> Ron Frank came in, looked at the copy on his desk, circled a
> digit towards the end of the long number and wrote, "Are you sure about
> this digit?" and put it back on the Desk Jeff was using.

This sounds very much like an anecdote in Richard Feynman's *Surely You're
Joking, Mr. Feynman?* But then my memory is going too.



On Thu, May 10, 2012 at 1:52 PM, Joey K Tuttle <[email protected]> wrote:

> Upon looking a little more at the page -
>
>   http://zenwerx.com/projects/pi-digits/pi/
>
> mentioned in the post below, I was amused to note that the last 138 of
> the 4,194,304 places of pi displayed on that page are random and/or
> wrong. Problems begin about decimal place 4194166 which is shown as 9
> but should be 7. I suppose such errors are not very high up on the list
> of misrepresentations on web pages, but in my case it caused a smile and
> memory of an event that took place some 40 years ago.
>
> Jeffery Shallit ( http://www.cs.uwaterloo.ca/~shallit/ ) was a summer
> intern at the IBM Philadelphia Scientific Center in 1972 (I think). He
> was working on some high precision computations using APL. I can't
> remember whether it was a precise evaluation of pi to several thousand
> places, or perhaps Mersenne prime #24  ( _1 + 2^19937x ) - but in any
> case, Jeffery had succeeded in printing out this lengthy number, and
> left a copy of the pages of digits on several people's desks. The next
> morning, Ron Frank came in, looked at the copy on his desk, circled a
> digit towards the end of the long number and wrote, "Are you sure about
> this digit?" and put it back on the Desk Jeff was using.
>
> That afternoon, Jeff came in and discovered the note and raced into
> Ron's office saying, "How do you know that digit is wrong??" To which
> Ron calmly replied, "I have no idea if it is correct or not - I was just
> asking if you were sure...."
>
> I was (am still) impressed with Jeffery's work because he did it in an
> 80Kbyte APL workspace. Hard to imagine these days when it is routine to
> work with a hundred thousand times that much memory.... Although I note
> that J only consumes 80704 bytes to calculate _1 + 2^19937x   that's
> impressive too.
>
> I've copied Jeffery on this note, and wish I had an address to pass my
> memory past Ron. In my experience, it seems likely to me that either
> they wouldn't have any memory of such an event, or might have versions
> quite different from my memory. It is strange/interesting how our
> memories work.
>
> On 2012/05/08 09:44 , Joey K Tuttle wrote:
> > Being a fan of things pi, I'm wondering about several things in your
> post.
> >
> > The URL you give points to a site purporting to have 4 (not 50) million
> > digits of pi. Maybe I missed a pointer to a larger dataset.
> >
> >
> >
> > On 2012/05/08 03:56 , Joe Bohart wrote:
> >
> > I've load 50 million integer digits of pi and trying to do a moving
> average
> > on them.
> >
> > NB. data from http://zenwerx.com/projects/pi-digits/pi/
> > NB. used perl to write each digits on 1 line of file data
> >
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to