sboxU

The logo of sboxU, showing a box being built or disassembled.

Description

sboxU is a SAGE/Python library that is intended to systematize knowledge about the black-box analysis of vectorial Boolean functions, p-ary fuctions, and in particular about the algorithms relevant to their study. To this end, it provides a wide variety of functions performing for instance the following tasks: - generating S-boxes from univariate polynomials, - multi-threaded computation of the Walsh spectrum, - display of the “Pollock representation” of a DDT, - computation of the automorphisms of the graph of an APN function, - identify all the bijections in the CCZ-equivalence class of a function, - computation of the non-linear invariants of a function, - … and much, much more!

At its core, sboxU is a C++ library providing convenient abstractions for S-boxe, affine maps, etc.; as well as the algorithms operating on them. Then, a cython layer exposes these functions to SAGE.

If you use sboxU in a published paper, please cite it using the following bibtex entry:

@misc{sboxU,
authors={Léo Perrin,
Jules Baudrin,
Aurélien Boeuf,
Xavier Bonnetain,
Alain Couvreur,
Merlin Fruchon,
Mathias Joly,
Pierre Galissant,
Lukas Stennes
},
year=2026,
title={sbox{U}: black-box analysis of discrete functions},
howpublished={Available online at \url{https://github.com/lpp-crypto/sboxU/}}
}

Documentation

Some tests/examples are provided in the tests folder. You must compile and install sboxU as explained below in order for them to work.

The SAGE API is documented here (note in particular the search box!).
If you need direct access to its C++ internals, a (still quite incomplete) API documentation is available here.

sboxU_CPP

The C++ component of sboxU can be used on its own, see the relevant folder.

Executable Scripts

During its installation, sboxU creates some executables that are installed in your path. To see their interfacce, simply run them with the -h argument.

sboxU_apn_db_generation uses known lists of APN fuctions to generate a local tinySQL database containing exactly one representstive per EA-class of functions

Installing SboxU

Dependencies

Most functions in sboxU only depend on a recent version of SAGE (it was tested for example with version 10.5). Some use openmp for multithreading, and you may need to install it in order to successfully compile. The python dependencies are installed automatically by pip.

Installing openmp can be done with:

sudo apt-get install libomp-dev

or, on macOS:

brew install libomp

Downloading and Compiling

Straight from Github

To be able to use sboxU in your scripts, simply run the following command. It will download and compile sboxU (and also create some executable scripts, see above).

sage -pip install git+https://github.com/lpp-crypto/sboxU

From a Local Copy of the Repository

Alternatively, especially if you want to work on sboxU, you can first clone this repository and then install sboxU from its content. To do this, simply cd to the directory containing this makefile and run

sage -pip install -e .

Once it has been installed, you can recompile it much faster using

sage setup.py build_ext --inplace -j 8