Our understanding of physics at the most fundamental scales is to a large degree based on our ability to make very precise measurements that can be compared with very precise predictions. In this talk, I will argue that -- in order to make such predictions -- a thorough understanding of the numbers and functions that occur in the result is crucial. In the simplest cases, the functions that occur are multiple polylogarithms, generalizations of logarithms. They are by now very well understood and allow us to reach precision that would be impossible using any other method; in one case up to the eighth subleading order in perturbation theory! In the next cases, we encounter elliptic integrals, where much progress was made in recent years. Beyond these, we encounter an infinite zoo of further functions involving integrals over Calabi-Yau manifolds, which are currently being explored.