riemann

The Lean code claims a proof of the Riemann Hypothesis. This site has not checked the claim yet. See what is checked.

Algebraic and complex geometry

Schemes and line bundles, the vanishing theorems and resolution of singularities, Dolbeault cohomology with the holomorphic Lefschetz formula and the Kodaira-Spencer map, étale and de Rham cohomology, and Shimura varieties with their toroidal compactifications.

Prerequisites: Elliptic curves and complex multiplication.

Algebraic geometry studies solution sets of polynomial equations as schemes, and complex geometry as complex manifolds. The Galois row of the proof uses both on towers of Shimura varieties, and the Weil bound on their integral models.

Schemes and varieties

Definition. A scheme is a locally ringed space that is locally isomorphic to Spec R\mathrm{Spec}\, R for commutative rings RR. The space of Spec R\mathrm{Spec}\, R is the set of prime ideals with the Zariski topology, whose closed sets are the sets V(I)V(I) of primes that contain an ideal II. A scheme over SS is a morphism X→SX \to S. A variety over a field kk is an integral separated scheme of finite type over kk.

Definition. A morphism f:X→Sf : X \to S is smooth of relative dimension dd if it is locally of finite presentation and flat, and its fibres are regular of pure dimension dd after each extension of the field. It is étale if it is smooth of relative dimension 00, proper if it is separated, of finite type and universally closed, and projective if it factors as a closed immersion into some ℙSN\mathbb{P}^N_S followed by the projection.

A smooth projective scheme over SS is one that is smooth and projective. Over ℂ\mathbb{C}, X(ℂ)X(\mathbb{C}) is then a compact complex manifold, and Serre's GAGA compares the two geometries.

Definition. A product of projective spaces over SS is ℙSn1×S⋯×SℙSnr\mathbb{P}^{n_1}_S \times_S \cdots \times_S \mathbb{P}^{n_r}_S. Its line bundles are the 𝒪(a1,…,ar)\mathcal{O}(a_1, \dots, a_r), tensor products of the pullbacks of 𝒪(ai)\mathcal{O}(a_i), of multidegree (a1,…,ar)(a_1, \dots, a_r). The Segre map embeds it in ℙSN\mathbb{P}^N_S with N+1=∏(ni+1)N + 1 = \prod (n_i + 1).

Definition. A subset of a scheme is Zariski dense if its Zariski closure is the whole space. For example, ℤn\mathbb{Z}^n is Zariski dense in 𝔸ℂn\mathbb{A}^n_{\mathbb{C}}, since a polynomial that vanishes on ℤn\mathbb{Z}^n is 00.

The Weil bound works with subvarieties of a product of projective spaces, and Zhang's upper bound there needs a Zariski dense set of curves.

Line bundles and divisors

Let XX be a normal projective variety of dimension nn over a field.

Definition. A Cartier divisor is a global section of 𝒦×/𝒪×\mathcal{K}^\times/\mathcal{O}^\times, where 𝒦\mathcal{K} is the sheaf of rational functions. It gives a line bundle 𝒪(D)\mathcal{O}(D). A Weil divisor is a finite sum ∑aiDi\sum a_i D_i of prime divisors with ai∈ℤa_i \in \mathbb{Z}, and a ℚ\mathbb{Q}-divisor is such a sum with ai∈ℚa_i \in \mathbb{Q}. It is ℚ\mathbb{Q}-Cartier if some multiple mDmD is Cartier.

Line bundles have intersection numbers (L1⋯Ln)(L_1 \cdots L_n), and degrees (Ldim⁡Y⋅Y)(L^{\dim Y} \cdot Y) on subvarieties YY. Divisors D,D′D, D' are numerically equivalent if (D⋅C)=(D′⋅C)(D \cdot C) = (D' \cdot C) for each curve CC.

Definition. A line bundle LL is ample if the sections of L⊗mL^{\otimes m} give a closed immersion into a projective space for some m>0m > 0. It is nef if (L⋅C)≥0(L \cdot C) \geq 0 for each curve C⊂XC \subset X, and big if h0(X,L⊗m)≥c mnh^0(X, L^{\otimes m}) \geq c\, m^n for some c>0c > 0 and all large divisible mm.

