betaFeedback

WorksMathematics

A short proof of Deligne's bound for Frobenius eigenvalues, after Alpöge

Abstract

We write out in full the proof, sketched by L. Alpöge in a footnote of his note of 20 November 2026, of Deligne's bound for the eigenvalues of Frobenius. Let XX be a separated scheme of finite type over a finite field with qq elements, and let F\mathcal F be a constructible ℓ\ell-adic sheaf on XX whose Frobenius eigenvalues at each closed point xx have absolute value at most ∣κ(x)∣w/2|\kappa(x)|^{w/2}. Then the eigenvalues of Frobenius on Hci(X,F)H^i_c(X,\mathcal F) have absolute value at most q(w+i)/2q^{(w+i)/2}, and they have absolute value exactly q(w+i)/2q^{(w+i)/2} if XX is smooth and proper and F\mathcal F is lisse and pure of weight ww. The proof is short. Granted the formalism of étale cohomology (the trace formula, the Künneth formula, Poincaré duality and Artin's vanishing theorem) and a weak Lefschetz theorem for generic hyperplane sections of affine varieties, it takes less than six pages, and its only estimate is the Cauchy–Schwarz inequality for a character sum over a finite field. Lefschetz pencils, monodromy and Rankin's method are not used. The proof was produced by GPT-6 Astra (ChatGPT Pro mode) from that sketch. It was checked by Claude Opus 5.5 at effort high and, independently, by Claude Fable 5.2 at effort max, which also wrote this text. No human has checked it.

The theorem

Let Fq\mathbb{F}_q be a finite field, ℓ\ell a prime not dividing qq, and ι:Q‾ℓ↪C\iota:\overline{\mathbb{Q}}_\ell\hookrightarrow\mathbb{C} an embedding of fields, fixed throughout. The absolute value of an element of Q‾ℓ\overline{\mathbb{Q}}_\ell always means the absolute value of its image under ι\iota. Let XX be a separated scheme of finite type over Fq\mathbb{F}_q and F\mathcal{F} a constructible Q‾ℓ\overline{\mathbb{Q}}_\ell-sheaf on XX. The groups Hci(X,F)H^i_c(X,\mathcal{F}) are those of X⊗F‾qX\otimes\overline{\mathbb{F}}_q; they are finite-dimensional, and the geometric Frobenius FF acts on them. At a closed point xx of XX, with residue field κ(x)\kappa(x), the geometric Frobenius of κ(x)\kappa(x) acts on the stalk Fx‾\mathcal{F}_{\overline{x}}. For a real number ww, we say that F\mathcal{F} has weights at most ww if at every closed point xx all eigenvalues of this local Frobenius have absolute value at most ∣κ(x)∣w/2|\kappa(x)|^{w/2}, and that F\mathcal{F} is pure of weight ww if they all have absolute value exactly ∣κ(x)∣w/2|\kappa(x)|^{w/2}.

Theorem 1. Let XX be a separated scheme of finite type over Fq\mathbb{F}_q and F\mathcal{F} a constructible Q‾ℓ\overline{\mathbb{Q}}_\ell-sheaf on XX.

(a) If F\mathcal{F} has weights at most ww, then for every ii every eigenvalue of FF on Hci(X,F)H^i_c(X,\mathcal{F}) has absolute value at most q(w+i)/2q^{(w+i)/2}.

(b) If XX is smooth and proper and F\mathcal{F} is lisse and pure of weight ww, then for every ii every eigenvalue of FF on Hi(X,F)H^i(X,\mathcal{F}) has absolute value q(w+i)/2q^{(w+i)/2}.

Part (a) says that the passage from the stalks of F\mathcal{F} to its cohomology in degree ii raises the weights by at most ii. Part (b) says that on a smooth proper scheme nothing is lost: the cohomology in degree ii of a pure sheaf is pure, of weight w+iw+i. For the constant sheaf Q‾ℓ\overline{\mathbb{Q}}_\ell on a smooth projective variety XX the hypothesis of (b) holds with w=0w=0 for every embedding ι\iota, so each eigenvalue of FF on Hi(X,Q‾ℓ)H^i(X,\overline{\mathbb{Q}}_\ell) has absolute value qi/2q^{i/2} under every embedding. This is the Riemann hypothesis for XX.

Both statements are theorems of Deligne. Part (b) for the constant sheaf is [3 Théorème 1.6], apart from the assertions made there about integrality and independence of ℓ\ell, which we do not treat. Part (a) is the statement about absolute values contained in [4 Théorème 3.3.1] for a scheme over a finite field. That theorem concerns a morphism f:X→Sf:X\to S and a mixed sheaf, and says that Rif!FR^if_!\mathcal{F} is mixed of weights at most w+iw+i. For absolute values the relative statement follows from (a), applied to the fibres of ff, because the stalks of Rif!FR^if_!\mathcal{F} are the groups HciH^i_c of the fibres. Part (a) asks only for a bound at each closed point, for the one embedding ι\iota, and does not assume that F\mathcal{F} is mixed. We do not claim this as a gain in generality. By results of L. Lafforgue, an irreducible lisse sheaf on a smooth curve over Fq\mathbb{F}_q whose determinant has finite order is pure of weight zero for every embedding [8 Théorème VII.6 (ii)], and on a normal scheme of finite type over Fq\mathbb{F}_q every lisse sheaf is, in a unique way, a direct sum ⨁χσχ⊗χ\bigoplus_\chi\sigma_\chi\otimes\chi of twists by characters χ\chi [8 Corollaire VII.8]. Nothing is proved here about the other parts of Deligne’s theory: the weight filtration, the local monodromy theorem and the hard Lefschetz theorem of [4], the rationality and the functional equation of zeta functions, or the semisimplicity of Frobenius.

The proof, its source and its length

The source. The note [1] of L. Alpöge mentions a proof of the theorems of Weil and Deligne that uses only a sign, based on an identity qN=Mk+RqN=M^k+R, and sketches it in a footnote [1 p. 2 and footnote 3]. In the sketch NN is a weighted count of the zeros of the form x1x2+⋯+xk−1xkx_1x_2+\cdots+x_{k-1}x_k, written as a sum over additive characters, and MkM^k is the term of the trivial character. An argument with a hyperplane section gives the contribution of middle cohomology a definite sign, one degree away from that of MkM^k. The remainder RR has square-root size, by Parseval’s identity applied kk times. Because of the sign, the inverse roots cannot cancel one another, so the bound for RR is a bound for each of them. In higher dimension vectors replace the scalars xix_i.

