IntervalMDPAbstractions.jl
Abstraction-based verification and synthesis via IMDP targets
IntervalMDPAbstractions.jl provides abstraction-based methods for verifying stochastic systems and synthesizing correct-by-construction controllers via IMDP-inspired target models.
Features
- Abstractions from several stochastic system classes (linear, nonlinear, uncertain PWA, Gaussian process, and switched)
- Target models including IMDPs, odIMDPs, and mixtures of odIMDPs
- Integration with IntervalMDP.jl for verification and synthesis on abstract models
- End-to-end abstraction and policy synthesis workflows
Workflow
- Model the stochastic dynamics and property specification
- Discretize state and input spaces and construct a target abstraction model
- Build the abstract model and solve verification or synthesis problems
- Evaluate lower and upper satisfaction probability bounds when needed