The structure of algebras admitting well-agreeing near weights (Q2464527)

\textit{T. Høholdt, J. van Lint} and \textit{R. Pellikaan} [V.S. Pless and W.C. Huffman (Eds), Handbook of Coding Theory, 1, 871--961 (1998; Zbl 0922.94015)] introduced the notions of order and weight functions over an algebra \(R\) defined over a finite field \(F_q\)\, to give a setting which can be used for the construction of algebraic geometric codes (AG codes) `without algebraic geometry'. As is well known the AG codes \(C(D,G)\)\, and their duals \(C(D,G)^*\)\, are based on a curve \({\mathcal X}\)\, over \(F_q\)\, and two rational divisors \(D, G\). The approach of Hohold, van Lint and Pellikaan allows to deal with one-point AG codes (when \(G=aP,\, P\in {\mathcal X}\)). In order to be able to give a similar construction for the case of two-points AG codes (\(G=a_1P_1+a_2P_2\)), \textit{C. Carvalho, C. Munuera, E. Silva} and \textit{F. Torres} [IEEE Trans. Inform. Theory 53 (5), 1919--1924 (2007)], introduce the notions of near order and near weight functions. In the present paper the authors characterize the algebras admitting two well-agreeing near weights. Section 2 of the paper recalls definitions and basic facts about weights and near weight over an algebra \(R\)\, and in particular the condition of well-agreement of two normal near weights \(\rho,\sigma\): \(\rho\) and \(\sigma\) agree well if \(\sharp(\mathbb{N}^2-H\))\, is finite and \({\mathcal U}_{\rho}\bigcap {\mathcal U}_{\sigma}=F_q\),\, where \({\mathcal U}_{\rho}=\{r\in R^*| \rho(r)\leq\rho(1) \}\bigcup\{0\}\) and \(H=\{(\rho(f),\sigma(f))| f\in R^*\}\). Section 3 shows that \(H\)\, is in fact a semigroup. The wanted characterization of the algebras \(R\)\, equipped with two well agreeing near weight is established in Sections 4 and 5. Section 4 proves that \(R\)\, should be a finitely generated integral domain over the finite field \(F_q\)\, (so \(R\)\, is an affine \(F_q\)-algebra) and that \(K\),\, the field of quotients of \(R\),\, is an algebraic function field of one variable over \(F_q\),\, while Section 5 shows (theorem 5.6) that there are two places \(P,Q\) of \(K\)\, such that \(\rho\) and \(\sigma\)\, are derived from the valuations associated to \(P\)\, and \(Q\). Then \(\bigcap_{S\in \mathbb{P}(K)-\{P,Q\}}\mathcal{O}_S \) is precisely the integral closure \(\bar{R}\)\, of \(R\)\, in \(K\).