Some new algorithms for high-precision computation of Euler's constant

49. R. P. Brent and E. M. McMillan, Some new algorithms for high-precision computation of Euler's constant, Mathematics of Computation 34 (1980), 305-312. MR 82g:10002.

Abstract: dvi (3K), pdf (80K), ps (28K),

Paper: pdf (617K).

Abstract

We describe several new algorithms, more efficient than those of [40], for the high-precision computation of Euler's constant gamma = 0.577...

Using one of the algorithms, which is based on an identity involving Bessel functions, gamma has been computed to 30,100 decimal places. By computing their regular continued fractions, we show that, if gamma or exp(gamma) is of the form P/Q for integers P and Q, then |Q| > 1015000. The computations were performed using the first author's MP package.

Erratum

On page 310, in the formula for Vp(z) after equation (17), (z/k!)p should be (zk/k!)p.

Comments

It is not known whether gamma or exp(gamma) is rational; hence the interest in lower bounds on Q.

A nice introductory paper on Euler's constant by Gourdon and Sebah is available here. In 1999, Demichel and Gourdon used formula (13) of [49] to compute Euler's constant to 108,000,000 decimal digits. In December 2006, Yee found 116,580,041 decimal digits using the same formula (evaluated with binary splitting). This and other record computations are summarised here.

More recently, Kreckel found 900,000,000 decimal digits. As at July 2007 the record appears to be 5,000,000,000 decimal digits by Shigeru Kondo.

Formula (13) of [49] is implemented in Zimmermann's MPFR package.

An interesting connection with the work of Ramanujan is described in Brent [139].

McMillan is better known for the discovery of plutonium: see Jackson and Panofsky, "Edwin Mattison McMillan 1907-1991", Biographical Memoirs of the National Academy of Sciences (USA), 69 (1996).

Go to next publication

Return to Richard Brent's index page