Chris Beall
|
675b30d1c1
|
updates to BTree
|
2010-02-06 15:53:31 +00:00 |
Alex Cunningham
|
f88438bab4
|
Removed SQP optimizer and moved remaining SQP optimizer tests into testSQP. All equality constraints should be fully functional now. Inequality constraints still to come.
|
2010-02-06 14:48:46 +00:00 |
|
Alex Cunningham
|
219dfd262d
|
SQP now works with single configs using the TupleConfigs, without needing a separate optimizer.
|
2010-02-06 05:14:52 +00:00 |
|
Alex Cunningham
|
d8a8575d26
|
Removed a fixme comment
|
2010-02-06 05:12:09 +00:00 |
|
Alex Cunningham
|
d1267d1ef3
|
Added more constructors, print functionality and formatting for TupleConfig
|
2010-02-06 05:10:25 +00:00 |
|
Alex Cunningham
|
bb74b5c882
|
Added a config template parameter for ProjectionFactors
|
2010-02-06 05:08:52 +00:00 |
|
Alex Cunningham
|
af9f45ff24
|
Used template parameters in simulated2D to make it possible to template factors on a config type with typedefs for the previous usage so no other code needs to change
|
2010-02-05 14:49:40 +00:00 |
|
Alex Cunningham
|
040d45cb45
|
Added wrapper classes for TupleConfig for ease of use.
|
2010-02-05 01:33:33 +00:00 |
|
Chris Beall
|
8fe0795dd0
|
BTree moved into gtsam
|
2010-02-04 21:02:05 +00:00 |
|
Alex Cunningham
|
5c1c8ee76f
|
Changed PairConfig interface so that the internal configs need to be accessed with first() and second(). This is in anticipation of switching PairConfig to TupleConfig2.
|
2010-02-04 17:15:09 +00:00 |
|
Alex Cunningham
|
2068477e32
|
Added pragma once to LieConfig to prevent errors if included multiple times
|
2010-02-04 16:08:59 +00:00 |
|
Alex Cunningham
|
67744a5f07
|
Changed NonlinearConstraint to correctly use new keys
|
2010-02-04 16:08:11 +00:00 |
|
Alex Cunningham
|
01bbd3cf8d
|
Disabling SQP tests while transitioning config types and SQP implementation
|
2010-02-04 16:07:20 +00:00 |
|
Alex Cunningham
|
843f11678b
|
Recursive TupleConfig now has typedefs for 2-6 configs
|
2010-02-04 03:40:03 +00:00 |
|
Alex Cunningham
|
4e70f14097
|
The recursive TupleConfig should be ready for use now.
|
2010-02-04 02:04:45 +00:00 |
Kai Ni
|
550121013f
|
fix const string problem in print
|
2010-02-03 20:12:16 +00:00 |
|
Alex Cunningham
|
e5a7d4c878
|
More functionality in TupleConfigs
|
2010-02-03 14:08:09 +00:00 |
|
Alex Cunningham
|
94d50ddfb9
|
Added a liberal sprinkling of 'typename' to make TupleConfig work better
|
2010-02-03 13:57:34 +00:00 |
|
Alex Cunningham
|
a66a405dd5
|
Partially implemented version of a recursive TupleConfig that is still very much under testing.
|
2010-02-03 13:47:13 +00:00 |
John Rogers
|
2f16d8f6a1
|
Changed the order of constructor initializers in GaussianFactor to remove compiler warnings
|
2010-02-01 16:04:14 +00:00 |
Frank Dellaert
|
5554f6fc7e
|
Faster non-GSL versions of BLAS 2-style calls
|
2010-01-31 22:56:06 +00:00 |
|
Alex Cunningham
|
efe51c8419
|
Fixed typo in function name
|
2010-01-31 21:09:03 +00:00 |
|
Alex Cunningham
|
2889a25aeb
|
Fixed autotools files for GSL inclusion so that they don't include the wrong BLAS implementation when ATLAS is enabled
|
2010-01-31 18:26:18 +00:00 |
|
Alex Cunningham
|
c266012dcd
|
Added GSL/ATLAS version of multiplyAdd() and transposeMultiplyAdd()
|
2010-01-31 18:25:03 +00:00 |
justinca
|
74dd7128ac
|
Fixup cholesky decomposition - rename to LLt and RtR to make convention clear
|
2010-01-31 18:07:29 +00:00 |
|
Alex Cunningham
|
87cc573d34
|
Added a new combine process for GaussianFactors that allocates only one matrix when combining, rather than using append factor to make a large number of smaller matrices. There is a flag to switch between these modes, which currently defaults to the older approach. Currently, there does not appear to be a performance improvement, however.
