# Change log for the `lapack` package

## 0.4

 * Unified `Matrix` type that provides the same type parameters
   across all special types.
   This reduces the use of type functions and improves type inference.

 * Unified `transpose` and `adjoint` functions enabled by the new `Matrix` type.

 * `Unpacked` format:
   We now support data type and according functions
   for unpacked triangular, symmetric and Hermitian matrices.
   Enables declaration e.g. of Hessenberg matrices.

 * There are now two types of square matrices:

    * `Square`: height and width shapes match exactly

    * `LiberalSquare`: only the sizes of height and width match

 * `Square.eigensystem`:
   Use liberal square as transformation matrix,
   such that the eigenvalue array has `ShapeInt` shape.
   The dimension of the input square matrix does not make sense
   as shape for the eigenvalue array.

 * `Square.fromGeneral` -> `fromFull`

 * `Orthogonal.affineKernelFromSpan` -> `affineFiberFromFrame`,
   `Orthogonal.affineSpanFromKernel` -> `affineFrameFromFiber`

## 0.3.2

 * `Orthogonal`: `project`, `affineKernelFromSpan`, `affineSpanFromKernel`,
   `leastSquaresConstraint`, `gaussMarkovLinearModel`

 * `Symmetric.fromHermitian`, `Hermitian.fromSymmetric`

 * `instance Monoid Matrix`, especially `mempty`
   for matrices with static shapes.

 * `Extent.Dimensions`: turn from type family to data family

 * Start using `doctest-extract` for simple tests

## 0.3.1

 * `Matrix.Symmetric`:
   You can now import many functions for symmetric matrices from this module.
   This is more natural than importing them from `Triangular`.

## 0.3

 * Matrix data family

 * `Matrix`: `ZeroInt` -> `ShapeInt`, `zeroInt` -> `shapeInt`

 * `Hermitian`, `BandedHermitian`: `covariance` -> `gramian`

 * `Square.eigensystem`:
   Return left eigenvectors as rows of the last matrix.
   This is adjoint with respect to the definition in `lapack-0.2`
   but it is consistent
   with the other eigenvalue and singular value decompositions.