btf_q_permutation¶

sksparse.btf.btf_q_permutation(q)[source]¶

Convert a raw BTF column permutation vector to a valid permutation.

Parameters:: q ((N,) ndarray of int) – The raw BTF column permutation vector. Contains negative entries for unmatched columns.
Returns:: q_perm ((N,) ndarray of int) – The valid BTF column permutation vector. Contains only non-negative entries, where unmatched columns are replaced with their shifted absolute values.

Notes

In C, the values of q are converted using j = BTF_UNFLIP(Q[k]), which is a macro for:

j = (Q[k] < 0) ? -Q[k] - 2 : Q[k]

This function is a Python equivalent of that macro.

Added in version 0.5.0.

Examples

>>> import numpy as np
>>> from sksparse.btf import btf_q_permutation
>>> q = np.array([0, 1, 2, -5, 10, 6, 5, 7, 8, 4, 9], dtype=np.int32)
>>> btf_q_permutation(q)
array([ 0,  1,  2,  3, 10,  6,  5,  7,  8,  4,  9], dtype=int32)