|
2010-01-31 17:49:33 +00:00 |
|
Alex Cunningham
|
9f16d7b07f
|
Added in-place sub insert function to Vector
|
2010-01-31 17:24:01 +00:00 |
|
Alex Cunningham
|
48d2dabe43
|
Added in-place insertion functions to Matrix, as well as a #define flag to enable or disable GSL code without a reconfigure.
|
2010-01-31 17:21:07 +00:00 |
|
Alex Cunningham
|
b5ca322d21
|
Added default tolerance value to equals in Key classes so that equals() works properly
|
2010-01-31 17:00:27 +00:00 |
|
Frank Dellaert
|
fa6a515bfd
|
SAM was substantially sped up by inlining VectorConfig::insert
|
2010-01-31 16:05:16 +00:00 |
|
Frank Dellaert
|
5e4b23df59
|
Matrix::multiplyAdd and transposeMultiplyAdd are "level 2" BLAS and speed up the numeric part of the code substantially. Alex might be able to speed them up even more by making them use real BLAS code within Matrix.cpp.
|
2010-01-31 16:04:24 +00:00 |
|
Frank Dellaert
|
3b6c4917a9
|
GaussianBayesNet::backSubstituteInPlace
|
2010-01-31 04:39:41 +00:00 |
|
Frank Dellaert
|
ac870bce4c
|
GaussianFactor::transposeMultiplyAdd
|
2010-01-31 03:33:53 +00:00 |
|
Frank Dellaert
|
4f998e5ecd
|
Matrix::transposeMultiplyAdd
|
2010-01-31 02:53:03 +00:00 |
|
Frank Dellaert
|
9b4ff5e099
|
transposeMultiplyAdd provied BLAS-style call for iterative speed
|
2010-01-30 23:59:29 +00:00 |
|
Frank Dellaert
|
6ef09583b9
|
multiplyInPlace shaves a few seconds off but is fairly dangerous - I feel ambivalent....
|
2010-01-30 17:31:05 +00:00 |
|
Frank Dellaert
|
161a47caa7
|
|
2010-01-30 05:22:17 +00:00 |
|
Frank Dellaert
|
881d739371
|
BLAS level 1 style "scal" saves even more time in PCG
|
2010-01-30 04:01:49 +00:00 |
|
Frank Dellaert
|
4913326c22
|
BLAS level 1 style axpy calls in Vector and VectorConfig shave some seconds off PCG
|
2010-01-30 02:04:37 +00:00 |
|
Frank Dellaert
|
65bc90bf15
|
timing VectorConfig to speed up iterative
|
2010-01-29 13:57:45 +00:00 |
|
Alex Cunningham
|
eb88a149a1
|
NonlinearFactors handle constrained noise models properly now, as before they would try to whiten and fail
|
2010-01-28 17:21:24 +00:00 |
Viorela Ila
|
06a062c931
|
Added special lines for static variables needed for template classes
|
2010-01-27 19:45:31 +00:00 |
|
Kai Ni
|
ed45d436be
|
add gsl and atlas to configure
|
2010-01-27 05:15:52 +00:00 |
|
Alex Cunningham
|
bc2a25385f
|
Added updated timing results.
|
2010-01-27 05:08:54 +00:00 |
|
Alex Cunningham
|
59c7ce7e29
|
Integrated householder-based QR into NoiseModel. Note that the examples for GFG have changed to ensure that they are actually a linearized version of the nonlinear example.
|
2010-01-27 04:39:35 +00:00 |
|
Frank Dellaert
|
98b825ddbd
|
Fixed loading of datasets
|
2010-01-27 03:41:23 +00:00 |
|
Frank Dellaert
|
ae073c0120
|
Comments, and better test of composePoses
|
2010-01-27 02:49:58 +00:00 |
|
Frank Dellaert
|
8b87eebba6
|
commented GSl bit
|
2010-01-27 01:03:49 +00:00 |
Richard Roberts
|
02210cc7e0
|
Argument order of load changed
|
2010-01-26 20:31:25 +00:00 |
|
Richard Roberts
|
9b62b326c9
|
One more fix, inverse was actually needed in one case.
