Home Geometric main conjectures in function fields
Article
Licensed
Unlicensed Requires Authentication

Geometric main conjectures in function fields

  • Werner Bley and Cristian D. Popescu EMAIL logo
Published/Copyright: October 16, 2024

Abstract

We prove an equivariant main conjecture in Iwasawa theory along any rank-one, sign-normalized Drinfeld modular, split-at-∞ Iwasawa tower of a general function field of characteristic 𝑝, for the Iwasawa modules recently considered by Greither and Popescu in their proof of the classical equivariant main conjecture along the (arithmetic) cyclotomic Iwasawa tower. As a consequence, we prove an equivariant main conjecture for a projective limit of certain Ritter–Weiss type modules, along the same Drinfeld modular Iwasawa towers. This generalizes the results of Anglès, Bandini, Bars, Coscelli and Longhi for the split-at-∞ piece of the Iwasawa towers considered in their work, and refines the results by Greither and Popescu.

Funding statement: Supported by a Simons Foundation collaboration grant.

A Appendix (𝑝-adic Ritter–Weiss modules and Tate sequences for small 𝑆)

Let 𝐿 be a finite, separable extension of F q ( t ) . Denote by 𝑍 a smooth, projective curve defined over F q , whose field of rational functions is isomorphic to 𝐿. We let Z ̄ : = Z × F q F q ̄ , Γ : = G ( F q ̄ / F q ) and let 𝛾 be the 𝑞-power arithmetic Frobenius automorphism, viewed as a canonical topological generator of Γ. Note that Z ̄ may not be connected. Consequently, L ̄ : = L F q F q ̄ (the F q ̄ -algebra of rational functions on Z ̄ ) could be a finite direct sum of isomorphic fields (the fields of rational functions of the connected components of Z ̄ .)

Next, we consider a finite, non-empty set 𝑆 of closed points on 𝑍 and let S ̄ be the set of closed points on Z ̄ sitting above points in 𝑆. We let Div S ̄ 0 ( L ̄ ) (respectively Div S ̄ ( L ̄ ) ) and Div S 0 ( L ) (respectively Div S ( L ) ) denote the divisors of degree 0 (respectively arbitrary degree) on Z ̄ and 𝑍, supported at S ̄ and 𝑆, respectively. Note that the degree of a divisor on 𝑍, denoted by deg , is computed relative to the field of definition F q . Also, the degree of a divisor on Z ̄ is in fact a multidegree, computed on each connected component on Z ̄ separately. Further, X S ( L ) denotes the 𝑆-supported divisors on 𝑍 of formal degree 0, where the formal degree, denoted fdeg below, takes each place 𝑣 to 1. Also, U S ( L ) denotes the group of 𝑆-units inside L × and

Pic S 0 ( L ) : = Pic 0 ( L ) Div S 0 ( L ) ̂ = Div 0 ( L ) Div S 0 ( L ) + div ( L × )

is the 𝑆-Picard group associated to 𝐿, obtained by taking the quotient of the usual Picard group Pic 0 ( L ) by the subgroup Div S 0 ( L ) ̂ of classes of all 𝑆-supported divisors of degree 0.

Finally, let M S denote the Picard 1-motive associated as in [8] to the data ( Z ̄ , F q ̄ , S ̄ , ) . As usual, T p ( M S ) Γ and T p ( M S ) Γ denote the Γ-invariants, respectively Γ-coinvariants of the 𝑝-adic Tate module of M S . In what follows, if 𝑁 is a ℤ-module, we let N p : = N Z Z p .

Definition A.1

The set 𝑆 is called 𝑝-large if the following are satisfied.

  1. Pic S 0 ( L ) p = 0 .

  2. 𝑆 contains at least one place of degree (relative to F q ) coprime to 𝑝.

Remark A.2

It is easily seen that 𝑆 is 𝑝-large if and only if Pic S ( L ) p = 0 , where

Pic S ( L ) : = Div ( L ) Div S ( L ) + div ( L × )

is the quotient of the full Picard group Pic ( L ) of 𝐿 by its subgroup of 𝑆-supported divisor classes. The equivalent formulation is perhaps a more natural definition of 𝑝-largeness, but we prefer to use the definition above because Pic S 0 ( L ) (as opposed to Pic S ( L ) ) is much more naturally related to the 1-motive M S .

The following result was obtained in [7]. (See [7, Proposition 2.1].)

Proposition A.3

Proposition A.3 (Greither–Popescu [7])

If 𝑆 is 𝑝-large, then the following hold.

  1. There is a canonical isomorphism T p ( M S ) Γ U S ( L ) p .

  2. There is a canonical isomorphism T p ( M S ) Γ X S ( L ) p .

Remark A.4

In fact, in [7], the authors describe the modules T p ( M S , T ) Γ , T p ( M S , T ) Γ , where M S , T is the Picard 1-motive associated to ( Z ̄ , F q ̄ , S ̄ , T ̄ ) , where 𝑇 is a finite, non-empty set of closed points on 𝑍, disjoint from 𝑆. However, by [8, Remark 2.7] and the proof of [7, Lemma 3.2], we have for any such 𝑇 equalities

T p ( M S , T ) = T p ( M S ) , U S , T ( L ) p = U S ( L ) p ,

where U S , T ( L ) is the group of 𝑆-units in L × , congruent to 1 modulo all primes in 𝑇.

