INTLAB - The Matlab Toolbox for Reliable Computing

The focus of INTLAB is to produce rigorously correct results based on floating-point operations. Rigorous means taking all possible sources of errors into account. INTLAB consists of various toolboxes including Hessians, Taylor expansions, Affine arithmetic and more. The goal of all algorithms is to produce correct results - in a mathematical sense - in a computing time not too far from ordinary numerical algorithms. In particular input data may be afflicted with tolerances. In that case all problems within the tolerances are proved to be solvable together with an inclusion of all solutions.

We will present basic design concepts of INTLAB, a little about the mathematics behind it, and, of course, examples. INTLAB is entirely written in Matlab and thus portable; the recent Version 9 runs on Octave as well.

Note: Visitors from outside NIST must contact Cathy Graham; (301) 975-3800; at least 24 hours in advance.