Towards Normalization by Evaluation for Erlang

Abstract

Normalization by Evaluation (NbE) is a technique to extract the normal form of a 𝜆-calculus term. Originally targeting pure 𝜆-calculus, the algorithm can be extended and adjusted to work on more practical call-by-value programming languages. We implement NbE for a fragment of Erlang, starting at the Moggi’s semantics framework of computational lambda calculus, and largely inspired by Filinski’s research. The result shows that our normalizer can be applied on programs with rich semantics, and can potentially be extended to perform partial evaluation.