|
2010-01-26 20:10:45 +00:00 |
|
Richard Roberts
|
b82d054aaa
|
Definitions of Lie::between, Pose2::compose, expmap, and derivatives were not correct. Fixed, but changes some behavior (iSAM and CitySLAM already fixed), will send email about this change.
|
2010-01-26 20:00:17 +00:00 |
Manohar Paluri
|
a0b96e7684
|
corrected case of Pose2SLAMoptimizer
|
2010-01-25 14:50:25 +00:00 |
|
Kai Ni
|
1685920d52
|
fixed Ab2 bug
|
2010-01-23 05:29:04 +00:00 |
|
Frank Dellaert
|
afa964b8db
|
Fixed sparse bug
|
2010-01-23 05:16:29 +00:00 |
|
Kai Ni
|
744269343f
|
add print
|
2010-01-23 04:46:00 +00:00 |
|
Kai Ni
|
b092fee64b
|
pose2slamoptimizer unit tests worked
|
2010-01-23 03:49:05 +00:00 |
|
Richard Roberts
|
21c1af2b9f
|
Collecting more statistics in isam2, logmap for configs
|
2010-01-23 01:53:04 +00:00 |
|
Kai Ni
|
4491db99be
|
add beijing data set
|
2010-01-23 01:10:12 +00:00 |
|
Frank Dellaert
|
7644ff0080
|
optimize returns Vector, does not update
|
2010-01-23 01:08:42 +00:00 |
|
Frank Dellaert
|
ada844300d
|
Fixed compile issue in tests
|
2010-01-23 01:03:47 +00:00 |
|
Frank Dellaert
|
807cffbd61
|
Pose2SLAMOptimizer prototype for use in MATLAB
|
2010-01-23 00:57:54 +00:00 |
Michael Kaess
|
7bc4ee65da
|
allow pure linear steps (disabled)
|
2010-01-23 00:21:34 +00:00 |
|
Manohar Paluri
|
80d335ed77
|
added saveGraph function to BayesTree which writes out a dot file which can be visualized using graphviz dot command
|
2010-01-22 23:52:56 +00:00 |
|
Kai Ni
|
e87faa078d
|
write correct information matrix to data file
|
2010-01-22 23:20:12 +00:00 |
|
Frank Dellaert
|
2150b24e29
|
Fixed memory hog problem with Alex's help. SAM marginally faster (3-4% with 1000 poses, might be bigger effect for full dataset)
|
2010-01-22 23:00:35 +00:00 |
|
justinca
|
16c55975c1
|
Fix inverse_square_root, add cholesky decomposition options
|
2010-01-22 22:28:03 +00:00 |
|
Kai Ni
|
e88ae4a944
|
add dataset.h/cpp
|
2010-01-22 20:18:40 +00:00 |
|
Michael Kaess
|
ab155999cf
|
simplified alg, but no measureable speed improvement
|
2010-01-22 18:29:27 +00:00 |
|
Frank Dellaert
|
351cdd18c2
|
case change: SharedGaussian and SharedDiagonal are now classes with their own header file. Needed for MATLAB TORO hail Mary
|
2010-01-22 17:36:57 +00:00 |
|
Kai Ni
|
490791cd48
|
add writer module to dump the errors and time
|
2010-01-22 08:13:54 +00:00 |
|
Kai Ni
|
4b778a4e82
|
fixed a bug in matrix solve function
|
2010-01-22 06:45:01 +00:00 |
|
Michael Kaess
|
1d093e388d
|
changed main algorithm to allow recovery of exact solution
|
2010-01-22 06:28:12 +00:00 |
|
Frank Dellaert
|
75e29dc015
|
failing test
|
2010-01-22 05:45:59 +00:00 |
|
Richard Roberts
|
5367e5a157
|
All std::map<Symbol,T> are now SymbolMap<T>, which is just a thin wrapper around std::map. at(Key) is used instead of first checking with find when an exception should be thrown for non-existent keys. This does not change any behavior or timing currently. This check-in also includes some functions in BayesTree for gathering clique statistics.
