Division of Floating Point Expansions with an Application to the Computation of a Determinant
Marc Daumas (Laboratoire de l'Informatique du Parallelisme, France)
Claire Finot (Laboratoire de l'Informatique du Parallelisme, France)
Abstract: Floating point expansion is a technique for implementing multiple precision using a processor's floating point unit instead of its integer unit. Research on this subject has arised recently from the observation that the floating point unit becomes a more and more efficient part of modern computers. Many simple arithmetic operators and some very useful geometric operators have already been presented on expansions. Yet previous work included only a very simple division algorithm. We present in this work a new algorithm that allows us to extend the set of geometric operators with Bareiss' determinant on a matrix of size between 3 and 10. Running times with different determinant algorithms on different machines are compared with GMP, a very common multi-precision package.
Keywords: computational geometry, division, exact arithmetic, expansion, floating point, library, multiple precision
Categories: B.2, G.4