d6267c0440
a preliminary version of SparseQR routines that passes the unit tests but has slightly different results as the old method when optimizing large nonlinear problems.
2010-05-31 02:21:37 +00:00
e6e93475cf
added sizeOfA
2010-05-26 08:05:31 +00:00
5ef0400e06
put Factorization class in its own file, and added checks in constructors that take shared pointers, so we get some more meaningful output rather than *segmentation error*
2010-03-05 15:09:09 +00:00
9d704bb977
Added flag to default Factorization for old or new elimination, defaulting to old code.
2010-02-25 21:07:35 +00:00
5b60511505
Reverted default elimination code to older implementation - use switch in optimize() or eliminate() to use newer code.
2010-02-25 20:43:59 +00:00
022f706d94
Changed default elimination algorithm to newer one with fewer matrix allocations to solve crashing bug from LM discovered in MAST
2010-02-24 14:19:52 +00:00
694f6e4219
linearize now always returns a shared pointer
2010-02-21 21:17:47 +00:00
517c82f62f
Small details, like using a reference in FOREACH
2010-02-21 17:06:11 +00:00
3247751b5d
Major check-in: there are now two interchangeable implementations of VectorConfig.
...
VectorMap uses a straightforward stl::map of Vectors. It has O(log n)
insert and access, and is fairly fast at both. However, it has high overhead
for arithmetic operations such as +, scale, axpy etc...
VectorBTree uses a functional BTree as a way to access SubVectors
in an ordinary Vector. Inserting is O(n) and much slower, but accessing,
is O(log n) and might be a bit slower than VectorMap. Arithmetic operations
are blindingly fast, however. The cost is it is not as KISS as VectorMap.
Access to vectors is now exclusively via operator[]
Vector access in VectorMap is via a Vector reference
Vector access in VectorBtree is via the SubVector type (see Vector.h)
Feb 16 2010: FD: I made VectorMap the default, because I decided to try
and speed up conjugate gradients by using Sparse FactorGraphs all the way.
2010-02-17 03:29:12 +00:00
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
9b4ff5e099
transposeMultiplyAdd provied BLAS-style call for iterative speed
2010-01-30 23:59:29 +00:00
6ef09583b9
multiplyInPlace shaves a few seconds off but is fairly dangerous - I feel ambivalent....
2010-01-30 17:31:05 +00:00
afa964b8db
Fixed sparse bug
2010-01-23 05:16:29 +00:00
b092fee64b
pose2slamoptimizer unit tests worked
2010-01-23 03:49:05 +00:00
807cffbd61
Pose2SLAMOptimizer prototype for use in MATLAB
2010-01-23 00:57:54 +00:00
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
5f588031bc
Merged NoiseQR back into trunk
2010-01-20 18:32:48 +00:00
3bf15333af
move ordering into the solver, and the nonlinear optimizer is now exact <G, T, L>
2010-01-20 02:28:23 +00:00
3806125096
add linear system as a template parameter in nonlinear optimizer
...
fixed a nasty bug and change the internal data type of subgraph preconditioner from reference to boost shared pointer. reference is not a good idea for class members, because no type checking will happen
2010-01-19 10:46:12 +00:00
05b07d443e
add linear system solver as a template class parameter in nonlinear optimizer.
2010-01-18 05:51:19 +00:00
aef0b42562
Refactoring to use a new Symbol key instead of strings in Bayes*, Gaussian*, Ordering, Symbolic*, VectorConfig. Renamed existing type-checking key Symbol<C,T> to TypedSymbol<C,T>
2010-01-17 19:34:57 +00:00
19a3e228d7
Large refactoring - made several Lie group functions global, which used to be member functions, to treat Lie groups more uniformly. Also created Lie.h, and a preprocessor flag in numericalDerivative to change the coordinate frame derivatives are reported in. gtsam and easylib build and pass unit tests, but this will probably break other projects, which will require a few small changes to work again. Email coming in a few minutes to describe the changes.
2010-01-08 00:40:17 +00:00
9845a5ae37
add findMinimumSpanningTree to FactorGraph
2010-01-06 11:15:37 +00:00
8d4a029665
rhs method
2009-12-28 18:46:01 +00:00
5ac304aff3
correct convergence criterion, verbose flag
2009-12-28 16:15:26 +00:00
674ae9d030
steepest descent now all implemented in iterative.h/cpp
2009-12-28 12:37:34 +00:00
bc27afc49f
CGD method now uses CGD template in iterative.cpp
2009-12-28 10:48:48 +00:00
266fc56dea
Gradient using new operator^ and errors method
2009-12-27 12:13:31 +00:00
7d1428de60
Added Errors class and operator* for GaussianFactor and GaussianFactorGraph. Also moved a few functions to cpp.
2009-12-26 22:48:41 +00:00
d5d70d7751
gradientDescent_ and conjugateGradientDescent_ works in matlab
2009-12-13 03:02:14 +00:00
c38683cd64
Sped up and fixed (?) sparse
2009-12-12 06:18:29 +00:00
99533f286f
Vanilla Conjugate Gradient Descent works
2009-12-12 04:44:34 +00:00
726858145a
gradient descent (with fixed nr. of iterations, choosing optimal step size)
2009-12-11 04:59:05 +00:00
e7a912bd3b
Calculate gradient of factor graph objective function.
2009-12-10 20:19:15 +00:00
d26664d975
Convenience functions for adding factors
2009-12-10 15:33:52 +00:00
c7b86cec97
Renamed ConditionalGaussian -> GaussianConditional
2009-11-12 16:41:18 +00:00
77a1754b69
Renamed LinearFactor -> GaussianFactor, LinearFactorGraph -> GaussianFactorGraph
2009-11-12 16:16:32 +00:00
Kai Ni
Frank Dellaert
Alex Cunningham
Richard Roberts
Viorela Ila