utils.orthogonalize_split#

Back to Module page

Summary#

Function to compute orthogonalized scores to those in the L subset

Signature#

def orthogonalize_split(scores, proj_l, submask_l, submask_h = None)

Input variables#
Name	Type	Default	Description
`scores`			nchunk x nscore matrix with split scores (can be vector for nscores=1)
`proj_l`			Matrix to multiply scores in the L subset before subtracting from those in H to orthogonalize the latter
`submask_l`			Boolean mask into nchunk to select splits in L
`submask_h`		None	Boolean mask into nchunk to select splits in H (if not given, assume complement of submask_l)

Output variables#

Return annotation	Docstring type	Description
`None`		np.count_nonzero(submask_h) x nscore matrix with orthogonalized split scores (can be vector for nscores=1)

Docstring#

Function to compute orthogonalized scores to those in the L subset
:param scores: nchunk x nscore matrix with split scores
               (can be vector for nscores=1)
:param proj_l:
    Matrix to multiply scores in the L subset before subtracting from those
    in H to orthogonalize the latter
:param submask_l: Boolean mask into nchunk to select splits in L
:param submask_h:
    Boolean mask into nchunk to select splits in H (if not given, assume
    complement of submask_l)
:return:
    np.count_nonzero(submask_h) x nscore matrix with orthogonalized split
    scores (can be vector for nscores=1)