Re: continued fractions

The continued fraction domain is using Moebius transforms internally
(z -> (az+b)/(cz+d))  like Gosper.  I wasn't specifically aware of his
work at the time, though.    

The main issue is not in converting a particular value to a  continued
fraction, which is easy, but rather to construct a c.f. out of a stream
of convergents, e.g. obtained as the result of arithmetic on two other c.f.s.

-- Stephen


On Thu, Dec 06, 2007 at 09:52:19PM +0100, Martin Rubey wrote:
> Stephen Watt <[email protected]> writes:
> 
> > Dear Martin,
> > 
> > The quick answer is that I don't remember off the top of my head what 
> > restrictions exist on the domain parameter. 
> > 
> > I'm on the road now (just 10 minutes ago checked into my hotel) and so 
> > don't have access to a couple of things that would refresh my memory.
> 
> Stephen, do you happen to remember whether you have been aware of Gosper's
> algorithms for continued fractions?  I cannot be sure, but it seems that
> CONTFRAC is not using them, is it?
> 
> Martin