On Sat, Nov 26, 2011 at 08:52, Oliver Kullmann <O.Kullmann at swansea.ac.uk>wrote:
> ...So "is(equal(a,b))" is a form of "simplify(t)" for t : a=b.
>
is(equal(a,b)) is simply equivalent to is(ratsimp(a-b)=0). For example,
is(equal(sin(x)^2+cos(x)^2,1)), is Unknown. This could surely be improved.
For example, for a large class of expressions, it should be possible to
determine the appropriate N such that if taylor(expr) of order N is 0, then
expr must be identically 0. Trigsimp or exponentialize/radcan would also
handle many cases. Unfortunately, the running time of radcan can be very
large in some cases.
And I would never claim that ratsimp(a-b) is a good general simplifier in
the sense of putting expressions in a human-friendly form.
-s