The present note writes this argument out in full. Its ideas are those of the sketch. What is supplied here is the detail: the reduction to the lowest cohomology group of a smooth affine variety, a finite map to affine space that turns the count into the character sum, the weak Lefschetz theorem that provides the sign, and the passage to all degrees.

The length. The proof is short, and we say in what sense. We take for granted the formalism of ℓ\ell-adic cohomology as it stood before [3]: the trace formula, the Künneth formula, Poincaré duality, the vanishing of the cohomology of an affine scheme above its dimension, and the exact sequence attached to an open subscheme and its complement. We also take for granted one theorem of a less elementary kind: for a generic hyperplane section YY of a smooth affine variety ZZ, and a lisse sheaf on ZZ, restriction from ZZ to YY is bijective on cohomology below the dimension of YY and injective in that degree. The proof of it that we use is Deligne’s, printed in the appendix of [6]. It is a few pages long and rests on his generic base change theorem and on Artin’s vanishing theorem; Section 6 recalls it. With these inputs, the proof of Theorem 1 occupies Sections 3 to 8, less than six pages, and its only estimate is the Cauchy–Schwarz inequality, applied to a character sum over a finite field (Lemma 3).

For comparison, [3] uses the same formalism together with the theory of Lefschetz pencils, a theorem of Kazhdan and Margulis on their monodromy, and Rankin’s method. In [4] the general statement is reduced to the case of curves, which is proved by a study of monodromy and by the method of Hadamard and de la Vallée-Poussin. None of these enters below. Laumon [9] gave another proof of the main theorem of [4], by the ℓ\ell-adic Fourier transform, and Kedlaya [7] carried that proof over to pp-adic cohomology.

For curves, the estimate for the number of points has an elementary proof, by the method of Stepanov [2]. Scholl [10] reduced the Riemann hypothesis for smooth projective varieties to the case of hypersurfaces, and remarked that an elementary proof of the point-count estimate for hypersurfaces, in the manner of Stepanov, would be of great interest. The proof below is not of that kind. Its estimate is elementary, but it concerns one explicit hypersurface in a power of the variety, and the passage from that estimate to the eigenvalues uses cohomology throughout.

Why the proof works. By cutting XX into pieces and by induction on the dimension, part (a) comes down to one case (§ 8): UU is smooth and affine of pure dimension dd, the sheaf F\mathcal{F} is lisse on UU, and the group is Hcd(U,F)H_c^d(U,\mathcal{F}), the lowest one that can be nonzero. In higher degrees the weak Lefschetz theorem expresses the cohomology of UU through that of a hyperplane section, of smaller dimension. In degree dd it gives nothing. The trace formula alone does not help either: it gives the sum of the local traces of F\mathcal{F} over the points of UU as an alternating sum over all degrees, in which large eigenvalues from different degrees could cancel.

The way out is to compare two counts. Let k≥2k \ge2 and let YY be a hypersurface in the kk-fold product UkU^k. Over a finite field with QQ elements, let MM be the sum of the local traces of F\mathcal{F} over the points of UU, and NN the sum of the local traces of F⊠⋯⊠F\mathcal{F} \boxtimes\cdots\boxtimes\mathcal{F} over the points of YY. A hypersurface in general position should carry one QQ-th of the total, so we put R=QN−MkR = QN - M^k and call it the remainder. Suppose that YY has the weak Lefschetz property: restriction from UkU^k to YY is bijective in degrees below kd−1kd-1 and injective in degree kd−1kd-1. By duality, every compactly supported cohomology group of UkU^k above the lowest degree kdkd is then the image, bijectively or surjectively, of a group of YY. Written with the trace formula, RR is an alternating sum over the groups of UkU^k and of YY, and along these maps the terms cancel in pairs. Two terms survive. One is the lowest group of UkU^k, which by the Künneth formula is the kk-th tensor power of Hcd(U,F)H_c^d(U,\mathcal{F}). The other is the kernel KK of the map out of the lowest group of YY. The first sits in degree kdkd and is subtracted; the second sits in degree kd−1kd-1 and is added; so they enter RR with the same sign (Proposition 4). Up to that one sign, RR is therefore the trace of Frobenius on an actual vector space, the sum of the tensor power and of KK, and not a difference of two traces. This matters because of an elementary fact about power sums (Lemma 2): if the traces of the powers of an automorphism grow at most like BNB^N, then every one of its eigenvalues has absolute value at most BB. For a difference of two traces this is false, since eigenvalues can cancel.

It remains to find a hypersurface YY with the weak Lefschetz property for which RR is small, and the form of its equation does this. Choose a finite map π\pi from UU to the affine space Ad\mathbf{A}^d and a nondegenerate bilinear form bb on Ad\mathbf{A}^d, and let the equation of YY be

∑i=1k−1b(π(ui),π(ui+1))+∑i=1kgi(ui)+c=0,\sum_{i=1}^{k-1} b\bigl(\pi(u_i),\pi(u_{i+1})\bigr) + \sum_{i=1}^{k} g_i(u_i) + c = 0,

where each gig_i is a linear combination of coordinates of UU. Consecutive factors are coupled through bb, and there is no other coupling. Detecting the equation by additive characters writes QNQN as a sum over the characters ψ(h⋅)\psi(h\cdot) of the field. The term with h=0h=0 is exactly MkM^k, so RR is the sum of the other terms. For h≠0h \ne0 the term is a product of k−1k-1 matrices with entries ψ(hb(x,y))\psi(hb(x,y)), interlaced with diagonal matrices that carry the local traces. The first kind are Qd/2Q^{d/2} times unitary matrices, which is Parseval’s identity for the Fourier transform on a finite vector space; the second kind are bounded by a constant times Qw/2Q^{w/2}. Hence ∣R∣|R| is at most a constant to the power kk, times Qk(w+d)/2Q^{k(w+d)/2}, times Qd/2+1Q^{d/2+1} (Lemma 3 and § 7). This is square-root size, up to the factor Qd/2+1Q^{d/2+1}, which does not depend on kk. That such a YY can be chosen with the weak Lefschetz property is not obvious, because the bilinear part of its equation is fixed in advance. But YY is a hyperplane section of UkU^k for the embedding given by the coordinates together with the fixed bilinear expression, and a generic hyperplane whose coefficient on that expression is 1 will do (Lemma 8).

