Publication Date

2007

Degree Type

Master's Project

Degree Name

Master of Science (MS)

Department

Computer Science

Abstract

The aim of computer proof checking is not to find proofs, but to verify them. This is different from automated deduction, which is the use of computers to find proofs that humans have not devised first. Currently, checking a proof by computer is done by taking a known mathematical proof and entering it into the special language recognized by a proof verifier program, and then running the verifier to hopefully obtain no errors. Of course, if the proof checker approves the proof, there are considerations of whether or not the proof checker is correct, and this has been complicated by the fact that so many systems have sprung into being. The two main challenges in using a proof checker today are the time needed to learn the syntax and general usage of the system and the time needed to formalize a proof in the system even when the user is already proficient with it. As mathematicians are not yet using proof checkers regularly, we wanted to evaluate the validity of this reluctance by analyzing these main obstacles. Judging by Dr. Wiedijk’s Formalizing 100 Theorems list, which gives an overview of the headway various proof systems have made in mathematics, Coq and Mizar are two of the most successful systems in use today (Wiedijk, 2007). I simultaneously formalized two fairly involved theorems in these two systems while I was at approximately the same level of familiarity with each. I kept track of my experiences with learning the systems and analyzed their comparative strengths and weaknesses. The analysis and summary of experiences should also give a general idea of the current state of computer-aided proof checking.

Share

COinS