Inverse R-factor of QR decomposition

NGSolve
1

Dear all,

I am trying to implement reduced rank extrapolation (RRE) methods in NGSolve.
A MultiVector.Orthogonalize() can be used to returns R-factor of QR decomposition.

I was wondering how can I inverse the matrix R in NGSolve because ‘ngsolve.bla.MatrixC’ object has no attribute ‘Inverse’. I need inverse of R and R.H for certain calculations.

I can convert it to scipy or numpy but I have a huge complex matrix R. I included a simple code for my problem below.

Thank you so much for your help and suggestion.

h = CreateVVector(5, complex=True)
vecs = MultiVector(h, 5)
for i in range(0,5):
    h[:] = np.random.rand(5) * (2 + 5j)
    vecs[0:5][i][:] = h.data
R_ = vecs.Orthogonalize()

Cheers,
VC

2

R.I gives the inverse matrix

1 Like