Remove suitesparse / arpack caveat?

[MasonProtter]

I just ran a test

julia> using LinearAlgebra; BLAS.vendor()
:mkl

julia> using Arpack, SparseArrays

julia> ⊗(A, B) = kron(A, B);

julia> function Hub_ED(U, μ, L,)
           id = [1 0; 0 1]
           c = [0 0; 1 0]

           c_up = c ⊗ id
           c_dn = id ⊗ c
           id² = id ⊗ id

           n_up = c_up' * c_up
           n_dn = c_dn' * c_dn

           Û =  U*(n_up * n_dn) - μ*(n_up + n_dn)

           c_dg_up(i) = foldl(⊗, sparse.([i==j ? c_up' : id² for j in 1:L]))
           cup(i)     = foldl(⊗, sparse.([i==j ? c_up  : id² for j in 1:L]))
           c_dg_dn(i) = foldl(⊗, sparse.([i==j ? c_dn' : id² for j in 1:L]))
           cdn(i)     = foldl(⊗, sparse.([i==j ? c_dn  : id² for j in 1:L]))
           Ûf(i)      = foldl(⊗, sparse.([i==j ? Û     : id² for j in 1:L]))

           function c_dg_c(i)
               out = c_dg_up(i)*cup(i+1) + c_dg_dn(i)*cdn(i+1)
               out + out'
           end

           H = -sum(c_dg_c, 1:(L-1)) + sum(Ûf, 1:L)
           @show typeof(H)
           (λ, ϕ), t = @timed Arpack.eigs(H, nev=1, which=:SR)
           λ, ϕ, t
       end
Hub_ED (generic function with 1 method)

julia> Hub_ED(1, 1, 8)
typeof(H) = SparseMatrixCSC{Int64,Int64}
([-15.813000555255263], [-1.949487752371787e-18; -2.0668833887408264e-18; … ; 2.035974575802704e-18; 1.216026630767711e-18], 0.340870967)

and it looks like things are running fine with SuiteSparse and Arpack.

[ViralBShah]

Do we know if the SuiteSparse and ARPACK in these cases are using MKL or do they end up using openblas?

[andreasnoack]

We now have https://github.com/JuliaLinearAlgebra/ArpackMKL.jl which is required for Arpack to use MKL. We might still need a solution for SuiteSparse. @arnavs have you looked into SuiteSparse?

[ViralBShah]

Is there a better way to do this? Perhaps we should be using a shim library that allows us to replace the BLAS implementation dynamically? There's a bunch of other JuMP solvers too, for example.

In theory I would have thought that if a library linked against openblas, and if instead of openblas, we LD_PRELOADed MKL, things would keep working. Is that possible? Also asking @staticfloat and @Keno

[aminya]

Is there any update on this?
#17

[andreasnoack]

See #17