Problem with eig in Octave 3.2.4

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

Problem with eig in Octave 3.2.4

vliviu
I have installed Octave 3.2.4 and fails some tests at make check.
Some are at eig decomposition, some at imscale etc.
For instance, trying
eig(A) for A=[1 3 -2; 3 5 1; -2 1 4]
gives aberrant results.

Any workaround? Thanks in advance.

--



_______________________________________________
Help-octave mailing list
[hidden email]
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
Reply | Threaded
Open this post in threaded view
|

Re: Problem with eig in Octave 3.2.4

tmacchant
Hello

Perhaps (an) external libraries (library) are (is) broken.
(blas, lapack etcs. )

Please describe your platform (e.g. Debian) and how did you prepare (or obtain) the external
libraries.

Without detailed information you cannot get anything from the ML.

Regards

Tatsuro

--- Liviu Vladutu  wrote:

> I have installed Octave 3.2.4 and fails some tests at make check.
> Some are at eig decomposition, some at imscale etc.
> For instance, trying
> eig(A) for A=[1 3 -2; 3 5 1; -2 1 4]
> gives aberrant results.
>
> Any workaround? Thanks in advance.
>
> --
> > _______________________________________________
> Help-octave mailing list
> [hidden email]
> https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
>


--------------------------------------
Get the new Internet Explorer 8 optimized for Yahoo! JAPAN
http://pr.mail.yahoo.co.jp/ie8/
_______________________________________________
Help-octave mailing list
[hidden email]
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
Reply | Threaded
Open this post in threaded view
|

Re: Problem with eig in Octave 3.2.4

martin_helm
In reply to this post by vliviu
Am Freitag, 30. Juli 2010 10:56:48 schrieb Liviu Vladutu:

> I have installed Octave 3.2.4 and fails some tests at make check.
> Some are at eig decomposition, some at imscale etc.
> For instance, trying
> eig(A) for A=[1 3 -2; 3 5 1; -2 1 4]
> gives aberrant results.
>
> Any workaround? Thanks in advance.
>
> --
>
This does not really sound like a general problem for which a workaround
exists, but that somehow your octave is screwed.

Please post more information about your environment (linux, windows, compiler,
did you get warnings in the configure script?, since you use make check I
guess you compiled octave yourself, maybe you have incompatible libraries or
something like that).

Also post the "aberrant results", otherwise everything is just wild guessing.

- mh
_______________________________________________
Help-octave mailing list
[hidden email]
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
Reply | Threaded
Open this post in threaded view
|

Problem with eig in Octave 3.2.4

vliviu
In reply to this post by tmacchant
Hello,
Platform is  Linux 2.6.34-12 -desktop SMP-PREEMPT.
I 've got the Octave sources, and folowed ./configure, make, make check..
I don't belive that any external library is broken.
For the given A matrix, the following:
[V,D]= eig(A) gives:
 
V = [-0.818395 -0.315302 -0.480434
0.434733 0.207055 00.876433
-0.375818 0.926128 0.032380] !!!




_______________________________________________
Help-octave mailing list
[hidden email]
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
Reply | Threaded
Open this post in threaded view
|

Re: Problem with eig in Octave 3.2.4

David Bateman
In reply to this post by vliviu
vliviu wrote
I have installed Octave 3.2.4 and fails some tests at make check.
Some are at eig decomposition, some at imscale etc.
For instance, trying
eig(A) for A=[1 3 -2; 3 5 1; -2 1 4]
gives aberrant results.

Any workaround? Thanks in advance.
It works for me (3.2.4 MingW version from OctaveForge)

octave:1> [v,l] = eig(A);
V =

 -0.818394  -0.315302  -0.480434
  0.434733   0.207055  -0.876433
 -0.375818   0.926128  0.032380

l =

Diagonal Matrix

 -1.5120        0          0
       0   4.9045           0
       0           0   6.6076

octave:2> norm(A*v(:,1) - l(1,1)*v(:,1))
ans = 6.2804e-016
octqve:3> norm(A*v(:,2) - l(2,2)*v(:,2))
ans = 6.2804e-016
octqve:4> norm(A*v(:,3) - l(3,3)*v(:,3))
ans = 2.0351e-015

Please be more specific in what you find "aberrant"

D.
Reply | Threaded
Open this post in threaded view
|

Re: Problem with eig in Octave 3.2.4

martin_helm
In reply to this post by vliviu
Am Freitag, 30. Juli 2010 12:00:10 schrieb Liviu Vladutu:
> V = [-0.818395 -0.315302 -0.480434
> 0.434733 0.207055 00.876433
> -0.375818 0.926128 0.032380]
>
I cannot really see why this should be wrong

 v*l*v'  
ans =            

   1.0000   3.0000  -2.0000
   3.0000   5.0000   1.0000
  -2.0000   1.0000   4.0000

(with the v you posted) gives the original matrix as it should.

