ranksvd

SVD at rank K for a weight matrix


'''import numpy

u,s,v = np.linalg.svd(weight, full_matrices=False)'''

where weight is matrix of 3 channels (rgb). I would like to find SVD at certain rank k. Can anyone help ? Thanks


Solution

  • As i read here, the rank k is not selective in Numpy library. Instead you can cut output matr1ices to keep top elements of matrix sigma.