Re: Ill-conditioned linear system
- To: mathgroup at smc.vnet.net
- Subject: [mg104328] Re: Ill-conditioned linear system
- From: guerom00 <guerom00 at gmail.com>
- Date: Wed, 28 Oct 2009 04:05:30 -0500 (EST)
- References: <hc6gl0$5r2$1@smc.vnet.net>
For what it's worse, it seems that instead of solving Ax=b, solving Transpose[A].Ax=Transpose[A]b is more stable numerically=85 Maybe it has to do with the fact that Transpose[A].A has some properties (symetric and positive definite) I'll try to explore this lead=85