Daniel Karrasch

I'm a bit surprised the cost computation isn't compiled away, actually, given that sizes of SMatrices are known at compile time and that this is all that is required to...

Another option you may wish to consider is to use [LinearMaps.jl](https://github.com/JuliaLinearAlgebra/LinearMaps.jl), which has limited "AbstractMatrix" support, though.

Inverses and solves are not within the scope of LinearMaps.jl, but we do have an `InverseMap` type, whose multiplication corresponds to solves. Feel free to open an issue there to...

Sorry for my ignorance, but what exactly is the difference between the "DiagonalBlockArray" and `BlockDiagonal`? I thought that the initial stimulus for this package was that BlockArrays.jl had some "undesirable"...

> What's the status of this pull-request? It requires a careful review. 😉 Maybe you could try to use the online review interface, so that comments are directly linked to...

For some reason, all triangles are identified as external, and hence omitted from the iteration.

As in #76, I recommend doing "higher-order" manipulations like linear combinations and concatenation of Kronecker products with `LinearMaps.jl`. You are not restricted to use its own Kronecker product, but can...

This functionality exists in [LinearMaps.jl](https://github.com/Jutho/LinearMaps.jl), which has, by the way, it's own lazy Kronecker product implemented. So you have two options: ```julia # use kronecker from Kronecker.jl and the linear...

You can implement it with [`LinearMaps.jl`](https://github.com/JuliaLinearAlgebra/LinearMaps.jl) as follows: ```julia using LinearMaps khatri_rao(A::AbstractMatrix, B::AbstractMatrix) = hcat(map(⊗, eachcol(A), eachcol(B))...) # alternatively using Kronecker khatri_rao(A::AbstractMatrix, B::AbstractMatrix) = hcat(map(LinearMap∘kronecker, eachcol(A), eachcol(B))...) ``` I strongly...

> First is that you removed the restriction to a subtype of `Number`. Hm, I see. Dispatching on type parameters is just not recommended (especially when we have so many...