Static inference in applicative languages

Static inference involves the compile-time analysis of programs, either with a view to optimizing program execution or ensuring that errors do not arise during program execution (e.g., type checking). We present a general framework for static inference of first-order applicative programs as well as several important instances of practical inference schemes.; This dissertation is in three parts: * We extend the techniques of Abstract Interpretation to applicative programs defined on nonflat domains. A semantic characterization of an "acceptable" abstract interpretation is developed. Two practical inference problems, that of minor signature and relevant clause inference, are solved within this framework.; We develop a type inference system for applicative programs that carries out type inference in the complete absence of any declarative information. A calculus of types called regular trees forms the basis of the type inference system. We relate the regular tree based type of a function to a canonical type extracted from the function; denotation.; We describe an implementation of strongly typed super-set of FEL, which incorporates a regular tree based type-checker, thereby demonstrating the practical utility of the type system.