Richard Fateman wrote:
> The reason not to use the general formula is that it won't do some examples
> that can be done by expansion. At least that is my supposition. Reading
> Moses' online
> thesis at MIT could be a check on this.
A quick read through the thesis, doesn't point to anything special,
especially wrt to 6.
At this point, there is nothing really suggesting one way or another, so
I'm inclined to leave everything as is.
>
> The notion that you have a form that looks like u*du or u^n*du in general
> can demand more than simple pattern matching. You might need to factor the
> integrand in some cases. Or maybe some effective method can be substituted
> in the program for the hackish "test division" used by SIN. I once assigned
> such a task in a grad class; I wrote a solution too; perhaps I can recreate
> it.
This would be interesting if you still have it.
Ray