Dr. Dobb's Python-URL! - weekly Python news and links (Mar 17)
Michael Hudson
mwh at python.net
Thu Mar 20 06:09:48 EST 2003
Tim Peters <tim.one at comcast.net> writes:
> [Tim]
> > Also, AFAIK, no absolutely normal number is known (constructed or
> > not). This is curious because the set of non-normal reals has
> > measure 0 ("almost all" reals are absolutely normal).
>
> [Michael Hudson]
> > <hand-waving>
> > But for a number to be "known", it must be comptuable in some sense
> > and there are only countable many Turing machines...
> > </hand-waving>
>
> That's an excellent argument for why we can't show an example of an
> uncomputable real, but really no more convincing wrt normality than wrt
> irrationality.
Well, I guess. But rationality seems a much more tractable notion
than normality -- it seems "finite" in some sense. I'm not sure
rationality is that good an example, as having irrationals is somehow
the point of the reals... Transcendence would be a better example,
but I'm an algebraist -- what's a transcendent number, again? <wink>.
> After all, we do have examples of numbers normal wrt specific bases,
> and there's no obvious way in which absolute normality "should be"
> intractable.
> > Why does anyone care about normality, anyway?
>
> I think because attempts to define what random means lead naturally
> to normality. If, e.g., pi is shown *not* to be normal, then it
> would be very hard to believe it's random in any reasonable sense.
> If it is shown to be absolutely normal, I expect reasonable people
> would disagree about whether that's strong enough to conclude it's
> random, though.
Except that pi clearly *isn't* random... but this leads very quickly
into the "the smallest integer than can't be described in less than
100 characters" type paradox, and I don't care about them, either :-)
Cheers,
M.
--
Well, you pretty much need Microsoft stuff to get misbehaviours
bad enough to actually tear the time-space continuum. Luckily
for you, MS Internet Explorer is available for Solaris.
-- Calle Dybedahl, alt.sysadmin.recovery
More information about the Python-list
mailing list