Open
Description
isapprox
is defined for Number
s and AbstractArray
s.
Factorization
s are very closely related to AbstractArray
s.
I think there is a strong case to be made for them also defining isapprox
.
Probably the way to do this is something like
isapprox(f1::Factorization, F2::Factorization; kwargs...) = (f1 == f2) || isapprox(Matrix(f1), Matrix(f2); kwargs...)
I think we need to densify the factorization if we are going to support rtol
/atol
using norm
.
But we can avoid that if they are exactly equal.
Maybe also someone is smart and can workout some bounds that less us avoid it if the factors are close enough
(Though right now == is pretty broken: JuliaLang/julia#41228, JuliaLang/julia#41363)
Metadata
Metadata
Assignees
Labels
No labels