Am Montag, den 02.08.2010, 23:39 -0600 schrieb Robert Dodier:
> On 8/2/10, Paul Bowyer <pbowyer at olynet.com> wrote:
>
> > So it looks like there is a partial fix for the bug because 0^0 produces
> > an error message but when 0^i or when
> > assume i >= 0 is given followed by 0^i, the error still shows up. Am I
> > correct in this assumption?
>
> I seem to recall from a CVS log message that the 0^foo
> problem was only partially solved, because it turned out
> to be more complicated than anticipated.
> I don't know what are the plans of Dieter Kaiser who was
> working on it.
Yes, the problem is partially solved. I had to revert parts because of a
problem in the the routine timesin. Meanwhile I have solved this problem
and it seems to work. But as I have posted already we get new problems
too.
Maxima will handle 0^a like any other symbolic expression. Therefore, we
get e. g. 0^a-0^a --> 0. We get some more correct results, but we get
wrong results too.
Furthermore, the initial problem of sum(0^i, i, 0, n) -> 0 is not
solved. We will get again 0. Perhaps this happens because of a
simplification like 0^a-0^a --> 0.
We might neglect this problem and commit the code, but we might get more
subtle problems too. It is always problematic to work on the main
simplifier.
So, at the moment I do not know what is the best.
Dieter Kaiser