2560ddef99
Fix building with MKL on Mac OS
2014-10-14 00:04:28 -04:00
5c68e9e755
silence unused variable warning on Mac
2014-10-13 15:48:17 -04:00
fdf1dfffa8
Now in nanoseconds, and added timing of localCoordinates.
2014-10-12 18:21:17 +02:00
3c42fbacb9
Update tests for testPoseRotationPrior
...
Adds new test for wraparound handling in Rot2.
Removes tests that are not quite equal for Rot3.
2014-10-09 14:13:24 +11:00
2000dbb868
Merged in pose_rotation_prior_logmap_fix (pull request #20 )
...
Fixed PoseRotationPrior error calculation.
2014-10-07 21:12:42 +02:00
7c8716f746
Fixed PoseRotationPrior error calculation.
...
Use measured_.localCoordinates not difference between Logmap
values to handle wraparound properly.
2014-10-07 11:33:46 +11:00
0c7368f408
Merge branch '3DInitialization-2'
2014-10-06 17:06:57 -04:00
d57ca93b7a
Merge branch 'develop'
...
Conflicts:
.cproject
examples/Pose3SLAMExample_g2o.cpp
examples/Pose3SLAMExample_initializePose3Chordal.cpp
examples/Pose3SLAMExample_initializePose3Gradient.cpp
gtsam/slam/InitializePose3.cpp
gtsam/slam/InitializePose3.h
gtsam/slam/tests/testInitializePose3.cpp
2014-10-06 17:02:08 -04:00
f3bcf305c2
merge with smartFactor branch
2014-10-06 14:45:50 -04:00
94fcf69106
Merge branch 'develop'
2014-10-06 14:37:51 -04:00
f99a28d3bd
merge with LevenbergMarquardt brank (no changes)
2014-10-06 14:30:28 -04:00
e041b9de6f
Merge branch 'develop'
2014-10-06 13:18:19 -04:00
07967d0b6e
merge with lago branch
2014-10-06 13:15:03 -04:00
98c507d70d
Merge branch 'feature/GPSFactor'
2014-10-06 13:12:29 -04:00
7a4f307342
moved biasedGPS and GaussMarkov to unstable
2014-10-06 13:11:40 -04:00
9662f8670d
Merge branch 'develop'
2014-10-06 13:00:03 -04:00
495fe90a41
Merge branch 'develop'
2014-10-06 12:10:47 -04:00
b4e225f1b8
Merged in feature/catkin-compatibility (pull request #19 )
...
Adding a minimal version of catkin compatibility
2014-10-02 18:00:50 +02:00
7ce2ec545c
Adding a minimal version of catkin compatibility
2014-10-02 16:51:29 +02:00
9bb6beed92
Fix OpenMP configuration when MKL is enabled
2014-09-30 16:13:15 -04:00
f049c45863
Handle SVD sign ambiguities. Fixes issue #112
2014-09-30 15:27:19 -04:00
4fecf8a172
Add path for new MKL installation folder in latest parallel studio release
2014-09-30 13:09:13 -04:00
dd33344744
Fixed unit test - might be platform specific though :-(
2014-09-30 12:04:38 +02:00
e408159545
Merged in feature/allowMap (pull request #18 )
...
Replaced range adaptor pattern with more explicit loop to create dimensions, which also allows maps to be supplied as TERMS containers.
2014-09-30 11:33:35 +02:00
34dcfa4e89
Added example from Pull Request #18 discussion
2014-09-30 11:31:17 +02:00
156c1abf4e
Added reserve as suggested by Richard
2014-09-30 11:27:19 +02:00
7c4a557011
Latest version of MKL no longer defines MKL_BLAS so I added this as a fix, following http://eigen.tuxfamily.org/bz/show_bug.cgi?id=874
2014-09-29 13:49:37 +02:00
0d05a1c206
Some doxygen documentation
2014-09-26 17:41:43 +02:00
78d9a8cab8
fix empty factors bug
2014-09-25 11:18:51 -04:00
1013ba83c9
Updated examples to conform to gtsam document. Re-factored iSAM example a bit.
