Higher-Order and Symbolic Computation, 16(3)
Editorial
David Basin, Olivier Danvy, Julian Padget and Carolyn Talcott
The present issue is composed of three contributed articles.
"Formal Foundations of Operational Semantics" presents a
formalization, using the PVS proof system, of a very general form of
the CIU context lemma for higher-order languages with effects. The
paper describes techniques for formalizing higher-order concepts such
as variable binding that closely parallel informal reasoning. Several
hidden assumptions in the original informal proof were discovered. The
paper serves both as a roadmap for the full PVS axiomatization and
proof script, and as a guide for developing further formal tools for
reasoning about higher-order languages with effects.
"Comparing Parallel Functional Languages: Programming and Performance"
presents a detailed evaluation of three state-of-the-art parallel
functional languages: PMLS, GpH, and Eden. The authors consider both
the different ways in which the three languages support communication
and the specification and control of parallelism, as well as pure
performance, by means of three applications: matrix multiplication, a
linear-equation solver, and a ray tracer, executed on a Beowulf
cluster. While moderate improvements result without changing the
sequential code, tuning load distribution and thread granularity are
shown to behighly beneficial.
"Strong Normalization from Weak Normalization" provides a systematic
account of techniques for inferring strong normalization from weak
normalization in the lambda-calculus. Common to all these techniques
is the fact that they all use syntactic translations from lambda-terms
to lambda-I-terms. The authors identify that all these translations
can be factored through Klop's i translation and they provide in all
cases the corresponding complementary translation. More generally,
they show that all these translations are instances of Sørensen's
permutative inner interpretations, and they present a detailed
comparison of Gandy, Klop, and Girard's techniques for reducing strong
normalization to weak normalization.
|
|
|