Consider rewriting this algorithm to use piecewise functions ala pw.mac. pw.mac can do this pretty easily although
there is no function to do it in one step.
http://maxima-project.org/wiki/index.php?title=Pw.mac
Rich
--------------------------------------------------
From: "Richard Fateman" <fateman at cs.berkeley.edu>
Sent: Friday, June 11, 2010 1:36 PM
To: "Chris Nassar" <chris.nassar at gmail.com>
Cc: <maxima at math.utexas.edu>
Subject: Re: [Maxima] 2D Discrete Convolution
> I suspect that you could encode your problem into a polynomial multiplication.
> It might be worth examining the cost of doing this computation using exact rationals
> rather than bfloats.
>
> It is also possible that you could do this by a bfloat version of an FFT.
>
> RJF
>
>
> _______________________________________________
> Maxima mailing list
> Maxima at math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima
>