Stefan Muller
Assistant Professor of Computer Science at Rice University
Expressive and Efficient Functional Semantics for Interaction
Abstract:
Prior work in
Functional Reactive Programming has sought to reconcile the seemingly
imperative notion of interaction with the coding conventions and
guarantees of functional programming, but this has often been at the
cost of efficiency or expressiveness. We seek to combine the best of
both worlds. To ensure predictable time and space performance, we use
an imperative implementation. To regain a functional semantics, we must
restrict the way in which streams can be used, but we wish to do so
without restricting the syntax, thus hindering the expression of safe
programs. Instead, we develop a linear type system that treats streams
as consumable resources whose contents can be split up but not
duplicated. We hypothesize that well-typed programs in this system can
be simulated by a functional, though nondeterministic, semantics.
Bio: Stefan
Muller completed his undergraduate at Harvard, where his senior thesis
advised by Stephen Chong developed a static analysis for reasoning
about information flow in concurrent programs. He is currently a
2nd-year PhD student in the Computer Science Department advised by Umut
Acar, focusing on interactive computation.