Investigate distributed linear algebra in Julia about acefit.jl

[wcwitt]

Useful links:

• Basic DistributedArrays package https://juliaparallel.github.io/DistributedArrays.jl/stable/
• An old discourse post requesting distributed QR (https://discourse.julialang.org/t/distributed-memory-qr-tsqr-linear-algebra/9849) suggested https://github.com/JuliaParallel/Elemental.jl)
• Julia ScaLAPACK bindings https://github.com/JuliaParallel/ScaLAPACK.jl
• ScaLAPACK and DArrays JuliaParallel/ScaLAPACK.jl#3

[wcwitt]

Here is a toy example showing how to set up IterativeSolvers/lsqr with DistributedArrays

using Distributed
@everywhere using Pkg
@everywhere Pkg.activate(".")
@everywhere using DistributedArrays, IterativeSolvers, Revise

function solve_lsqr(nrows, ncols)
    # solves Ax+b where x=ones()+noise
    A = drand((nrows,ncols))
    b = A*dones((ncols,), [1]) + 0.1*drand((nrows,))
    x0 = dzeros((ncols,), [1])
    return lsqr!(x0, A, b; damp=1e-3, atol=1e-6, verbose=true)
end

@time x = solve_lsqr(1000, 750)

[wcwitt]

Closing this as we now have a decent idea of what's available.