The LAPACK forum has moved to https://github.com/Reference-LAPACK/lapack/discussions.

Solving Large Matrix with Deficient Rank with Dgelsd

Open discussion regarding features, bugs, issues, vendors, etc.

Solving Large Matrix with Deficient Rank with Dgelsd

Postby yohan » Thu Nov 30, 2006 1:15 pm

Hi all,
I am a beginner of Lapack. I was trying to solve a matrix which is rank deficient. The matrix I am going to solve is overdetermined , and huge i.e. approximately 2000x2000. The first and the third colums' values are the same. The number of variables I am trying to solve is 4.
I had implemented dgelsd using C++ and managed to solve a small matrix , ie 20 x20. But when the matrix gets bigger, i.e 100x100, I am getting rubbish results, i.e. infinite solutions.

Could somebody tell me if there is a limitation on the matrix size (which I doubt) or it just simply could not converge?

I would appreciate very much your help/advice on this.....

Have a nice day
Yohan
yohan
 
Posts: 1
Joined: Thu Nov 30, 2006 11:05 am

Postby Julien Langou » Thu Nov 30, 2006 1:53 pm

Hello,
how have you set RCOND ? Set RCOND to 1e-10 or to a negative value, just for a try.
Julien.
Julien Langou
 
Posts: 835
Joined: Thu Dec 09, 2004 12:32 pm
Location: Denver, CO, USA


Return to User Discussion

Who is online

Users browsing this forum: No registered users and 6 guests