On good infinite families of toric codes or the lack thereof (Q6596265)

In this paper, the authors (as part of an REU project led by K. Jabbusch) study polytope constructions of toric codes \(C\) and their affects on the parameters of \(C\).\N\NMore precisely, fix a prime power \(q > 2\), the associated finite field \(\mathrm{GF}(q)\) and let \(P\subset \mathbb{R}^n\) be an integral convex polytope which is contained in the \(n\)-dimensional box \([0,q-2]^n\) and has \(k=|P\cap \mathbb{Z}^n|\) lattice points. Such polytopes can be used to construct a toric code \(C=C_P(\mathrm{GF}(q))\) as the image of the following evaluation map. Let\N\[\NL_P = \operatorname{span}_{\mathrm{GF}(q)}\{x^p \mid p \in P\cap \mathbb{Z}^n\} \subset \mathrm{GF}(q)[x_1,\dots,x_n],\N\]\Nwhere \(x = (x_1, \dots , x_n)\) are variables, \(p = (p_1, \dots , p_n)\in P\) a lattice point on the polytope, and \(x^p =x_1^{p_1}\cdot \dots \cdot x_n^{p_n}\). If we choose an indexing of the elements of \((\mathrm{GF}(q)^\times)^n\), then the evaluation map\N\[\Ne_P : L_P \to \mathrm{GF}(q)^{(q-1)^n}, \qquad f \mapsto (f(a) \mid a \in (\mathrm{GF}(q)^\times)^n),\N\]\Nis a linear transformation from \(L_P\) to Euclidean space \(\mathrm{GF}(q)^N\), where \(N=|(\mathrm{GF}(q)^\times)^n|\). The image of this map defines the toric code \(C=C_P(\mathrm{GF}(q))\subset \mathrm{GF}(q)^N\).\N\NThe block length \(N=N(C)\) and the dimension \(k=k(C)\) are parameters of the code \(C\) that are relatively easy to compute, but more work is required to compute or even estimate the minimum distance \(d=d(C)\). The author's goal is to use polytope operations, such as join and sum, to construct new, infinite families of toric codes and investigate whether or not these families have good parameters (where ``good'' is defined precisely in the paper under review). Long story short, none of these new families of toric codes have good parameters. They conjecture, based on this evidence, that no such good families of toric codes exist.\N\NFor further details of their results, we refer the reader to the very clearly written paper itself.