High precision numerical accuracy in physics research

Concerns arise that the current standard of double-precision floating-point may no longer be sufficient for today's large-scale numerical simulations. One approach to solve this problem will be to switch to a wider floating-point format: the upcoming quadruple-precision standard is introduced and compared to currently available software-based approaches. Another complimentary approach is to use mathematical and algorithmic techniques to improve the accuracy of large floating-point programs and the confidence in the quality of the result. (c) 2005 Elsevier B.V. All rights reserved.