InformedOverApproximations

This repository contains all the code necessary to implement the experiments in

• Antoine Aspeel, Antoine Girard, and Thiago Alves Lima. Title.

The code uses Julia-1.10 and requires a Gurobi license. Free Gurobi licenses are available for academics. More information here.

Use the Julia REPL to activate the environment by running:

import Pkg; Pkg.activate("Project.toml")

Then, to reproduce the experiments, run:

include("src/experiments.jl").

Other files:

• LinearSystems.jl contains a structure to represent linear systems.
• overapproximations.jl contains the functions to compute a linear over-approximation of a nonlinear function.
• systemLevelSynthesis.jl contains functions related to policy synthesis using system level synthesis. It also contains the methods to concretize an informed policy.
• utils.jl contains functions to sample a polytope. These are used in overapproximations.jl.