The number of cuspidal representations over a function field and its behavior under base changes

Takuro Fukayama

REVIEW 3 major objections 3 minor 13 references

The number of cuspidal representations over a function field and its behavior under base changes

T0 review · 3 major / 3 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Assuming Gross's trace-formula conjectures, the paper proves that for $G=\mathrm{SL}_\ell$ with $\ell$ prime and for $G=\mathrm{Sp}_4$ or $\mathrm{Sp}_6$, the sum over semisimple conjugacy classes of the L-functions…

desk verdict Careful, honest computation in Gross's program; the Sp6 stress-test worry dissolves on reading, but the paper remains conditional on Gross's conjectures. read the letter →

arxiv 2504.20564 v1 pith:VURWTQY3 submitted 2025-04-29 math.NT

Takuro Fukayama This is my paper · ORCID

classification math.NT MSC 11F7011G4011R5814G10

keywords cuspidalrepresentationsLefschetztypefunctionsbasechangefunctionfieldstraceformulaArtin-TatemotivessimplesupercuspidalSteinbergrepresentation

topics Birch and Swinnerton-Dyer

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper is trying to establish a rigid law for how counts of cuspidal representations of a split simply connected group over a function field change when the base field $\mathbb{F}_q$ is enlarged to $\mathbb{F}_{q^m}$. The law, called Lefschetz type, says the count is a finite $\mathbb{Z}$-linear combination of exponentials $\alpha_i^m$; this is exactly the shape of a Frobenius trace on a finite-dimensional object. The route goes through Gross's conjectural trace formula, which converts a sum of representation multiplicities into a sum of L-functions attached to centralizers of semisimple elements. The paper proves unconditionally that for $G=\mathrm{SL}_\ell$ with $\ell$ prime and for $G=\mathrm{Sp}_4,\mathrm{Sp}_6$, the relevant L-function sums are of Lefschetz type, and that the transfer to cuspidal multiplicities follows if Gross's conjectures are accepted. If true, this places representation counting in the same framework as counting $\ell$-adic local systems, where the same Lefschetz law was previously known.

What carries the argument

The central object is the Artin–Tate motive $M_G=\bigoplus_{d\geq 1}V_d(1-d)$ attached to a reductive group, where $V_d$ is the degree-$d$ part of the $\Gamma_q$-representation on $X^*(T_0)\otimes\mathbb{Q}$, together with its global L-functions $L_{S,T}(M,s)$; the $S$-modified value $L_S(M_{G_\gamma})$ is what the trace formula produces for each semisimple class. The argument is carried by three devices: the Springer–Steinberg classification of semisimple conjugacy classes by characteristic polynomial, which reduces each centralizer $G_\gamma$ to products of Weil restrictions of $\mathrm{GL}_a$, unitary groups, and symplectic groups; explicit determinant formulas $\det(1-t\,\mathrm{Fr}_q\mid M_{G_\gamma})$; and cyclotomic-polynomial identities plus Möbius inversion that turn the base-change sum into a polynomial in $q^m$ and the Frobenius eigenvalue set $J^m$.

What would settle it

For the unconditional theorems, a direct check would be to compute $L(G,m)$ for $G=\mathrm{Sp}_8$ or $G=\mathrm{SL}_4$ over a concrete curve with $\#S\geq 2$ for several small $m$, and see whether the sequence is a finite integer-linear combination of exponentials; a negative answer disproves Conjecture 4.6. For the conditional claim, one could compute both sides of Gross's trace formula in an explicit case, for example $G=\mathrm{Sp}_4$ over $\mathbb{F}_q(t)$ with $S=\{0,\infty\}$; a mismatch would show that the multiplicity-sum conclusion does not follow from the conjectures.

