Squares of matrix-product codes (Q2302565): Difference between revisions

Given two linear codes \(C\) and \(C'\) the Schur product is defined by \[ C*C'=\left\langle\{c*c'\mid c\in C, c' \in C'\}\right\rangle, \] where \(c*c'=(c_1c_1',\ldots,c_nc_n')\). It is well know that for some cryptographic applications, private information retrieval or multiparty computations among others, the knowledge of \(C*C'\) is of particular interest. In certain protocols for multiparty computations, both a large minimum distance for \(C^{*2}=C*C\) and a large dimension for \(C\) are required. Depending on the protocol, sometimes a large minimum distance for \(C^{\perp}\) is also demanded. According to previous motivations, they study the structure of \(C^{*2}\) when the code \(C\) is a matrix product code. In the particular case of the \((u,u+v)\)-construction, they provide a lower bound for the minimum distance which is sharp in case that the codes used in the \((u,u+v)\)-construction are nested. Furthermore, when the constituent codes of the \((u,u+v)\)-construction are binary cyclic codes they use the cyclotomic coset to control at the same time the dimension of \(C\) and a lower bound of the minimum distance of \(C\) and \(C^{*2}\), actually they notice that considering large cyclotomic cosets one can obtain the desired codes. They are able to obtain new codes with large dimension of \(C\) and large minimum distance of \(C^{*2}\) simultaneously. Finally they study matrix-product codes where the defining matrix \(A\) is a Vandermonde matrix. Thanks to it, they can provide a better algebraic structure for \(C^{*2}\), i.e., \(C^{*2}\) is also a matrix-product code and a formula for the dimension and a lower bound for the minimum distance are given. In the particular case where the constituent codes of the matrix-product code are AG-codes, then more precise parameters are given.