Gauss-Legendre algorithm

From Wikipedia, the free encyclopedia

The Gauss-Legendre algorithm is an algorithm to compute the digits of π. It is notable for being rapidly convergent, with only 25 iterations producing 45 million correct digits of π. However, the drawback is that it is memory intensive and it is therefore sometimes not used over Machin-like formulas.

The method is based on the individual work of Carl Friedrich Gauss (1777-1855) and Adrien-Marie Legendre (1752-1833) combined with modern algorithms for multiplication and square roots. It repeatedly replaces two numbers by their arithmetic and geometric mean, in order to approximate their arithmetic-geometric mean.

The version presented below is also known as the Brent-Salamin (or Salamin-Brent) algorithm; it was independently discovered in 1975 by Richard Brent and Eugene Salamin. It was used to compute the first 206,158,430,000 decimal digits of π on September 18 to 20, 1999, and the results were checked with Borwein's algorithm.

1. Initial value setting:

a_0 = 1\qquad b_0 = \frac{1}{\sqrt{2}}\qquad t_0 = \frac{1}{4}\qquad p_0 = 1\!

2. Repeat the following instructions until the difference of a_n\! and b_n\! is within the desired accuracy:

a_{n+1} = \frac{a_n + b_n}{2}\!,
b_{n+1} = \sqrt{a_n b_n}\!,
t_{n+1} = t_n - p_n(a_n - a_{n+1})^2\!,
p_{n+1} = 2p_n\!.

3. π is approximated with a_n\!, b_n\! and t_n\! as:

\pi \approx \frac{(a_n+b_n)^2}{4t_n}\!.

The first three iterations give:

3.140...\!
3.14159264...\!
3.14159265358979...\!

The algorithm has second-order convergent nature, which essentially means that the number of correct digits doubles with each step of the algorithm.

Contents

[edit] Mathematical background

[edit] Limits of the arithmetic-geometric mean

The arithmetic-geometric mean of two numbers, a_0\! and b_0\!, is found by calculating the limit of the sequences a_{n+1}={a_n+b_n \over 2}\!, b_{n+1}=\sqrt{a_n b_n}\!, which both converge to the same limit. If a_0=1\! and b_0=\cos\phi\! then the limit is {\pi \over 2K(\sin\phi)}\! where K(k)\! is the complete elliptic integral of the first kind

K(k) = \int_0^{\frac{\pi}{2}} \frac{d\theta}{\sqrt{1-k^2 \sin^2\theta}}\!.

If c_0 = \sin\phi\!, c_{i+1} = a_i - a_{i+1}\!. then

\sum_{i=0}^{\infty} 2^{i-1} c_i^2 = 1 - {E(\sin\phi)\over K(\sin\phi)}\!

where E(k)\! be the complete elliptic integral of the second kind:

E(k) = \int_0^{\frac{\pi}{2}}\sqrt {1-k^2 \sin^2\theta}\ d\theta\!.

Gauss knew of both of these results.[1]

[edit] Legendre’s Identity

For \phi\! and \theta\! such that \phi+\theta={1 \over 2}\pi\! Legendre proved the identity:

K(\sin \phi) E(\sin \theta ) + K(\sin \theta ) E(\sin \phi) - K(\sin \phi) K(\sin \theta) = {1 \over 2}\pi\![1]

[edit] Gauss-Legendre Method

The values \phi=\theta={\pi\over 4}\! can be substituted into Legendre’s Identity and the approximations to K, E can be found by terms in the sequences for the arithmetic geometric mean with a_0=1\! and b_0=\sin{\pi \over 4}=\frac{1}{\sqrt{2}}\!.[1]

[edit] See also

[edit] References

  1. ^ a b c Brent, Richard, Traub, J F, ed., “Multiple-precision zero-finding methods and the complexity of elementary function evaluation”, Analytic Computational Complexity (New York: Academic Press): 151–176, <http://wwwmaths.anu.edu.au/~brent/pub/pub028.html>. Retrieved on 8 September 2007