aims of + better separation of concerns and thus better testability and + reliability. - The Android.mk and Application.mk files have also been updated - to use a newer NDK revision which ships with LLVM's libc++. + The call to Solver::Solve() now consists of - Change-Id: I25161fb3ddf737be0b3e5dfd8e7a0039b22548cd - -commit 8e0991381ea3a2baddea017cd07b333f0c5de595 -Author: Joydeep Biswas <joydeep.biswas@gmail.com> -Date: Tue Apr 22 10:40:47 2014 -0400 - - Added a simplified robotics example for DynamicAutoDiffCostFunction. + 1. Presummarize - summarize the given state of the problem and solver + options. + 2. Preprocess - Setup everything that is needed to call the minimizer. + This includes, removing redundant parameter and residual blocks, + setting up the reordering for the linear solver, creating the + linear solver, evaluator, inner iteration minimizer etc. + 3. Minimize. + 4. Post summarize - summarize the result of the preprocessing and the + solve. - Change-Id: I9520e0a9a8d9743285c5114523fbafa6ffa5b0bd + Change-Id: I80f35cfc9f2cbf78f1df4aceace27075779d8a3a -commit cc9d3bba1008066e51502cabd956985c6bdedfe8 +commit bd90384226a7f8629467f72fc410a9e8086a2dff Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri Apr 18 22:58:09 2014 -0700 +Date: Mon Aug 18 11:27:06 2014 -0700 - Remove a comment from conf.py + Lint comments from William Rucklidge. + + Also some minor refactoring of the trust_region_preprocessor_test.cc - Change-Id: I675f7e8fc5dd2143eab74901bc7241e02e37285f + Change-Id: Ica28002254c95722faf93a7ef35bf3deab557f0b -commit c4cd29dd7c80ade5b3ac7a1f6ee7df22c8869ab5 +commit 3150321db4a0cb1bb4894961a030d95dacae3591 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Wed Apr 16 23:40:12 2014 -0700 +Date: Tue Aug 12 22:46:51 2014 -0700 - Merge landing page with introduction. - - The existing introduction was a bit redundant and also - was not really an introduction. Also updated the build - instructions to reflect the new reality on Mac OSX. + Preprocessor for the LineSearchMinimizer. - Also updated the beginning of the tutorial to be a bit - gentler and updated the history to be more consistent - - Change-Id: Ife38c1949252cf9f4c6301856957f2d38365f313 + Change-Id: Ieb5dfe1c0b96ef323c1130edd0c3a8a8b2c644cc -commit 46ccfb376ac52ac159f9187e0f7384ef68c1cbdd +commit f7da411ef0d0067e269629887d64cdb769368800 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Sat Apr 12 21:56:51 2014 -0700 +Date: Thu Aug 7 14:30:33 2014 -0700 - Cleanup block_structure.h/cc + Preprocessor for the TrustRegionMinimizer. - 1. Remove obsolete Proto conversion functions. - 2. Fix a strict weak ordering bug. + 1. Base class for preprocessors. + 2. A preprocessor for problems that will be solved using + the trust region minimizer. + 3. Added sanity tests to the program reordering options + for Schur type linear solvers. + 4. Tests for the TrustRegionPreprocessor. - Change-Id: I1ce6d4b06e29cf475df1d5bd37c79f66f20f8d93 + Change-Id: I88cd926f0053bbbf2bd6b11e03ec55b8bf473cf1 -commit 7d489fdb073937ac05c0693c1902fbcb9eeb7dfc -Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Mon Apr 7 11:14:51 2014 -0700 +commit 54893ba523106e38ab06eb72fb5d8748685c7797 +Author: Alex Stewart <alexs.mac@gmail.com> +Date: Mon Aug 11 19:04:18 2014 +0100 - Refactor the landing page to be a bit more compact. - - Also minor changes to the introduction. + Add missing #include of <limits> for loss functions. - Change-Id: Iaa71f576b95c869f075d6837dbb60ba4bb608ee7 + Change-Id: Id632451429e03031a1533a9be795270debc70706 -commit 406ac7816730c15425db20d994ac0d60d932ab6c -Author: Keir Mierle <mierle@gmail.com> -Date: Mon Apr 7 08:36:07 2014 +0000 +commit 4a2a888905fd1ce7203e45df15762d52740bb240 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Thu Aug 7 11:48:03 2014 -0700 - Rework Ceres documentation as new website + Change ownership of pointers in Minimizer::Options. - This reworks the Ceres Sphinx documentation such that it can - function as the main Ceres website, now hosted at - ceres-solver.org. This also changes