Algorithms for Arbitrary Precision Floating Point Arithmetic

  • Douglas M. Priest

Tenth Symposium on Computer Arithmetic |

Published by IEEE Computer Society Press

The author presents techniques for performing computations of very high accuracy using only straightforward floating-point arithmetic operations of limited precision. The validity of these techniques is proved under very general hypotheses satisfied by most implementations of floating-point arithmetic. To illustrate the applications of these techniques, an algorithm is presented which computes the intersection of a line and a line segment. The algorithm is guaranteed to correctly decide whether an intersection exists and, if so, to produce the coordinates of the intersection point accurate to full precision. The algorithm is usually quite efficient; only in a few cases does guaranteed accuracy necessitate an expensive computation.