Skip to content

Latest commit

 

History

History
43 lines (30 loc) · 2.59 KB

README.md

File metadata and controls

43 lines (30 loc) · 2.59 KB
Raw

Proof Repair across Quotient Type Equivalences

This repo contains the artifact for "Proof Repair across Quotient Type Equivalences" by Cosmo Viola, Max Fan, and Talia Ringer. The paper is currently under submission (preprint). This repo contains the artifacts as-submitted.

There are two parts to this artifact:

  • An internal view of proof across quotient type equivalences in Cubical Agda (located in cubical). This is currently manual due to engineering and theoretical limitations of cubical type theory.
  • An external view of proof repair across quotient type equivalences (via setoids) in Coq (located in /pumpkin-pi). This is automated by a prototype that extends the original pumpkin-pi automation.

Manual Case Studies in Cubical Agda

To run our Cubical Agda case studies, you should install Cubical Agda. The main corresponding files for our case studies in Cubical Agda can be found in the following locations:

Automated Case Studies in Coq (extending Pumpkin Pi with quotient types/setoids)

To fetch and build everything, make sure you have Coq 8.9.1 installed (later versions are not currently supported).

Then, to build everything, run:

cd pumpkin-pi/plugin
./build.sh

To run the relevant case studies files, run:

./setoids.sh

which will first run the setoid repair tests contained within pumpkin-pi/plugin/coq/ToSetoidTest.v, then build the following two case studies:

  • pumpkin-pi/plugin/coq/playground/grothendieck_int_equivalence_repair_tool.v (for the integer case study in Coq)
  • pumpkin-pi/plugin/coq/playground/TwoListQueue/two_list_queue_equivalence_repair_tool.v (for the queue case study in Coq)

The script will also build UIPList.v, as a dependency of the queue case study. In our queue case study, we parameterize over lists/queues of type A, where UIP holds on A. UIPList.v proves that if UIP holds on A, it will hold on list A. We do not assume it for all types, so we are not adding an axiom to our theory.

Funding Acknowledgements

This research was developed with funding from the Defense Advanced Research Projects Agency. The views, opinions, and/or findings expressed are those of the author(s) and should not be interpreted as representing the official views or policies of the Department of Defense or the U.S. Government.