Huge residues in aaa approximation ?!

[Calin Gheorghiu]

Dear  YUJI, dear Nick, dear all,

By solving a degenerate parabolic initial/boundary value problem and using the aaa code, at the final stage, I obtain huge residues, i.e. O(10^190)!? Moreover, zeros disappear from the phase diagram, making it impossible to interpret! Do you have any idea what these enormous residues mean and how they could be avoided? Thank you in advance for a valid suggestion!

Best,

Calin

Dr. C. I. Gheorghiu

Senior research fellow
"T. Popoviciu" Institute of Numerical Analysis
P. O. Box 68-1
401010 Cluj-Napoca 
Romania
Phone +40 (0)744 70 18 96

http://www.ictp.acad.ro/gheorghiu

https://www.springer.com/gp/book/9783319062297

[nakatsu...@gmail.com]

Dear Calin, 

We have just updated the way we compute the residues; this *might* resolve your issue, could you try it out (I think the link on the website is up-to-date, containing this:

% Compute more accurate residues, if nargout > 2:
if ( nargout > 2 )
    deg = max(0, length(zer)-length(pol) );
    A = [Z.^(0:deg) 1./(Z-pol.')];
    c = A\F;
    res = c(deg+2:end);
end

if not, please try the latest version on github.

(TBH I would not be too surprised if an issue persists, but I hope this is a step in the right direction).

Best wishes, 
Yuji