1. Generalized theory of data flow analysis
(Joint work with Uday Khedker)

The generalized theory for bit vector data flow analysis develops the notion of information flow and shows that it is equally applicable to classical unidirectional data flows as well as to bi-directional data flows used in some advanced optimizations. The notion of an information flow path provides deep insights into the nature of data flows and their computational complexity. Tight complexity bounds are obtained by analysing information flow paths for a data flow problem. The classical theory of data flow analysis is shown to be a special case of the generalized theory.

The generalized theory has applications in designing efficient data flows. In one particular application, the solution cost of an efficient data flow designed using this theory was found to be 80 percent lower than that of a comparable data flow. The theory has also motivated interesting applications of data flow analysis (see Applications of program analysis techniques).

Selected publications

 POPL'93    TOPLAS'94    SN'99