Example. On ℙn1×⋯×ℙnr\mathbb{P}^{n_1} \times \cdots \times \mathbb{P}^{n_r}, 𝒪(a1,…,ar)\mathcal{O}(a_1, \dots, a_r) is ample exactly when all ai>0a_i > 0, and nef exactly when all ai≥0a_i \geq 0.

Theorem (Kleiman). A nef line bundle LL has (Ldim⁡Y⋅Y)≥0(L^{\dim Y} \cdot Y) \geq 0 for each subvariety YY, and it is a limit of ample ℚ\mathbb{Q}-divisors in the space of numerical classes. It is big exactly when (Ln)>0(L^n) > 0 (Lazarsfeld, Theorem 2.2.16).

Definition. A reduced divisor D=∑DiD = \sum D_i on a smooth variety has simple normal crossings (snc) if each DiD_i is smooth and, near each point, DD is z1⋯zk=0z_1 \cdots z_k = 0 in local coordinates.

Definition. Let XX be normal and Δ\Delta an effective ℚ\mathbb{Q}-divisor with KX+ΔK_X + \Delta ℚ\mathbb{Q}-Cartier, where KXK_X is the canonical divisor. A log resolution is a proper birational μ:Y→X\mu : Y \to X with YY smooth and the exceptional locus plus μ−1(Δ)\mu^{-1}(\Delta) an snc divisor. Write KY+ΔY=μ∗(KX+Δ)K_Y + \Delta_Y = \mu^*(K_X + \Delta) with μ∗KY=KX\mu_* K_Y = K_X. The pair (X,Δ)(X, \Delta) is klt (Kawamata log terminal) if all coefficients of ΔY\Delta_Y are <1< 1 for one, equivalently each, log resolution.

Definition. For XX smooth and DD an effective ℚ\mathbb{Q}-divisor with log resolution μ\mu, the multiplier ideal is 𝒥(D)=μ∗𝒪Y(KY/X−⌊μ∗D⌋)\mathcal{J}(D) = \mu_* \mathcal{O}_Y(K_{Y/X} - \lfloor \mu^* D \rfloor). The pair (X,D)(X, D) is klt exactly when 𝒥(D)=𝒪X\mathcal{J}(D) = \mathcal{O}_X.

Vanishing theorems and resolution

Riemann-Roch and vanishing

Theorem (asymptotic Riemann-Roch). Let LL be a nef line bundle on a projective variety XX of dimension nn. Then h0(X,L⊗m)=(Ln)n!mn+O(mn−1)h^0(X, L^{\otimes m}) = \frac{(L^n)}{n!} m^n + O(m^{n-1}), and hi(X,L⊗m)=O(mn−i)h^i(X, L^{\otimes m}) = O(m^{n-i}) for i>0i > 0 (Lazarsfeld, Corollary 1.4.41).

The first formula follows from the second, since χ(X,L⊗m)\chi(X, L^{\otimes m}) is a polynomial in mm of degree nn with leading coefficient (Ln)/n!(L^n)/n! (Hirzebruch-Riemann-Roch when XX is smooth). The Weil bound states it as NefRR, for XX in a product of projective spaces and a multidegree nef on the generic fibre.

Theorem (Kodaira vanishing). Let XX be a smooth projective variety over a field of characteristic 00 and LL an ample line bundle. Then Hi(X,ωX⊗L)=0H^i(X, \omega_X \otimes L) = 0 for i>0i > 0.

Theorem (Kawamata-Viehweg vanishing). The same holds for XX of dimension nn and LL nef with (Ln)>0(L^n) > 0 (Kawamata, 1982; Viehweg, 1982).

Theorem (Nadel vanishing). Let XX be a smooth complex projective variety, DD an effective ℚ\mathbb{Q}-divisor and LL a line bundle with L−DL - D nef and big. Then Hi(X,ωX⊗L⊗𝒥(D))=0H^i(X, \omega_X \otimes L \otimes \mathcal{J}(D)) = 0 for i>0i > 0 (Nadel, 1990).

