Iwasawa main conjecture for the Carlitz cyclotomic extension and applications (Q2290810)

In this paper, the authors prove an Iwasawa main conjecture for the Carlitz cyclotomic extension of a global function field and an analogue of the `\(\mu = 0\)' theorem of \textit{B. Ferrero} and \textit{L. C. Washington} [Ann. Math. (2) 109, 377--395 (1979; Zbl 0443.12001)]. Let \(F = \mathbb F_q(\theta)\) be a function field with ring of integers \(A = \mathbb F_q[\theta]\) and denote by \(\infty\) the place of \(F\) with uniformizer \(1/\theta\). Let \(\mathfrak p\) be a prime ideal in \(A\) of positive degree \(d\) and let \(F_n\) be the field obtained from \(F\) by adjoining all \(\mathfrak p^{n+1}\)-torsion points of the Carlitz module \(\Phi\). Recall that the latter is just an \(\mathbb F_q\)-linear ring homomorphism \[ \Phi: A \rightarrow F \left\{\tau \right\}, \, \theta \mapsto \theta + \tau, \] where \(F \left\{\tau \right\}\) denotes the skew polynomial ring with \(\tau f = f^q \tau\) for every \(f \in F\). Then \(F_n\) is an abelian extension of \(F\) with Galois group \[ \mathrm{Gal}(F_n/F) \simeq (A / \mathfrak p^{n+1})^{\times}. \] Let \(F_{\infty}\) be the union of the \(F_n\) so that \[ G_{\infty} := \mathrm{Gal}(F_{\infty}/F) = \Delta \times \Gamma, \] where \(\Delta = \mathrm{Gal}(F_0/F) \simeq (A / \mathfrak p)^{\times}\) and \(\Gamma\) is canonically isomorphic to the \(1\)-units in \(A_{\mathfrak p}\), and in particular non-canonically isomorphic to \(\mathbb Z_p^{\infty}\). This is the \(\mathfrak{p}\)-cyclotomic extension of \(F\). The Iwasawa algebra in this context is the (non-Noetherian) completed group ring \(\Lambda := \mathbb Z_p[\mu_{q^d-1}][[\Gamma]]\). Let \(\chi\) be a non-trivial character of \(\Delta\). On the algebraic side the authors study the inverse limit of the \(p\)-\(\chi\)-parts of the group of classes of degree zero divisors of \(F_n\). This is a finitely generated torsion \(\Lambda\)-module. The main result then states that its Fitting ideal is generated by a Stickelberger element, denoted by \(\Theta_{\infty}^{\sharp}(1, \chi)\) in the paper. The latter is constructed as follows, where we assume that the cardinality of \(\mathrm{ker}(\chi)\) is not divisible by \(q-1\) for simplicity. Let \(F_S\) be the maximal abelian extension of \(F\) unramified outside \(S = \left\{\mathfrak p, \infty \right\}\). Consider the Stickelberger series \[ \Theta_{F_S/F}(X) := \prod_{\mathfrak q \not\in S} (1 - \mathrm{Fr}_{\mathfrak q}^{-1} X^{\mathrm{deg} (\mathfrak q)})^{-1}\in \mathbb Z[\mathrm{Gal}(F_S/F)][[X]]. \] Then \(\Theta_{\infty}^{\sharp}(X, \chi)\) is just the \(\chi\)-component of the projection of this series to \(\mathbb Z[G_{\infty}][[X]]\). The main input of the proof of this `main conjecture' is a result of \textit{C. Popescu} and \textit{C. Greither} [Int. Math. Res. Not. 2012, No. 5, 986--1036 (2012; Zbl 1254.11063)] on the Galois module structure, including the computation of certain Fitting ideals, of \(p\)-adic realizations of Picard \(1\)-motives and class groups of global function fields. It is also important to note that the above Stickelberger series might be seen as a universal object for \(L\)-functions in this context. It can be related to \(\mathfrak{p}\)-adic and to complex \(L\)-functions and even to the Carlitz-Goss \(\zeta\)-function. The second main result (which might be seen as an analogue of the Ferrero-Washington theorem) states that for non-trivial \(\chi\) one has that \[ \Theta_{\infty}^{\sharp}(1, \chi) \not\equiv 0\, \bmod p. \] This is achieved by relating \(\Theta_{\infty}^{\sharp}(X, \chi)\) with a \(\mathfrak p\)-adic \(L\)-function. Certain special values of the latter are given in terms of Bernoulli-Goss numbers and the result essentially follows from observing that these numbers are nonzero.