Verifying a Semantic βη-Conversion Test for Martin-Löf Type Theory

  • Authors:
  • Andreas Abel;Thierry Coquand;Peter Dybjer

  • Affiliations:
  • Department of Computer Science, Ludwig-Maximilians-University, Munich,;Department of Computer Science, Chalmers University of Technology,;Department of Computer Science, Chalmers University of Technology,

  • Venue:
  • MPC '08 Proceedings of the 9th international conference on Mathematics of Program Construction
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

Type-checking algorithms for dependent type theories often relyon the interpretation of terms in some semantic domain ofvalueswhen checking equalities. Here we analyze a versionof Coquand's algorithm for checking theβη-equality of such semantic values in a theorywith a predicative universe hierarchy and large elimination rules.Although this algorithm does not rely on normalization byevaluation explicitly, we show that similar ideas can be employedfor its verification. In particular, our proof uses the new notionsof contextualreification and strong semanticequality.The algorithm is part of a bi-directional type checkingalgorithm which checks whether a normal term has a certain semantictype, a technique used in the proof assistants Agda and Epigram. Wework with an abstract notion of semantic domain in order toaccommodate a variety of possible implementation techniques, suchas normal forms, weak head normal forms, closures, and compiledcode. Our aim is to get closer than previous work to verifying thetype-checking algorithms which are actually used in practice.