behavior of ind, = and equal, also fiddling with limit
Subject: behavior of ind, = and equal, also fiddling with limit
From: Richard Fateman
Date: Thu, 24 Sep 2009 10:09:19 -0700
Stavros Macrakis wrote:
> On Thu, Sep 24, 2009 at 11:45 AM, Richard Fateman
> <fateman at cs.berkeley.edu <mailto:fateman at cs.berkeley.edu>> wrote:
>
> is (ind=ind) returns true
> is (equal(ind,ind)) returns false.
>
>
> I believe this is correct, though confusing.
I sort of agree, except that the documentation says that is
(equal(a,b)) is computed by testing (0=ratsimp(a-b)). And that would
result in true, not false.
The limit questions were really asking about ind [for indefinite] and
und [for undefined]. If we need both, are we handling them correctly?
{I think not. (und-und -> 0, ind-ind->0, and for that matter, inf-inf
-> 0 in the simplifier)}
We also have, at least in some Lisps, not-a-numbers ... NaNs -- single,
double, maybe extended, and there are actually a huge number
of distinguishable NaNs.
NaN is specified by a reserved exponent, but the fraction part can be
used to store info.
I'm still thinking about limit sets vs. intervals :)