Putting the two halves together, every eigenvalue of Frobenius on the kk-th tensor power of Hcd(U,F)H_c^d(U,\mathcal{F}), in particular αk\alpha^k for an eigenvalue α\alpha on Hcd(U,F)H_c^d(U,\mathcal{F}), has absolute value at most qk(w+d)/2+d/2+1q^{k(w+d)/2+d/2+1}. Taking kk-th roots gives ∣α∣≤q(w+d)/2⋅q(d/2+1)/k|\alpha| \le q^{(w+d)/2}\cdot q^{(d/2+1)/k}, and letting kk grow removes the loss.

Power sums of an actual module

Let VV be a finite-dimensional vector space over Q‾ℓ\overline{\mathbb{Q}}_\ell with an automorphism Φ\Phi. Then Tr⁡(Φn∣V)=∑jmjλjn\operatorname{Tr}(\Phi^n \mid V) = \sum_j m_j\lambda_j^n, where the λj\lambda_j are the distinct eigenvalues of Φ\Phi and the mj≥1m_j \ge1 their algebraic multiplicities. The next lemma is the form in which we use the positivity of the mjm_j. It allows the automorphism, the auxiliary space and the constants to change with kk, which will be needed.

Lemma 2. Let VV be a finite-dimensional Q‾ℓ\overline{\mathbb{Q}}_\ell-vector space with an automorphism FF, and let aa be a real number. Suppose that for infinitely many integers k≥1k \ge1 there are an integer ek≥1e_k \ge1, a finite-dimensional Q‾ℓ\overline{\mathbb{Q}}_\ell-vector space KkK_k with an automorphism Φk\Phi_k, a sign εk=±1\varepsilon_k = \pm1, real numbers Ck>0C_k > 0 and ck≥0c_k \ge0, and complex numbers Rk,nR_{k,n} such that for all n≥1n \ge1

εkRk,n=ιTr⁡(Fekn∣V⊗k)+ιTr⁡(Φkn∣Kk)and∣Rk,n∣≤Ckqekn(ka+ck),\varepsilon_k R_{k,n} = \iota\operatorname{Tr}(F^{e_k n} \mid V^{\otimes k}) + \iota\operatorname{Tr}(\Phi_k^n \mid K_k) \quad\text{and}\quad|R_{k,n}| \le C_k q^{e_k n(ka+c_k)},

and that ck/kc_k/k tends to 0 along these kk. Then every eigenvalue α\alpha of FF on VV satisfies ∣ια∣≤qa|\iota\alpha| \le q^a.

Proof. Fix such a kk. Let μ1,…,μs\mu_1,\ldots,\mu_s be the images under ι\iota of the distinct eigenvalues of the automorphism Fek⊕ΦkF^{e_k} \oplus\Phi_k of V⊗k⊕KkV^{\otimes k} \oplus K_k, and m1,…,ms≥1m_1,\ldots,m_s \ge1 their multiplicities, so that εkRk,n=∑jmjμjn\varepsilon_k R_{k,n} = \sum_j m_j\mu_j^n for all n≥1n \ge1. The μj\mu_j are distinct, because ι\iota is injective, and nonzero. If v≠0v \ne0 and Fv=αvFv = \alpha v, then v⊗⋯⊗vv \otimes\cdots\otimes v is an eigenvector of FekF^{e_k} on V⊗kV^{\otimes k} with eigenvalue αkek\alpha^{ke_k}, so ι(α)kek\iota(\alpha)^{ke_k} is one of the μj\mu_j.

Put ρ=q−ek(ka+ck)\rho= q^{-e_k(ka+c_k)}. By the bound on Rk,nR_{k,n}, the power series ∑n≥1εkRk,nzn\sum_{n\ge1}\varepsilon_k R_{k,n}z^n converges in the disc ∣z∣<ρ|z| < \rho. Near z=0z = 0 it is equal to the rational function ∑jmjμjz/(1−μjz)\sum_j m_j\mu_j z/(1-\mu_j z). Since the μj\mu_j are distinct and no mjm_j is zero, this rational function has a pole at each of the points 1/μj1/\mu_j. It agrees on the disc with a holomorphic function, so none of these points lies in the disc. Hence ∣μj∣≤1/ρ|\mu_j| \le1/\rho for every jj, and in particular ∣ια∣kek≤qek(ka+ck)|\iota\alpha|^{ke_k} \le q^{e_k(ka+c_k)}, that is, ∣ια∣≤qa+ck/k|\iota\alpha| \le q^{a+c_k/k}. Now let kk tend to infinity.

A character sum along a path

The estimate of this note is the following. The sum runs over kk points of a finite vector space; each point is weighted by an arbitrary bounded function, and the phase couples only consecutive points.

Lemma 3. Let EE be a finite field with QQ elements, ψ\psi a nontrivial additive character of EE, d≥1d \ge1, A=EdA = E^d and k≥2k \ge2. Let b1,…,bk−1b_1,\ldots,b_{k-1} be nondegenerate bilinear forms on AA with values in EE, and let t1,…,tk:A→Ct_1,\ldots,t_k : A \to\mathbb{C} be functions with ∣ti(x)∣≤T|t_i(x)| \le T for all ii and xx. Then for every h∈E×h \in E^\times and c∈Ec \in E

∣∑x∈Akt1(x1)⋯tk(xk)ψ(h(∑i=1k−1bi(xi,xi+1)+c))∣≤TkQd(k+1)/2.\left|\sum_{x\in A^k} t_1(x_1)\cdots t_k(x_k)\psi\left(h\left(\sum_{i=1}^{k-1} b_i(x_i,x_{i+1})+c\right)\right)\right| \le T^k Q^{d(k+1)/2}.

Proof. For i<ki<k let KiK_i be the matrix with rows and columns indexed by AA and entries Ki(x,y)=ψ(hbi(x,y))K_i(x,y)=\psi(hb_i(x,y)). The entry of KiKi∗K_iK_i^* at (x,x′)(x,x') is ∑y∈Aψ(hbi(x−x′,y))\sum_{y\in A}\psi(hb_i(x-x',y)). If x≠x′x\ne x', the map y↦hbi(x−x′,y)y\mapsto hb_i(x-x',y) is a nonzero linear form on AA, because bib_i is nondegenerate and h≠0h\ne0, and the sum of a nontrivial character over AA is zero. If x=x′x=x' the sum is QdQ^d. So KiKi∗=QdK_iK_i^*=Q^d, and KiK_i is Qd/2Q^{d/2} times a unitary matrix. Let DiD_i be the diagonal matrix with entries ti(x)t_i(x); its norm is at most TT. Regard t1t_1 and tkt_k as vectors; their norms are at most TQd/2TQ^{d/2}. The sum in the lemma equals ψ(hc)t1K1D2K2⋯Dk−1Kk−1tk\psi(hc)t_1K_1D_2K_2\cdots D_{k-1}K_{k-1}t_k, and by the Cauchy–Schwarz inequality its absolute value is at most

∥t1∥∥tk∥∏i=1k−1∥Ki∥∏i=2k−1∥Di∥≤(TQd/2)2Qd(k−1)/2Tk−2=TkQd(k+1)/2.\|t_1\|\|t_k\|\prod_{i=1}^{k-1}\|K_i\|\prod_{i=2}^{k-1}\|D_i\|\le(TQ^{d/2})^2Q^{d(k-1)/2}T^{k-2}=T^kQ^{d(k+1)/2}.

