4a3dc51f85
more tests work, except for serialization based tests
2014-10-24 23:47:02 +02:00
e7ec6b3fa5
Fixed size
2014-10-23 12:04:16 +02:00
49ff33602d
Undid change
2014-10-23 01:44:04 +02:00
0501750c7c
Fixed accuracy and size issues in Quaternion mode
2014-10-22 22:54:25 +02:00
48a6777935
Some refactoring
2014-10-22 16:26:40 +02:00
0f53c8d5ec
Timing of Ceres AutoDiff adaptor
2014-10-22 12:49:18 +02:00
f44e6f0187
Moved AdaptAutoDiff template in its own header file
2014-10-22 11:41:04 +02:00
516bb4b0b1
Isolated Snavely example
2014-10-22 11:23:35 +02:00
439f51ec7f
test out invoke
2014-10-22 10:10:58 +02:00
1eb5e185e5
New numericalDerivatives with traits an Charts - still some segfaults, *and* there should be no need for (a) multiple prototypes to match against c++ pointers, (b) the use of explicit template arguments. A task for someone...
2014-10-21 18:50:52 +02:00
224b71d696
Created testManifold
2014-10-21 12:54:28 +02:00
0acffe5ae9
Fixed bug in DefaultChart: keeping a reference s never a good idea.
2014-10-21 11:13:14 +02:00
25ad9ade05
Moved AdaptAutoDiff into its own test
2014-10-21 10:42:30 +02:00
bf16446f92
Deal with traits changes
2014-10-21 01:26:17 +02:00
e0841fb3e6
No more Ceres dependecy, copied relevant Ceres files here (for now)
2014-10-20 23:53:56 +02:00
bce84ca4db
Successfully created Expression from AutoDiff function!
2014-10-20 15:38:27 +02:00
bf5580d518
AdaptAutoDiff now works with RowMajor Eigen matrices
2014-10-20 14:39:28 +02:00
df5e584412
Compiles, but Jacobains not correct yet
2014-10-20 14:32:20 +02:00
a423f284e9
Canonical coordinates prototype, works for Snavely
2014-10-20 14:23:08 +02:00
821f776118
Wrapper works to some extent
2014-10-20 09:43:32 +02:00
7ebc8e969f
Charts with default constructors
2014-10-20 09:29:45 +02:00
63ae33088e
Some success on the way to autodiff
2014-10-19 23:47:59 +02:00
8ee16c9018
Comments for Paul
2014-10-19 11:19:09 +02:00
eac76cd0f0
Some progress on defining interface
2014-10-19 11:18:36 +02:00
6e142184cc
Implemented is_manifold and dimension for all types in testExpressionFactor
2014-10-19 00:35:25 +02:00
c32d2bb3b2
Fixed comments
2014-10-18 23:48:51 +02:00
d436d99146
Moved stuff to Manifold.h
2014-10-18 23:25:25 +02:00
fcda501ee2
double as manifold. No more LieScalar !
2014-10-18 18:13:33 +02:00
ed6a2b6eff
Charts !!!!
2014-10-18 17:11:28 +02:00
9c97b1d8a0
Some more refactoring
2014-10-18 16:45:04 +02:00
66b3081603
localCoordinates and retract
2014-10-18 15:02:22 +02:00
10cfd47404
TangentVector meta-function
2014-10-18 14:29:40 +02:00
ec69949f43
Point2 specialized
2014-10-18 14:22:01 +02:00
f08dc6c031
More boost-style traits
2014-10-18 14:16:24 +02:00
4c33444415
Snavely tested
2014-10-18 13:16:44 +02:00
bdf12b14b9
Add Snavely cost function
2014-10-18 12:35:02 +02:00
7018afdd58
Slight refactor of numerical derivatives
2014-10-18 12:34:05 +02:00
2972671064
Use boost::bind to avoid code duplication
2014-10-18 12:29:14 +02:00
db037c96c5
Implemented manifold_traits to allow numerical derivatives wrpt Matrix arguments
2014-10-18 12:12:25 +02:00
2cbba15573
ceres style functor
2014-10-18 00:29:18 +02:00
02d25f6658
New tests on traceSize
2014-10-16 15:07:05 +02:00
0bcca2c386
Drastic reduction in allocations at ExpressionFactor construction by having dims constructed imperatively, and using it for both keys_ and dimensions_
2014-10-16 14:16:58 +02:00
944422e295
Only ExpressionFactor needs dimensions! Also, add dimensions at construction -> speeds up linearize.
2014-10-16 11:15:47 +02:00
3413b98331
New storage sizes
2014-10-15 14:30:36 +02:00
79ff0c54f9
createUnknowns
2014-10-15 10:38:54 +02:00
9b1c9bbf37
Allocate temporary matrix on the stack rather tahn on heap, and give VerticalBlockMatrix a view on it.
2014-10-15 00:56:06 +02:00
c88b2a5902
Fixed bizarre link erro as well as off-by-1 bug
2014-10-15 00:34:28 +02:00
0f055f7910
Pass matrix to VerticalBlockMatrix constructor
2014-10-15 00:28:53 +02:00
625b939b66
Another very significant speed-up of reverseAD pipeline, by template specialization of the leaf case for fixed matrices. Unfortunately, while this sped up reverse AD for our SFM kernel by 300%, reverseAD was only 6%, and is now 2% of total time. So, time to look elsewhere.
...
Oh, and, it is clear that the Identity matrix for Leaf only expressions is completely known at compile time: Eigen::Matrix<double, T::dimension, T::dimension>::Identity(). That should nicely speed up many a PriorFactor (replacement).
2014-10-14 23:40:21 +02:00
baaeaacabe
Made dimensions constant property. Now performance is ***blazing***, way past custom factors.
2014-10-14 17:46:57 +02:00
Mike Bosse
dellaert