>> I see a lot of non-reduced exponents of the form:
>> 
>>                      a log(z)
>>                    %e        
>
>This is "by definition" equal to z^a.  I'd think it is rather a question of
>output aesthetics whether we want to write z^a instead -- I'd say yes, we
>should output z^a if a does not have a log factor.  For computation I'd believe
>that %e^(a*log z) is better, especially because of
>
>%e^(log x*log z) = x^log z = z^log x

Barry Trager has pointed out to me that the integrator always wants
expressions like x^n to be represented as exp(n*log(x)) since the
integration algorithm recurses on the tower of generators involving
exps and logs. The infinite recursion bug is some kind of a bug in
the implementation but not related to simplification.

So, obviously, I don't have a clue. Ah, if only the sources were
literate. We could read what it was intended to do.... sigh.

Tim


_______________________________________________
Axiom-developer mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/axiom-developer

Reply via email to