No label defined (Q4370164)

The paper gives an overview of symbolic and numerical methods for finding the real affine solutions of a system of polynomial equations. The special cases treated are \(0\)-dimensional and \(1\)-dimensional systems. The symbolic methods are primarily variants of the resultant approach and the Gröbner bases approach. Exact computation guarantees topological correctness, but it takes its toll in the form of high computation times. The main numerical method in the \(0\)-dimensional case is the homotopy method, in which the system of equations is perturbed to a system for which solutions are readily available, and then these solutions are transformed back to solutions of the original system. Geometric problems can occur when these continuation paths are not isolated. For \(1\)-dimensional systems, the approach described is to first map the problem in \(n\)-space to a problem in \(2\)-space. This is possible by a birational transformation derived from multivariant resultant methods. Now, in some rectangle in the plane, one has to find starting points for all the real components of the curve, and then to trace every component. Applications considered are the inverse kinematics problem in robotics, and the problem of changing representations (from boundary representation to constructive solid geometry) in solid modelling, which requires the evaluation of curves of intersection. The paper seems to be derived from lecture notes on this topic. The positive effect of this is that the approach is pedagogical and understandable for the non-expert, on the negative side it means quite some typos. For the entire collection see [Zbl 0880.00039].