Closed devmotion closed 1 year ago
Base: 89.93% // Head: 90.11% // Increases project coverage by +0.18%
:tada:
Coverage data is based on head (
b4c05b6
) compared to base (cc7c5c4
). Patch coverage: 100.00% of modified lines in pull request are covered.
:umbrella: View full report at Codecov.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.
This PR adds
AbstractPDMat(::AbstractMatrix)
,AbstractPDMat(::Cholesky)
andAbstractPDMat(::CholTypeSparse)
.The main motivation is that this allows to specialize on the matrix type, which allows downstream packages such as Distributions (e.g. in https://github.com/JuliaStats/Distributions.jl/blob/9f9c5caf6cd99c5205a11ca1c8d6b6471fc9ce29/src/multivariate/mvnormal.jl#L201-L203) to just use
AbstractPDMat(A)
instead of adding custom dispatches.Unfortunately, we can't add something like
AbstractPDMat(A::UniformScaling)
(size unknown) orAbstractPDMat(A::Diagonal{<:Real,<:FillArrays.Fill{<:Real}})
(FillArrays dependency).Fixes #136 (without lying about the return type).