2014-09-25 14:30:41 +02:00
11969b32f6
fix gradient and gradientAtZero for linear constrained Jacobian, optionally scaled by the dual variables
2014-09-24 22:19:47 -04:00
e0a9eb63ce
Fix for Windows: renamed some tests
2014-09-24 16:47:07 -04:00
968bea0a58
removed odprint which used varargs to fix issue 91
2014-09-24 14:56:38 -04:00
4a1d04a4d0
Applying patch from https://bitbucket.org/eigen/eigen/commits/d336163ce058/ to fix serialization on Mac with Boost 1.56.
...
https://forum.kde.org/viewtopic.php?f=74&t=122194
Matlab still crashes, however.
2014-09-22 16:51:09 -04:00
bc5f1c4a46
upgrade to Eigen 3.2.2
2014-09-22 16:20:53 -04:00
d34ba9c8bf
The range adaptor scheme did not work for std::map TERMS in creating a JacobianFacor. Hence, I removed it - which I think is more readable - and replaced it with an explicit creationg of dimensions. I also added a test with std::map, which works.
2014-09-21 13:40:38 +02:00
81645cb1f4
Added list_of.hpp dependence and fixed header order in doing so...
2014-09-21 13:38:12 +02:00
3046d52d6e
Fix bugs in metis re building on Windows. Mostly version issues were new c math functions were redefined poorly in gklib and metis.
2014-09-18 10:45:01 -04:00
f9f6fadba4
Updating code documentation per issue #114
2014-09-15 11:52:40 -04:00
be68cc0047
add given marginals to update_noiseModel to improve speed
2014-09-15 03:03:25 -04:00
43a8de01c1
Merge remote-tracking branch 'origin/feature/quadratic_programming' into feature/quadratic_programming
...
Conflicts:
.cproject
.gitignore
gtsam/linear/GaussianFactorGraph.cpp
gtsam/linear/GaussianFactorGraph.h
gtsam/linear/HessianFactor.cpp
gtsam/linear/JacobianFactor.h
gtsam/linear/QPSolver.cpp
gtsam/linear/tests/testQPSolver.cpp
gtsam/nonlinear/NonlinearFactor.h
2014-09-14 01:45:27 -04:00
1a7653730c
fix headers for TransBtwRobotFactorEM
2014-09-13 18:12:28 -04:00
df1eede30c
Produce a graph of dual-scaled Hessians of each factor: -lambda*H, used for solving nonlinear equality constraints with SQP.
2014-09-13 01:37:17 -04:00
20fb8ab77d
Build a dual graph to compute dual values for equality constrained factors
2014-09-13 01:36:04 -04:00
4ca9d5757f
Support optional dual key for constrained Jacobian and nonlinear factors. Default boost::none for unconstrained factors.
2014-09-13 01:34:33 -04:00
5e8c36b3ca
compute gradient wrt a key
2014-09-13 01:31:49 -04:00
0dcb3e209d
add macro TEST_DISABLED for disabling tests without commenting them out
2014-09-13 01:29:25 -04:00
3ad83e6394
small improvements
2014-09-11 20:54:46 -04:00
6d697f2c92
heuristic in EliminationPreferCholesky to work around the Indeterminant exception while solving linear constrained systems.
...
Instead of turning Hessian factors into Jacobian factors -- so that they can be eliminated with constrained Jacobian factors using the special QR in Constrained's noise model -- we combine all Hessian factors, eliminate the variable first to have a conditional and a new factor 1, then combine the constrained Jacobians with this conditional (also a Jacobian) to eliminate again, producing the final conditional, and a new factor 2. The two new factors are then combined into a new Hessian factor to be returned.
2014-09-11 14:51:35 -04:00
661862f3e3
small comment
2014-09-11 12:08:42 -04:00
cbeall3
dellaert
Lachlan Toohey
Luca
Paul Furgale
thduynguyen
Andrew Melim
Jing Dong