Is something broken?



Thanks for the reply.  I've replied in context below.


On Mon, 2006-02-06 at 12:03 -0500, Stavros Macrakis wrote:
> Without looking at your problem closely, I see a couple of superficial
> problems:
>         trigexpandplus:true,expand,all; 
>         trigexpandtimes:true,expand,all;
>         trigexpand:true,expand,all;
> 
> What are you trying to do here?  These have the same effect as
> 
>        trigexpandplus:true;
>        trigexpandtimes:true;
>        trigexpand:true;
> 
> The "expand,all" part has no effect.

I'd seen some of these in the info documentation examples; as you say
they have no effect (and AFAIK "true" is the default for some of these
anyhow).  I've removed the extra stuff.

> 
>         g(x,t,x0):=sqrt(w/(2*%pi*sinh(w*t))) 
>                   *exp(-w/(2*sinh(w*t))*((x**2
>         +x0**2)*cosh(w*t)-2*x*x0));
>         
>         a:integrate(g(x1,t,x0)*x1**2*g(x0,b-t,x1),x1,minf,inf);
>         b:integrate(g(x1,t,x0)*g(x0,b-t,x1),x1,minf,inf);
>         c:trigsimp(a/b);
>         ev(%,x0=0);
> 
> You seem to be using "b" both as the name of a parameter and as an
> intermediate result in your calculation.  I doubt that this is what
> you intend.
> 
> In particular, it has the effect in the ev(...) expression of
> RE-EVALUATING b and substituting it back into the expression. 
> 
Oops.  You're quite right.  I have changed the b:integrate... as you
suggested and now I get:

 trigsimp(%o11);
			   sinh((b - t) w) sinh(t w)
(%o13) 	   ---------------------------------------------------------
	   w cosh((b - t) w) sinh(t w) + w sinh((b - t) w) cosh(t w)

The numerator is what I want; however, the denomenator will simplify to 
w sinh (b w) if the cosh((b - t)w) and sinh((b -t)w) factors are
expanded.

David