Regular Inference for Reactive Systems

Models of reactive systems play a central role in many techniques for verification and analysis of reactive systems. Both a specification of the system and the abstract behavior of the system can be expressed in a formal model. Compliance with the functional parts in the specification can be controlled in different ways. Model checking techniques can be applied to a model of the system or directly to source code. In testing, model-based techniques can generate test suites from specification. A bottleneck in model-based techniques is however to construct a model of the system. This work concerns a technique that automatically constructs a model of a system without access to specification, code or internal structure. We assume that responses of the system to sequences of input can be observed. In this setting, so called regular inference techniques build a model of the system based on system responses to selected input sequences. There are three main contributions in this thesis. The first is a surve