A note on a tower by Bassa, Garcia and Stichtenoth (Q1746234)

Let \(\mathbb{F}_q\) be a finite field with \(q\) elements, and let \(F\) be a function field with full constant field \(\mathbb{F}_q\). The number \(N(F)\) of \(\mathbb{F}_q\)-rational places of \(F\) is bounded in terms of the genus \(g(F)\) and the cardinality of the finite field \(q\). Namely, according to the well-known Hasse-Weil upper bound: \(N(F) \leq 1+q+2g(F)\sqrt{q}\). It was noted by Ihara and Manin that this bound is not optimal when \(g(F)\) is large compared with \(q\). This motivated the study of the asymptotic behaviour of the number of rational places of a function field over its genus \(g\) as \(g\) goes to infinity, giving rise to the Ihara's constant: \(A(q):=\limsup_{g(F) \rightarrow \infty} \frac{N(F)}{g(F)}\), where ``limsup'' runs over function fields with constant field \(\mathbb{F}_q\). To analyze this constant, towers of function fields over \(\mathbb{F}_q\) are investigated, that is, sequences \(\mathcal{F}/\mathbb{F}_q=(F_1 \subset F_2 \subset \ldots)\) of function fields with constant field \(\mathbb{F}_q\) such that \(F_1=\mathbb{F}_q(x_1)\) for some \(x_1 \in F_1\) trascendental over \(\mathbb{F}_q\), for each \(i \geq 1\) \(F_{i+1}=F_i(x_{i+1})\) with \([F_{i+1}:F_i]>1\) and \(\varphi(x_i,x_{i+1})=0\) for some separable polynomial \(\varphi \in \mathbb{F}_q[T]\) (the recursion of the tower), and \(g(F_i) \rightarrow \infty\) as \(i \rightarrow \infty\). In this setting a tower \(\mathcal{F}/\mathbb{F}_q\) is good if its limit \(\lambda(\mathcal{F}/\mathbb{F}_q):=\lim_{i \rightarrow \infty} \frac{N(F_i)}{g(F_i)}\) is a positive real number. Since \(\lambda(\mathcal{F}/\mathbb{F}_q)\) gives a lower bound for Ihara's constant \(A(q)\), interesting constructions are those for which this limit is as large as possible. Drinfeld, Vladut, Ihara, Tsfasman and Zink showed that \(A(q)= \sqrt{q}-1\) for square \(q\). Garcia and Stichtenoth gave another proof for the exact value of \(A(q)\) for square \(q\) by constructing an explicitly defined recursive tower. The exact value of \(A(q)\) is still an open problem for non-square \(q\), but there are many bounds for it in the literature. An example is given by the Zink's bound \(A(p^3) \geq 2(p^2-1)/(p+2)\), for \(p\) prime. This bound was then generalized by Bezerra, Garcia and Stichtenoth in \(A(q^3) \geq 2(q^2-1)/(q+2)\) for arbitrary prime powers \(q\) giving an explicit tower \(\mathcal{A}/ \mathbb{F}_{q^3}\) meeting the Zink's bound. After that, a simpler proof of this bound was given by Bassa, Garcia and Stichtenoth with another explicitly defined tower \(\mathcal{C}/\mathbb{F}_{q^3}\), which was then shown to be a partial Galois closure of \(\mathcal{A}/ \mathbb{F}_{q^3}\) by Zieve. Another tower \(\mathcal{X}\) was introduced in 2017 by Anbar, Beelen and Nguyen over cubic fields resulting in Zink's bound. It was noticed that all steps in \(\mathcal{X}/ \mathbb{F}_{q^3}\) are Galois except the first one, and that \(\mathcal{X}/ \mathbb{F}_{q^3}\) contains \(\mathcal{A}/ \mathbb{F}_{q^3}\) as a subtower. In this article the authors show that \(\mathcal{C}/ \mathbb{F}_{q^3}\) is also a subtower of \(\mathcal{X}/ \mathbb{F}_{q^3}\). This completes the determination of the relations between the cubic towers constructed in the literature.