Orthonormalize a matrix in place

General Usage Performance
4

If you orthogonalize with classical Gram-Schmidt twice you’ll be fine. The GMRES code in ```SIAMFANLEquations.jl```` does it that way. That code, and the orthogonalizer live in

https://github.com/ctkelley/SIAMFANLEquations.jl/tree/master/src/Solvers/LinearSolvers

Here is a quick and dirty qr code that uses the orthogonalizer and overwrites A with Q.
The default option, cgs2 is classical Gram-Schmidt twice, which is the best choice on any modern computer.

function qrctk!(A, orth = "cgs2")
    T = typeof(A[1, 1])
    (m, n) = size(A)
    R = zeros(T, n, n)
    @views R[1, 1] = norm(A[:, 1])
    @views A[:, 1] /= R[1, 1]
    @views for k = 2:n
        hv = vec(R[1:k, k])
        Qkm = view(A, :, 1:k-1)
        vv = vec(A[:, k])
        Orthogonalize!(Qkm, hv, vv, orth)
    end
    return (Q = A, R = R)
end
1 Like