Commit Graph

7215 Commits (24e1334e5a584a5dd9708fedcaf78902a5adf8d0)

Author SHA1 Message Date
Paul Furgale 24e1334e5a Trying to update the concepts definition to reflect reality 2014-12-11 12:41:16 +01:00
Mike Bosse 376dec5103 cleaned up a bit; but not really working yet. Trouble with partial specialization of lie_group::expmap<Quaternion<> >() 2014-12-11 03:15:45 +01:00
dellaert 21b97ddedf Removed obsolete methods (but link error!) 2014-12-09 01:04:21 +01:00
dellaert 449ba1f37a LieGroupChart makes use of Exponential map (and its inverse) around identity 2014-12-09 00:52:53 +01:00
dellaert 3077d03568 Header should also be included in non-quaternion mode 2014-12-09 00:11:28 +01:00
dellaert 8fcf9d9866 SO(3) is a *Matrix* Lie group. This is it. Rot3 will inherit from this *or* Quaternion. 2014-12-09 00:11:28 +01:00
dellaert 118d561e96 Created GTSAM_GROUP_IDENTITY as it handles templates differently 2014-12-09 00:11:28 +01:00
Frank Dellaert 473bc1b703 Derivatives of Charts, and special Lie group treatment 2014-12-08 13:08:39 +00:00
dellaert 9dc3d28bf2 Now Rot3 uses QuaternionChart Expmap/Logmap in quaternion mode 2014-12-08 13:16:30 +01:00
dellaert 8cbab779b3 Slight refactor 2014-12-08 13:16:30 +01:00
dellaert ccd843040e Small change in file that might disappear anyway 2014-12-08 13:16:30 +01:00
Paul Furgale 2af04a8e41 Typo... 2014-12-08 10:26:04 +00:00
Frank Dellaert fd9f5ae1ec Clarified derivatives of group actions 2014-12-08 09:42:56 +00:00
Paul Furgale 2b1d907a74 More indentation fixes. 2014-12-08 08:55:14 +00:00
Paul Furgale 21596e1894 Making the indentation work in Markdown view. 2014-12-08 08:51:27 +00:00
dellaert fd59783fae Forgot to check in how to use macros... 2014-12-07 21:03:54 +01:00
dellaert 949a6b7f4c Found solution for template arguments in macros. Requires a bit more from caller, but is better. I quite like the MACROS: It's compact... 2014-12-07 19:55:06 +01:00
dellaert 04d3457b45 First working Manifold macro 2014-12-07 19:30:46 +01:00
dellaert 62ae58e7ea Quaternion header 2014-12-07 19:13:02 +01:00
dellaert 10ae9f5505 Easier to read with Cyclic<3> 2014-12-07 18:55:49 +01:00
dellaert 481be10509 A bit more compact 2014-12-07 18:55:33 +01:00
dellaert 9194b92cf6 Now work with MACROS instead - but get linking error. Upside (if we can fix that): uniform treatment between foreign types (see Quaternion) and GTSAM types (Cyclic). Downside: seems I had to create a different macro for different number of template arguments. Help? 2014-12-07 18:41:18 +01:00
dellaert 38a0842090 Macros. Too ugly? 2014-12-07 18:01:30 +01:00
dellaert 19c38b91ee Finish CRTP in Cyclic 2014-12-07 15:59:31 +01:00
dellaert ef58a8a56a Attempt at satisfying Group concept by deriving from base class. Needs to be fixed, also, test does not link :-( 2014-12-07 13:24:59 +01:00
dellaert 022e930085 Fixed typo 2014-12-07 13:23:59 +01:00
dellaert e0e2a9b063 Eigen::Quaternions now satisfy IsManifold, IsGroup, and IsLieGroup! Also reverted back to a simpler treatment of Charts.
Merge branch 'traits/quaternion' into feature/tighteningTraits
2014-12-07 13:06:32 +01:00
dellaert cdc0029158 Reverted on Chart base class. But no chart-specific traits needed, as assumed created by us. 2014-12-07 13:01:25 +01:00
dellaert a31e596448 Working local/Logmap (taken from Rot3Q) 2014-12-07 12:47:26 +01:00
dellaert 36da8702f9 Retract works 2014-12-07 12:35:17 +01:00
dellaert 8db8cb54b0 Some chart refinement (early check does not work) 2014-12-07 12:35:05 +01:00
dellaert e2f250c160 Added Manifold, Lie Group, and Vector Space concepts back in 2014-12-07 11:52:09 +01:00
dellaert 01aab77504 Made Testable a boost concept 2014-12-07 11:51:40 +01:00
dellaert 187760ce95 Refactor for readability 2014-12-07 10:20:54 +01:00
dellaert 3a6b89e840 Now all Eigen::Quaternion flavors are certified as IsGroup 2014-12-07 10:19:38 +01:00
dellaert 06640cc414 New naming convention for concept (IsGroup), moved invariant checking out to namespace 2014-12-07 09:48:31 +01:00
dellaert e1c1d788c0 Quaternions are a group (test compiles, at least) 2014-12-06 18:01:13 +01:00
dellaert 491fd91af6 Fixed compile problem. 2014-12-06 14:28:39 +01:00
dellaert d1fec1a90d Reverted moste changes in .cproject 2014-12-06 13:17:44 +01:00
dellaert 2e86da1096 Merge branch 'feature/groups' into feature/tighteningTraits 2014-12-06 13:12:44 +01:00
dellaert 614bfeb9d3 Moved to headers 2014-12-06 13:11:56 +01:00
dellaert 4d2a0aefc1 Group checks out ! 2014-12-06 12:59:35 +01:00
dellaert 6c2b28aa8b between tested 2014-12-06 12:50:40 +01:00
dellaert 5407232e36 between and inverse, as well as tag dispatching 2014-12-06 11:51:44 +01:00
dellaert ddeb764912 group::traits, and group::compose 2014-12-06 11:37:14 +01:00
dellaert 07ef30039a structure_tag works 2014-12-06 11:17:54 +01:00
dellaert 71e77b8c87 Compiles 2014-12-06 10:58:16 +01:00
dellaert 6f82458bee Trying to firm things up by having a small unit test for cyclic groups. 2014-12-06 10:17:35 +01:00
dellaert 0ea2e5efe4 Merge remote-tracking branch 'origin/develop' into feature/tighteningTraits 2014-12-06 09:20:13 +01:00
Frank Dellaert 8cc26c759d BIG CHANGE: the OptionalJacobian<M,N> obviates the need for the `optional<Matrix&> Jacobian arguments. They will continue to exist, for backwards compatibility, in the old-style factors, but everywhere else they should disappear. This PR by Natesh has eradicated all but one in gtsam/geometry. Great job !!!!
Merged in feature/fixedSizeDerivatives (pull request #54)

Proposed way to do Jacobians from now on via Eigen::Ref like type
2014-12-05 22:54:36 +01:00