Talk:Lehmer's GCD algorithm

Draft

Hope I've got the algorithm right. Maybe someone could write up a program to test it. We should also make note of the numerous variants on this algorithm eg. Jebelean etc. sometime. Jafet 17:02, 21 December 2006 (UTC)

This looks like an algorithm well-suited for Matlab or some functional programming language, with multiple assignment and matrices. My question is, how on earth do you tell whether two numbers have the same number of digits in constant time, unless you're keeping track of the number of digits independently somehow (e.g., if this is part of a bignum library)? Without some kind of "similar_size(a,b)" primitive, we're not going to be able to write concise code for this algorithm, are we? --Quuxplusone 03:18, 23 December 2006 (UTC)

Well, if you're trying to get the GCD of some really large numbers, chances are you are using some sort of bignum library. If not, your numbers are small, and the Euclidean algorithm will do for you; its speed won't be such a big concern with small numbers (and the simplicity of the Euclidean algorithm is welcome). -- Why Not A Duck 02:25, 31 July 2007 (UTC)