Re: Confidence in eigs answers

[CdeMills]

My guess is that you're asking for the smallest eigenvalues of a nearly
singular matrix, and that, according to termination criteria (how small is
small) used in various algorithms, you observe small differences between the
results. You should have a look at your matrix condition number, the ratio
between the smallest and the greatest singular value (see "help svd"). If it
is below the machine numerical accuracy, the differences you're observing
should be considered as pure numerical noise.

Regards

Pascal
-- 
View this message in context: 
http://octave.1599824.n4.nabble.com/Confidence-in-eigs-answers-tp2142541p2165106.html
Sent from the Octave - General mailing list archive at Nabble.com.