The goal of this appendix is to remove the hypothesis “𝑆 is 𝑝-large” in the proposition above. More precisely, we sketch the proof of the following.

Proposition A.5

With notation as above, the following hold for all sets 𝑆.

  1. There is a canonical isomorphism T p ( M S ) Γ U S ( L ) p .

  2. There are canonical exact sequences of Z p -modules,

    0 Pic S 0 ( L ) p T p ( M S ) Γ X ̃ S ( L ) 0 , 0 Z p / d S Z p X ̃ S ( L ) X S ( L ) p 0 ,

    where d S Z = deg ( Div S ( L ) ) and

    X ̃ S ( L ) : = ( Div S ̄ 0 ( L ̄ ) p ) Γ .

    In particular, if 𝑆 contains a prime of degree not divisible by 𝑝, then d S Z p × and X ̃ S ( L ) X S ( L ) p .

Proof

We will give only a brief sketch of the proof, as the techniques and main ideas are borrowed from [7]. First, we consider the exact sequence of Z p [ [ Γ ] ] -modules

0 Div S ̄ 0 ( L ̄ ) p Div S ̄ ( L ̄ ) p deg Z p 0

and take Γ-invariants and Γ-coinvariants to obtain a long exact sequence

0 Div S 0 ( L ) p Div S ( L ) p deg Z p ( Div S ̄ 0 ( L ̄ ) p ) Γ Div S ( L ) p fdeg Z p 0 .

The fact that the Γ-invariant of the complex [ Div S ̄ ( L ̄ ) p deg Z p ] is [ Div S ( L ) p deg Z p ] and its Γ-coinvariant is

[ Div S ( L ) p fdeg Z p ]

follows immediately from the definitions and is explained in [7, §2]. Now, in the long exact sequence above, we have

ker ( fdeg ) = X S ( L ) p , coker ( deg ) = Z p / d S Z p .

Therefore, if we let X ̃ S ( L ) : = ( Div S ̄ 0 ( L ̄ ) p ) Γ , we have a canonical exact sequence

(A.1) 0 Z p / d S Z p X ̃ S ( L ) X S ( L ) p 0 .

If 𝐽 denotes the Jacobian of Z ̄ , there is a canonical exact sequence of Z p -modules

0 T p ( J ) T p ( M S ) Div S ̄ 0 ( L ̄ ) p 0 .

(See [8, §2] for the exact sequence above.) Since we have a canonical isomorphism

T p ( J ) Γ Pic 0 ( L ) p

(see [8, Corollary 5.7]) and T p ( J ) is Z p -free of finite rank, we also have T p ( J ) Γ = 0 . Therefore, when taking Γ-invariants and Γ-coinvariants in the above exact sequence, we obtain a canonical long exact sequence of Z p -modules

0 T p ( M S ) Γ Div S 0 ( L ) p δ Pic 0 ( L ) p T p ( M S ) Γ X ̃ S ( L ) 0 ,

where the connecting morphism 𝛿 is the usual divisor-class map. (See [7, §1] for this fact.) As there is a canonical isomorphism U S ( L ) p ker ( δ ) , where U S ( L ) p injects into Div S 0 ( L ) p via the divisor map, we obtain a canonical isomorphism of Z p -modules T p ( M S ) Γ U S ( L ) p , which concludes the proof of part (1) of the proposition.

To conclude the proof of part (2), observe that, by definition, coker ( δ ) = Pic S 0 ( L ) p . Therefore, the last four non-zero terms of the long exact sequence above lead to a canonical short exact sequence of Z p -modules

(A.2) 0 Pic S 0 ( L ) p T p ( M S ) Γ X ̃ S ( L ) 0 .

In combination with (A.1), this concludes the proof of part (2). ∎

Remark A.6

Remark A.6 (Ritter–Weiss modules and Tate sequences)

Let L / K be a finite Galois extension with group 𝐺 and assume that F q ( t ) K L . Further, assume that the set 𝑆 is 𝐺-equivariant. Then all the Z p -modules involved in the proof of the above proposition carry natural Z p [ G ] -module structures. Most importantly, due to their canonical constructions, all the exact sequences above are exact in the category of Z p [ G ] -modules.

Exact sequence (A.2) is the 𝑝-adic, function field analogue of the Ritter–Weiss exact sequence (see [19]), defining a certain extension class S of a module of 𝑆-divisors by an 𝑆-ideal class group, in the number field setting. This is what prompts the notation

S ( L ) p : = T p ( M S ) Γ .

Further, since T p ( M S ) is Z p [ G ] -projective, the exact sequence of Z p [ G ] -modules

(A.3) 0 U S ( L ) p T p ( M S ) 1 γ T p ( M S ) S ( L ) p 0

is the 𝑝-adic, function field analogue of a Tate exact sequence (see [7] and also [19] for more details), in the case where 𝑆 is not necessarily 𝑝-large.