Watch

Extended reading notes

Core claim

The central discovery is Theorem 5.4: for $G=\mathrm{SL}_\ell$ with $\ell$ prime, the function $m\mapsto L(G,m)=\sum_{[\gamma]}L_S(M_{G_\gamma})$, summed over semisimple conjugacy classes in $G(\mathbb{F}_{q^m})$, is of Lefschetz type; and Theorems 6.5 and 6.6: the same holds for $G=\mathrm{Sp}_4$ and $G=\mathrm{Sp}_6$. Here $M_{G_\gamma}$ is the Artin–Tate motive of the centralizer and $L_S$ is its $S$-modified L-function at $s=0$. In the case where the finite set $S$ has at least two places and the set $T$ is empty, the paper's Theorem 4.2 (conditional on Gross's Conjectures 1.1 and 1.2) identifies $1+(-1)^{\# S\cdot r(G)}\sum_\pi m(\pi)$ with exactly this sum, so these theorems transfer the Lefschetz property to sums of multiplicities of cuspidal representations that are Steinberg at $S$ and unramified elsewhere. Theorems 5.4 and 6.5–6.6 are unconditional statements about L-functions; the representation-theoretic conclusion carries the conjectural trace formula as a hypothesis.

Load-bearing premise

The whole transfer from L-function sums to representation counts depends on an unproved trace-formula conjecture of Gross, which says that a certain global test function can be decomposed into finitely many orbital integrals; if that fails, the representation-counting conclusion fails, though the L-function theorems survive.

Editorial extensions

If this is right

For $G=\mathrm{SL}_\ell$ with $\ell$ prime and for $G=\mathrm{Sp}_4,\mathrm{Sp}_6$, the function $m\mapsto L(G,m)$ is a finite integer-linear combination of exponentials, so its values for all $m$ are determined by finitely many initial values.
If Gross's Conjectures 1.1 and 1.2 hold, the same Lefschetz law transfers to sums of multiplicities of cuspidal representations that are Steinberg at $S$ and unramified elsewhere, in the $T=\emptyset$, $\#S\geq 2$ case.
When $S$ and $T$ are both nonempty, Proposition 4.4 gives the Lefschetz law conditionally for every split simply connected almost simple group, not only the groups in Theorems 5.4 and 6.5–6.6.
The explicit polynomial expressions for $L(G,m)$ mean the base-change counts can be computed in finite terms from $q$, the Frobenius eigenvalues on $H^1(X)$, and the places in $S$.
The truth of Conjecture 4.6 for all split simply connected $G$ is reduced to proving that the associated rational function $P(x,J)$ lies in $\mathbb{Z}[x,J]$.
Editorial extension: if the Lefschetz law holds for all $G$, then the counting function $m\mapsto\sum m(\pi)$ is morally the trace of $\mathrm{Fr}_{q^m}$ on a virtual motive; this suggests that cuspidal counts should admit a geometric, motive-like realisation rather than only a trace-formula computation.
Editorial extension: because Theorems 5.4, 6.5, and 6.6 are unconditional, the polynomial identities they prove constrain any future spectral expansion: the geometric side of the trace formula must reproduce these explicit Lefschetz identities, so they can serve as test cases for proposed trace-formula refinements.
Editorial extension: a direct computational check of Conjecture 4.6 for the next open cases, such as $G=\mathrm{SL}_4$ or $G=\mathrm{Sp}_8$, would either extend the pattern or find the first counterexample; the paper's tables and polynomial recipes make such a check concrete.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the Lefschetz law holds for all split simply connected $G$, then the counting function $m\mapsto\sum m(\pi)$ is morally the trace of $\mathrm{Fr}_{q^m}$ on a virtual motive; this suggests a geometric construction of a motive or sheaf whose cohomology counts cuspidal representations, in the spirit of counting local systems.
Because Theorems 5.4, 6.5, and 6.6 are unconditional, they constrain any future trace formula: whatever the correct geometric side is, its sum over semisimple classes must reproduce the explicit Lefschetz identities proved here, so those identities can be used as test cases for proposed spectral expansions.
The methods suggest a concrete computational check of Conjecture 4.6: evaluate the rational function $P(x,J)$ constructed in Section 6 for $\mathrm{Sp}_8$ or $\mathrm{SL}_4$; if its denominator cannot be cancelled for all $q$, the conjecture fails for that group.
The sharp distinction between the $S,T\neq\emptyset$ case and the $T=\emptyset$ case suggests that the hardest part of extending this work is not the Lefschetz law itself but the classification and counting of semisimple conjugacy classes that contribute to $L_S(M_{G_\gamma})$.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Fast take: this is a careful, honest computation paper in Gross's program, and the specific stress-test worry about the Sp6 cancellation does not survive contact with the paper. The proofs are mostly long elementary manipulations, but they are real proofs.

