Lecture Notes in Computer Science Volume 1563, 1999, pp 302-312
Date: 12 Apr 2002

Fast Computations of the Exponential Function

* Final gross prices may vary according to local VAT.

Get Access

Abstract

In this paper we present an algorithm which shows that the exponential function has algebraic complexity O(log2 n), i.e., can be evaluated with relative error O(2-n ) using O(log2 n) infinite-precision additions, subtractions, multiplications and divisions. This solves a question of J. M. Borwein and P. B. Borwein [9].

The best known lower bound for the algebraic complexity of the exponential function is Ω(log n).

The best known upper and lower bounds for the bit complexity of the exponential function are O(μ(n) log n) [10] and Ω(ν(n)) [4], respectively, where μ(n) denotes an upper bound and ν(n) denotes a lower bound for the bit complexity of n-bit integer multiplication.

The presented algorithm has bit complexity O(μ(n) log n).