|
2010-01-22 04:41:40 +00:00 |
|
Frank Dellaert
|
41a6e64bbb
|
Fixed insert to use IndexTable, a new class
|
2010-01-22 02:27:26 +00:00 |
|
Frank Dellaert
|
2cc777228b
|
coments and variables only
|
2010-01-21 20:15:52 +00:00 |
|
justinca
|
7575996d52
|
Remove leftover commented out unused code
|
2010-01-21 19:49:36 +00:00 |
|
justinca
|
f89a53ed91
|
Bugfix in Rot3::logmap. The function checked for a trace of 3, but the check could fail due to rounding errors, causing the function to try to take acos(something larger than 1) resulting in NaNs being returned.
|
2010-01-21 19:49:12 +00:00 |
|
Alex Cunningham
|
ac746ccead
|
Improved performance of updateAb in NoiseModel with GSL/ATLAS. Various other small optimizations were made.
|
2010-01-21 18:51:59 +00:00 |
|
Michael Kaess
|
2183e09c67
|
fluid relinearization works
|
2010-01-21 17:41:54 +00:00 |
|
Michael Kaess
|
523c8630d3
|
works at least on already solved data...
|
2010-01-21 16:09:57 +00:00 |
|
Frank Dellaert
|
0dc4f417e9
|
sampling from NoiseModel
|
2010-01-21 08:31:18 +00:00 |
|
Michael Kaess
|
8fd0c2ae72
|
new fluid relinearization algorithm, in sync with lyx
|
2010-01-21 07:38:37 +00:00 |
|
Alex Cunningham
|
9db7623f80
|
Improved instructions for GSL/ATLAS
|
2010-01-21 01:03:34 +00:00 |
|
Alex Cunningham
|
063aa14118
|
Reimplemented matrix::householder_update using GSL and BLAS, you'll need to edit Makefile.am to enable the new version -- see email
|
2010-01-21 00:59:33 +00:00 |
|
Michael Kaess
|
188561d925
|
removeTop/Path now returns a BayesNet to avoid converting the conditionals to a factor graph
|
2010-01-21 00:38:22 +00:00 |
|
Richard Roberts
|
eb03f78e7d
|
Removing dependency hack speeds up compilation
|
2010-01-20 20:47:15 +00:00 |
|
Alex Cunningham
|
588c119bf3
|
Added timing script for Matrix::householder_()
|
2010-01-20 18:52:31 +00:00 |
|
Alex Cunningham
|
5f588031bc
|
Merged NoiseQR back into trunk
|
2010-01-20 18:32:48 +00:00 |
|
Alex Cunningham
|
9c9007920a
|
Changed the interface on Matrix::column() so that it uses the one in our library (now called column_() ) rather than the boost default. Currently, our implementation just uses the boost default, but this may change due to timing results.
|
2010-01-20 16:08:14 +00:00 |
|
Alex Cunningham
|
98b143cd22
|
Switched all loop indices in Matrix/Vector to size_t, with improved results in timeGaussianFactorGraph
|
2010-01-20 14:42:05 +00:00 |
|
Alex Cunningham
|
b4f9163296
|
Added more labels and results (with higher numbers of repetitions) for timeGaussianFactorGraph
|
2010-01-20 14:31:24 +00:00 |
|
Alex Cunningham
|
8db99db57e
|
Small performance improvement for matrix_augmented, with accompanying test in timeGaussianFactor
|
2010-01-20 14:09:44 +00:00 |
|
Alex Cunningham
|
6bc0462669
|
Checked Matrix::collect() with different timing metric, and memcpy has approx 25% speedup over original version. Matrix_range is slower or the same speed as the original version.
|
2010-01-20 13:45:22 +00:00 |
|
Kai Ni
|
977a97eacf
|
rename variable
|
2010-01-20 09:47:09 +00:00 |
|
Michael Kaess
|
aae49e43f0
|
code factorization, speedup
|
2010-01-20 06:49:19 +00:00 |
|
Kai Ni
|
ae908316e0
|
add comments
|
2010-01-20 06:34:25 +00:00 |
|
Richard Roberts
|
19dc8bf4b1
|
Fixed size() crash when BayesTree empty, added clear() function to BayesTree, added and removed #includes to fix link errors
|
2010-01-20 04:23:35 +00:00 |
|
Michael Kaess
|
e767996b45
|
same bug fix as in ISAM2
|
2010-01-20 03:54:20 +00:00 |
|
Alex Cunningham
|
55e414cbdd
|
Tried noalias() for collect(), but no real improvement
|
2010-01-20 03:25:55 +00:00 |