Inferring Symbolic Automata

We study the learnability of symbolic finite state automata (SFA), a model shown useful in many applications software verification. The state-of-the-art literature on this topic follows query learning paradigm, and so far all obtained results are positive. provide necessary condition for efficient SFAs from which we obtain first negative result. main focus our work lies under paradigm identification limit using polynomial time data, its strengthening identifiability, concerned with existence systematic set characteristic samples learner can correctly infer target language. sufficient SFAs. From these conditions derive positive performance algorithm is typically bounded as function size representation Since SFAs, general, do not have canonical form, there trade-offs between complexity predicates transitions number transitions, start by defining measures revisit procedures analyze them according to measures, paying attention special forms SFAs: normalized neat well over monotonic effective Boolean algebra. This an extended version paper same title published CSL'22.