Stable sort algorithms in Rocq

This library provides a characterization of stable mergesort functions using relational parametricity, and deduces several functional correctness results, including stability, solely from the characteristic property. This library allows the users to prove their mergesort correct just by proving that the mergesort in question satisfies the characteristic property. The functional correctness lemmas are overloaded using a canonical structure (StableSort.function) that bundles the characteristic property, and automatically apply to any declared instance of this structure.

As instances of the characteristic property, this library provides two kinds of optimized mergesorts. The first kind is non-tail-recursive mergesort. In call-by-need evaluation, they compute the first k smallest elements of a list of length n in O(n + k log k) time, which is known to be the optimal time complexity of the partial and incremental sorting problems. However, the non-tail-recursive merge function linearly consumes the call stack and triggers a stack overflow in call-by-value evaluation. The second kind is tail-recursive mergesorts and thus solves the above issue in call-by-value evaluation. However, it does not allow us to compute the output incrementally regardless of the evaluation strategy. In addition, each of the above two kinds of mergesort functions has a smooth (also called natural) variant of mergesort, which takes advantage of sorted slices in the input.

Meta

• Author(s):
  • Kazuhiko Sakaguchi (initial)
  • Cyril Cohen
• License: CeCILL-B Free Software License Agreement
• Compatible Rocq/Coq versions: 8.19 or later
• Additional dependencies:
  • MathComp ssreflect 2.3.0 or later
  • Paramcoq 1.1.3 or later
  • Mczify (required only for icfp25/)
  • Equations (required only for icfp25/)
• Rocq/Coq namespace: stablesort
• Related publication(s):
  • A bargain for mergesorts — How to prove your mergesort correct and stable, almost for free doi:10.1145/3747505

Files

The theories/ directory is the main part of the library. The icfp25/ directory contains Rocq files corresponding more closely to the paper. The latter files are not a part of the installation (see below), and explained further in the dedicated README file.

Building and installation instructions

The easiest way to install the development version of Stable sort algorithms in Rocq and its dependencies is via OPAM:

git clone https://github.com/pi8027/stablesort.git
cd stablesort
opam repo add rocq-released https://rocq-prover.org/opam/released

To build and install the theories/ files:

opam install ./rocq-stablesort.opam

Alternatively, to build and install only the dependencies:

opam install ./rocq-stablesort.opam --deps-only --with-test

Given that the dependencies are installed, you can use one of the following make targets to manually build the Rocq files:

• The default target: builds the theories/ files.
• build-icfp25: builds the icfp25/ files.
• validate: checks the compiled theories/ files and their dependencies and prints a summary about their context (such as axioms), which should show that the theories/ files are axiom-free.
• validate-icfp25: checks the compiled icfp25/ files and their dependencies and prints a summary about their context, which should print the axiom of dependent functional extensionality (functional_extensionality_dep) on which the Equation plugin relies.

Credits

The technique to make mergesort functions structurally-recursive and the functional correctness proofs of mergesort, except for the use of parametricity, are largely based on the path library of Mathematical Components, to which the authors made significant contributions (in pull requests #328, #358, #601, #650, #680, #727, #1174, and #1186). The authors would like to thank other MathComp developers and contributors who contributed to the discussion: Yves Bertot, Christian Doczkal, Georges Gonthier, Assia Mahboubi, and Anton Trunov.