to the theme sphinx_rtd_theme - used by Read The Docs; this theme has strong mobile support and is - well enough designed. + This is a intermediate change to clean things up + in preparation for a broader refactoring of the SolverImpl. - Change-Id: I63232d985859a6dac94ff58f08bf81eb2b9e7f99 - -commit 3e60a998ac970da659d590bac2ff892ee619aa1b -Author: Richard Bowen <rsbowen@google.com> -Date: Tue Apr 1 16:22:49 2014 -0700 - - Added support and tests: row and column blocks for sparse matrix - transpose. + Essentially we are replacing raw pointers in Minimizer::Options + with shared_ptr objects. For now this only makes things a bit + more complicated looking inside solver_impl.cc, but going + forward this will lead to considerable simplifications in + tracking ownership of various pointers. - Change-Id: Ife641b08a9e86826478521a405f21ba60667f0e8 + Change-Id: I21db8fc6763c29b0d15e834d7c968a0f514042a0 -commit 5ecb1c3f1dfde6e8ed4b493eafef7b43dad19e72 +commit 0d4e3bd664d442b700fee2895c7a8ac37717dc08 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Tue Apr 1 09:20:35 2014 -0700 +Date: Thu Aug 7 12:19:10 2014 -0700 - Add Problem::IsParameterBlockPresent. + GradientCheckingProblem's parameter blocks are initialized correctly. + + Ensure that when a new problem object is constructed for validing + gradients, the parameter blocks have their data pointers point to + the user's parameter blocks. - This allows the user to query the Problem to see if a - parameter block is already present or not. + We used to do this inside solver_impl.cc, but doing this at + construction is the right thing to do. - Change-Id: If786f6c008cc644f3398597901d718d12a6d865d + Change-Id: I3bfdc89bb0027c8d67cde937e8f2fa385d89c30c -commit 75e2232b29ff2ea42c8406c9d45b138a7e7a0048 +commit cfb36463f9c1f806121779d651c7105ad899bb20 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Fri Mar 28 11:21:02 2014 -0700 +Date: Tue Aug 5 14:42:33 2014 -0700 - Fix spacing in building.rst + Small fixes from William Rucklidge. - Change-Id: I4c68d732c80d7ff2bdbc812bf0b7c7fb98c43957 + Change-Id: I0be52f0f1e53cedccffe4807dc664a2f3fb4a8e4 -commit b555b489b8447434294a8a6676272289140d6a1d -Author: Richard Bowen <rsbowen@google.com> -Date: Thu Mar 27 15:51:28 2014 -0700 +commit 9a41132a0523af407b53644c07900f86aa6fceac +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Mon Aug 4 22:45:53 2014 -0700 - Changes documentation to reflect changes in output format. + Small changes from Jim Roseborough. - Change-Id: Ic0ba234283e791edcad29aec067905dcb2130813 + Change-Id: Ic8b19ea5c5f4f8fd782eb4420b30514153087d18 -commit 1cfb600bfc3be8342f85f155b2b219a595ee58da +commit a521fc3afc11425b46992388a83ef07017d02ac9 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Wed Mar 26 12:55:11 2014 -0700 +Date: Fri Aug 1 08:27:35 2014 -0700 - Add the (2,4,9) template specialization for PartitionedMatrixView - and SchurEliminator. + Simplify, cleanup and instrument SchurComplementSolver. + + The instrumentation revealed that EIGEN_SPARSE can be upto + an order of magnitude slower than CX_SPARSE on some bundle + adjustment problems. + + The problem comes down to the quality of AMD ordering that + CXSparse/Eigen implements. It does particularly badly + on the Schur complement. In the CXSparse implementation + we got around this by considering the block sparsity structure + and computing the AMD ordering on it and lifting it to the + full matrix. - Also update the comment inside generate_partitioned_matrix_view_specializations.py + This is currently not possible with the release version of + Eigen, as the support for using preordered/natural orderings + is in the master branch but has not been released yet. - Change-Id: I99a7ab4256091b1da48553da3076e5996a5757ed + Change-Id: I25588d3e723e50606f327db5759f174f58439e29 -commit 195e49351b386ffc23020d406883eaa6511e29b3 -Author: Alex Stewart <alexs.mac@gmail.com> -Date: Wed Mar 26 11:36:11 2014 +0000 +commit b43e73a03485f0fd0fe514e356ad8925731d3a81 +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Fri