With D=0D = 0 this is Kawamata-Viehweg vanishing over ℂ\mathbb{C}. Both are fields (KawamataViehweg, NadelVanishingAndNonKltDivisor) of the bundle of published arithmetic results.

Point separation

Theorem (Angehrn-Siu). Let XX be a smooth complex projective variety of dimension nn, LL an ample line bundle and x∈Xx \in X. If (Ld⋅Z)>(12n(n+1))d(L^d \cdot Z) > (\tfrac12 n(n+1))^d for each subvariety Z∋xZ \ni x of dimension dd, 1≤d≤n1 \leq d \leq n, then ωX⊗L\omega_X \otimes L has a section that does not vanish at xx (Angehrn and Siu, 1995).

EveryCharacter from the literature uses a two-point version for klt pairs, PointSeparationPair: under similar degree bounds, a section vanishes at exactly one of two given points.

Resolution of singularities

Theorem (resolution of singularities). Let XX be a reduced separated scheme of finite type over a field of characteristic 00. A finite sequence of blow-ups in smooth centres gives a smooth X′→XX' \to X that is an isomorphism over the smooth locus of XX, such that the reduced preimage of the singular locus is an snc divisor (Hironaka, 1964).

Theorem (principalisation). Let XX be smooth and separated of finite type over a field of characteristic 00, II an ideal sheaf whose support contains no component of XX, and EE an snc divisor. A sequence of blow-ups in smooth centres with normal crossings with EE, an isomorphism outside the support of II, makes the pullback of II an effective Cartier divisor with snc support (Hironaka, 1964).

Resolution of singularities and principalisation are inputs of the Weil bound, as ResolutionCharZero and PrincipalisationCharZero.

Complex manifolds

Let XX be a compact complex manifold of dimension nn.

Definition. A smooth form on XX has type (p,q)(p, q) if locally it is ∑fIJ dzI∧dzˉJ\sum f_{IJ}\, dz_I \wedge d\bar z_J with ∣I∣=p|I| = p, ∣J∣=q|J| = q. Then d=∂+∂ˉd = \partial + \bar\partial, and the Dolbeault cohomology is

Hp,q(X)=ker⁡(∂ˉ:𝒜p,q→𝒜p,q+1)im⁡(∂ˉ:𝒜p,q−1→𝒜p,q).H^{p,q}(X) = \frac{\ker(\bar\partial : \mathcal{A}^{p,q} \to \mathcal{A}^{p,q+1})}{\operatorname{im}(\bar\partial : \mathcal{A}^{p,q-1} \to \mathcal{A}^{p,q})}.

Theorem (Dolbeault). Hp,q(X)≅Hq(X,ΩXp)H^{p,q}(X) \cong H^q(X, \Omega^p_X), the sheaf cohomology of the sheaf of holomorphic pp-forms.

Theorem (Hodge decomposition). If XX is compact Kähler, for example X(ℂ)X(\mathbb{C}) for a smooth projective XX, then Hk(X,ℂ)=⨁p+q=kHp,q(X)H^k(X, \mathbb{C}) = \bigoplus_{p+q = k} H^{p,q}(X) and Hp,q‾=Hq,p\overline{H^{p,q}} = H^{q,p}.

Theorem (Atiyah-Bott fixed-point theorem). Let XX be a compact smooth manifold with an elliptic complex Γ(E0)→⋯→Γ(Em)\Gamma(E_0) \to \cdots \to \Gamma(E_m), and f:X→Xf : X \to X smooth with det⁡(1−dfx)≠0\det(1 - df_x) \neq 0 at each fixed point. Let φi:f∗Ei→Ei\varphi_i : f^*E_i \to E_i be bundle maps that commute with the complex and induce TT on its cohomology. Then

∑i(−1)itr⁡(T∣Hi)=∑f(x)=x∑i(−1)itr⁡(φi,x)∣det⁡(1−dfx)∣\sum_i (-1)^i \operatorname{tr}(T \mid H^i) = \sum_{f(x) = x} \frac{\sum_i (-1)^i \operatorname{tr}(\varphi_{i,x})}{|\det(1 - df_x)|}

(Atiyah and Bott, 1967 and 1968).

