small problem with the manual.



Richard Fateman wrote:
>  
>
>   
>>> RJF --- take a look at this, does this look reasonable?
>>>       
>
> Not really.  Ratdiff(expr,a+b) is just something random.  It happens to be
> the same as ratdiff(expr,b). I don't know why you picked it, unless it was
> to illustrate the lack of error checking.
>
>
> Here's an explanation.
>
> Ratdiff is a function that is intended for use only on polynomials or ratios
> of polynomials in simple independent symbolic variables.  It does not use
> any knowledge of exponential, trigonometric, or other functions. It only
> knows about +, *, -, / and integer powers. Its second argument should be a
> simple symbolic variable that occurs in expr.  In exchange for your
> agreement to use ratdiff on this simple class of functions, ratdiff will
> compute the derivative blazingly fast. It is especially fast if expr is in
> CRE form, and you want it simplified to CRE form, because that is what it
> will return.
> If ratdiff is given a general form input, it will convert it to CRE form,
> compute the derivative, and convert back to general form.
>
> To illustrate its use, try this:
> showtime:all$
> z: rat( (a+b+c+d)^40)$ 
> ze: ratexpand(z)$
> ratdiff(z,c)$   observe the time.. I see 0.22 seconds on my system
> diff(ze,c)$     observe the time.. I gave up waiting.
>
> .......................
>
>
>
>   


Thanks for pointing me to the updated manual.
and Richard Fateman's explanation makes it clear as to what it is for. 
This should be in the manual.
Doug