Talk:Generalized inverse/Archive 1

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

Multiple linear regression

I am reading the articles about the pseudoinverse and wondering about what is the difference between that and multiple linear regression. In another word, if I have a same set of linear equations, using the pseudoinverse method or linear regression, should I get same solution? Thanks. Simonliang (talk) 00:55, 29 September 2008 (UTC)

Construction formulae: unhelpful?

I'm thinking of changing or removing the construction formulae for the left and right inverses, for three reasons:

The formulae only apply when the matrix $A$ has full rank;
When $A$ is poorly conditioned, the formulae suffer from numerical issues;
The formulae, if they apply at all and give sufficiently accurate results, give only the Moore-Penrose pseudo-inverse, and do not constitute general formulae for the construction of generalized inverses (which are the subject of the article).

Any comments before I press on with my edits? Cooperised (talk) 22:35, 3 June 2013 (UTC)