MATLAB Central Newsreader - eigenvalues

eigenvalues

munir — Fri, 06 Nov 2009 18:21:59 -0500

dear all,
i want to find eigenvalues for a dynamic problem.
but somthing is going wrong. as i m doing it first time i cant figure
it out.
when i give command as
fsol=eig(kuufinal,muufinal)
error is
??? Error using ==> sym.eig
Too many input arguments.

and when i give as
fsol=eig(kuufinal,muufinal,qz)
error is
??? Error using ==> qz
Not enough input arguments.

what is the problem???
when i indivisually find eigenvalues of matrix it gives results.
also muufinal is not singular.
both matrices are symmetric.
plz plz help soon.....
thanks n regards.

Re: eigenvalues

Wayne King — Fri, 06 Nov 2009 18:48:02 -0500

munir <munirpansare@gmail.com> wrote in message <d7d30128-4368-4ff6-97e4-32d1984d23d1@h14g2000pri.googlegroups.com>...
> dear all,
> i want to find eigenvalues for a dynamic problem.
> but somthing is going wrong. as i m doing it first time i cant figure
> it out.
> when i give command as
> fsol=eig(kuufinal,muufinal)
> error is
> ??? Error using ==> sym.eig
> Too many input arguments.
>
> and when i give as
> fsol=eig(kuufinal,muufinal,qz)
> error is
> ??? Error using ==> qz
> Not enough input arguments.
>
> what is the problem???
> when i indivisually find eigenvalues of matrix it gives results.
> also muufinal is not singular.
> both matrices are symmetric.
> plz plz help soon.....
> thanks n regards.
Hi, I think you need to tell us more about what you are trying to input into eig(). If you type

>>help sym/eig

it does not support two input arguments. eig() supports two input arguments but not the symbolic variant.

Hope that helps,
wayne

Re: eigenvalues

Bruno Luong — Fri, 06 Nov 2009 19:17:04 -0500

Re: eigenvalues

munir — Fri, 06 Nov 2009 19:53:11 -0500

On Nov 7, 12:17 am, "Bruno Luong" <b.lu...@fogale.findmycountry>
wrote:
> munir <munirpans...@gmail.com> wrote in message <d7d30128-4368-4ff6-97e4-32d1984d2...@h14g2000pri.googlegroups.com>...
> > dear all,
> > i want to find eigenvalues for a dynamic problem.
> > but somthing is going wrong. as i m doing it first time i cant figure
> > it out.
> > when i give command as
> > fsol=eig(kuufinal,muufinal)
>
> I hope you know eigenvalues of A with respect to B is eigenvalues of inv(B)*A. So that might help for your symbolic call.
>
> Bruno

i tried whatever u suggested,but nothing is working. my matrix size is
40*40. is there any effect of diagonal terms of very small
magnitude???
thank u for ur replies.

Re: eigenvalues

munir — Sat, 07 Nov 2009 05:52:00 -0500

On Nov 7, 12:53 am, munir <munirpans...@gmail.com> wrote:
> On Nov 7, 12:17 am, "Bruno Luong" <b.lu...@fogale.findmycountry>
> wrote:
>
> >munir<munirpans...@gmail.com> wrote in message <d7d30128-4368-4ff6-97e4-32d1984d2...@h14g2000pri.googlegroups.com>...
> > > dear all,
> > > i want to find eigenvalues for a dynamic problem.
> > > but somthing is going wrong. as i m doing it first time i cant figure
> > > it out.
> > > when i give command as
> > > fsol=eig(kuufinal,muufinal)
>
> > I hope you know eigenvalues of A with respect to B is eigenvalues of inv(B)*A. So that might help for your symbolic call.
>
> > Bruno
>
> i tried whatever u suggested,but nothing is working. my matrix size is
> 40*40. is there any effect of diagonal terms of very small
> magnitude???
> thank u for ur replies.

dear all,
any suggestion plz...

Re: eigenvalues

Bruno Luong — Sat, 07 Nov 2009 07:31:30 -0500

>
> dear all,
> any suggestion plz...

Yes: drop trying to compute symbolic eigenvalues for matrix of 40 in size. You might get - if -ever - a formula that not fits on 1000 pages of paper. Before you will be asked what to do with it, might be you might think about saving a tree not to print out such formula.

Now the chance that Matlab or any software that can solve for eigenvalues problem for you is zero. You might know that this problem is equivalent to finding the roots of 40th order polynomials. Take a look at Galois's theory, who has shown two centuries ago that - in general - such formula does not exist for polynomial or order 5th or greater, e.g. 40th order.

Bruno

Re: eigenvalues

munir — Sat, 07 Nov 2009 11:06:23 -0500

On Nov 7, 12:31 pm, "Bruno Luong" <b.lu...@fogale.findmycountry>
wrote:
> > dear all,
> > any suggestion plz...
>
> Yes: drop trying to compute symbolic eigenvalues for matrix of 40 in size. You might get - if -ever - a formula that not fits on 1000 pages of paper. Before you will be asked what to do with it, might be you might think about saving a tree not to print out such formula.
>
> Now the chance that Matlab or any software that can solve for eigenvalues problem for you is zero. You might know that this problem is equivalent to finding the roots of 40th order polynomials. Take a look at Galois's theory, who has shown two centuries ago that - in general - such formula does not exist for polynomial or order 5th or greater, e.g. 40th order.
>
> Bruno

as per ur suggestion i took inv(B)*A and then took its eigenvalues and
as i converted all symbolic values to numerical values by giving
symbols their numeric values. i m getting solution. so can i now find
eigenvalues of any order of a matrix??
thank u very much.