What is new: the paper proves that for G = SL_ℓ (ℓ prime), Sp4, Sp6, the base-change function m ↦ L(G,m) = Σ L_S(M_{Gγ}) is of Lefschetz type. That is new as far as I can tell; the framework comes from Gross and Yu, but the explicit centralizer-motive bookkeeping is the contribution. Theorem 5.5 with Proposition 5.8 (the polynomiality statement for SL_n when gcd(n, q^m−1) = 1) is a substantial algebraic result, and the S,T nonempty case is handled cleanly via Proposition 3.6.

Where it is soft: the title's cuspidal-representation claim is conditional on Gross's Conjectures 1.1 and 1.2, and the author says so plainly. If those conjectures fail, the unconditional Lefschetz-type results for the L-function sums still stand. The Sp6 proof is dense; I did not machine-check every table entry or congruence, and there are minor typos (e.g., the duplicated (c) label in Theorem 6.6's proof). But I did check the specific cancellation at x = √−1 flagged in the stress-test. The paper's grouping is correct: Hτ1, Hτ2, Hτ7 all evaluate to ∏(1−iα)(1+α²), while Hτ3 and Hτ11 evaluate to ∏(1+α)(1+α²); the coefficients for the first group sum to zero and the coefficients for the second group are i and −i. So P4(i) = 0 does hold. The stress-test simply misread which H's land in which product.

Bottom line: this deserves a serious referee. The claims are honestly delimited, the computations are substantial, and the conditional interpretation is clearly labeled. The citation pattern is standard for the area. I would send it to peer review, asking a referee to verify the Sp6 tables and the congruence checks in Theorem 5.5 carefully, but I do not see a load-bearing flaw.

Referee Report

3 major / 3 minor

Summary. The paper assumes Gross's conjectural trace formula to express sums of multiplicities of cuspidal representations with prescribed local conditions as sums of Artin-Tate L-functions, and studies the behavior of these sums under the base change F_q ↦ F_{q^m}. The main unconditional results are Theorem 5.4 for SL_ℓ (ℓ prime) and Theorems 6.5 and 6.6 for Sp4 and Sp6, which assert that the corresponding sums over semisimple conjugacy classes are Lefschetz type functions of m; Proposition 4.4 treats the case where both S and T are nonempty. The connection to cuspidal representations is conditional on Gross's conjectures as stated in Theorem 4.2.

Significance. The SL_ℓ result is a clean, self-contained verification of Conjecture 4.6 in a non-trivial case, and the paper is transparent about the conditional nature of the trace-formula applications. If the Sp4 and Sp6 proofs were correct, they would provide further strong evidence for the expected Lefschetz-type behavior. However, as detailed in the major comments, the proofs of Theorems 6.5 and 6.6 contain a load-bearing error in the evaluation at x = -1, and the resulting expressions appear to fail for an explicit elliptic curve; hence the main new claims are not established.

major comments (3)