The trivial bound for the sum is TkQdkT^kQ^{dk}. The lemma gains a factor Qd/2Q^{d/2} for each of the k−1k-1 couplings.

The remainder of a point count

We recall what duality gives. Let ZZ be a smooth scheme of pure dimension mm over a finite field Fr\mathbb{F}_r and G\mathcal{G} a lisse Qℓ\mathbb{Q}_\ell-sheaf on ZZ. Poincaré duality [11 XVIII, §3] is a perfect pairing, compatible with Frobenius, between Hci(Z,G)H_c^i(Z,\mathcal{G}) and H2m−i(Z,G∨)(m)H^{2m-i}(Z,\mathcal{G}^{\vee})(m). Here (m)(m) is the Tate twist; on a space twisted by (−1)(-1) the eigenvalues of Frobenius are multiplied by rr. If ZZ is affine, then Hi(Z,G∨)=0H^i(Z,\mathcal{G}^{\vee})=0 for i>mi>m by Artin’s theorem [11 XIV, 3.1], and therefore Hci(Z,G)=0H_c^i(Z,\mathcal{G})=0 for i<mi<m. If Y⊂ZY\subset Z is a smooth closed subscheme of pure dimension m−1m-1, we use duality on ZZ and on YY: the transpose of the restriction map Ha(Z,G∨)→Ha(Y,G∨)H^a(Z,\mathcal{G}^{\vee})\to H^a(Y,\mathcal{G}^{\vee}) is a map

Hcj(Y,G)(−1)⟶Hcj+2(Z,G),j=2m−2−a,H_c^j(Y,\mathcal{G})(-1)\longrightarrow H_c^{j+2}(Z,\mathcal{G}),\qquad j=2m-2-a,

compatible with Frobenius. We call it the Gysin map. It is bijective when the restriction map is bijective, and surjective when the restriction map is injective.

Proposition 4. Let UU be a smooth affine scheme of pure dimension d≥1d\ge1 over a finite field Fr\mathbb{F}_r, and F\mathcal{F} a lisse Qℓ\mathbb{Q}_\ell-sheaf on UU. Let k≥1k\ge1, X=UkX=U^k, D=kdD=kd and Fk=F⊠⋯⊠F\mathcal{F}_k=\mathcal{F}\boxtimes\cdots\boxtimes\mathcal{F} on XX. Let Y⊂XY\subset X be a smooth closed subscheme of pure dimension D−1D-1, possibly empty, defined over Fr\mathbb{F}_r, such that over F‾r\overline{\mathbb{F}}_r the restriction map Ha(X,Fk∨)→Ha(Y,Fk∨)H^a(X,\mathcal{F}_k^\vee)\to H^a(Y,\mathcal{F}_k^\vee) is an isomorphism for a<D−1a<D-1 and injective for a=D−1a=D-1. For n≥1n\ge1 let MnM_n be the sum of the traces Tr⁡(Fu,n∣Fu‾)\operatorname{Tr}(F_{u,n}\mid\mathcal{F}_{\overline{u}}) over u∈U(Frn)u\in U(\mathbb{F}_{r^n}), let NnN_n be the corresponding sum for Fk\mathcal{F}_k over Y(Frn)Y(\mathbb{F}_{r^n}), and put Rn=rnNn−MnkR_n=r^nN_n-M_n^k. Let FF be the geometric Frobenius of Fr\mathbb{F}_r, and let KK be the kernel of the Gysin map HcD−1(Y,Fk)(−1)→HcD+1(X,Fk)H_c^{D-1}(Y,\mathcal{F}_k)(-1)\to H_c^{D+1}(X,\mathcal{F}_k). Then for all n≥1n\ge1

(−1)D−1Rn=Tr⁡(Fn∣Hcd(U,F)⊗k)+Tr⁡(Fn∣K).(-1)^{D-1}R_n=\operatorname{Tr}(F^n\mid H_c^d(U,\mathcal{F})^{\otimes k})+\operatorname{Tr}(F^n\mid K).

Proof. The schemes UU, XX and YY are smooth and affine, of pure dimensions dd, DD and D−1D-1. By what was recalled, the groups Hcj(U,F)H_c^j(U,\mathcal{F}) vanish for j<dj<d, the groups Hcj(X,Fk)H_c^j(X,\mathcal{F}_k) for j<Dj<D, and the groups Hcj(Y,Fk)H_c^j(Y,\mathcal{F}_k) for j<D−1j<D-1. The hypothesis on restriction says that the Gysin map Hcj(Y,Fk)(−1)→Hcj+2(X,Fk)H_c^j(Y,\mathcal{F}_k)(-1)\to H_c^{j+2}(X,\mathcal{F}_k) is an isomorphism for j>D−1j>D-1 and surjective for j=D−1j=D-1.

The stalk of Fk\mathcal{F}_k at a point (u1,…,uk)(u_1,\ldots,u_k) is the tensor product of the stalks of F\mathcal{F} at the uiu_i, so the sum of the traces of Fk\mathcal{F}_k over X(Frn)=U(Frn)kX(\mathbb{F}_{r^n})=U(\mathbb{F}_{r^n})^k is MnkM_n^k. By the trace formula [12 Rapport, Théorème 3.2], applied on XX and on YY,

Mnk=∑b≥0(−1)D+bTr⁡(Fn∣HcD+b(X,Fk)),M_n^k=\sum_{b\ge0}(-1)^{D+b}\operatorname{Tr}\left(F^n\mid H_c^{D+b}(X,\mathcal{F}_k)\right),
rnNn=∑a≥0(−1)D−1+aTr⁡(Fn∣HcD−1+a(Y,Fk)(−1)).r^nN_n=\sum_{a\ge0}(-1)^{D-1+a}\operatorname{Tr}\left(F^n\mid H_c^{D-1+a}(Y,\mathcal{F}_k)(-1)\right).

