1995-07-11 - Re: Moby ints [Re: Num Rat]

Header Data

From: Ray Cromwell <rjc@clark.net>
To: jim@acm.org
Message Hash: 962a942c73a4e62f8eb9b204a0720eab4c00df505938c696edea43997ace3b70
Message ID: <199507112331.TAA12573@clark.net>
Reply To: <199507111851.LAA18222@mycroft.rand.org>
UTC Datetime: 1995-07-11 23:32:29 UTC
Raw Date: Tue, 11 Jul 95 16:32:29 PDT

Raw message

From: Ray Cromwell <rjc@clark.net>
Date: Tue, 11 Jul 95 16:32:29 PDT
To: jim@acm.org
Subject: Re: Moby ints [Re: Num Rat]
In-Reply-To: <199507111851.LAA18222@mycroft.rand.org>
Message-ID: <199507112331.TAA12573@clark.net>
MIME-Version: 1.0
Content-Type: text/plain



The state of the art in multiprecision integer arithmetic is Scho"nhage.
Schonhage invented the all-integer Fast-Fourier-Transform based 
big-int multiplication method. An n-bit can be multiplied in O(n ln n) 
operations. This is a big improvement over the Karatsuba method which is
O(n^1.5) and the classical method O(n^2). Surprisingly, the constant
factor isn't that large. This can be combined with modmult techniques
for fast modexp routines. However, it's only worthwhile for large
numbers (>512 bits). At n=512, if your bigints are stored as polynomials
with a 32-bit radix, then N=512/32=16. 16^1.5 = 64, 16 * lg(16) = 64
(so the FFT method and the Karatsuba method are equivalent for numbers
of that size)

If you are dealing with 2048 or 4096 bit keys, it starts to look attractive.


Schonhage published a book in the last year, the result of more than 10 
years of research into this area. It's hard to get a hold of though, you
have to order it from germany.

95-133299: Schonhage, Arnold.  Fast algorithms : a multitape Turing
     machine implementation /  Mannheim : B.I. Wissenschaftsverlag, 
     c1994.  x,
     297 p. : ill. ; 25 cm.

 





Thread