For the de Rham complex this is the Lefschetz fixed-point theorem. For holomorphic ff and the Dolbeault complex of Ωp\Omega^p, a fixed point contributes tr⁡(∧p tdfx)/det⁡(1−dfx)\operatorname{tr}(\wedge^p\, {}^t df_x)/\det(1 - df_x), with dfxdf_x on the holomorphic tangent space: the holomorphic Lefschetz formula. The proof needs it for correspondences (defined below): let ZZ be smooth projective of dimension dd over ℂ\mathbb{C}, c1,c2:Y→Zc_1, c_2 : Y \to Z finite étale, FF the finite set of yy with c1(y)=c2(y)c_1(y) = c_2(y), and AyA_y the d×dd \times d matrices at these points, with det⁡(1−Ay)≠0\det(1 - A_y) \neq 0. Then for each aa,

∑b=0d(−1)btr⁡(c1∗∘c2∗∣Ha,b)=∑y∈F[Xa]det⁡(1+XAy)det⁡(1−Ay).\sum_{b=0}^{d} (-1)^b \operatorname{tr}\big(c_{1*} \circ c_2^* \mid H^{a,b}\big) = \sum_{y \in F} \frac{[X^a] \det(1 + X A_y)}{\det(1 - A_y)}.

The code proves this as HolomorphicLefschetzFormula, an input of GeoOne in the Galois row.

Definition. Let π:𝒳→B\pi : \mathcal{X} \to B be a proper submersion of complex manifolds with fibres XtX_t, and Θ\Theta the holomorphic tangent bundle. The Kodaira-Spencer map κt:TtB→H1(Xt,ΘXt)\kappa_t : T_t B \to H^1(X_t, \Theta_{X_t}) is the connecting map of 0→ΘXt→Θ𝒳∣Xt→𝒪Xt⊗TtB→00 \to \Theta_{X_t} \to \Theta_{\mathcal{X}}|_{X_t} \to \mathcal{O}_{X_t} \otimes T_t B \to 0.

It vanishes when the family is locally trivial. For abelian varieties it is a map TtB→Hom⁡(ωAt,Lie⁡(At∨))T_t B \to \operatorname{Hom}(\omega_{A_t}, \operatorname{Lie}(A_t^\vee)). The record KSAll of the Galois row holds six statements about it; one, KSHodge, uses the Lefschetz formula.

Theorem (projective bundle formula). Let EE be a vector bundle of rank rr on XX, π:ℙ(E)→X\pi : \mathbb{P}(E) \to X its projective bundle and ξ=c1(𝒪(1))\xi = c_1(\mathcal{O}(1)). Then H∗(ℙ(E))H^*(\mathbb{P}(E)) is a free H∗(X)H^*(X)-module with basis 1,ξ,…,ξr−11, \xi, \dots, \xi^{r-1}.

The same holds for étale and de Rham cohomology and for Chow groups.

Étale cohomology

Definition. A morphism is finite étale if it is finite and étale. Over ℂ\mathbb{C} this is a finite covering map on complex points. A correspondence from XX to itself is a pair of maps c1,c2:Y→Xc_1, c_2 : Y \to X. When c1c_1 is finite étale, it acts on cohomology by c1∗∘c2∗c_{1*} \circ c_2^*.

Example. For a Shimura variety ShK\mathrm{Sh}_K (defined below) and g∈G(𝔸f)g \in G(\mathbb{A}_f), the maps ShK∩gKg−1→ShK\mathrm{Sh}_{K \cap gKg^{-1}} \to \mathrm{Sh}_K given by 11 and by gg form a Hecke correspondence, finite étale for small KK.

