Re: Algorithmic Differentiation in Octave

[Brad Bell]

First order forward mode is probably best done using the complex step 
method; see
http://blogs.mathworks.com/cleve/2013/10/14/complex-step-differentiation/


On 01/25/2017 12:36 AM, Olaf Till wrote:
> On Tue, Jan 24, 2017 at 04:25:44PM -0500, John W. Eaton wrote:
> > Are you aware of the (now unmaintained and probably obsolete) "ad" package
> > from Octave Forge?  An old version is still accessible here:
> >
> >
> > https://octave.sourceforge.io/ad/
> >
> > This may be the last released version:
> >
> > https://sourceforge.net/projects/octave/files/Octave%20Forge%20Packages/R2009-06-07/ad-1.0.6.tar.gz/download
> >
> > Even though it is old and probably won't work out of the box with current
> > versions of Octave, it might be worth looking at.
> I wasn't aware of it, thanks. This package seems to have implemented
> the overloading even without the scripted class functionality of newer
> Octave. It seems to be for first order derivatives only.
>
> Olaf