Theseus Files

This code release is also associated with arXiv v3 paper release.

Major Updates

• Add sparse solvers support to LM adaptive damping by @luisenp in https://github.com/facebookresearch/theseus/pull/360
• Add a differentiable sparse matrix vector product on top of our ops by @luisenp in https://github.com/facebookresearch/theseus/pull/392
• Dogleg by @luisenp in https://github.com/facebookresearch/theseus/pull/371
• Add support for masking jacobians of zero weights in the batch by @luisenp in https://github.com/facebookresearch/theseus/pull/398
• Add a labs package for experimental stuff by @luisenp in https://github.com/facebookresearch/theseus/pull/424

Other Changes

• Minor vmap fix in SO2 by @luisenp in https://github.com/facebookresearch/theseus/pull/362
• Change Objective.error() so it can be vectorized w/o changing vectorized cache by @luisenp in https://github.com/facebookresearch/theseus/pull/363
• Added missing square in test_theseus_layer loss. by @luisenp in https://github.com/facebookresearch/theseus/pull/372
• Add proper accept/reject logic for LM optimizer by @luisenp in https://github.com/facebookresearch/theseus/pull/364
• Cleaned up sparse solvers code by @luisenp in https://github.com/facebookresearch/theseus/pull/386
• Remove softmax from end to end test and do some clean up by @luisenp in https://github.com/facebookresearch/theseus/pull/389
• Add a workaround for NonlinearOptimizer rejecting all batch steps by @luisenp in https://github.com/facebookresearch/theseus/pull/388
• Add code to automatically pick number of LUCuda contexts by @luisenp in https://github.com/facebookresearch/theseus/pull/390
• Bug fix in NonlinearOptimizer.reset() when using LUCudaSparseSolver by @luisenp in https://github.com/facebookresearch/theseus/pull/396
• Allow float data to be used with our sparse solver extensions by @luisenp in https://github.com/facebookresearch/theseus/pull/391
• Add diagonal scaling method and re-enable LM adaptive + ellipsoidal by @luisenp in https://github.com/facebookresearch/theseus/pull/393
• Add a timer util that adapts to a torch.device. by @luisenp in https://github.com/facebookresearch/theseus/pull/399
• Made torch's CUDA_GCC_VERSION check optional in setup.py. by @luisenp in https://github.com/facebookresearch/theseus/pull/387
• Fix black and wheel script errors by @luisenp in https://github.com/facebookresearch/theseus/pull/421
• Fix bug in Vectorization of autodiff costs. by @luisenp in https://github.com/facebookresearch/theseus/pull/400
• Objective.error() no longer updates vectorization if also_update=False. by @luisenp in https://github.com/facebookresearch/theseus/pull/401
• update flake8 to github by @bamos in https://github.com/facebookresearch/theseus/pull/428
• Use IMPLICIT for test_theseus_layer and fix related bugs by @luisenp in https://github.com/facebookresearch/theseus/pull/431
• Remove nox dependency by @luisenp in https://github.com/facebookresearch/theseus/pull/436
• Bumped version to 0.1.4. by @luisenp in https://github.com/facebookresearch/theseus/pull/435

Development in labs package

• Add theseus.geometry.functional.so3 module with exp() and jexp() implementations by @fantaosha in https://github.com/facebookresearch/theseus/pull/365
• Add theseus.geometry.functional.so3.adjoint() by @fantaosha in https://github.com/facebookresearch/theseus/pull/373
• Add theseus.geometry.functional.so3.inverse() by @fantaosha in https://github.com/facebookresearch/theseus/pull/374
• Add hat() and vee() operators to theseus.geometry.so3 by @fantaosha in https://github.com/facebookresearch/theseus/pull/378

Full Changelog: https://github.com/facebookresearch/theseus/compare/0.1.3...0.1.4