Re: zp2sos error: cplxpair: could not pair all complex numbers

[hale812]

Of course, I forgot to copypaste 
[sos,gain] = zp2sos(z,p,k);
But it is in the header, anyway. 

So yes, the frequency is close, but not critically.

Definitely this has little physical meaning.

But it has meaning for the optimizer (fsolve) that does not care about
meanings. Only about numbers.

The goal was to optimize the filter stack (DSP blocks) to certain passband.
AFAIK, there is no explicit formula, so I was using numerical solver of
arbitrary nonlinear cost function.

And this bug just does not allow me to optimize higher order circuit.

It is just a common sense, that the signal toolbox SHOULD work from 0 to
Nyquist, at least! 
And from my opinion, it should work in all the Euclidean space for combined
computations with "negative time" and "negative frequency", which happens
quite often. That depends on return value, of course.



--
View this message in context: 
http://octave.1599824.n4.nabble.com/zp2sos-error-cplxpair-could-not-pair-all-complex-numbers-tp4681260p4681263.html
Sent from the Octave - General mailing list archive at Nabble.com.