This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: forwarded message from Peng-Sheng Chen
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: gsl-discuss at sources dot redhat dot com
- Date: Thu, 23 May 2002 21:21:08 +0100 (BST)
- Subject: Re: forwarded message from Peng-Sheng Chen
- References: <15596.1967.148989.805388@debian>
> If I want to get all solutions (not only one solution), how
> to do it ?
> (some equations exist many solutions.)
> Thanks very much.
>
>
There should be some information on the web about solving
"underdetermined systems" using the SVD. If not I think there is a
section in Golub and Van Loan's book on 'Matrix Computations' about
it.
regards,
Brian Gough
Ps. Chen
> > -----Original Message-----
> > From: Brian Gough [mailto:bjg@network-theory.co.uk]
> > Sent: Wednesday, May 08, 2002 5:23 AM
> > To: Peng-Sheng Chen
> > Cc: gsl-discuss@sources.redhat.com
> > Subject: Re: FW: GSL support "left division \ " ?
> >=20
> > Peng-Sheng Chen writes:
> > > Hello: =A0 =A0=A0 There is an equation: Ax =3D B. =A0=A0 A is =
> m-by-n matrix.
> > > =A0=A0 x is n vector. =A0=A0 B is m vector. =A0=A0 The condition =
> is m >=3D n,
> > > that is =A0=A0 we have equations more than unknown variable. =A0 =
> =A0=A0 In
> > > matlab, it support a operator =93left divide \ =94 to manipulate
> > > =A0=A0 the equations. In GSL, how to compute x ? (multidimention
> > > root-finding?) =A0=A0 Thanks very much.
> >=20
> > For overdetermined systems you can find the least-squares solution
> > using gsl_linalg_QR_lssolve or gsl_linalg_SV_solve, described in the
> > "Linear Algebra" chapter of the manual.
> >=20
> > regards
> > Brian Gough
> >=20
>