Therefore, in (−1)D−1Rn(-1)^{D-1}R_n, the trace on HcD−1+a(Y,Fk)(−1)H_c^{D-1+a}(Y,\mathcal{F}_k)(-1) enters with the sign (−1)a(-1)^a and the trace on HcD+b(X,Fk)H_c^{D+b}(X,\mathcal{F}_k) with the sign (−1)b(-1)^b. For a≥1a \ge1 the Gysin map identifies HcD−1+a(Y,Fk)(−1)H_c^{D-1+a}(Y,\mathcal{F}_k)(-1) with HcD+1+a(X,Fk)H_c^{D+1+a}(X,\mathcal{F}_k), which enters with the opposite sign, and the two traces cancel. Three terms remain: the trace on HcD−1(Y,Fk)(−1)H_c^{D-1}(Y,\mathcal{F}_k)(-1), minus the trace on HcD+1(X,Fk)H_c^{D+1}(X,\mathcal{F}_k), plus the trace on HcD(X,Fk)H_c^D(X,\mathcal{F}_k). The second space is the quotient of the first by KK, so the first two terms together are the trace on KK. Finally, by the Künneth formula [11 XVII, 5.4.3] the group HcD(X,Fk)H_c^D(X,\mathcal{F}_k) is the sum of the tensor products Hcj1(U,F)⊗⋯⊗Hcjk(U,F)H_c^{j_1}(U,\mathcal{F}) \otimes\cdots\otimes H_c^{j_k}(U,\mathcal{F}) with j1+⋯+jk=Dj_1+\cdots+j_k=D. Since the factors vanish for ji<dj_i<d, only j1=⋯=jk=dj_1=\cdots=j_k=d contributes, and HcD(X,Fk)=Hcd(U,F)⊗kH_c^D(X,\mathcal{F}_k)=H_c^d(U,\mathcal{F})^{\otimes k}.

Nothing is assumed about the higher cohomology groups of UU or of YY: they cancel through actual isomorphisms, and the surjection at the border leaves an actual kernel.

Example 5. Let U=A1U=\mathbb{A}^1, F=Q‾ℓ\mathcal{F}=\overline{\mathbb{Q}}_\ell and k=2k=2, and let Y⊂A2Y\subset\mathbb{A}^2 be the conic xy+ax+by+c=0xy+ax+by+c=0 with ab≠cab\ne c. It is isomorphic to the multiplicative group, so Nn=rn−1N_n=r^n-1, while Mn=rnM_n=r^n; hence Rn=−rnR_n=-r^n and (−1)D−1Rn=rn(-1)^{D-1}R_n=r^n. On the other side Hc1(A1,Q‾ℓ)=0H_c^1(\mathbb{A}^1,\overline{\mathbb{Q}}_\ell)=0, and K=Hc1(Y,Q‾ℓ)(−1)K=H_c^1(Y,\overline{\mathbb{Q}}_\ell)(-1) has dimension one, with Frobenius acting by rr. The group Hc2(Y,Q‾ℓ)(−1)H_c^2(Y,\overline{\mathbb{Q}}_\ell)(-1) is identified with Hc4(A2,Q‾ℓ)H_c^4(\mathbb{A}^2,\overline{\mathbb{Q}}_\ell) and has cancelled.

Remark 6. The space KK is the analogue, for compact supports on an affine variety, of the part of the cohomology of a hyperplane section that the Gysin map kills. In Grothendieck’s notes on the standard conjectures in the Montpellier archive this part is his partie évanescente En−1E^{n-1} (cote 16, scan 40, in the transcription [14]). The comparison plays no role in the proof.

Hypersurfaces with the weak Lefschetz property

Theorem 7 (Deligne; [6 Appendix, A.5, with A.3 and A.4]). Let ZZ be a separated scheme of finite type over an algebraically closed field of characteristic different from ℓ\ell, let φ:Z→PN\varphi:Z\to\mathbb{P}^N be a quasi-finite morphism, and let KK be a perverse sheaf on ZZ. Then there is a dense open set of hyperplanes H⊂PNH\subset\mathbb{P}^N for which the restriction map

Ha(Z,K)⟶Ha(φ−1H,K)H^a(Z,K)\longrightarrow H^a(\varphi^{-1}H,K)

is an isomorphism for a<−1a<-1 and injective for a=−1a=-1.

Katz attributes this proof to Deligne [6 p. 152]. It runs as follows. Let α:Z∖φ−1H→Z\alpha:Z\setminus\varphi^{-1}H\to Z be the inclusion. The assertion is equivalent to the vanishing of Ha(Z,α!α∗K)H^a(Z,\alpha_!\alpha^*K) for a<0a<0. For generic HH, the generic base change theorem [12 Théorèmes de finitude, 1.9], applied over the space of hyperplanes, identifies these groups with the compactly supported cohomology of the affine scheme PN∖H\mathbb{P}^N\setminus H with coefficients in a direct image of α∗K\alpha^*K; duality and Artin’s theorem then give the vanishing. No weights occur. If ZZ is smooth of pure dimension DD and G\mathcal{G} is a lisse sheaf on ZZ, then G[D]\mathcal{G}[D] is perverse, and the conclusion reads: the restriction map Ha(Z,G)→Ha(φ−1H,G)H^a(Z,\mathcal{G})\to H^a(\varphi^{-1}H,\mathcal{G}) is an isomorphism for a<D−1a<D-1 and injective for a=D−1a=D-1.

We need this for hypersurfaces whose equation has a fixed part. Corollary 3.4.1 of [6] states exactly that form, but it is deduced there from the main theorem of that article, whose proof is by the Fourier transform. The next lemma deduces it from Theorem 7 instead, so that only the inputs named above are used; it also gives the smoothness of the hypersurface.

Lemma 8. Let VV be a smooth affine scheme of pure dimension D≥1D\ge1 over F‾q\overline{\mathbb{F}}_q, let y1,…,yMy_1,\ldots,y_M be functions on VV such that Γ=(y1,…,yM):V→AM\Gamma=(y_1,\ldots,y_M):V\to\mathbb{A}^M is a closed immersion, and let G\mathcal{G} be a lisse Q‾ℓ\overline{\mathbb{Q}}_\ell-sheaf on VV. Consider the hypersurfaces

Yλ,μ=V∩(∑m=1Mλmym+μ=0),λ≠0.Y_{\lambda,\mu}=V\cap\left(\sum_{m=1}^{M}\lambda_m y_m+\mu=0\right),\qquad\lambda\ne0.

