Many years ago, dealing with special functions was a tedious, time-consuming, and
errorprone task, which required long training, skillful manipulations and structural
insight. This is now mostly obsolete: today there is an increasing number of symbolic
algorithms available which are capable of dealing with special functions~\cite{AAR}.
In particular, questions about orthogonal polynomials which often arise in numerical
mathematics can be answered by packages for holonomic functions. With these programs,
dealing with special functions is straight-forward, fast, and reliable.
Recently we used an algorithm developed by Manuel Kauers~\cite{MK} for deriving a three
term recurrence for the polynomial of best uniform approximation to $1/x$ on a finite
interval~\cite{JM,Rivlin}. This recurrence relation entered the analysis of an algebraic
multilevel iteration method. We will give an introduction to the underlying symbolic
algorithm and its scope, and sketch our application to algebraic multigrid methods.