Reduction mod p p of semi-stable representations of some super-Breuil weights
Anand Chitrao, Eknath Ghate · 2026
Read Reduction mod p p of semi-stable representations of some super-Breuil weights by Anand Chitrao, Eknath Ghate (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-16867.txt.
Introduction
Chitrao and Ghate compute the reductions modulo a prime p of certain semi-stable representations. Their results cover weights from p plus 5 to 2p and from 2p plus 6 to 3p plus 1, for primes p at least 5. This extends earlier methods beyond the weight range where they previously worked, and sharpens a bound on the valuation of the parameter given by Bergdall, Levin, and Liu.
Catalog record
| CREATOR | Anand Chitrao, Eknath Ghate |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51437 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.16867
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Reduction mod p p of semi-stable representations of some super-Breuil weights Anand Chitrao, Eknath Ghate Abstract We determine the mod p p reductions of the semi-stable representations V k , ℒ V_{k,\mathcal{L}} of weight k ∈ [ p + 5 , 2 p ] ∪ [ 2 p + 6 , 3 p + 1 ] k\in[p+5,2p]\cup[2p+6,3p+1] and v p ( ℒ ) < 1 − k / 2 v_{p}(\mathcal{L})<1-k/2 for primes p ≥ 5 p\geq 5 . In particular, this shows that the techniques introduced in [ 5 ] involving the p p -adic and mod p p local Langlands correspondences can be used to compute the reduction of V k , ℒ V_{k,\mathcal{L}} outside the range k ∈ [ 3 , p + 1 ] k\in[3,p+1] . Moreover, this shows that the bound on v p ( ℒ ) v_{p}(\mathcal{L}) given by Bergdall-Levin-Liu [ 1 ] can be improved, at least for weights k ∈ [ 2 p + 6 , 3 p + 1 ] k\in[2p+6,3p+1] . 1 Introduction Fix a prime p\geq 5 . Let E be a finite extension of \mathbb{Q}_{p} containing \sqrt{p} . For k\geq 3 , let V_{k,\mathcal{L}} be the two-dimensional irreducible semi-stable representation with Hodge-Tate weights (0,k-1) and \mathcal{L} -invariant \mathcal{L}\in E . In our earlier paper [ 5 ] , we computed the reduction mod p of V_{k,\mathcal{L}} for k\in[3,p+1] and for arbitrary \mathcal{L} , extending the work of [ 2 ] and [ 9 ] for k\in[3,p-1] . To achieve this, we computed the reduction mod p of the Banach space associated to V_{k,\mathcal{L}} by the p -adic local Langlands correspondence (see [ 3 , 4 , 7 ] ). As was noted in [ 5 ] , our method works, in theory, for arbitrary k\geq 3 . It was put into practice by computing the reductions completely for the weights k=p and p+1 . In [ 1 ] , Bergdall-Levin-Liu used the theory of Breuil and Kisin modules to compute the reductions of V_{k,\mathcal{L}} for arbitrary k\geq 4 but for \mathcal{L} satisfying v_{p}(\mathcal{L})<2-k/2-v_{p}((k-2)!) (even for p=3 ). In Section 1.3 of their paper, the authors mention that a computation using R. Pollack’s code shows that the conclusion of their result holds for p=3 , k=6 , and v_{p}(\mathcal{L})=-2 . Note that 2-k/2-v_{p}((k-2)!)=-2 , hinting at the fact that their bound could possibly be improved. Indeed, in [ 1 , Remark 1.2] , the authors wonder to what extent their bound is optimal. In [ 5 ] , we showed that their bound is indeed optimal for k\in[3,p+1] . For k\in[3,p-1] , this was already known by [ 2 ] and [ 9 ] . The goal of this paper is twofold. On the one hand, we wish to recover some of the results in [ 1 ] using our methods. On the other hand, we wish to see whether the bound given in [ 1 ] can be improved. In order to do this, we compute the reduction of V_{k,\mathcal{L}} for k belonging to [p+5,2p]\cup[2p+6,3p+1] and \mathcal{L} satisfying v_{p}(\mathcal{L})<-r/2 , where r=k-2 . When k\in[p+5,2p] , we recover the result in [ 1 ] . Since for k\in[2p+6,3p+1] , the expression 2-k/2-v_{p}((k-2)!) equals -1-r/2 , which is strictly smaller than -r/2 , our result improves the bound in [ 1 ] for k\in[2p+6,3p+1] . We use the notation in [ 5 ] . We write most of this paper assuming that p\leq r\leq p^{2}-p-1 and that v_{p}(\mathcal{L}) is slightly more general than v_{p}(\mathcal{L})<-r/2 . We only specialize to r\in[p+3,2p-2]\cup[2p+4,3p-1] and v_{p}(\mathcal{L})<-r/2 in the last section, where we prove the