We study a type system equipped with universal types and equire-cursive types, which we refer to as F͘. We show that type equality may be decided in time O(nlog n), an improvement over the previous known bound of O(n2 ). In fact, we show that two more general problems, namely entailment of type equations and type unification, may be decided in time O(nlog n), a new result. To achieve this bound, we associate, with every F͘ type, a first-order canonical form, which may be computed in time O(nlogn). By exploiting this notion, we reduce all three problems to equality and unification of first-order recursive terms, for which efficient algorithms are known.

Defunctionalization is a program transformation that aims to turn a higher-order functional program into a first-order one, that is, to eliminate the use of functions as first-class values. Its purpose is thus identical to that of closure conversion. It differs from closure conversion, however, by storing a tag, instead of a code pointer, within every closure. Defunctionalization has been used both as a reasoning tool and as a compilation technique.Defunctionalization is commonly defined and studied in the setting of a simply-typed λ-calculus, where it is shown that semantics and well-typedness are preserved. It has been observed that, in the setting of a polymorphic type system, such as ML or System F, defunctionalization is not type-preserving. In this paper, we show that extending System F with guarded algebraic data types allows recovering type preservation. This result allows adding defunctionalization to the toolbox of type-preserving compiler writers.

This paper investigates the cooperation of online and off-line handwriting word recognition systems. Our goal is to improve a mature online recognition system by, exploiting the complementary information present in the off-line representation built from online signal. After describing the online and off-line HMM based handwriting recognition systems, we propose a formal framework, which allows describing different strategies for combining the two HMM. These schemes are then evaluated on the UNIPEN database, both for isolated character and word recognition tasks.