Of course, in order to cement these analogies, one would have to compute the extension classes of (A.2) and (A.3) in Ext Z p [ G ] 1 ( X ̃ S ( L ) , Pic S 0 ( L ) p ) and Ext Z p [ G ] 2 ( S ( L ) p , U S ( L ) p ) and show that they coincide with the class-field theoretically meaningful Ritter–Weiss and Tate classes, respectively. In [7], this was done ℓ-adically, for p , for the exact sequence (A.3), in the case where 𝑆 is ℓ-large. (See [7, Theorem 2.2].) A proof of the 𝑝-adic analogue of that theorem (even in the case where 𝑆 is 𝑝-large) is still missing in the literature, unless | G | is not divisible by 𝑝, in which case this was proved in [7]. (See [7, Theorem 2.2].)

Acknowledgements

We would like to thank the referees for their careful reading of the manuscript and for their invaluable suggestions.

References

[1] B. Anglès, A. Bandini, F. Bars and I. Longhi, Iwasawa main conjecture for the Carlitz cyclotomic extension and applications, Math. Ann. 376 (2020), no. 1–2, 475–523. Search in Google Scholar

[2] A. Bandini, F. Bars and E. Coscelli, Fitting ideals of class groups in Carlitz–Hayes cyclotomic extensions, J. Number Theory 232 (2022), 261–282. Search in Google Scholar

[3] A. Bandini and E. Coscelli, Stickelberger series and main conjecture for function fields, Publ. Mat. 65 (2021), no. 2, 459–498. Search in Google Scholar

[4] D. Burns, A. Daoud and D. Liang, On non-Noetherian Iwasawa theory, preprint (2024), https://arxiv.org/abs/2401.02946. Search in Google Scholar

[5] E. Coscelli, Stickelberger series and Iwasawa main conjectures for function fields, Ph. D. Thesis, University of Milan, Milan 2017. Search in Google Scholar

[6] V. G. Drinfeld, Elliptic modules, Mat. Sb. (N. S.) 94(136) (1974), 594–627, 656. Search in Google Scholar

[7] C. Greither and C. Popescu, Picard 1-motives and Tate sequences for function fields, J. Théor. Nombres Bordeaux 33 (2021), no. 3, 835–852. Search in Google Scholar

[8] C. Greither and C. D. Popescu, The Galois module structure of ℓ-adic realizations of Picard 1-motives and applications, Int. Math. Res. Not. IMRN 2012 (2012), no. 5, 986–1036. Search in Google Scholar

[9] R. Hartshorne, Algebraic geometry, Grad. Texts in Math. 52, Springer, New York 1977. Search in Google Scholar

[10] D. R. Hayes, Stickelberger elements in function fields, Compos. Math. 55 (1985), no. 2, 209–239. Search in Google Scholar

[11] K. Iwasawa, Local class field theory, Oxford Math. Monogr., Oxford University, New York 1986. Search in Google Scholar

[12] I. Kaplansky, Projective modules, Ann. of Math. (2) 68 (1958), 372–377. Search in Google Scholar

[13] J. A. Kiehlmann, Classifications of countably-based abelian profinite groups, J. Group Theory 16 (2013), no. 1, 141–157. Search in Google Scholar

[14] H. Kisilevsky, Multiplicative independence in function fields, J. Number Theory 44 (1993), no. 3, 352–355. Search in Google Scholar

[15] S. Lang, Cyclotomic fields I and II, 2nd ed., Grad. Texts in Math. 121, Springer, New York 1990. Search in Google Scholar

[16] J. Neukirch, Algebraic number theory, Grundlehren Math. Wiss. 322, Springer, Berlin 1999. Search in Google Scholar

[17] H. Nishimura, On the unique factorization theorem for formal power series, J. Math. Kyoto Univ. 7 (1967), 151–160. Search in Google Scholar

[18] C. D. Popescu, Integral and 𝑝-adic refinements of the abelian Stark conjecture, Arithmetic of 𝐿-functions, IAS/Park City Math. Ser. 18, American Mathematical Society, Providence (2011), 45–101. Search in Google Scholar

[19] J. Ritter and A. Weiss, A Tate sequence for global units, Compos. Math. 102 (1996), no. 2, 147–178. Search in Google Scholar

[20] M. Rosen, Formal Drinfeld modules, J. Number Theory 103 (2003), no. 2, 234–256. Search in Google Scholar

[21] J. Tate, Les conjectures de Stark sur les fonctions 𝐿 d’Artin en s = 0 , Progr. Math. 47, Birkhäuser, Boston 1984. Search in Google Scholar

[22] L. C. Washington, Introduction to cyclotomic fields, 2nd ed., Grad. Texts in Math. 83, Springer, New York 1997. Search in Google Scholar

Received: 2022-10-18
Revised: 2024-05-23
Published Online: 2024-10-16
Published in Print: 2025-01-01

© 2024 Walter de Gruyter GmbH, Berlin/Boston

Downloaded on 8.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/crelle-2024-0069/html
Scroll to top button