Commit Graph

29 Commits (2c37c94b5d0f5c357d887e2e37ec0f20770d7133)

Author SHA1 Message Date
Alex Cunningham 2c37c94b5d Replaced the householder transform with the weighted system
Removed constrained components from makefile, they will disappear shortly
2009-11-09 21:34:20 +00:00
Frank Dellaert a3de1964d7 BIG CHANGE:
1) eliminate methods no longer return a shared pointer. Shared pointers are good for Factors and Conditionals (which are also non-copyable), because these are often passed around under the hood. However, a BayesNet is simple a list of shared pointers and hence does not cost a lot to return as an object (which is compiler-optimized anyway: there is no copy). So, the signature of all eliminate methods changed to simply return a BayesNet<> object (not a shared pointer).

2) GaussianBayesNet::optimize is now replaced by optimize(GaussianBayesNet) and returns a VectorConfig and not a shared pointer

3) GaussianBayesNet and SymbolicBayesNet are now simply typedefs, not derived classes. This is desirable because the BayesTree class uses templated methods that return BayesNet<Conditional>, not a specific BayesNet derived class.
2009-11-09 07:04:26 +00:00
Frank Dellaert cc5a2c3183 Renamed double-templated functions to _eliminate and _eliminateOne, and created FactorGraph-specific eliminateOne methods to make life easier 2009-11-07 21:03:30 +00:00
Frank Dellaert df3e5f2416 BIG: eliminate and eliminateOne now doubly templated functions, not methods.
Minor: Standardized on new shared_ptr naming convention:
shared_factor -> sharedFactor
conditional_ptr -> sharedConditional
node_ptr -> sharedClique
2009-11-07 19:31:39 +00:00
Frank Dellaert e1716a39cd Two changes: LinearFactor::sparse and LinearFactorGraph:sparse, and renamed VariableSet -> Dimensions, which is now a map from keys to integer variable dimensions. Merged in from the "sparse" branch created with Viorela. 2009-11-06 05:43:03 +00:00
Frank Dellaert e87c19ed7a ConditionalGaussian now stores sigmas 2009-11-05 06:59:59 +00:00
Frank Dellaert 88b8f4e16b fixed unit test 2009-11-05 06:34:43 +00:00
Frank Dellaert 3d334401f5 Constructor from BayesNet now in FactorGraph base class
FactorGraph::eliminate
Removed eliminate_partially
2009-11-05 04:56:59 +00:00
Richard Roberts e2414561be Merged r895:900 from branch weightedQR - LinearFactorGraph now has sigmas and ConditionalGaussian now has precisions 2009-11-04 20:59:16 +00:00
Frank Dellaert eab038651e Renamed BayesNet::insert -> push_back. BayesTree now uses Bayes nets as nodes. 2009-11-02 05:17:44 +00:00
Frank Dellaert a8d267c4ca Small change necessitating lots of edits: Conditionals now include key of random variable
This simplifies Bayes nets quite a bit. Also created a Conditional base class, derived classes ConditionalGaussian and SymbolicConditional
Finally, some changes were needed because I moved some headers to .cpp
2009-11-02 03:50:30 +00:00
Frank Dellaert 943b692a6b BIG CHANGE: I got rid of the BayesChain/ChordalBayesNet classes and we now simply have a BayesNet class. It will just happen to be chordal when it is the result of an elimination. This will simplify a lot of things.
The main renaming that happened is

BayesChain -> BayesNet
ChordalBayesNet -> GaussianBayesNet == BayesNet<ConditionalGaussian>
SymbolicBayesChain -> SymbolicBayesNet == BayesNet<SymbolicConditional>
2009-10-31 19:53:20 +00:00
Frank Dellaert 68d2f81f0a Smoother now creates x1...xT, not x0 anymore 2009-10-31 16:54:38 +00:00
Frank Dellaert fd5f43092f Ordering is now a list and Testable 2009-10-31 15:24:22 +00:00
Frank Dellaert 0d7d0e2161 Explicit template instantiation in .cpp files, so we can link with the code without having to include -inl.h everywhere... 2009-10-30 05:45:22 +00:00
Frank Dellaert 80b162a412 LinearFactorGraph::eliminate_one is now FactorGraph::eliminateOne<ConditionalGaussian>
Symbolic version FactorGraph::eliminateOne<SymbolicConditional> also implemented and tested
2009-10-29 14:34:34 +00:00
Frank Dellaert a513ae0287 Changed names and moved two LinearFactorGraph functions to FactorGraph
Added SymbolicFactorGraph compiulation unit and unit tests
Added symbolic combine constructor
2009-10-29 05:39:13 +00:00
Frank Dellaert b6cee73571 Significant change: Made FactorGraph templated on Factor only, and moved error and probPrime to derived classes
Moved find_and_remove_factors to base class
Added and tested symbolic factor graph constructor and conversion from any factor graph type
2009-10-29 04:11:23 +00:00
Frank Dellaert 4d9ff77249 moved timing example here from EasySLAM 2009-10-27 13:34:36 +00:00
Frank Dellaert f54ba387fe FactorGraph, ChordalBayesNet, and ConditionalGaussian now Testable 2009-10-24 23:14:14 +00:00
Chris Beall baef89ccf0 updated find_factors_and_remove to use map of factor indices to gather factors connected to a variable. factors are set to null instead of being erased. also updated size() to count non-NULL factors, and print() to only print non-NULL factors. added new unit test which tries to remove the same variable twice. 2009-10-23 00:56:40 +00:00
Chris Beall 52bedcad3a order 1 factors by using map 2009-10-22 21:33:00 +00:00
Frank Dellaert c3b8a20b5c Added new test for find_factors_and_remove and deleted obsolete one 2009-10-22 13:28:37 +00:00
Frank Dellaert 1b199a4d3b Added a "factors" function that returns indices to all factors connected to a variable. 2009-10-22 05:02:31 +00:00
Alex Cunningham 485f8d003b Added notes concerning the ordering issue when handling sets of LinearFactor::shared_ptrs 2009-10-14 21:06:44 +00:00
Alex Cunningham 7d0a30c20f Renamed FGConfig to VectorConfig in gtsam, easylib, EasySLAM, and mast. 2009-10-14 20:39:59 +00:00
Frank Dellaert 68e20eec2c 2 BIG changes:
(1) FactorGraph and NonlinearOptimizer now no longer have a .cpp file, but a -inl.h file as in [http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml Google's C++ Style Guide]. This means if you expect to instantiate one of the functions in a cpp file, you have to include the -inl.h file.
(1) getOrdering is now in FactorGraph, and the non-linear version does *not* take a config anymore. 
Long version: I made this change because colamd works on the graph structure alone, and should not depend on the type of graph. Instead, because getOrdering happened to implemented in LinearFactorGraph first, the non-linear version converted to a linear factor graph (at the cost of an unnecessary linearization), and then threw all that away to call colamd. To implement this in a key-neutral way (a hidden agenda), i had to modify the keys_ type to a list, so a lot of changes resulted from that.
2009-09-13 04:13:03 +00:00
Frank Dellaert fd11a1e2c7 Changed setBCN to take a reference 2009-08-22 22:01:17 +00:00
Richard Roberts d80fa24a9f Fixing directory structure 2009-08-21 22:23:24 +00:00