Generating Compiler Optimizations from Proofs

We present an automated technique for generating compiler optimizations from examples of concrete programs before and after improvements have been made to them. The key technical insight of our technique is that a proof of equivalence between the original and transformed concrete programs informs us which aspects of the programs are important and which can be discarded. Our technique therefore uses these proofs, which can be produced by translation validation or a proof-carrying compiler, as a guide to generalize the original and transformed programs into broadly applicable optimization rules. We present a category-theoretic formalization of our proof generalization technique. This abstraction makes our technique applicable to logics besides our own. In particular, we demonstrate how our technique can also be used to learn query optimizations for relational databases or to aid programmers in debugging type errors. Finally, we show experimentally that our technique enables programmers ...

• Type: Conference
• Year: 2010
• Where: POPL
• Authors: Ross Tate, Michael Stepp, Sorin Lerner
• Keywords: POPL 2010, Programming Languages, Proof Generalization Technique, Technique Enables Programmers, Transformed Concrete Programs