At 08:04 PM 5/15/01 +0100, you wrote:
>From: "Jean-Louis Couturier"
>
> > Robert Shaw wrote:
> >
> > >When I was a student we spent one rainy afternoon working
> > >out what numbers we could get from our calculators using
> > >only the unitary buttons, starting fom 0 and never touching
> > >another number afterwards.
> >
> > >It took a dozen maths students twenty minutes to work
> > >out how to generate all the integers, then we went after
> > >the rationals.
> >
>
>
> > Now THAT sounds like fun!
> >
> > When you say never touching another number, does that include
> > zero?
>
>Yes.
>
>When you switch the calculator on you get a zero, but after that
>you never touch another number key. These were scientific
>calculators so we had ^2,^3,^-1,cos, sin, tan, cosh, sinh, tanh,
>ln, log and their inverses plus factorial.



For anyone who is interested, this is what I was playing with last night:

MirCalc, 9x, 06 Nov 98, 31K, Multi-digit calculator, up to 10,000 digits.
<http://www.winsite.com/edu/calculator/page10.html>

It has all the above functions.  For some reason when I was closing my 
e-mail because I really needed to finish grading finals and go to bed, I 
saw the MirCalc icon staring at me and started playing with it, and noticed 
the interesting behavior of cos( cos( cos( ...


> >  If so, 3 is going to take some work.
>
>Actually, there's a trivial way of doing it. By composing
>the allowed functions you can produce the function x+1.
>Repeat indefinitely and you can produce all integers.



Which is what you do when developing the nonnegative integers in a course 
on set theory or other fundamentals:  starting with 0 and using a 
"successor of" function, e.g., "ss0 + ss0 = ssss0" is the way of writing "2 
+ 2 = 4."  By the time you get that far, though, you're near the end of the 
book and the semester is almost over.



>Similarly ln(sqrt(exp(x)))=x/2 so once you've got all
>integers you can easily get all halves, quarters, eights, etc.
>
>That means you can get arbitarily close to any number.
>
>Pi is ln(square(square(exp(arctan(exp(0))))))
>though there may be quicker ways to get it.
>
> > I have been a mth student in the past.  I can imagine that this
> > might sound boring to most of you.



Or that you and I (and apparently others here) are the kind of folk who 
find this kind of thing a fun distraction when we are bored.



>  Kind of like when I tried to
> > solve Fermat's theorem...
> >
>--
>Robert



Of course, the thing I found interesting was that it was oscillating back 
and forth but fairly rapidly converging to

lim (cos^n (0)) = 0.73908 51332 15160 64165 53120 87673 87340 40134 11758 
90075 74649 65680 63577 ...


I guess the next step is to try to express the exact value of the above in 
closed form.  One starting place might be

cos^n (0) = (2^-n) Sum C(n,j)  (summation over j | j = 0, 1, ..., n)



-- Ronn!  :)


Reply via email to