On the automorphisms of generalized algebraic geometry codes (Q2672296)

Let \(F/GF(q)\) denote an algebraic function field over the finite field \(GF(q)\) with \(q\) elements, where \(q\) is a power of a prime number. For each place \(P\) of \(F\) of degree \(n\) there is a \(GF(q)\)-isomomorphism of vector spaces, \(\phi_P:F_P\to GF(q^n)\cong GF(q)^n\). For any \(z\in F\), \(z(P)\) denotes the value of \(z\) at \(P\), regarded as an element of \(GF(q)^n\) if desired. A divisor of \(F/GF(q)\) is an element of the free abelian group generated by the places. Let the divisor \(\Phi=\sum_{i=1}^N P_i\) be a sum of distinct places, where \(P_i\) has degree \(n_i\) (\(1\leq i\leq N\)). Let \(G\) be a divisor with \(supp(G)\cap \{P_1,\dots,P_N\}=\emptyset\). Consider the evaluation map \[ ev_\Phi:{\mathcal{L}}(G)\to \prod_{i=1}^N GF(q^{n_i})\cong GF(q)^{\sum_{i=1}^N n_i}, \] defined by \(ev_\Phi(z)=(z(P_1),\dots,z(P_N))\), for each \(z\) in the Riemann-Roch space \({\mathcal{L}}(G)\). The image of \(ev_\Phi\) will be denoted \(C(\Phi, G)\). This is a linear block code, called a \textit{generalized algebraic geometry code}, or \textit{GAG code}, for brevity. The authors compute the length, dimension, and give a lower bound for the minimum distance under the condition that the degree of the divisor \(G\) is ``not too large,'' \(\deg(G)<\sum_{i=1}^N n_i\). The GAG codes considered in this paper form a subclass of those codes described above. Assume that the \(N=N_1+N_2\) distinct places \(P_i\) satisfy the following: (a) all the places \(P_1\), \dots, \(P_{N_1}\) are of degree \(n_1\), (b) all the places \(P_{N_1+1}\), \dots, \(P_{N_1+N_2}\) are of degree \(n_2\). In this case, \(\sum_{i=1}^N n_i = n_1N_1+n_2N_2\). The subgroup \(H_{N_1,N_2}(\Phi,G)\) of the automorphism group \(\mathrm{Aut}(C(\Phi, G))\) of \(C(\Phi, G)\) considered in this paper is defined next. Write each \(c\in C(\Phi,G)\) as an \(N_1+N_2\)-tuple \(c=(a_1,a_2,\dots,a_{N_1+N_2})\), where \(a_i=z(P_i)\in GF(q^{n_i})\). If \(\pi \in S_{N_1}\times S_{N_2}\subset S_{N_1+N_2}=S_N\) and if, for each \(c\in C(\Phi, G)\), we have \(\pi(c)\in C(\Phi, G)\), then we say \(\pi\) is an \(N_1N_2\)-automorphism of \(C(\Phi,G)\). The subgroup of \(N_1N_2\)-automorphisms of \(C(\Phi,G)\) is denoted \(H_{N_1,N_2}(\Phi,G)\). We will now summarize, in rough terms, the main results of the paper. The hypothesis of Theorem 3.2 of the paper assumes that \(\sigma\in \mathrm{Aut}(F/GF(q))\) fixes each place \(P_1\), \dots, \(P_{N_1}\), \(P_{N_1+1}\), \dots, \(P_{N_1+N_2}\). The theorem says that if, in addition, a certain condition (too bulky to state precisely here) holds among the parameters of \(F\) then \(\sigma=1\). The hypothesis of Theorem 3.3 assumes that \(\sigma\in\mathrm{Aut}(F/GF(q))\). Among other things, this theorem gives conditions (too technical to state here) under which \(\sigma \in H_{N_1,N_2}(\Phi,G)\). The hypothesis of Theorem 4.5 assumes certain technical conditions (such as the degree of \(G\) is ``not too large'' in some sense). In this situation, this theorem describes the group \(H_{N_1,N_2}(\Phi,G)\) as the subgroup of Aut\((F/GF(q))\) that is ``as large as reasonable possible,'' in a rough sense. An example with \(q=5\) is given at the end of the paper. For precise details, the reader is referred to the paper itself.