`scipy.linalg._flapack:ctrsen`

Summary

Reorder a Schur decomposition to gather selected eigenvalues into a leading cluster (LAPACK ctrsen).

select : ndarray: Boolean mask of length n: the eigenvalues it marks are gathered into the leading block. A bool array works; LAPACK reads it as a Fortran LOGICAL.
t : ndarray: Upper (quasi-)triangular Schur form of shape (n, n).
q : ndarray: Matrix of Schur vectors; updated when wantq is nonzero.
job : str, optional: What to compute besides the reordering: 'N' nothing, 'E' the cluster condition number s, 'V' the separation sep, 'B' both. Default is 'B'.
wantq : int, optional: If nonzero, the transformation is accumulated into q. Default is 1.
lwork : int, optional: Workspace size; -1 requests the query. Default is max(1, n). The default is a bare minimum for the cheapest job only -- what is actually needed depends on the cluster size, so query the matching _lwork routine.
overwrite_t : int, optional: If nonzero, t may be overwritten in place. Default is 0.
overwrite_q : int, optional: If nonzero, q may be overwritten in place. Default is 0.

ts : ndarray: The reordered Schur form.
qs : ndarray: The updated Schur vectors.
w : ndarray: Reordered eigenvalues, length n.
m : int: Size of the selected cluster.
s : float: Condition number of the selected cluster; 0 unless job asked for it.
sep : float: Separation of the cluster; 0 unless job asked for it.
info : int: 0 on success; if negative, the -info-th argument had an illegal value.