A note on depth preservation
Manish Mishra, Basudev Pattanayak

TL;DR
This paper demonstrates that for certain wildly ramified tori, the depth characteristic is not maintained under the local Langlands correspondence, highlighting a specific limitation in the current understanding.
Contribution
It reveals that depth preservation fails for wildly ramified tori in the local Langlands correspondence, providing new insights into ramification behavior.
Findings
Depth is not preserved for wildly ramified tori.
The failure of depth preservation is specific to certain ramification conditions.
This challenges assumptions about the general behavior of the local Langlands correspondence.
Abstract
We show that for a wildly ramified torus, depth is not preserved in general under local Langlands correspondence for tori.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Algebra and Geometry · Geometry and complex manifolds · Algebraic Geometry and Number Theory
A note on depth preservation
Manish Mishra
Department of Mathematics
Indian Institute for Science Education and Research
Dr. Homi Bhabha Road, Pashan
Pune 411 008
India
and
Basudev Pattanayak
Abstract.
We show that for a wildly ramified torus, depth is not preserved in general under local Langlands correspondence for tori.
1. Introduction
Let be a non-archimedean local field and let denote its Weil group. Local class field theory (LCFT) tells us that there is a canonical isomorphism and this isomorphism respects the numbering on the filtration subgroups of and the upper numbering on the filtration subgroups of . Local Langlands Correspondence (LLC) stipulates a vast generalization of the LCFT isomorphism. In LLC, irreducible representations of the -points of a reductive -group are expected to be parametrized by arithmetic objects called Langlands parameters in a certain natural way. For each , Moy-Prasad theory associates an invariant called depth . Also for each , one defines the notion of depth . It is the smallest number such that is trivial on for all . If associates to under LLC, then one expects that quite fairly . This is known in many cases (see the introduction in [ABPS2016] for a survey). However, counter examples have been constructed for inner forms of [ABPS2016] and in the case of when has characteristic [AMPS2017].
Now let where is a finite separable extension of and denotes the Weil restriction and let under LLC. In this note, we show that where is the Hasse-Herbrand function and is the ramification index of . Thus for all postitive depth characters , . When is a tamely induced wildly ramified torus (see Sec. 7.1), we show that admits characters for which depth is not preserved under LLC. In Section 8, we compute Hasse-Herbrand function for a certain wildly ramified extension of a cyclotomic field to illustrate the failure of depth preservation.
The proofs in Section 7 follow closely the proofs in [Yu] and [MM2015].
2. Review of ramification groups
Let be a non-archimedean local field and let be a finite Galois extension of . Write , for the ring of integers of and the maximal ideal of . For , define to be the set of all such that operates trivially on . Then . The groups are called ramification groups. They form a decreasing filtration of normal subgroups. Extend the definition of for all real numbers by setting
[TABLE]
This numbering of ramification groups is called lower numbering. Lower numbering behaves well with respect to intersections, i.e., if is a subgroup of , then .
Upper numbering of ramification groups
Define to be the map where for . The function is called the Hasse-Herbrand function. It has the basic properties [Serre]:
- (a)
is continuous, piecewise linear, increasing and concave.
- (b)
.
- (c)
is a homeomorphism of onto itself.
- (d)
If is a normal subgroup of , then .
If an extension is not Galois, define , where is a Galois extension of containing . The inverse is denoted .
Define an *upper numbering *on ramification groups by setting if . Upper numbering behaves well with respect to quotients, i.e., if is a normal subgroup of , then
[TABLE]
For an infinite Galois extension of , define the ramification groups on by:
[TABLE]
Now let be Galois extension of local fields and let be a finite extension of contained in . Write and .
Lemma 1**.**
For all , .
Proof.
Let be a fintie Galois extension of in containing . Write . Then
[TABLE]
The lemma now follows by taking inverse limit over . ∎
3. Notion of depth
Let where and are local fields and let be a -module. Define the depth of to be:
[TABLE]
Define the depth of a co-cycle to be:
[TABLE]
4. Depth change under induction
Let and where and finite Galois. Let be an -module.
Proposition 2**.**
**
Proof.
By Mackey theory,
[TABLE]
Here denotes the restriction functor and denotes the -twisted module . By Lemma 1, . Thus
[TABLE]
∎
5. Depth change under Shapiro’s isomorphism
Again and where and is any finite extension and let be an -module.
Shapiro’s lemma states that the map
[TABLE]
defined by
[TABLE]
is an isomorphism. We wish to relate the depth of co-cycles under this isomorphism. We first observe the following:
Lemma 3**.**
Let A be a group, and subgroups of with being normal in . Let be a -module. Then there is a canonical isomorphism of -modules:
[TABLE]
Proof.
The map is easily verified to be the required isomorphism. ∎
Lemma 4**.**
For , Shapiro’s lemma induces an isomorphism
Proof.