l =

Diagonal Matrix

  -1.5120        0        0
        0   4.9045        0
        0        0   6.6076


- mh
_______________________________________________
Help-octave mailing list
[hidden email]
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
Reply | Threaded
Open this post in threaded view
|

Re: Problem with eig in Octave 3.2.4

martin_helm
Am Freitag, 30. Juli 2010 12:44:42 schrieb Liviu Vladutu:

> Yes it worked, sorry Martin .
> But I still have problems reading (with imread) a valid bmp
> file. I get:
>
> error: imread: invalid image file:
> /usr/local/libexec/octave/3.2.4/oct/i686-pc-linux-gnu/__magick_read__.oct:
> undefined symbol: _ZTIN6Magick9ExceptionE
> error: called from:
> error:   /usr/local/share/octave/3.2.4/m/image/imread.m at line 74, column
>  7
>
> On Fri, Jul 30, 2010 at 1:27 PM, Martin Helm <[hidden email]> wrote:
> > Am Freitag, 30. Juli 2010 12:00:10 schrieb Liviu Vladutu:
> > > V = [-0.818395 -0.315302 -0.480434
> > > 0.434733 0.207055 00.876433
> > > -0.375818 0.926128 0.032380]
> >
> > I cannot really see why this should be wrong
> >
> >  v*l*v'
> > ans =
> >
> >   1.0000   3.0000  -2.0000
> >   3.0000   5.0000   1.0000
> >  -2.0000   1.0000   4.0000
> >
> > (with the v you posted) gives the original matrix as it should.
> >
> > l =
> >
> > Diagonal Matrix
> >
> >  -1.5120        0        0
> >        0   4.9045        0
> >        0        0   6.6076
> >
> >
> > - mh
>

Can you check which graphicsmagick version is installed on your system? I
remember there is a problem with the 1.3 versions (on my machine I work with
1.2.10).
There is also an old thread with that "undefined symbol" problem which
mentions libjpeg and libpng, can you check this libs as well

http://old.nabble.com/imread-imwrite-td18142037.html

- mh
_______________________________________________
Help-octave mailing list
[hidden email]
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
Reply | Threaded
Open this post in threaded view
|

Re: Problem with eig in Octave 3.2.4

martin_helm
Am Freitag, 30. Juli 2010 14:48:19 schrieb Liviu Vladutu:

> Yes is Graphics Magick 1.3.12 and is a known issue (bug) with Octave. 3.2x.
> Probably I'll build external C++ code for accessing images inside Octave.
> I'll get back to you.
> Have a nice weekend.
> L.
>
> On Fri, Jul 30, 2010 at 2:28 PM, Martin Helm <[hidden email]> wrote:
> > Am Freitag, 30. Juli 2010 12:44:42 schrieb Liviu Vladutu:
> > > Yes it worked, sorry Martin .
> > > But I still have problems reading (with imread) a valid bmp
> > > file. I get:
> > >
> > > error: imread: invalid image file:
> >
> > /usr/local/libexec/octave/3.2.4/oct/i686-pc-linux-gnu/__magick_read__.oct:
> > > undefined symbol: _ZTIN6Magick9ExceptionE
> > > error: called from:
> > > error:   /usr/local/share/octave/3.2.4/m/image/imread.m at line 74,
> >
> > column
> >
> > >  7
> > >
> > > On Fri, Jul 30, 2010 at 1:27 PM, Martin Helm <[hidden email]> wrote:
> > > > Am Freitag, 30. Juli 2010 12:00:10 schrieb Liviu Vladutu:
> > > > > V = [-0.818395 -0.315302 -0.480434
> > > > > 0.434733 0.207055 00.876433
> > > > > -0.375818 0.926128 0.032380]
> > > >
> > > > I cannot really see why this should be wrong
> > > >
> > > >  v*l*v'
> > > > ans =
> > > >
> > > >   1.0000   3.0000  -2.0000
> > > >   3.0000   5.0000   1.0000
> > > >  -2.0000   1.0000   4.0000
> > > >
> > > > (with the v you posted) gives the original matrix as it should.
> > > >
> > > > l =
> > > >
> > > > Diagonal Matrix
> > > >
> > > >  -1.5120        0        0
> > > >        0   4.9045        0
> > > >        0        0   6.6076
> > > >
> > > >
> > > > - mh
> >
> > Can you check which graphicsmagick version is installed on your system? I
> > remember there is a problem with the 1.3 versions (on my machine I work
> > with
> > 1.2.10).
> > There is also an old thread with that "undefined symbol" problem which
> > mentions libjpeg and libpng, can you check this libs as well
> >
> > http://old.nabble.com/imread-imwrite-td18142037.html
> >
> > - mh
>
Please do not forget to keep the list on cc (use reply to all).

_______________________________________________
Help-octave mailing list
[hidden email]
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave