Polyhedral invariants have long been used in static program analysis, and are either computed in restricted classes of polyhedra, or obtained through heuristics, as though it were impossible to find a general algorithm for finding such invariants.
There have been attempts at proving that no such algorithm exists. The goal of this internship is to think further about the issue.