Re: [Fwd: Possible contribution]

On Friday 07 February 2003 02:38 am, Orestes Mas wrote:
> Appreciated Mr. Davis,
>
> A couple of months ago I sent you an email about a possible contribution
> to gnucap. As I've had no response since then, I assume the email did
> not reach you. So I resend it to you herein.
>
> Thanks.

I remember reading it, but apparently didn't reply, or lost it.

> > Some years ago, our group developed a piece of software, called ARIADNA,
> > intended to compute the transfer function of any (linear or linearized)
> > circuit in a semi-symbolic manner. Basically, the algorithm works by
> > finding the frequency response of the circuit in a set of (complex)
> > frequencies equally spaced over the unit circle in the complex plane,
> > and then solving a set of linear equations to find the coefficients of
> > the transfer function's numerator and denominator.
> >
> > This software, originally written in FORTRAN, was later rewritten in
> > C++, so we think it would be relatively easy to merge the code with the
> > gnucap one, adding to it the ability to compute network functions.
> >
> > So how do you consider it? We're relatively new to the GNU/Linux world
> > (and to the UNIX world too), and we don't master programming in this
> > environment, nor the gnucap internals, but being our software mainly a
> > calculation routine, and with major mathematical algorithms already
> > implemented in gnucap, we think this wouldn't be a very huge task if
> > we've your assistance.

Yes, it is a good idea.

I assume that since you offered, that you are willing to release it under GPL.

I am familiar with unit circle interpolation.  It is described in a book by 
Vlach and Singhal.

I have been thinking of another algorithm to compute poles and zeros, based 
on LU decomposition with a matrix of polynomials.  It would be interesting to 
compare them.