Daniel Kühlwein

Overview and evaluation of premise selection techniques for large theory mathematics

In this paper, an overview of state-of-the-art techniques for premise selection in large theory mathematics is provided, and new premise selection techniques are introduced. Several evaluation metrics are introduced, compared and their appropriateness is discussed in the context of automated reasoning in large theory mathematics. The methods are evaluated on the MPTP2078 benchmark, a subset of the Mizar library, and a 10% improvement is obtained over the best method so far.

The naproche project controlled natural language proof checking of mathematical texts

This paper discusses the semi-formal language of mathematics and presents the Naproche CNL, a controlled natural language for mathematical authoring. Proof Representation Structures, an adaptation of Discourse Representation Structures, are used to represent the semantics of texts written in the Naproche CNL. We discuss how the Naproche CNL can be used in formal mathematics, and present our prototypical Naproche system, a computer program for parsing texts in the Naproche CNL and checking the proofs in them for logical correctness.

A Learning-Based Fact Selector for Isabelle/HOL

Sledgehammer integrates automatic theorem provers in the proof assistant Isabelle/HOL. A key component, the fact selector, heuristically ranks the thousands of facts (lemmas, definitions, or axioms) available and selects a subset, based on syntactic similarity to the current proof goal. We introduce MaSh, an alternative that learns from successful proofs. New challenges arose from our “zero click” vision: MaSh integrates seamlessly with the users’ workflow, so that they benefit from machine learning without having to install software, set up servers, or guide the learning. MaSh outperforms the old fact selector on large formalizations.

Automated and human proofs in general mathematics: an initial comparison

First-order translations of large mathematical repositories allow discovery of new proofs by automated reasoning systems. Large amounts of available mathematical knowledge can be re-used by combined AI/ATP systems, possibly in unexpected ways. But automated systems can be also more easily misled by irrelevant knowledge in this setting, and finding deeper proofs is typically more difficult. Both large-theory AI/ATP methods, and translation and data-mining techniques of large formal corpora, have significantly developed recently, providing enough data for an initial comparison of the proofs written by mathematicians and the proofs found automatically. This paper describes such an initial experiment and comparison conducted over the 50000 mathematical theorems from the Mizar Mathematical Library.

Premise selection in the naproche system

Automated theorem provers (ATPs) struggle to solve problems with large sets of possibly superfluous axiom. Several algorithms have been developed to reduce the number of axioms, optimally only selecting the necessary axioms. However, most of these algorithms consider only single problems. In this paper, we describe an axiom selection method for series of related problems that is based on logical and textual proximity and tries to mimic a human way of understanding mathematical texts. We present first results that indicate that this approach is indeed useful.

MaLeS: A Framework for Automatic Tuning of Automated Theorem Provers

MaLeS is an automatic tuning framework for automated theorem provers. It provides solutions for both the strategy finding as well as the strategy scheduling problem. This paper describes the tool and the methods used in it, and evaluates its performance on three automated theorem provers: E, LEO-II and Satallax. On a representative subset of the TPTP library a MaLeS-tuned prover solves on average 8.67 % more problems than the prover with its default settings.

E-MaLeS 1.1

Picking the right search strategy is important for the success of automatic theorem provers. E-MaLeS is a meta-system that uses machine learning and strategy scheduling to optimize the performance of the first-order theorem prover E. E-MaLeS applies a kernel-based learning method to predict the run-time of a strategy on a given problem and dynamically constructs a schedule of multiple promising strategies that are tried in sequence on the problem. This approach has significantly improved the performance of E 1.6, resulting in the second place of E-MaLeS 1.1 in the FOF divisions of CASC-J6 and CASC@Turing.