We have
[TABLE]
The first isomorphism follows from Lemma 3, second from Lemma 1 and the last from Shapiro’s lemma. ∎
Write
Corollary 5**.**
If , then .
Proof.
Let . Then if . By Lemma 4, this implies if . Therefore . The argument is reversible showing that . Therefore .
∎
6. Langlands correspondence for tori
We review here the statement of local Langlands correspondence for tori as stated and proved in [Yu].
6.1. Special case
Let where is a finite separable extension of and denotes the Weil restriction. Then and the group of characters is canonically a free -module with basis where (resp. ) denotes the Weil group of (resp. ). From this, it follows that the complex dual of is canonically isomorphic to . We get,
[TABLE]
The isomorphism 6.1 follows by class field theory and the isomorphism 6.1 by Shapiro’s lemma.
6.2. The LLC for tori in general
Theorem**.**
*[L97]*There is a unique family of homomorphisms
[TABLE]
with the following properties:
- (1)
* is additive functorial in , i.e., it is a morphism between two additive functors from the category of tori over to the category of abelian groups;* 2. (2)
For , where is a finite separable extension, is the isomorphism described in Section 6.1.
7. Depth change for tori under LLC
We keep the notations as in Section 6. Let be a local field. Recall that admits a filtration where is the units of the ring of integers and for , . Here is the valuation of normalised so that . Under local class field theory isomorphism
[TABLE]
We recall that carries a Moy-Prasad filtration [MP96]. The depth of a character is defined to be
[TABLE]
The group is called the Iwahori subgroup of . It is a subgroup of finite index in the maximal compact subgroup of . When , then for ,
[TABLE]
Here is the valuation on normalised so that and is the ramification index of . The equality 7.1 follows from [Yu03, Sec. 4.2] and the equality 7 follows from the fact that for all
Theorem 6**.**
Let , where is a finite separable extension of local fields of ramification index . Then for , the local Langlands correspondence for tori induces an isomorphism:
[TABLE]
Proof.
The case is a special case of [MM2015, Theorem 7]. For , this follows by
[TABLE]
Here, the isomorphism 7 follows from Lemma 4. ∎
Corollary 7**.**
For as in Theorem 6
[TABLE]
Proof.
This follows from an argument analogous to the argument in the proof of the Corollary 5. ∎
Remark 8*.*
The slope of the map at a differentiable point is . Thus, when is a wildly ramified extension, and consequently .
When is a tamely ramified extension, . Therefore in this case, Corollary 7 simplifies to,
[TABLE]
This is a special case of Depth-preservation Theorem of Yu for tamely ramified tori [Yu, Sec. 7.10].
7.1. Case of a tamely induced tori
Recall that a -torus is called induced if it is of the form , where are finite separable extensions of . A -torus is called tamely induced if is an induced torus for some tamely ramified extension of . In this section, we compare depths under LLC for such tori following the proof in [Yu, Sec. 7.10].
Let be a tamely induced -torus. Then there exists an induced torus such that and is connected. Further (see proof in [Yu03, Lemma 4.7.4]). Let and let denote its lift to . Then
[TABLE]
Here denotes and . By functoriality, is the image of under and therefore . But
[TABLE]
Here denotes the ramification index of . Thus
[TABLE]
Now assume is wildly ramified. We will now produce a character of for which the inequality (7.6) is strict. We can assume without loss of generality that is wildly ramified. Let be a positive depth character of which is trivial on . Extend trivially to a character of . Then since is divisible, the character lifts to a character of . By Remark 8, . Since and , it follows that the inequality (7.6) is strict for this choice of .
8. An Example
Let , , where denotes a primitive th root of unity, . Then is a totally ramified extension of degree Consider the intermediate extension of of degree over . Then is a wildly ramified extension. Write and .
Lemma 9**.**
For and .
Proof.
We first note that since we considering abelian extensions, the jumps in filtration occur at integer values. We have for ,
[TABLE]
The last equality holds because for and . Thus
[TABLE]
∎
Write and let . By [Serre, Chap IV, Prop. 17], . Define
[TABLE]
Then . The ramification groups of are [Serre, Chap IV, Prop. 18]:
[TABLE]
We now calculate
Proposition 10**.**
The Hasse-Herbrand function of the wildly ramified extension is given by
[TABLE]
Proof.
We consider various cases:
- •
Case
[TABLE]
- •
Case
[TABLE]
Therefore, .
- •
Case with
[TABLE]
Therefore, .
- •
Case
[TABLE]
Therefore,
∎
Now write and let be as denoted in Sec. 6. It then immediately follows from Prop. 10:
Lemma 11**.**
* Consequently, for all positive depth , .*
9. Acknowledgement
The authors would like to thank Anne-Marie Aubert for carefully going over this article and suggesting several improvements.
References