following result. Theorem 1.1 . Let p\geq 5 and p+3\leq r\leq 2p-2 or 2p+4\leq r\leq 3p-1 . Then, for any \mathcal{L} with v_{p}(\mathcal{L})<-r/2 , we have \overline{V}_{k,\mathcal{L}}\simeq\mathrm{ind}\>\omega_{2}^{r+1} . To prove Theorem 1.1 , we recall that by [ 5 , (9)] , there is a surjection \mathrm{ind}_{IZ}^{G}\>\mathrm{Sym}^{r}\mathbb{F}_{q}^{2}\twoheadrightarrow\overline{\tilde{\Theta}(k,\mathcal{L})}. The canonical IZ -equivariant filtration on \mathrm{Sym}^{r}\mathbb{F}_{q}^{2} induces a G -equivariant filtration on \overline{\tilde{\Theta}(k,\mathcal{L})} . We name the sub-quotients of this filtration on \overline{\tilde{\Theta}(k,\mathcal{L})} in such a way that the sub-quotient \mathrm{ind}_{IZ}^{G}\>a^{i}d^{r-i} of \mathrm{ind}_{IZ}^{G}\>\mathrm{Sym}^{r}\mathbb{F}_{q}^{2} surjects onto the sub-quotient F_{2i,2i+1} of \overline{\tilde{\Theta}(k,\mathcal{L})} for 0\leq i\leq r . The proof of the theorem proceeds by eliminating all but one of the sub-quotients F_{2i,2i+1} . The remaining sub-quotient is the one contributing towards \overline{\tilde{\Theta}(k,\mathcal{L})} . We then use the Iwahori theoretic version of the mod p local Langlands correspondence given in [ 6 , Theorem 2.2] to see that this sub-quotient must be supersingular. We explain the strategy of the proof in detail now. First, using [ 5 , Lemma 6.1] , we may eliminate F_{2i,2i+1} for i>r/2 . Next, consider the following diagram of the F_{2i,2i+1} for 0\leq i\leq\lfloor r/2\rfloor \begin{array}[]{ccccccc}p^{r/2-\lceil r/2\rceil}z^{\lceil r/2\rceil}&\ldots&p^{r/2-(r-2)}z^{r-2}&p^{r/2-(r-1)}z^{r-1}&p^{r/2-r}z^{r}\\ \vskip 5.69046ptF_{2\lfloor r/2\rfloor,2\lfloor r/2\rfloor+1}&\ldots&F_{4,5}&F_{2,3}&F_{0,1}\\ z^{\lfloor r/2\rfloor}&\ldots&z^{2}&z&z^{0}.\end{array} The bottom row displays the usual generator of F_{2i,2i+1} , which is the image of the function \llbracket\mathrm{id},X^{i}Y^{r-i}\rrbracket under the surjection above (although we must multiply by the common function \mathbbm{1}_{p\mathbb{Z}_{p}} ), and the top row displays -\beta=-\left(\begin{smallmatrix}0&1\\ p&0\end{smallmatrix}\right) applied to the usual generator (again with \mathbbm{1}_{p\mathbb{Z}_{p}} everywhere). In our earlier paper [ 5 ] , the sub-quotient which contributed in the \nu<1-r/2 case was F_{0,1} . This time around, however, F_{0,1} will be eliminated. This requires a new idea, which is described in Section 3 . It was a surprise to us that F_{0,1} can be eliminated without using the poly \cdot log functions (so called in [ 8 ] ) so prevalent in [ 5 ] ! It turns out that for i\geq 0 , the sub-quotient F_{2i,2i+1} can be eliminated if r\geq(i+1)(p+1)-1 . The shallowest sub-quotient that is not eliminated by this easy method is the contributing one. Therefore, we skip that sub-quotient and move on to eliminating the remaining ones, which we henceforth refer to as the “deeper” sub-quotients. In Section 4 , we prove a master congruence in the form of Proposition 4.4 , which will be used multiple times to eliminate all the deeper sub-quotients. One can view this congruence as an extension of the \nu<1+r/2-n case of [ 5 , Proposition 9.6 ] to higher values of r . Stirling numbers of the second kind also make an appearance. In Section 5 , we eliminate the deeper sub-quotients by proving three mutually exclusive and exhaustive theorems depending on the power of z in the top row in the diagram above. Theorems 5.1 and 5.2 use the congruence in Proposition 4.4 for a single value of n satisfying v_{p}([n]_{b+1})=0 and 1 , respectively, where b=\lfloor n/p\rfloor . We refer to the arguments given in these theorems as the good method and the bad method, respectively. Theorem 5.3 requires invoking Proposition 4.4 twice, once for n satisfying v_{p}([n]_{b+1})=1 and the other for n satisfying v_{p}([n]_{b+1})=0 . We shall refer to this as the ugly method 1 1 1 With apologies to the 1966 cult classic. . Our main theorem then follows from a simple application of the Iwahori mod p LLC. 2 Preliminaries First, we state some combinatorial lemmas. We begin with the mod p^{2} version of Lucas’ theorem, proved by Rowland. Lemma 2.1 ( [ 10 , Theorem 4] ) . If 0\leq