Section: New Results

TreeKs: A Functor to Make Numerical Abstract Domains Scalable

Participant : Mehdi Bouaziz.

Relational numerical abstract domains do not scale up. To ensure a linear cost of abstract domains, abstract interpretation-based tools analyzing large programs generally split the set of variables into independent smaller sets, sometimes sharing some non-relational information. In [14] , we present a way to gain precision by keeping fully expressive relations between the subsets of variables, whilst retaining a linear complexity ensuring scalability.