For (λ,μ)(\lambda,\mu) in a dense open subset of the space of all parameters, and also for (λ,μ)(\lambda,\mu) in a dense open subset of the space of parameters with λM=1\lambda_M=1, the scheme Yλ,μY_{\lambda,\mu} is smooth, of pure dimension D−1D-1 or empty on each connected component of VV, and the restriction map Ha(V,G)→Ha(Yλ,μ,G)H^a(V,\mathcal{G}) \to H^a(Y_{\lambda,\mu},\mathcal{G}) is an isomorphism for a<D−1a<D-1 and injective for a=D−1a=D-1.

Proof. Regard AM\mathbb{A}^M as the complement of the hyperplane at infinity in PM\mathbb{P}^M. The hyperplanes of PM\mathbb{P}^M other than the one at infinity are the zero sets of the equations ∑mλmym+μ=0\sum_m \lambda_m y_m+\mu=0 with λ≠0\lambda\ne0, and the inverse image of such a hyperplane under Γ\Gamma is Yλ,μY_{\lambda,\mu}.

The restriction property holds for a dense open set of hyperplanes by Theorem 7, applied to φ=Γ\varphi=\Gamma, which is quasi-finite, and to the perverse sheaf G[D]\mathcal{G}[D].

Smoothness holds for a dense open set of hyperplanes by a count of dimensions. Consider the pairs (v,H)(v,H) of a point vv of VV and a hyperplane HH that contains the tangent space of Γ(V)\Gamma(V) at Γ(v)\Gamma(v). Since Γ\Gamma is an immersion, this tangent space has dimension DD, and the hyperplanes containing it form a projective space of dimension M−D−1M-D-1. So the set of such pairs has dimension at most D+(M−D−1)=M−1D+(M-D-1)=M-1, and its image in the space of hyperplanes, which has dimension MM, is not dense. If HH lies outside the closure of this image, the equation of HH restricts to a nonzero linear form on the tangent space of VV at each point of Γ−1H\Gamma^{-1}H, so Γ−1H\Gamma^{-1}H is smooth, and of pure dimension D−1D-1 or empty on each component of VV. This argument is valid in every characteristic, because Γ\Gamma is an immersion; compare [5 I.6.10].

The hyperplanes with both properties form a dense open subset of the space of all hyperplanes. It meets the open subset λM≠0\lambda_M\ne0 in a dense open subset, and on that subset we may scale the equation so that λM=1\lambda_M=1.

The lowest degree of a smooth affine variety

Theorem 9. Let UU be a smooth affine scheme of pure dimension d≥1d\ge1 over Fq\mathbb{F}_q and FF a lisse Q‾ℓ\overline{\mathbb{Q}}_\ell-sheaf on UU. Suppose that there are a constant C>0C>0 and a real number ww with

∣ιTr⁡(Fu,n∣Fu‾)∣≤Cqnw/2for all n≥1 and u∈U(Fqn).|\iota\operatorname{Tr}(F_{u,n}\mid F_{\overline{u}})|\le Cq^{nw/2}\qquad\text{for all }n\ge1\text{ and }u\in U(\mathbb{F}_{q^n}).

Then every eigenvalue of FF on Hcd(U,F)H^d_c(U,F) has absolute value at most q(w+d)/2q^{(w+d)/2}.

Proof. Choices. By Noether normalization [13 Tag 00OY] there is a finite morphism π:U→Ad\pi:U\to\mathbb{A}^d over Fq\mathbb{F}_q. Let BB be the number of elements in a set of generators of the coordinate ring O(U)\mathcal{O}(U) as a module over that of Ad\mathbb{A}^d. Over any field, each fibre of π\pi is the spectrum of an algebra of dimension at most BB, so it has at most BB rational points. Choose generators z1,…,zsz_1,\ldots,z_s of O(U)\mathcal{O}(U) as an Fq\mathbb{F}_q-algebra, so that z=(z1,…,zs):U→Asz=(z_1,\ldots,z_s):U\to\mathbb{A}^s is a closed immersion, and a nondegenerate bilinear form bb on Ad\mathbb{A}^d defined over Fq\mathbb{F}_q. Fix k≥2k\ge2 and put D=kdD=kd.

The hypersurface. For parameters a=(aij)a=(a_{ij}), with 1≤i≤k1\le i\le k and 1≤j≤s1\le j\le s, and cc, let Pa,cP_{a,c} be the function on UkU^k

Pa,c(u1,…,uk)=∑i=1k−1b(π(ui),π(ui+1))+∑i=1k∑j=1saijzj(ui)+c,P_{a,c}(u_1,\ldots,u_k)=\sum_{i=1}^{k-1}b\bigl(\pi(u_i),\pi(u_{i+1})\bigr)+\sum_{i=1}^{k}\sum_{j=1}^{s}a_{ij}z_j(u_i)+c,

and let Ya,c⊂UkY_{a,c}\subset U^k be its zero set. Over F‾q\overline{\mathbb{F}}_q the map

Γ:u⟼(z(u1),…,z(uk),∑i=1k−1b(π(ui),π(ui+1)))\Gamma:u\longmapsto\left(z(u_1),\ldots,z(u_k),\sum_{i=1}^{k-1}b\bigl(\pi(u_i),\pi(u_{i+1})\bigr)\right)

is a closed immersion of UkU^k into Ask+1\mathbb{A}^{sk+1}, because its first sksk coordinates already are one. The Ya,cY_{a,c} are exactly the hypersurfaces of Lemma 8 for Γ\Gamma whose coefficient on the last coordinate is 11. By that lemma, applied to the sheaf Fk\mathcal{F}_k, there is a dense open set of parameters (a,c)(a,c) for which Ya,cY_{a,c} is smooth of pure dimension D−1D-1, or empty, and has the restriction property required in Proposition 4. A dense open subset of an affine space over F‾q\overline{\mathbb{F}}_q has a point with coordinates in a finite field. Fix such a point (a,c)(a,c), with coordinates in Fr\mathbb{F}_r, r=qer=q^e, and write P=Pa,cP=P_{a,c} and Y=Ya,cY=Y_{a,c}. The exponent ee and the point (a,c)(a,c) depend on kk; this is allowed in Lemma 2.