Aug 1 12:09:09 2014 -0700 - Check validity of residual block before removal in RemoveResidualBlock. + Simplify the Eigen code in SparseNormalCholeskySolver. - - Breaking change: Problem::Options::enable_fast_parameter_block_removal - is now Problem::Options::enable_fast_removal, as it now controls - the behaviour for both parameter and residual blocks. + Simplifying some of the template handling, and remove the use + of SelfAdjointView as it is not needed. The solver itself takes + an argument for where the data is actually stored. - - Previously we did not check that the specified residual block to - remove in RemoveResidualBlock actually represented a valid residual - for the problem. - - This meant that Ceres would die unexpectedly if the user passed an - uninitialised residual_block, or more likely attempted to remove a - residual block that had already been removed automatically after - the user removed a parameter block upon on which it was dependent. - - RemoveResidualBlock now verifies the validity of the given - residual_block to remove. Either by checking against a hash set of - all residuals maintained in ProblemImpl iff enable_fast_removal - is enabled. Or by a full scan of the residual blocks if not. + The performance of SparseNormalCholesky with EIGEN_SPARSE + seems to be on par with CX_SPARSE. - Change-Id: I9ab178e2f68a74135f0a8e20905b16405c77a62b + Change-Id: I69e22a144b447c052b6cbe59ef1aa33eae2dd9e3 -commit 74762b60332d4a1c08ec5aef75ec718da9d305a2 -Author: Alex Stewart <alexs.mac@gmail.com> -Date: Thu Mar 20 14:50:25 2014 +0000 +commit 031598295c6b2f061c171b9b2338919f41b7eb0b +Author: Sameer Agarwal <sameeragarwal@google.com> +Date: Thu Jul 17 14:35:18 2014 -0700 - Allow construction of an AutoDiffLocalParameterization with a functor. + Enable Eigen as sparse linear algebra library. + + SPARSE_NORMAL_CHOLESKY and SPARSE_SCHUR can now be used + with EIGEN_SPARSE as the backend. + + The performance is not as good as CXSparse. This needs to be + investigated. Is it because the quality of AMD ordering that + we are computing is not as good as the one for CXSparse? This + could be because we are working with the scalar matrix instead + of the block matrix. + + Also, the upper/lower triangular story is not completely clear. + Both of these issues will be benchmarked and tackled in the + near future. + + Also included in this change is a bunch of cleanup to the + SparseNormalCholeskySolver and SparseSchurComplementSolver + classes around the use of the of defines used to conditionally + compile out parts of the code. - - Previously AutoDiffLocalParameterization would internally instantiate - a functor instance whenever one was required. This prohibits the - user passing arguments to the constructor of the functor. - - Now AutoDiffLocalParameterization can take over ownership of an - allocated functor which the user created. This mimics the behaviour - of AutoDiffCostFunction. + The system_test has been updated to test EIGEN_SPARSE also. - Change-Id: I264e1face44ca5d5e71cc20c77cc7654d3f74cc0 + Change-Id: I46a57e9c4c97782696879e0b15cfc7a93fe5496a -commit 4f603fb0d82317a53fa9d96abe6a97b2e69bff36 +commit 1b17145adf6aa0072db2989ad799e90313970ab3 Author: Sameer Agarwal <sameeragarwal@google.com> -Date: Wed Mar 19 17:16:43 2014 -0700 +Date: Wed Jul 30 10:14:15 2014 -0700 - Grammer fixes from William Rucklidge. + Make canned loss functions more robust. + + The loss functions that ship with ceres can sometimes + generate a zero first derivative if the residual is too + large. + + In such cases Corrector fails with an ugly undebuggable + crash. This CL is the first in a series of fixes to + take care of this. + + We clamp the values of rho' from below by + numeric_limits<double>::min(). + + Also included here is some minor cleanup where the constants + are treated as doubles rather than integers. + + Thanks to Pierre Moulon for reporting this problem. - Change-Id: Ia40df7a1d141eb2552694510453d1431bb0c8dce + Change-Id: I3aaf375303ecc2659bbf6fb56a812e7dc3a41106 |