of little minds — Ralph Waldo Emerson Lecture 8 DEGENERACY AND PERTURBATION

The inputs that satisfy these assumptions may be said to be non-degenerate; otherwise they are degenerate. Implementors are left to their own devices to handle degenerate cases. Some papers may not even list the degenerate inputs. Thus, Forrest [2] deplores the “plethora of special cases ... glossed over by theoretical algorithms”. Forrest, and also Sedgewick [6], calls this the “bugbear” of geometric algorithm implementation. In this lecture, we explore some general techniques that can help bridge this particular gap between theoretical algorithms and their implementation. Notice we do not address other possible gaps, such as the under-specification of data representation.