1. Introduction
Fix a positive integer. To any degree extension of number fields , we may canonically associate an element of the ideal class group of , namely where as an -module. The class is called the Steinitz class of and is denoted . By a theorem of Hecke, the square of the Steinitz class is the the class of the relative discriminant . The Steinitz realization problem asks: does every element of occur as the Steinitz class of a degree extension of ?
When , there is an affirmative answer to the Steinitz realization problem; in these cases the Steinitz classes of -extensions are equidistributed in when degree number fields are ordered by discriminant. The case was proven by Kable and Wright [kab] and the case was proven by Bhargava, Shankar, and Wang [bsw]. The proofs rely on the parametrizations of degree number fields; for , there are no such parametrizations. The purpose of this article is to give an affirmative answer to the Steinitz realization problem for all .
Theorem 1.1.
Every element of occurs as the Steinitz class of a degree extension of with squarefree discriminant.
We give a sketch of the proof. A rank n ring over is a ring that is locally free rank of as an -module. By work of Wood [wood], a binary -ic form
| | |
gives rise to a rank ring . If is an integral ideal of such that and , we construct a rank ring and an inclusion . When is irreducible, both rings are orders in a number fields. The Steinitz class of is and .
If is squarefree as an ideal and is irreducible, then is the maximal order in a degree extension of with Steinitz class . Thus, Theorem1.1 immediately follows from the existence of an appropriate .
Theorem 1.2.
For every nonzero proper prime ideal of coprime to , there are infinitely many monic degree polynomials
| | |
with coefficients in such that:
- (1)
is irreducible over ;
- (2)
and ;
- (3)
and is squarefree as an ideal in .
The proof of Theorem1.2 is a generalization of the proof given by Kedlaya when proving that there exist infinitely many irreducible monic polynomials over with squarefree discriminant [ked].
1.1. Acknowledgements
The author would like to thank Ravi Vakil and Kiran Kedlaya for valuable conversations. She would also like to thank the NSF and Stanford University, where this work was completed.
3. Proof of Theorem1.2
The proof given in this section is essentially that given by Kedlaya [ked]. The main differences are that we work over a number field and impose a few additional congruence conditions needed for our application. We repeat much of Kedlayaβs construction verbatim, and encourage the reader to refer to Kedlayaβs paper for more detail. Fix a class , and pick a representative prime ideal coprime to . For , put:
| | |
| | |
The integral in the definition of should be interpreted formally; is a polynomial. The discriminant of , up to sign, is the product of evaluated at the roots of . So, we obtain the following equality of ideals:
| | |
Suppose and . If we write , then and
| | |
Furthermore, divides and hence divides . Therefore, we wish to show that there exists a choice of with and such that has integral coefficients, is irreducible, and the ideal is squarefree. We will do this by fixing an appropriate choice of and then applying a squarefree sieve to choose .
We will first show that there exist such that:
- (1)
and for ;
- (2)
has integral coefficients;
- (3)
for every prime ideal , there exists such that is not divisible by ;
- (4)
there exists such that is not divisible by ;
- (5)
there exists a prime ideal and such that is irreducible modulo ;
- (6)
and the polynomials are pairwise coprime.
To do this, weβll need the following two lemmas.
Lemma 3.1 (Kedlaya, Lemma 2.3 [ked]).
Let be be such that is a principal prime ideal coprime to . Then there exist infinitely many primes modulo which the polynomial is irreducible and its derivative splits into linear factors.
Proof.
The polynomial has splitting field , in which does not ramify because does not divide . Thus is an Eisenstein polynomial with respect to any prime above in ; in particular, is irreducible over . By the Chebotarev density theorem, there exist infinitely many totally split prime ideals of modulo which is irreducible; the norm to of any such prime ideal is a prime ideal of the desired form.β
Lemma 3.2 (Kedlaya, Lemma 2.4 [ked]).
For any field of characteristic zero, there exist such that
| | |
are distinct.
Let , , and be as in Lemma3.1 and ensure that . Apply Lemma3.2 to obtain so that
| | |
have distinct values. Choose a prime ideal such the reductions have well-defined and distinct values modulo . Now use the Chinese remainder theorem to choose subject to the following congruence conditions:
- β’
is coprime to and is contained in and the ideal generated by ;
- β’
the integers are contained in and modulo for ;
- β’
the quantities are congruent to the roots of modulo ;
- β’
and modulo .
Weβll now show that this choice of satisfies conditions .
- (1)
Follows directly from the choice of .
- (2)
Observe that:
| | |
so has integer coefficients.
- (3)
Suppose that is a prime ideal dividing . Then for , we have for because . If , then
| | |
and if , then
| | |
Consequently, is not divisible by . Now suppose that is a prime ideal coprime to , so . There are at least choices of for which is indivisible by , because each linear factor rules out exactly one choice of .
- (4)
Choose so that . To conclude it suffices to show that for all . Because modulo for and is a nonzero polynomial with positive coefficients in , we can ensure this by choosing to have sufficiently large norm.
- (5)
Ensured by the fact that the quantities are congruent to the roots of modulo .
- (6)
Ensured by the fact that modulo .
Given such a choice of the above, let be the set of all such that amd . For any , the polynomial is irreducible with integral coefficients such that , , , and . For , put where are the embeddings of into . The proof of Theorem1.2 is concluded by the following lemma.
Lemma 3.3.
Let be a set in defined by congruence conditions at a finite set of primes . Let and put . Suppose that is squarefree away from . For , let
| | |
Then
| | |
If for all , then .
Proof.
If for all , then there exists a such that:
| | |
which proves the second claim. To prove the first claim it suffices to prove the following tail estimate:
| | |
We upper bound the tail estimate by the sum of the following expressions:
| | |
| | |
Both sums are upper bounded by , completing the proof.β