A generalization of the Weierstrass semigroup (Q2504381): Difference between revisions

Let \(X\) be a smooth algebraic projective curve defined over a finite field \textbf{F}. Let \(P \in X\) be a rational point; the Weierstrass semigroup at \(P\) is defined as \(H(P):= \{ k \in \mathbb{N} \; | \; \mathrm{div}_{\infty}(f) = k P, \mathrm{\;for\;some\;} f \in \mathbb{F}(X)\}\) (here \(\mathrm{div}_{\infty}(f)\) denotes pole divisor of the function \(f\) and \(\mathbb{N} = \{0, 1, \dots \}\)). This is a classical concept, and given distinct rational points \(P_1, \dots, P_n \in X\), a natural generalization of this concept is to define \(H(P_1, \dots, P_n) := \{ (k_1, \dots, k_n) \in \mathbb{N}^n \; | \; \mathrm{div}_{\infty}(f) = \sum_{i = 1}^n k_i P_i, \mathrm{\;for\;some\;} f \in \mathbb{F}(X)\}\) as being the Weierstrass semigroup at points \(P_1, \dots, P_n\). This has been done and this semigroup has been studied, specially in connection with the construction of geometric Goppa codes with good parameters (see e.g.\ results by \textit{G. Matthews} [in: Finite fields and applications. 7th int. conf., \(\mathbb{F}_{q^7}\), Toulouse, France, 2003. Lect. Notes Comp. Sci. 2948, 12--24 (2004; Zbl 1063.14041)], and by \textit{C. Carvalho} and \textit{F. Torres} [Des. Codes Cryptography 35, No.~2, 211--225 (2005; Zbl 1070.94029)]). In the present paper, the authors propose another approach for the definition of Weierstrass semigroup at several points. Let \(P_1, \dots, P_n\) be as above, and denote by \(R_{P_1, \dots, P_n}\) the ring of functions on \(X\) that are regular in \(X \setminus \{P_1,\dots, P_n\}\). Let \(\widehat{H}(P_1, \dots, P_n) := \{ -(v_{P_1}(f), \dots, v_{P_n}(f)) \in \mathbb{Z}^n \; | \; f \in R_{P_1, \dots, P_n} \setminus\{0\} \}\), where \(v_{P_i}\) stands for the valuation associated to the point \(P_i\) (\(i = 1, \dots, n\)). It's clear that \(\widehat{H}(P_1) = H(P_1)\) and \(H(P_1, \dots, P_n) = \widehat{H}(P_1, \dots, P_n) \cap \mathbb{N}^n\). The authors then present reasons that justify the following definition for the set of gaps: \(\widehat{G}(P_1, \dots, P_n) := \{(k_1, \dots, k_n) \in \mathbb{Z}^n \; | \; k_1 + \cdots + k_n \geq 0 \} \setminus \widehat{H}(P_1, \dots, P_n)\). Unlike the set of gaps of \(H(P_1, \dots, P_n)\), which is always finite, \(\widehat{G}(P_1, \dots, P_n)\) has cardinality either equal to zero or infinity, if \(n > 1\). This motivates the introduction of another semigroup, which is \(\widetilde{H}(P_1, \dots, P_n) := \widehat{H}(P_1, \dots, P_n)/\Lambda(P_1, \dots, P_n)\), where \[ \Lambda(P_1, \dots, P_n) := \{ -(v_{P_1}(f), \dots, v_{P_n}(f)) \in \mathbb{Z}^n \; | \; f \text{ is an invertible element of } R_{P_1, \dots, P_n}\}. \] Given the definition of \(\widehat{G}(P_1, \dots, P_n)\), it's now natural to define the set \(\widetilde{G}(P_1, \dots, P_n)\) of gaps of \(\widetilde{H}(P_1, \dots, P_n)\) as the complement of \(\widetilde{H}(P_1, \dots, P_n)\) in the set \(\{(k_1, \dots, k_n) \in \mathbb{Z}^n \; | \; k_1 + \cdots + k_n \geq 0 \} / \Lambda(P_1, \dots, P_n)\). The authors prove that \(\widetilde{G}(P_1, \dots, P_n)\) is a finite set, and find lower and upper bounds for its cardinality. They then turn to the case where \(n = 2\) and find a formula for \(\#\widetilde{G}(P_1, P_2)\), that depends on a certain function \(\sigma_{P_1, P_2}: \mathbb{Z} \rightarrow \mathbb{Z}\). In the last sections, they study \(\widetilde{G}(P_1, P_2)\) in the case where \(X\) is a certain maximal curve covered by the hermitian curve or is the Suzuki curve; among other results, they determine \(\#\widetilde{G}(P_1, P_2)\) in these cases.