Remove hacks for 6721 for typealias in symmetric.jl #21565
Conversation
base/linalg/symmetric.jl
Outdated
| HermOrSym{T,S} = Union{Hermitian{T,S}, Symmetric{T,S}} | ||
| RealHermSymComplexHerm{T<:Real,S} = Union{Hermitian{T,S}, Symmetric{T,S}, Hermitian{Complex{T},S}} | ||
| const HermOrSym{T,S} = Union{Hermitian{T,S}, Symmetric{T,S}} | ||
| const RealHermSymComplexHerm{T<:Real,S} = Union{Hermitian{T,S}, Symmetric{T,S}, Hermitian{Complex{T},S}} |
There was a problem hiding this comment.
these consts aren't necessary
There was a problem hiding this comment.
I know, but it's a lot easier to quickly glance a file and figure out it's a typealias with the const there
There was a problem hiding this comment.
This can also be written
const RealHermSymComplexHerm{T,S} = Union{Hermitian{T,S}, Symmetric{T,S}, Hermitian{Complex{T},S}} where T<:Realperhaps that is more clear ? as the right hand side matches something you would find in a function signature.
Probably but for a different PR. |
base/linalg/symmetric.jl
Outdated
| eigfact!(S != T ? convert(AbstractMatrix{S}, A) : copy(A)) | ||
| end | ||
|
|
||
| eigfact!(A::RealHermSymComplexHerm, irange::UnitRange) = Eigen(LAPACK.syevr!('V', 'I', A.uplo, A.data, 0.0, 0.0, irange.start, irange.stop, -1.0)...) |
There was a problem hiding this comment.
The <:StridedMatrix constraint seems necessary given the implementation via LAPACK? (Edit: Similarly, shouldn't the element type be constrained to BlasFloats?)
kshyatt
added
linear algebra
|
(Absent objections, requests for time, or someone else beating me to it, I plan to merge this pull request this evening or tomorrow morning. Best!) |
No description provided.