[Section 6, proof of Theorem 6.5, part (b)] The assertion that H_{τ_i}(-1) = ∏_{α∈J}(1+α)^2 for i = 1,...,5 is false for i = 1. From the definition H_{τ_1}(x) = ∏_{α∈J}(1-αx)(1-αx^3), one obtains H_{τ_1}(-1) = ∏_{α∈J}(1+α)(1+α^3), not A. Consequently P_2(-1) = ∏(1+α)(1+α^3) - ∏(1+α)^2, which is not identically zero; for J = {2} it equals 18. This invalidates the proof that P(x) ∈ Z[x], so the Lefschetz-type conclusion for Sp4 is not established.
[Section 6, proof of Theorem 6.6, part (c)] The same type of error occurs in the Sp6 proof. One has H_{τ_1}(-1) = ∏(1+α)(1+α^3)(1+α^5), and H_{τ_2}(-1) = H_{τ_3}(-1) = ∏(1+α)^2(1+α^3), none of which equals A = ∏(1+α)^3 for general J. Writing B = ∏(1+α), C = ∏(1+α^3), D = ∏(1+α^5), the value P_3(-1) becomes (1/3)BCD - B^2C + (2/3)B^3, which is nonzero for example when J = {2}. Thus the claimed vanishing at x = -1 is not proved, and Theorem 6.6 is unsupported.
[Section 6, q-even cases and a concrete counterexample] The error also affects the q-even parts of Theorems 6.5 and 6.6. For the elliptic curve E: y^2+y = x^3 over F_4, the Frobenius eigenvalues on H^1 are -2,-2; choosing S to be two F_4-rational points gives J = {-2,-2}. In the q-even part of Theorem 6.5 one computes Q_2(-1) = (B/2)(C-B) with B = 1 and C = 49, so Q_2(-1) = 24 ≠ 0. Since the proof identifies L(Sp4,m) with Q(q^m, J^m), a pole at x = -1 implies L(Sp4,m) cannot be of Lefschetz type as a function of m. This strongly indicates that the failure is not merely a proof gap but a genuine counterexample to the stated theorem if the identification is correct.

minor comments (3)

[Section 3, Proposition 3.6] The sentence 'Note that JX, JS, JT ⊂ Z' is false for JX: Frobenius eigenvalues on H^1(X) are Weil numbers and need not be rational integers. The subsequent argument treats J as indeterminates, so the incorrect claim is not used, but the sentence should be corrected.
[Section 6, proof of Theorem 6.6] The verification steps are mislabeled: the label '(c)' appears twice, and part (e) follows. The numbering should be corrected for readability.
[Section 6, proof of Theorem 6.6, part (d)] The evaluation at x = √-1 in part (d) is correct when the terms are grouped as τ1, τ2, τ7 for the first product and τ3, τ11 for the second; the alleged mis-evaluation in the stress-test note does not occur. The actual obstruction lies in the x = -1 evaluations discussed in the major comments.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the L-function sums are computed directly from explicit class-type data, and the cuspidal-representation conclusions are conditional on Gross's conjectures, which are independent premises rather than conclusions of the paper.

full rationale

The paper's central computations in Sections 5 and 6 are self-contained reductions to polynomial identities. For SL_l, Theorem 5.4 explicitly evaluates N_{l,1}(q) and N_{l,ell}(q), defines H_1(x) and H_ell(x) from the centralizer motives, and proves integrality via the cyclotomic identity H_1(zeta_l)=H_ell(zeta_l); no fitted parameter is renamed as a prediction, and no step assumes the Lefschetz-type property being proved. For Sp4 and Sp6, Theorems 6.5 and 6.6 list the finite semisimple types, compute R_tau(x) and H_tau(x) from the explicit centralizer motives and the class-counting formulas, and then verify the divisibility conditions needed to conclude P(x,J) lies in Z[x,J]. These are direct algebraic verifications, not circular reductions. The link to sums of multiplicities of cuspidal representations rests on Theorem 4.2, quoted from Gross's paper, and is explicitly conditional on Conjectures 1.1 and 1.2; relying on an unproved external conjecture is a conditionality issue, not a circularity issue. The citations to [Gro11], [Gro97], [Kot88], [SS70], and [GR10] supply standard or stated external inputs; none of the load-bearing identities is justified merely by the present authors' prior work. The skeptical concern about the Sp6 computation in Theorem 6.6 concerns the correctness of a displayed evaluation at sqrt(-1), which is a potential arithmetic error rather than a circularity: even if the divisibility argument fails, the derivation would be wrong, not circular. Accordingly, no specific circular step can be quoted, and the appropriate score is 0.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The central computational theorems rest on standard algebraic geometry (Weil conjectures, Grothendieck-Lefschetz trace formula), the motivic L-function machinery of Gross, the classification of centralizers from Springer-Steinberg, and Carlitz's count of reciprocal polynomials. The cuspidal-representation interpretation adds Gross's two conjectures as assumptions, which are clearly labeled. No free parameters are fitted to data and no new entities are introduced.

