Download PDFOpen PDF in browser

Better Proof Output for Vampire

15 pagesPublished: March 1, 2017

Abstract

Vampire produces highly usable and informative proofs, but now they are even better and this paper explains how. It is important that the proofs produced by automated theorem provers are both understandable and machine checkable. Producing something that satisfies both of these goals is challenging, especially when dealing with complex steps performed by the solver. The main areas where proof output has been improved for understanding include (i) introduction of new symbols (such as Skolem functions) in preprocessing, (ii) representation of unifiers (for example, in resolution steps), and (iii) presentation of AVATAR proofs. These improvements will be illustrated via a number of examples. For checkable proofs Vampire provides a mode that outputs the proof as a number of individual (TPTP) problems that can be independently checked. This process is explained and illustrated with examples.

Keyphrases: automated theorem proving, first-order logic, proof output, Vampire

In: Laura Kovács and Andrei Voronkov (editors). Vampire 2016. Proceedings of the 3rd Vampire Workshop, vol 44, pages 46--60

Links:
BibTeX entry
@inproceedings{Vampire2016:Better_Proof_Output_for,
  author    = {Giles Reger},
  title     = {Better Proof Output for Vampire},
  booktitle = {Vampire 2016. Proceedings of the 3rd Vampire Workshop},
  editor    = {Laura Kovacs and Andrei Voronkov},
  series    = {EPiC Series in Computing},
  volume    = {44},
  pages     = {46--60},
  year      = {2017},
  publisher = {EasyChair},
  bibsource = {EasyChair, http://www.easychair.org},
  issn      = {2398-7340},
  url       = {https://easychair.org/publications/paper/1DlL},
  doi       = {10.29007/5dmz}}
Download PDFOpen PDF in browser