Re: problem with consistency of root finding and sorting

[Ben Abbott]

On Oct 10, 2007, at 7:55 AM, David Bateman wrote:

> Ben Abbott wrote:
> > On Oct 10, 2007, at 3:45 AM, David Bateman wrote:
> >
> > > Ben Abbott wrote:
> > > > Thanks David,
> > > >
> > > > another attempt, if you don't mind
> > > >
> > > > p = roots([1 0 18 0 81])
> > > > p =
> > > >
> > > >   -0.0000 + 3.0000i
> > > >   -0.0000 - 3.0000i
> > > >    0.0000 + 3.0000i
> > > >    0.0000 - 3.0000i
> > > octave:1> p = roots([1 0 18 0 81])
> > > p =
> > >
> > >    0.0000 + 3.0000i
> > >    0.0000 - 3.0000i
> > >   -0.0000 + 3.0000i
> > >   -0.0000 - 3.0000i
> > >
> > > > [e, indx] = mpoles (p)
> > > > e =
> > > >
> > > >    1
> > > >    2
> > > >    1
> > > >    2
> > > >
> > > > indx =
> > > >
> > > >    1
> > > >    3
> > > >    2
> > > >    4
> > >
> > > octave:2> [e, indx] = mpoles (p)
> > > e =
> > >
> > >    1
> > >    2
> > >    1
> > >    2
> > >
> > > indx =
> > >
> > >    2
> > >    4
> > >    1
> > >    3
> > >
> > > > p(indx)
> > > > ans =
> > > >
> > > >   -0.0000 + 3.0000i
> > > >    0.0000 + 3.0000i
> > > >   -0.0000 - 3.0000i
> > > >    0.0000 - 3.0000i
> > > octave:3> p(indx)
> > > ans =
> > >
> > >    0.0000 - 3.0000i
> > >   -0.0000 - 3.0000i
> > >    0.0000 + 3.0000i
> > >   -0.0000 + 3.0000i
> >
> > Thanks again. I believe I've found the bug ... and it is mine :-(
> >
> > Using the attached mpoles.m, I get
> >
> > p = roots([1 0 18 0 81])
> > p =
> >
> >   -0.0000 + 3.0000i
> >   -0.0000 - 3.0000i
> >    0.0000 + 3.0000i
> >    0.0000 - 3.0000i
> >
> > [e, indx] = mpoles(p)
> > e =
> >
> >    1
> >    2
> >    1
> >    2
> >
> > indx =
> >
> >    1
> >    3
> >    2
> >    4
> >
> > I assume you will get the same.
> Ok with the attached mpoles at the latest CVS I get
>
> octave:1> p = roots([1 0 18 0 81])
> p =
>
>    0.0000 + 3.0000i
>    0.0000 - 3.0000i
>   -0.0000 + 3.0000i
>   -0.0000 - 3.0000i
>
> octave:2> [e, indx] = mpoles(p)
> e =
>
>    1
>    2
>    1
>    2
>
> indx =
>
>    3
>    1
>    4
>    2
>
> octave:3> p(indx)
> ans =
>
>   -0.0000 + 3.0000i
>    0.0000 + 3.0000i
>   -0.0000 - 3.0000i
>    0.0000 - 3.0000i
>
>
> D.

Thanks for anticipating the last part

 p(indx)
ans =

  -0.0000 + 3.0000i
   0.0000 + 3.0000i
  -0.0000 - 3.0000i
   0.0000 - 3.0000i

The problem appears to have been solved!

Thanks David
(and Henry too)