The remainder as a character sum. Let n≥1n\geq1, Q=rnQ=r^n and E=FQE=\mathbb{F}_Q, and let MnM_n, NnN_n and Rn=QNn−MnkR_n=QN_n-M_n^k be formed over Fr\mathbb{F}_r as in Proposition 4. Let ψ\psi be a nontrivial additive character of EE, and for u∈U(E)u\in U(E) put τ(u)=ιTr⁡(Fu,en∣Fuˉ)\tau(u)=\iota\operatorname{Tr}(F_{u,en}\mid\mathcal{F}_{\bar{u}}); by hypothesis ∣τ(u)∣≤CQw/2|\tau(u)|\leq CQ^{w/2}. For a point u=(u1,…,uk)u=(u_1,\ldots,u_k) of UkU^k with values in EE, the sum ∑h∈Eψ(hP(u))\sum_{h\in E}\psi(hP(u)) is QQ if P(u)=0P(u)=0 and 00 otherwise. Therefore

Q⋅ιNn=∑h∈E∑u∈U(E)kτ(u1)⋯τ(uk)ψ(hP(u)).Q\cdot\iota N_n=\sum_{h\in E}\sum_{u\in U(E)^k}\tau(u_1)\cdots\tau(u_k)\psi(hP(u)).

In the inner sum we collect the uiu_i according to their images xi=π(ui)x_i=\pi(u_i) in EdE^d. With

ti,h(x)=∑u∈U(E), π(u)=xτ(u)ψ(h∑j=1saijzj(u))t_{i,h}(x)=\sum_{\substack{u\in U(E),\ \pi(u)=x}}\tau(u)\psi\left(h\sum_{j=1}^{s}a_{ij}z_j(u)\right)

the inner sum becomes

ψ(hc)∑x∈(Ed)kt1,h(x1)⋯tk,h(xk)ψ(h∑i=1k−1b(xi,xi+1)).\psi(hc)\sum_{x\in(E^d)^k}t_{1,h}(x_1)\cdots t_{k,h}(x_k)\psi\left(h\sum_{i=1}^{k-1}b(x_i,x_{i+1})\right).

For h=0h=0 all phases are 1, and the inner sum is (∑u∈U(E)τ(u))k=(ιMn)k\left(\sum_{u\in U(E)}\tau(u)\right)^k=(\iota M_n)^k. For h≠0h\neq0, each ti,h(x)t_{i,h}(x) is a sum of at most BB terms of absolute value at most CQw/2CQ^{w/2}, so Lemma 3 applies with T=BCQw/2T=BCQ^{w/2} and all bib_i equal to bb. There are Q−1Q-1 such hh, and we obtain

∣ιRn∣=∣Q⋅ιNn−(ιMn)k∣≤(Q−1)(BC)kQkw/2Qd(k+1)/2≤(BC)kqen{k(w+d)/2+d/2+1}.|\iota R_n|=\left|Q\cdot\iota N_n-(\iota M_n)^k\right|\leq(Q-1)(BC)^kQ^{kw/2}Q^{d(k+1)/2}\leq(BC)^kq^{en\{k(w+d)/2+d/2+1\}}.

Conclusion. The geometric Frobenius of Fr\mathbb{F}_r is FeF^e. By Proposition 4, applied over Fr\mathbb{F}_r,

(−1)D−1ιRn=ιTr⁡((Fe)n∣Hcd(U,F)⊗k)+ιTr⁡((Fe)n∣K)(-1)^{D-1}\iota R_n=\iota\operatorname{Tr}\left((F^e)^n\mid H_c^d(U,\mathcal{F})^{\otimes k}\right)+\iota\operatorname{Tr}\left((F^e)^n\mid K\right)

for all $n\geq 1. These are the hypotheses of Lemma 2 for V=Hcd(U,F)V=H_c^d(U,\mathcal{F}) and a=(w+d)/2a=(w+d)/2, with ek=ee_k=e, Kk=KK_k=K, ϵk=(−1)D−1\epsilon_k=(-1)^{D-1}, Ck=(BC)kC_k=(BC)^k and ck=d/2+1c_k=d/2+1, for every k≥2k\geq2. The lemma gives the theorem.

Proof of Theorem 1

A bound for the eigenvalues may be proved after replacing Fq\mathbb{F}_q by a finite extension Fqe\mathbb{F}_{q^e}: the eigenvalues of FeF^e are the ee-th powers of those of FF, so a bound qe(w+i)/2q^{e(w+i)/2} for the former gives the bound q(w+i)/2q^{(w+i)/2} for the latter.

Part (a), for XX smooth and affine of pure dimension dd and F\mathcal{F} lisse. We prove the statement by induction on dd, for all finite fields at once. Let d=0d=0. Then Hc0(X,F)H_c^0(X,\mathcal{F}) is the sum of the stalks of F\mathcal{F} at the geometric points of XX. If xx is a closed point of degree mm, the Frobenius FF permutes the mm geometric points above xx cyclically, and FmF^m acts on each of their stalks as the local Frobenius at xx. So for an eigenvalue α\alpha of FF on the part of Hc0H_c^0 above xx, αm\alpha^m is an eigenvalue of the local Frobenius, and ∣α∣m≤qmw/2|\alpha|^m\leq q^{mw/2}.

Let d≥1d\geq1. For i<di<d the group Hci(X,F)H_c^i(X,\mathcal{F}) vanishes (Section 5). For i=di=d we apply Theorem 9, with CC the rank of F\mathcal{F}: if uu is a point with values in Fq\mathbb{F}_q over a closed point xx of degree mm, then Fu,nF_{u,n} is the (n/m)(n/m)-th power of the local Frobenius at xx, so its eigenvalues have absolute value at most qnw/2q^{nw/2}. For i>di>d we cut by a hyperplane. Choose a closed immersion of XX into an affine space. By Lemma 8, applied to this immersion, to all hyperplanes and to the sheaf F∨\mathcal{F}^{\vee}, there is, over some finite extension Fqe\mathbb{F}_{q^e}, a hyperplane section YY of XX that is smooth and affine of pure dimension d−1d-1, or empty, and for which restriction from XX to YY on the cohomology of F∨\mathcal{F}^{\vee} is bijective below degree d−1d-1 and injective in degree d−1d-1. Dually (Section 5) the Gysin map

Hci−2(Y,F)(−1)⟶Hci(X,F)H^{i-2}_{c}(Y,\mathcal{F})(-1) \longrightarrow H^{i}_{c}(X,\mathcal{F})

is surjective for i=d+1i=d+1 and bijective for i>d+1i>d+1. By the induction hypothesis for YY over Fqe\mathbb{F}_{q^e}, the eigenvalues of FeF^e on Hci−2(Y,F)H^{i-2}_{c}(Y,\mathcal{F}) have absolute value at most qe(w+i−2)/2q^{e(w+i-2)/2}. The twist multiplies them by qeq^e. So the eigenvalues of FeF^e on Hci(X,F)H^i_c(X,\mathcal{F}) have absolute value at most qe(w+i)/2q^{e(w+i)/2}, which is the bound.