assumptions (6)

domain assumption Gross's Conjecture 1.2: Tr(ϕ | L^2_d(G)) = Σ_{[γ]} O_γ(ϕ) for the global test function ϕ_{S,T,V}
Used in Theorem 4.2 to convert the trace of the test function into the sums of L-functions L_S(M_{Gγ}) that are the paper's main object. Location: Conjecture 1.2 and its use in Section 4.
domain assumption Gross's Conjecture 1.1: orbital integral identity for the Euler-Poincaré function f^EP
Used to evaluate the orbital integrals appearing in the trace formula expansion. Location: Conjecture 1.1, used in Section 4.
domain assumption The identity for the global L-function L(M_G,s) = det(1−q^{-s}Fr | H^1(X)⊗M_G) / (det(...|M_G) det(...|M_G)) from [Gro11, p.1250]
Used in Remark 3.4 and Section 4 to express L_{S,T}(M_G) in terms of Frobenius eigenvalues. Taken as an established result from Gross's theory of motives.
standard math Semisimple conjugacy classes in SL_n(F_q) and Sp_{2n}(F_q) are classified by characteristic polynomials (Lemmas 5.1 and 6.1, citing [SS70])
Basis for the type classification and counting formulas N_{n,d}(q) and N_τ(q) used throughout Sections 5 and 6.
standard math Carlitz's formula for the number of self-reciprocal irreducible monic polynomials over F_q (Lemma 6.4)
Used to compute N_τ(q) for the Sp4 and Sp6 tables.
standard math Grothendieck-Lefschetz trace formula for varieties over finite fields
Used in Sections 2 and 3 to assert that m ↦ #Y(F_{q^m}) is of Lefschetz type and that the eigenvalue multisets J transform as J^m under base change.

how reviews work

0 comments

Cite this review

Pith. "Pith review of The number of cuspidal representations over a function field and its behavior under base changes." pith.science (2026). https://pith.science/paper/VURWTQY3

@misc{pith2026250420564,
  author       = {Pith},
  title        = {Pith review of: The number of cuspidal representations over a function field and its behavior under base changes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VURWTQY3}},
  note         = {Machine review of arXiv:2504.20564}
}

abstract

Let $X$ be a smooth projective curve over a finite field $\mathbb{F}_q$, $k$ be its function field, and $G$ be a simply connected almost simple split group over $\mathbb{F}_q$. We also write $G$ for its structure over $k$. We calculate the sum of multiplicities of all cuspidal representations of $G$ satisfying a given condition assuming the conjectural trace formula. We also observe how the sum changes if we replace $X$ by its base change $X\otimes_{\mathbb{F}_q}\mathbb{F}_{q^m}$.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

13 extracted references · 12 canonical work pages

[1]

Carlitz, Some theorems on irreducible reciprocal polynomials over a finite field, J

L. Carlitz, Some theorems on irreducible reciprocal polynomials over a finite field, J. Reine Angew. Math. 227 (1967), 212--220

work page 1967
[2]

Deligne and Y

P. Deligne and Y. Z. Flicker, Counting local systems with principal unipotent local monodromy, Ann. of Math. (2) 178 (2013), no. 3, 921--982

work page 2013
[3]

