Commit Graph

6309 Commits (a4f8ead2a68c30eb5151f505e2cb2f0685cd0ce6)

Author SHA1 Message Date
thduynguyen 4730527481 testSerializationLinear passes. Fill in the lower triangle part of matrix_ so that boost::serialization won't throw the input_stream exception due to uninitialized data 2014-05-08 11:49:21 -04:00
djensen3 0f03cd736c added comments 2014-05-08 11:08:43 -04:00
Richard Roberts d725dd4816 Renamed unit tests that still had 'Unordered' in the name 2014-05-07 18:19:11 -07:00
cbeall3 fec73365d7 Warning for Clang 5.0+ 2014-05-07 10:42:15 -04:00
Chris Beall b108e1c94f Upgrade to Eigen 3.2.1 - minor fixes, Richard's patches carried forward 2014-05-07 00:04:22 -04:00
Chris Beall 1062e41de3 shut up some warnings on Windows 2014-05-06 22:57:26 -04:00
Frank Dellaert 3e3e919718 Merged in feature/update_example_project (pull request #14)
Better and up-to-date example project
2014-05-06 22:18:16 -04:00
Richard Roberts fe3ecc9f97 Fixed another comment 2014-05-06 19:14:03 -07:00
Richard Roberts bd6b7a63d9 Fixed comment 2014-05-06 19:12:59 -07:00
Richard Roberts d45fe95b8d Better and up-to-date example project 2014-05-06 19:05:59 -07:00
cbeall3 842554a230 Unit test fixes for quaternion mode and also rotation matrix + full expmap mode. 2014-05-06 20:03:45 -04:00
dellaert 2649b0fd7a print variants for KeyList, KeyVector 2014-05-06 13:21:49 -04:00
cbeall3 9395bb0913 typos 2014-05-06 12:31:03 -04:00
cbeall3 ba8db019da Adding unit test for skewSymmetric because there's a strange warning elsewhere 2014-05-06 12:06:07 -04:00
cbeall3 0450c58d66 silence warning 2014-05-06 12:06:07 -04:00
Frank Dellaert a3b16dfa2b Merged in feature/BNincompleteSolve (pull request #13)
Optimize an incomplete BayesNet
2014-05-06 11:09:32 -04:00
dellaert 3f125b46df New method merge (mainly for MATLAB) 2014-05-06 11:08:33 -04:00
dellaert 108357992c Optimize an incomplete BayesNet 2014-05-06 02:18:52 -04:00
dellaert 9373711a85 Finish TSAMFactors 2014-05-06 01:34:56 -04:00
dellaert 6c6c545d99 Comments 2014-05-06 01:33:09 -04:00
dellaert 779d6ad2af Added utilities to Contents.m 2014-05-06 00:23:34 -04:00
dellaert cea4aef9f2 New perturbPose2 utility 2014-05-06 00:23:20 -04:00
cbeall3 2f180d1d02 Remove using namespace std from header files 2014-05-05 16:35:22 -04:00
cbeall3 cf77189d71 Start a DEVELOP document and update INSTALL 2014-05-05 14:42:46 -04:00
dellaert 3b1f947909 Renamed to test 2014-05-05 11:29:02 -04:00
dellaert 9409357fe7 Works, but is really a test, not an example 2014-05-05 11:26:16 -04:00
dellaert f5e0a7f2b1 ignore 2014-05-05 10:15:23 -04:00
dellaert 0a2385711b MATLAB wrapping of TSAMFactors 2014-05-05 10:14:56 -04:00
dellaert 05c1e572b6 Moved TSAMFactors.h from tsam to gtsam 2014-05-05 10:14:36 -04:00
Luca 5c420f1aec added references 2014-05-04 18:16:40 -04:00
Frank Dellaert 0a5690dfb3 Merged in feature/UnifiedProjectionModel (pull request #11)
Unified projection model
2014-05-04 15:51:18 -04:00
Chris Beall 0704e82190 Fix GCC 4.6+ specific warning 2014-05-03 19:17:52 -04:00
thduynguyen 0fc1dc3658 add detailed comments for the new EliminatePreferCholesky 2014-05-03 18:08:33 -04:00
thduynguyen 3804057daf disable two warning options in METIS which are not understood by my clang compiler. 2014-05-03 18:05:55 -04:00
thduynguyen 1e3ae3b3d3 Support non positive definite Hessian factors while doing EliminatePreferCholesky with some constrained factors.
Currently, when eliminating a constrained variable, EliminatePreferCholesky converts every other factors to JacobianFactor before doing the special QR factorization for constrained variables. Unfortunately, after a constrained nonlinear graph is linearized, new hessian factors from constraints, multiplied with the dual variable  (-lambda*\hessian{c} terms in the Lagrangian objective function), might become negative definite, thus cannot be converted to JacobianFactors.

Following EliminateCholesky, this version of EliminatePreferCholesky for constrained var gathers all unconstrained factors into a big joint HessianFactor before converting it into a JacobianFactor to be eliminiated by QR together with the other constrained factors.

Of course, this might not solve the non-positive-definite problem entirely, because (1) the original hessian factors might be non-positive definite and (2) large strange value of lambdas might cause the joint factor non-positive definite [is this true?]. But at least, this will help in typical cases.
2014-05-03 18:04:37 -04:00
Chris Beall 2daeae2438 GeographicLib now works on Windows 2014-05-03 17:26:07 -04:00
jing f27cb3f317 few style bugs fixed 2014-05-03 17:04:07 -04:00
dellaert 219b2b4db7 Updated to 3.0 2014-05-03 15:50:38 -04:00
cbeall3 a6ae176a54 Upgraded to GeographicLib 1.35 2014-05-03 14:44:39 -04:00
Chris Beall 46859474cb Fix warning 2014-05-03 12:55:51 -04:00
dellaert 92c9e9c0aa Fixed "unused" warnings 2014-05-03 12:23:49 -04:00
dellaert 1ed07ca4ed Moved to unstable 2014-05-03 12:23:20 -04:00
cbeall3 267b86ec14 Remove svd inverse_square_root 2014-05-03 11:54:01 -04:00
cbeall3 312d3fac1e ignore /doc* 2014-05-03 11:53:10 -04:00
Luca e8a4767591 cleaned code to log LM iterations 2014-05-02 16:10:51 -04:00
Luca 9f499d2257 fixed key in unit test 2014-05-02 16:05:08 -04:00
Luca c1650f72fd added comments 2014-05-02 15:23:19 -04:00
Luca f1fb2c09a4 bug fix in LM 2014-05-02 15:01:42 -04:00
djensen3 5402f9fa20 Removed smartProjectionPoseFactor from matlab wrapper; it was causing matlab crashes 2014-05-02 11:12:10 -04:00
thduynguyen fc1f5ff6a8 unittest for QPSolver without initial point 2014-05-02 10:27:05 -04:00