Generating Data Flow Analysis Algorithms from Modal Specifications

The paper develops a framework that is based on the idea that modal logic provides an appropriate framework for the speciication of data ow analysis (DFA) algorithms as soon as programs are represented as models of the logic. This can be exploited to construct a DFA-generator that generates eecient implementations of DFA-algorithms from modal speciications by partially evaluating a speciic model checker with respect to the specifying modal formula. Moreover, the use of a modal logic as speciication language for DFA-algorithms supports the compositional development of speciications and structured proofs of properties of DFA-algorithms. { The framework is illustrated by means of a real life example: the problem of determining optimal computation points within ow graphs.