Definition. Let XX be a separated scheme of finite type over a field kk and ℓ\ell a prime different from the characteristic. The étale cohomology of XX is Hi(Xkˉ,ℚℓ)=(lim←⁡nHeti(Xkˉ,ℤ/ℓn))⊗ℚℓH^i(X_{\bar k}, \mathbb{Q}_\ell) = (\varprojlim_n H^i_{\mathrm{et}}(X_{\bar k}, \mathbb{Z}/\ell^n)) \otimes \mathbb{Q}_\ell, a finite-dimensional space with an action of Gal(kˉ/k)\mathrm{Gal}(\bar k/k).

For XX over ℂ\mathbb{C}, Artin's comparison theorem gives Hi(X,ℚℓ)≅Hi(X(ℂ),ℚ)⊗ℚℓH^i(X, \mathbb{Q}_\ell) \cong H^i(X(\mathbb{C}), \mathbb{Q}) \otimes \mathbb{Q}_\ell. They are the main examples of the Galois representations chapter.

Theorem (proper base change). Let f:X→Sf : X \to S be proper and ℱ\mathcal{F} a torsion étale sheaf on XX. For each geometric point sˉ\bar s of SS, (Rif∗ℱ)sˉ≅Hi(Xsˉ,ℱ∣Xsˉ)(R^i f_* \mathcal{F})_{\bar s} \cong H^i(X_{\bar s}, \mathcal{F}|_{X_{\bar s}}) (SGA 4, XII).

With smooth base change: if XX is smooth and proper over 𝒪K[1/N]\mathcal{O}_K[1/N], then Hi(XKˉ,ℚℓ)H^i(X_{\bar K}, \mathbb{Q}_\ell) is unramified at each v∤Nℓv \nmid N\ell, and Frob⁡v\operatorname{Frob}_v acts as on the cohomology of the reduction. The record Steps of GeoOne uses proper base change and a projective bundle formula.

De Rham cohomology and deformations

Definition. Let f:X→Sf : X \to S be smooth. The algebraic de Rham cohomology is HdRi(X/S)=Rif∗ΩX/S∙H^i_{\mathrm{dR}}(X/S) = R^i f_* \Omega^\bullet_{X/S}, with the Hodge filtration FpF^p from the truncations ΩX/S≥p\Omega^{\geq p}_{X/S}.

For ff smooth and proper over a base of characteristic 00, the Hodge to de Rham spectral sequence degenerates and Fp/Fp+1≅Ri−pf∗ΩX/SpF^p/F^{p+1} \cong R^{i-p} f_* \Omega^p_{X/S}. For an abelian scheme A→SA \to S this gives 0→ωA→HdR1(A/S)→Lie⁡(A∨)→00 \to \omega_{A} \to H^1_{\mathrm{dR}}(A/S) \to \operatorname{Lie}(A^\vee) \to 0, with ωA\omega_A the invariant differentials.

Definition. Let f:X→Sf : X \to S be smooth and proper with SS smooth over a field of characteristic 00. The Gauss-Manin connection is an integrable connection ∇\nabla on HdRi(X/S)H^i_{\mathrm{dR}}(X/S). Over ℂ\mathbb{C} its horizontal sections are the local system Rif∗ℂR^i f_* \mathbb{C}.

Theorem (Griffiths transversality). ∇(Fp)⊂Fp−1⊗ΩS1\nabla(F^p) \subset F^{p-1} \otimes \Omega^1_S, and the induced map Fp/Fp+1→(Fp−1/Fp)⊗ΩS1F^p/F^{p+1} \to (F^{p-1}/F^p) \otimes \Omega^1_S is the cup product with the Kodaira-Spencer class.

Definition. A deformation of X0X_0 over a local Artin ring RR with residue field kk is a flat X→Spec RX \to \mathrm{Spec}\, R with X⊗Rk≅X0X \otimes_R k \cong X_0. For X0X_0 smooth over kk, first-order deformations (over k[ε]/(ε2)k[\varepsilon]/(\varepsilon^2)) are classified by H1(X0,ΘX0)H^1(X_0, \Theta_{X_0}), and obstructions lie in H2(X0,ΘX0)H^2(X_0, \Theta_{X_0}).

