A language over the Cartesian product of a finite set of component alphabets is projective if it is closed under projection equivalence. Two languages are projection equivalent if their projections into each of the component alphabets are equivalent, up to stuttering (a finite subsequence of identical elements). The author shows that for regular languages (along with star-free and infinite languages), projective languages coincide precisely with separable languages (languages obtained using parallel composition and intersection of stuttering-closed component languages).
Peled also discusses how projective languages can be viewed as properties of various temporal logics that are used in the specification and verification of concurrent programs. The paper shows that projective properties are important for partial-order model-checking methods used in the analysis of concurrent systems. The properties and methods developed in this paper are significant and have implications and potential applications in concurrent computer architectures and distributed databases. The paper is clearly written and will be accessible to readers familiar with the theory of formal languages and temporal logic.