Changing almost perfect nonlinear functions on affine subspaces of small codimensions
Hiroaki Taniguchi, Alexandr Polujan, Alexander Pott, Razi Arshad
TL;DR
This work advances secondary constructions of APN functions by analyzing modifications on affine subspaces with small codimensions. It develops precise APN-ness criteria for hyperplane and codimension-two subspace modifications, introduces and leverages $H$-equivalence to unify quadratic APN classifications (notably showing that in dimension $6$ every quadratic APN can be represented as $x^3+\mathrm{Tr}(x)L(x)$ up to $H$-equivalence), and provides an exponential-sum condition for the APN-ness of such forms. The results yield numerous EA-inequivalent APN functions from a single representative and demonstrate concrete constructions (including a codimension-two subspace example in dimension $8$) with explicit univariate representations. Overall, the paper deepens understanding of the APN landscape, offering practical criteria for generating new APN families and enriching the taxonomy of equivalent and inequivalent classes in cryptographic settings.
Abstract
In this article, we study algebraic decompositions and secondary constructions of almost perfect nonlinear (APN) functions. In many cases, we establish precise criteria which characterize when certain modifications of a given APN function yield new ones. Furthermore, we show that some of the newly constructed functions are extended-affine inequivalent to the original ones.
