forum.alglib.net
http://forum.alglib.net/

Cholesky decomposition with semi-positive definite matrices
http://forum.alglib.net/viewtopic.php?f=2&t=47
Page 1 of 1

Author:  hayt [ Mon Aug 02, 2010 10:42 am ]
Post subject:  Cholesky decomposition with semi-positive definite matrices

Hello,

I am using the c++ alglib 2.6 to calculate the cholesky decomposition and call the hpdmatrixcholesky-function which always returns false on a spd(semi positive definte) matrix. (like 4x4 all filled with 1)

Now I read in the header that the library returns false on not positive finite matrices. But my math understanding (and wikipedia ^^) tells me you only need spd matrices. Is this only a wrong formulation is the header or is this a bug and cholesky does not work with spd-matrices?

Greetz
Ricardo

Author:  Sergey.Bochkanov [ Tue Aug 03, 2010 9:32 pm ]
Post subject:  Re: Cholesky decomposition with semi-positive definite matri

It work with strictly positive definite matrices only. Semidefinite matrices are not supported.

Can you explain in more details what is wrong with header? I don't get it :) it talks about positive definite matrices; it is well known that Cholesky decomposition is stable only for strictly positive definite matrices.

Author:  hayt [ Wed Aug 04, 2010 8:36 am ]
Post subject:  Re: Cholesky decomposition with semi-positive definite matri

Ah sorry you are right of course. I just remembered that you can use the cholesky decomposition for semi-positive definite matrices, but I forgot that there is no stable computation for those.

The explanation in the header is correct in that case, too. So I think this thread can be closed.

Thx for your time :)

Ricardo

Page 1 of 1 All times are UTC
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/