[SciPy-user] Bug in eig(A)

[Nils Wagner]

Hi,

There is an interesting discussion concerning a bug of eig(A) in octave.
I have attached the example already converted into scipy.

Any suggestion ?

Nils
-------------- next part --------------
from scipy import *
#
# http://www.octave.org/mailing-lists/bug-octave/2002/184
#
A = array(([0.00430344378328325,0.00775725778022706,0.00775725778022705,0.0145879135099477,0.0145879135099477],
    [0.00430344378328325,0.00775725778022706,0.00775725778022705,0.0145879135099477,0.0145879135099477],
    [0.00430344378328325,0.00775725778022706,0.00775725778022705,0.0145879135099477,0.0145879135099477],
    [0.00430344378328325,0.00775725778022706,0.00775725778022705,0.0145879135099477,0.0145879135099477],
    [0.00430344378328325,0.00775725778022706,0.00775725778022705,0.0145879135099477,0.0145879135099477]))
#
# Eigenvalues, eigenvectors
#
w, vr = linalg.eig(A)
print w
#
# Singular value decomposition
#
u, s, vh = linalg.svd(A)
print 'Singular values', s
#
# QR decomposition
#
q,r = linalg.qr(A)
print q 
print r