Tutorial: Exact Numerical Computation in Algebra and Geometry

Many problems in Computational Science & Engineering (CS&E) are defined on the continuum. Standard algorithms for these problems are numerical and approximate. Their computational techniques include iteration, subdivision, and approximation. Such techniques are rarely seen in exact or algebraic algorithms. In this tutorial, we discuss a mode of computation called exact numerical computation (ENC) that achieves exactness through numerical approximation. Through ENC, we can naturally incorporate iteration, subdivision and approximation into the design of exact algorithms for computer algebra and computational geometry. Such algorithms are both novel and practical. This tutorial on ENC is divided into three equal parts: (a) Zero Problems (b) Explicitization Problems (c) Techniques and Complexity Analysis of Adaptivity