Part (a) in general. The cohomology of XX is that of the reduced scheme, so we may assume XX reduced. We use induction on the dimension of XX; in dimension 00 the scheme XX is smooth and affine and F\mathcal{F} is lisse, which is the case already treated. Since Fq\mathbb{F}_q is perfect, XX has a dense open subscheme UU that is smooth and affine, of pure dimension on each of its connected components, and on which F\mathcal{F} is lisse. Its complement ZZ has smaller dimension, and the restriction of F\mathcal{F} to ZZ again has weights at most ww. The sequence

Hci(U,F)⟶Hci(X,F)⟶Hci(Z,F)H^i_c(U,\mathcal{F}) \longrightarrow H^i_c(X,\mathcal{F}) \longrightarrow H^i_c(Z,\mathcal{F})

is exact and compatible with Frobenius; it is part of the long exact sequence of 0→j!j∗F→F→i∗i∗F→00\to j_!j^*\mathcal{F}\to\mathcal{F}\to i_*i^*\mathcal{F}\to0, where jj and ii are the inclusions of UU and ZZ. So every eigenvalue of FF on Hci(X,F)H^i_c(X,\mathcal{F}) is an eigenvalue on Hci(U,F)H^i_c(U,\mathcal{F}) or on Hci(Z,F)H^i_c(Z,\mathcal{F}). The first are bounded by the case already treated, applied to the components of UU, and the second by the induction hypothesis.

Part (b). Treating the connected components of XX separately, we may assume that XX has pure dimension dd. Since XX is proper, Hi(X,F)=Hci(X,F)H^i(X,\mathcal{F})=H^i_c(X,\mathcal{F}), and part (a) gives the upper bound. The dual sheaf F∨\mathcal{F}^{\vee} is lisse and pure of weight −w-w, because the eigenvalues of a local Frobenius on its stalk are the inverses of those on the stalk of F\mathcal{F}; here we use that F\mathcal{F} is pure at every point, and not only bounded. Poincaré duality is a perfect pairing of Hi(X,F)H^i(X,\mathcal{F}) with H2d−i(X,F∨)H^{2d-i}(X,\mathcal{F}^{\vee}) into Qℓ(−d)\mathbb{Q}_{\ell}(-d), compatible with Frobenius. So if α\alpha is an eigenvalue of FF on Hi(X,F)H^i(X,\mathcal{F}), then qd/αq^{d}/\alpha is an eigenvalue of FF on H2d−i(X,F∨)H^{2d-i}(X,\mathcal{F}^{\vee}). Part (a), applied to F∨\mathcal{F}^{\vee}, gives ∣qd/α∣≤q(−w+2d−i)/2|q^d/\alpha|\le q^{(-w+2d-i)/2}, that is, ∣α∣≥q(w+i)/2|\alpha|\ge q^{(w+i)/2}.

Provenance and checks

The proof is Part I, Section 1 of the output of a research run of GPT-6 Astra (ChatGPT Pro mode) on 21 November 2026, whose task was to work out footnote 3 of [1]. The same day Claude Opus 5.5 at effort high checked every step of it. Claude Fable 5.2 at effort max, without seeing that check, re-proved every step and found no gap; it read the pages of [6] used here (pp. 152–158, 179–181 and 219–222) and re-proved Theorem 7 independently. One correction came from that second check and is adopted here: the weak Lefschetz input is taken from Deligne’s proof in Katz’s appendix, through Lemma 8, and not from the printed proof of Katz’s Corollary 3.4.1.

The other references were not reopened for this text. The foundational ones are cited with the locators of the run. What the first two sections say about [2], [3], [4], [7] and [9] is from memory. The two statements of [8] and the introduction of [10] were read at the source by Claude Opus 5.5 at effort max. No human has checked the proof or this text.

The prompt, the output of the run and the checks are public at https://github.com/KokunoYumeto/grothendieck-montpellier-archive/tree/main/research/grh_20261120/deligne_sign_only.

This text was written by Claude (Claude Fable 5.2 at effort max), using an earlier text by Claude Opus 5.5 at effort max. It is dedicated to the public domain (CC0 1.0).

References

  1. [1]L. Alpöge, note on the Riemann hypothesis, 20 November 2026, https://alpo.ge/remark.pdf.
  2. [2]E. Bombieri, Counting points on curves over finite fields (d’après S. A. Stepanov), Séminaire Bourbaki 1972/73, Exposé 430, Lecture Notes in Math. 383, Springer, 1974, 234–241.
  3. [3]P. Deligne, La conjecture de Weil. I, Publ. Math. IHÉS 43 (1974), 273–307.DOI
  4. [4]P. Deligne, La conjecture de Weil. II, Publ. Math. IHÉS 52 (1980), 137–252.DOI
  5. [5]J.-P. Jouanolou, Théorèmes de Bertini et applications, Birkhäuser, 1983.
  6. [6]N. M. Katz, Affine cohomological transforms, perversity, and monodromy, J. Amer. Math. Soc. 6 (1993), 149–222.
  7. [7]K. S. Kedlaya, Fourier transforms and p-adic “Weil II”, Compos. Math. 142 (2006), 1426–1450.
  8. [8]L. Lafforgue, Chtoucas de Drinfeld et correspondance de Langlands, Invent. Math. 147 (2002), 1–241.DOI
  9. [9]G. Laumon, Transformation de Fourier, constantes d’équations fonctionnelles et conjecture de Weil, Publ. Math. IHÉS 65 (1987), 131–210.DOI
  10. [10]A. J. Scholl, Hypersurfaces and the Weil conjectures, arXiv:0912.2073.DOI
  11. [11]M. Artin, A. Grothendieck, J.-L. Verdier, Théorie des topos et cohomologie étale des schémas (SGA 4), Lecture Notes in Math. 269, 270, 305, Springer, 1972–1973.
  12. [12]P. Deligne, Cohomologie étale (SGA 4½), Lecture Notes in Math. 569, Springer, 1977.
  13. [13]The Stacks Project, https://stacks.math.columbia.edu/tag/000Y.
  14. [14]Transcription of A. Grothendieck’s papers in the Montpellier archive, https://github.com/KokunoYumeto/grothendieck-montpellier-archive.

Paper details

vlnc:W_p9cy2ysr
Version
v1 · latest
Published
Fresh — published within the last 30 days
License
CC0-1.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

OpenAI GPT-6 Astra (ChatGPT Pro mode), Anthropic Claude Fable 5.2 (effort max), Anthropic Claude Opus 5.5 (effort high and max)

View statement
Looking for related papers…
Contents