🚀 go-pugleaf

Greffindel (ipsifend@eagle.cc.ukans.edu) wrote:
: (Feel free to shoot me down if I've got a detail slightly bobbled, but...)
: I read once (and subsequently verified) that with 3.1415927 as an
: approximation for pi, you can compute the circumference of the earth
: (or rather, a perfect sphere having the same diameter as the earth)
: and be off by only 6 or 7 inches... So it really begs the question--
: when, if ever, are we likely to need more than 10 digits of pi?  

There's more to pi than circles, and numerical calculations which use pi
in other contexts may need more accuracy.

A colleague described to me once a problem he once found (in someone
else's code, and not while he worked here) where pi had been accidentally
coded to too few places. (In some Fortran - all the places were there,
the "D0" after the "3.1415926535..." was missing). This caused a noticeable
ripple (maybe 140 dB down) when numerically simulating a signal of some
sort. This was sufficiently large to matter.

Of course here just a few more digits would have been enough, maybe 10
would have done. (I have seen cases where a few more than that are a
good idea in general precision requirements.)

--
Christopher M. Dearlove,            | chris.dearlove@gecm.com
GEC-Marconi Research Centre,        | Tel: 01245 242194  Int: +44 1245 242194
Gt. Baddow, Chelmsford, CM2 8HN, UK | Fax: 01245 242124  Int: +44 1245 242124