Newsgroups: sci.math,sci.answers,news.answers
Path: senator-bedfellow.mit.edu!bloom-beacon.mit.edu!hookup!multiverse!ragnarok.oar.net!malgudi.oar.net!news.sprintlink.net!howland.reston.ans.net!swrinde!cssun.mathcs.emory.edu!nntp.msstate.edu!news.memphis.edu!lamarck.sura.net!fconvx.ncifcrf.gov!news.mathworks.com!newsfeed.internetmci.com!in1.uu.net!spool.mu.edu!torn!watserv3.uwaterloo.ca!undergrad.math.uwaterloo.ca!neumann.uwaterloo.ca!alopez-o
From: [email protected] (Alex Lopez-Ortiz)
Subject: sci.math FAQ: How to compute Pi?
Summary: Part 12 of many, New version,
Originator: [email protected]
Message-ID: <[email protected]>
Sender: [email protected] (news spool owner)
Approved: [email protected]
Date: Fri, 17 Nov 1995 17:14:38 GMT
Expires: Fri, 8 Dec 1995 09:55:55 GMT
Reply-To: [email protected]
Nntp-Posting-Host: neumann.uwaterloo.ca
Organization: University of Waterloo
Followup-To: sci.math
Lines: 131
Xref: senator-bedfellow.mit.edu sci.math:124786 sci.answers:3474 news.answers:58051



Archive-Name: sci-math-faq/specialnumbers/computePi
Last-modified: December 8, 1994
Version: 6.2


How to compute digits of pi ?



  Symbolic Computation software such as Maple or Mathematica can compute
  10,000 digits of pi in a blink, and another 20,000-1,000,000 digits
  overnight (range depends on hardware platform).

  It is possible to retrieve 1.25+ million digits of pi via anonymous
  ftp from the site wuarchive.wustl.edu, in the files pi.doc.Z and
  pi.dat.Z which reside in subdirectory doc/misc/pi. New York's
  Chudnovsky brothers have computed 2 billion digits of pi on a homebrew
  computer.

  There are essentially 3 different methods to calculate pi to many
  decimals.

   1. One of the oldest is to use the power series expansion of atan(x)
      = x - x^3/3 + x^5/5 - ... together with formulas like pi =
      16*atan(1/5) - 4*atan(1/239) . This gives about 1.4 decimals per
      term.

   2. A second is to use formulas coming from Arithmetic-Geometric mean
      computations. A beautiful compendium of such formulas is given in
      the book pi and the AGM, (see references). They have the advantage
      of converging quadratically, i.e. you double the number of
      decimals per iteration. For instance, to obtain 1 000 000
      decimals, around 20 iterations are sufficient. The disadvantage is
      that you need FFT type multiplication to get a reasonable speed,
      and this is not so easy to program.

   3. The third, and perhaps the most elegant in its simplicity, arises
      from the construction of a large circle with known radius. The
      length of the circumference is then divided by twice the radius
      and pi is evaluated to the required accuracy. The most ambitious
      use of this method was successfully completed in 1993, when
      H. G. Smythe produced 1.6 million decimals using high-precision
      measuring equipment and a circle with a radius of a staggering
      nine hundred and fifty miles.



  References

  P. B. Borwein, and D. H. Bailey. Ramanujan, Modular Equations, and
  Approximations to pi American Mathematical Monthly, vol. 96, no. 3
  (March 1989), p. 201-220.



  J.M. Borwein and P.B. Borwein. The arithmetic-geometric mean and fast
  computation of elementary functions. SIAM Review, Vol. 26, 1984, pp.
  351-366.



  J.M. Borwein and P.B. Borwein. More quadratically converging
  algorithms for pi . Mathematics of Computation, Vol. 46, 1986, pp.
  247-253.



  Shlomo Breuer and Gideon Zwas Mathematical-educational aspects of the
  computation of pi Int. J. Math. Educ. Sci. Technol., Vol. 15, No. 2,
  1984, pp. 231-244.



  David Chudnovsky and Gregory Chudnovsky. The computation of classical
  constants. Columbia University, Proc. Natl. Acad. Sci. USA, Vol. 86,
  1989.



  Y. Kanada and Y. Tamura. Calculation of pi to 10,013,395 decimal
  places based on the Gauss-Legendre algorithm and Gauss arctangent
  relation. Computer Centre, University of Tokyo, 1983.



  Morris Newman and Daniel Shanks. On a sequence arising in series for
  pi . Mathematics of computation, Vol. 42, No. 165, Jan 1984, pp.
  199-217.



  E. Salamin. Computation of pi using arithmetic-geometric mean.
  Mathematics of Computation, Vol. 30, 1976, pp. 565-570



  David Singmaster. The legal values of pi . The Mathematical
  Intelligencer, Vol. 7, No. 2, 1985.



  Stan Wagon. Is pi normal? The Mathematical Intelligencer, Vol. 7, No.
  3, 1985.





  A history of pi . P. Beckman. Golem Press, CO, 1971 (fourth edition
  1977)



  pi and the AGM - a study in analytic number theory and computational
  complexity. J.M. Borwein and P.B. Borwein. Wiley, New York, 1987.




    _________________________________________________________________



   [email protected]
   Tue Apr 04 17:26:57 EDT 1995