Added - added the option for the Eigen layer to support semi-definite predictions - new option `n_zero_eigvals` for Eigen layer to specify how many zero eigenvalues the model should expect
Changed - Eigen layer now uses `torch.linalg.eigh` instead of `torch.linalg.eig`
0.2.0
Changed - add 'None' as a positivity function - set the default Cholesky positivity function to 'None' - add citations to methods used on README.md
0.1.1
Changed - add our preprint to the readme
0.1.0
Changed - `spdlayers.tools.in_shape_from` now should always output integers (as opposed to floats)
0.0.3
Changed - `spdlayers.tools.in_shape_from` now uses the sum of the first n natural numbers, see https://cseweb.ucsd.edu/groups/tatami/handdemos/sum/
0.0.2
Added - Add url homepage in setup.py to point to this github repo
Changed - tweak readme installation notes to just use pip to install
Fixed - bug in single source version, which prevented install of spdlayers if torch was not already installed - mistake in the pypi license classifier