B. H. Gross and M. Reeder, Arithmetic invariants of discrete L anglands parameters , Duke Math. J. 154 (2010), no. 3, 431--508

2010
[4]

B. H. Gross, On the motive of a reductive group, Invent. Math. 130 (1997), no. 2, 287--313

work page 1997
[5]

, Irreducible cuspidal representations with prescribed local behavior, Amer. J. Math. 133 (2011), no. 5, 1231--1258

work page 2011
[6]

R. E. Kottwitz, Tamagawa numbers, Ann. of Math. (2) 127 (1988), no. 3, 629--646

work page 1988
[7]

Lafforgue, Chtoucas de D rinfeld et correspondance de L anglands , Invent

L. Lafforgue, Chtoucas de D rinfeld et correspondance de L anglands , Invent. Math. 147 (2002), no. 1, 1--241

work page 2002
[8]

Lan, Arithmetic compactifications of PEL -type S himura varieties , Princeton University Press, 2013

K.-W. Lan, Arithmetic compactifications of PEL -type S himura varieties , Princeton University Press, 2013

work page 2013

Show all 13 references

[9]

T. A. Springer and R. Steinberg, Conjugacy classes, Seminar on A lgebraic G roups and R elated F inite G roups ( T he I nstitute for A dvanced S tudy, P rinceton, N . J ., 1968/69), Lecture Notes in Math., vol. Vol. 131, Springer, Berlin-New York, 1970, pp. 167--266

1968
[10]

Yu, Number of cuspidal automorphic representations and H itchin's moduli spaces , 2023, arXiv:2110.13858

H. Yu, Number of cuspidal automorphic representations and H itchin's moduli spaces , 2023, arXiv:2110.13858

2023 arXiv
[11]

, -adic local systems and H iggs bundles: the generic case , 2024, arXiv:2304.06637

2024 arXiv
[12]

, Rank 2 -adic local systems and H iggs bundles over a curve , 2024, arXiv:2301.13157

2024 arXiv
[13]

write newline

" write newline "" before.all 'output.state := FUNCTION output.nonempty.mrnumber duplicate missing pop "" 'skip if duplicate empty 'pop " " swap * " " * write if FUNCTION fin.entry add.period write newline INTEGERS nameptr namesleft numnames FUNCTION format.language language e...

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Carlitz, Some theorems on irreducible reciprocal polynomials over a finite field, J

L. Carlitz, Some theorems on irreducible reciprocal polynomials over a finite field, J. Reine Angew. Math. 227 (1967), 212--220

work page 1967

[2] [2]

Deligne and Y

P. Deligne and Y. Z. Flicker, Counting local systems with principal unipotent local monodromy, Ann. of Math. (2) 178 (2013), no. 3, 921--982

work page 2013

[3] [3]

B. H. Gross and M. Reeder, Arithmetic invariants of discrete L anglands parameters , Duke Math. J. 154 (2010), no. 3, 431--508

2010

[4] [4]

B. H. Gross, On the motive of a reductive group, Invent. Math. 130 (1997), no. 2, 287--313

work page 1997

[5] [5]

, Irreducible cuspidal representations with prescribed local behavior, Amer. J. Math. 133 (2011), no. 5, 1231--1258

work page 2011

[6] [6]

R. E. Kottwitz, Tamagawa numbers, Ann. of Math. (2) 127 (1988), no. 3, 629--646

work page 1988

[7] [7]

Lafforgue, Chtoucas de D rinfeld et correspondance de L anglands , Invent

L. Lafforgue, Chtoucas de D rinfeld et correspondance de L anglands , Invent. Math. 147 (2002), no. 1, 1--241

work page 2002

[8] [8]

Lan, Arithmetic compactifications of PEL -type S himura varieties , Princeton University Press, 2013

K.-W. Lan, Arithmetic compactifications of PEL -type S himura varieties , Princeton University Press, 2013

work page 2013