[SciPy-user] best way of matrix G => L, D: G = L D L^T

dmitrey openopt@ukr....
Fri Aug 24 03:07:08 CDT 2007


Hi all,
I need best way of getting from matrix G matrices L, D: G = L D L^T (L - 
unit lower triangular matrix, D - diagonal)
(IIRC it's related to LU-decomposition)

(it's for implementing Gill-Murray Stable Newton’s Method)
then I need to solve the equation

L^T d = e[t]
for d , where e[t] is a unit vector with the t-th component of e[t] being 1

What's the best way of doing it via numpy/scipy?

Has using something like colamd before LU any sense here?

Regards, D.





More information about the SciPy-user mailing list