fastquotient?



----- Original Message ----- 
From: "Fabrizio Caruso" <caruso at posso.dm.unipi.it>
To: <maxima at math.utexas.edu>
Sent: Wednesday, February 08, 2006 6:39 AM
Subject: fastquotient?


> Hi all again!
> 
> Is there in Maxima a function to divide
> polynoimals in CRE faster than "divide"?
Divide probably converts its arguments to CRE. Do
you want quotient or remainder?  Or exact division?
I guess if you are in a hurry for the answer, and you have
univariate polynomials you might consider a program that 
does nothing else, and may be faster.  NTL is one such
program.

> I mean something equivalent to "fasttimes",
> (it could be implemented by multiplication
> by the modular inverse).

I think this would work for exact division; knowing that
you can do exact division requires a GCD which may be,
in practice, the more expensive item.
> 
> Is fasttimes using Karatsuba or something else?
Karatsuba (division into two pieces, recursively)
> If so, why does the documentation say that the
> arguments of "fasttimes" must be multivariate?
I think that testing showed that for one variable it
was not faster, at least when it was written (in 1970 or so).
It may be different now.

RJF