Theorem (Grothendieck-Messing). Let S0↪SS_0 \hookrightarrow S be a closed immersion of schemes on which pp is locally nilpotent, whose ideal has nilpotent divided powers. Then lifting an abelian scheme A0A_0 over S0S_0 to SS is the same as lifting the Hodge filtration ωA0⊂HdR1(A0/S0)\omega_{A_0} \subset H^1_{\mathrm{dR}}(A_0/S_0) to a direct summand of the crystal of A0A_0 evaluated on SS (Messing, 1972).

These objects enter CM and abelian inputs and, as GrothendieckMessing, the bundle of published arithmetic results.

Shimura varieties and toroidal compactifications

Definition. A Shimura datum is a pair (G,X)(G, X) of a reductive group GG over ℚ\mathbb{Q} and a G(ℝ)G(\mathbb{R})-conjugacy class XX of homomorphisms Resℂ/ℝ𝔾m→Gℝ\mathrm{Res}_{\mathbb{C}/\mathbb{R}} \mathbb{G}_m \to G_{\mathbb{R}} that satisfies Deligne's axioms. For a compact open subgroup K⊂G(𝔸f)K \subset G(\mathbb{A}_f), the Shimura variety is

ShK(G,X)=G(ℚ)\X×G(𝔸f)/K.\mathrm{Sh}_K(G, X) = G(\mathbb{Q}) \backslash X \times G(\mathbb{A}_f)/K.

For KK neat it is a smooth quasi-projective complex variety (Baily and Borel), with a canonical model over the reflex field E(G,X)E(G, X) (Deligne; Borovoi and Milne in general). As KK shrinks through neat subgroups these form a tower with finite étale transition maps, on which G(𝔸f)G(\mathbb{A}_f) acts by Hecke correspondences.

Example. For G=GL2G = \mathrm{GL}_2 and X=ℂ∖ℝX = \mathbb{C} \setminus \mathbb{R}, ShK\mathrm{Sh}_K is a finite union of modular curves Γ\ℍ\Gamma \backslash \mathbb{H}. A unitary Shimura variety of signature (p,q)(p, q) comes from the unitary group of a Hermitian space over a CM field with signature (p,q)(p, q) at the infinite places.

In PEL type, ShK\mathrm{Sh}_K is a moduli space of abelian varieties with polarization, endomorphisms and level structure. For K=KpKpK = K^p K_p with KpK_p hyperspecial and pp unramified, the same moduli problem over 𝒪E⊗ℤ(p)\mathcal{O}_E \otimes \mathbb{Z}_{(p)} gives a smooth integral model (Kottwitz, 1992).

Definition. A toric variety is a normal variety with an action of a torus TT that has a dense open orbit isomorphic to TT. It is given by a fan of rational polyhedral cones.

Theorem (toroidal compactification). Let ShK\mathrm{Sh}_K be a Shimura variety with KK neat. For each admissible collection Σ\Sigma of rational polyhedral cone decompositions at the cusps there is a compactification ShKΣ\mathrm{Sh}_K^\Sigma that contains ShK\mathrm{Sh}_K as a dense open subset and is locally like a toric variety near its boundary. For suitable smooth projective Σ\Sigma it is smooth and projective with snc boundary (Ash, Mumford, Rapoport and Tai, 1975).

The literature chapter uses integral versions, SiegelToroidalIntegral for Siegel moduli spaces (Faltings and Chai, 1990) and ToroidalForFanKS for unitary Shimura varieties (Lan, 2013). Chapter 15 defines the heights that the Weil bound computes on these models.

Further reading

  • R. Hartshorne, Algebraic Geometry, Chapters II and III.
  • P. Griffiths and J. Harris, Principles of Algebraic Geometry, Chapters 0 and 3.
  • R. Lazarsfeld, Positivity in Algebraic Geometry I and II, Sections 1.4, 4.3, 9.4 and 10.4.
  • J. S. Milne, Étale Cohomology, Chapters I and VI.
  • K.-W. Lan, Arithmetic Compactifications of PEL-Type Shimura Varieties, Chapters 1 and 6.

The source of this page. Markdown. Something is wrong? Report it.