riemann

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

Analysis

The analysis that the proof uses: limits, lim sup and OO bounds, measures and Fubini, L2L^2 spaces, the Fourier transform and Schwartz functions, Sobolev norms, topological groups and covers, Haar measure, Lie groups and one-parameter subgroups.

Prerequisites: Algebra.

The proof uses real analysis and analysis on groups. The bounds of the arithmetic branches are stated with lim sups and OO notation, and the theta branch integrates over adelic quotients with a Haar measure and lets the Weil representation act on Schwartz functions.

Limits and the O notation

Definition. A sequence (an)(a_n) of real numbers has limit aa if for each ε>0\varepsilon > 0 there is n0n_0 with ∣an−a∣<ε|a_n - a| < \varepsilon for each n≥n0n \ge n_0. A function ff on ℝ\mathbb{R} has limit bb as x→x0x \to x_0 if for each ε>0\varepsilon > 0 there is δ>0\delta > 0 with ∣f(x)−b∣<ε|f(x) - b| < \varepsilon whenever 0<∣x−x0∣<δ0 < |x - x_0| < \delta.

Definition. The limit superior of a real sequence is lim sup⁡n→∞an=lim⁡n→∞sup⁡k≥nak∈[−∞,∞].\limsup_{n \to \infty} a_n = \lim_{n \to \infty} \sup_{k \ge n} a_k \in [-\infty, \infty]. It always exists. The sequence converges if and only if lim sup⁡an=lim inf⁡an\limsup a_n = \liminf a_n is finite, and then both equal the limit. For example, lim sup⁡(−1)n=1\limsup (-1)^n = 1 and lim inf⁡(−1)n=−1\liminf (-1)^n = -1.

Definition. A statement P(x)P(x) holds for all large xx if there is x0x_0 such that P(x)P(x) holds for each x≥x0x \ge x_0. Let g>0g > 0. One writes f(x)=O(g(x))f(x) = O(g(x)) if there is a constant CC, the implied constant, with ∣f(x)∣≤Cg(x)|f(x)| \le C g(x) for all large xx.

Example. log⁡x=O(xε)\log x = O(x^\varepsilon) for each ε>0\varepsilon > 0, and x2+5x=O(x2)x^2 + 5x = O(x^2).

The volume vol(P,Λ)\mathrm{vol}(P, \Lambda) of the Weil bound is a lim sup of normalized logarithms of counts, the hypotheses of HypsAlt hold for all large xx, and the literature bounds ∣L′(1,χ)/L(1,χ)−Z(χ)∣|L'(1, \chi)/L(1, \chi) - Z(\chi)| by Clog⁡(3q)C \log(3q).

Continuity and smooth functions

Definition. A map f:X→Yf : X \to Y of topological spaces is continuous if f−1(U)f^{-1}(U) is open for each open U⊆YU \subseteq Y. For metric spaces this is the ε\varepsilon-δ\delta condition at each point.

Definition. A space XX is compact if each cover of XX by open sets has a finite subcover, and locally compact if each point has a compact neighbourhood. A continuous real function on a nonempty compact space is bounded and attains its maximum.

Definition. Let U⊆ℝnU \subseteq \mathbb{R}^n be open. A function f:U→ℂf : U \to \mathbb{C} is smooth if all partial derivatives ∂αf=∂1α1⋯∂nαnf\partial^\alpha f = \partial_1^{\alpha_1} \cdots \partial_n^{\alpha_n} f exist and are continuous on UU. For a multi-index α\alpha, ∣α∣=α1+⋯+αn|\alpha| = \alpha_1 + \dots + \alpha_n.

The Galois representation rr of the trunk is continuous, and the Galois row works with smooth functions on GLa(𝔸E)\mathrm{GL}_a(\mathbb{A}_E).

Measures and integrals

Definition. A σ-algebra on a set XX is a family ℳ\mathcal{M} of subsets that contains XX and is closed under complements and countable unions. A measure is a map μ:ℳ→[0,∞]\mu : \mathcal{M} \to [0, \infty] with μ(∅)=0\mu(\emptyset) = 0 and μ(⋃iAi)=∑iμ(Ai)\mu(\bigcup_i A_i) = \sum_i \mu(A_i) for each countable family of disjoint sets. The Borel σ-algebra of a topological space is the one generated by the open sets.

Definition. Let (X,ℳ,μ)(X, \mathcal{M}, \mu) be a measure space. A measurable function f:X→ℂf : X \to \mathbb{C} is integrable if ∫X∣f∣ dμ<∞\int_X |f| \, d\mu < \infty. Then the Lebesgue integral ∫Xf dμ\int_X f \, d\mu is defined, and ∣∫f dμ∣≤∫∣f∣ dμ|\int f \, d\mu| \le \int |f| \, d\mu. Two functions are equal almost everywhere if they differ on a set of measure 00.

Theorem (dominated convergence theorem). Let fnf_n be measurable, with fn→ff_n \to f almost everywhere, and let gg be integrable with ∣fn∣≤g|f_n| \le g for each nn. Then ff is integrable and ∫fn dμ→∫f dμ\int f_n \, d\mu \to \int f \, d\mu.

Definition. A measure space is σ-finite if it is a countable union of sets of finite measure. For σ-finite (X,μ)(X, \mu) and (Y,ν)(Y, \nu), the product measure μ×ν\mu \times \nu is the unique measure on the product σ-algebra of X×YX \times Y with (μ×ν)(A×B)=μ(A)ν(B)(\mu \times \nu)(A \times B) = \mu(A) \nu(B).

Theorem (Tonelli). Let f:X×Y→[0,∞]f : X \times Y \to [0, \infty] be measurable. Then ∫X×Yf d(μ×ν)=∫X(∫Yf(x,y) dν(y))dμ(x)=∫Y(∫Xf(x,y) dμ(x))dν(y).\int_{X \times Y} f \, d(\mu \times \nu) = \int_X \Big( \int_Y f(x, y) \, d\nu(y) \Big) d\mu(x) = \int_Y \Big( \int_X f(x, y) \, d\mu(x) \Big) d\nu(y).

Theorem (Fubini's theorem). Let f:X×Y→ℂf : X \times Y \to \mathbb{C} be integrable for μ×ν\mu \times \nu. Then the same equalities hold, and the inner integrals are finite almost everywhere. By Tonelli, a measurable ff is integrable when one iterated integral of ∣f∣|f| is finite.

Example. For f(x,y)=(x2−y2)/(x2+y2)2f(x, y) = (x^2 - y^2)/(x^2 + y^2)^2 on (0,1]2(0, 1]^2 the two iterated integrals are π/4\pi/4 and −π/4-\pi/4, so ff is not integrable.

The proof upLast of the Bessel seesaw writes both sides of the seesaw identity as one integral over [U(W0)]×[U(T)][U(W_0)] \times [U(T)] and applies Fubini. Step 4 of block determinants proves that a product of two matrix coefficients is integrable.

L2 spaces and the Cauchy-Schwarz inequality

Definition. Let (X,μ)(X, \mu) be a measure space. The space L2(X,μ)L^2(X, \mu) is the set of measurable f:X→ℂf : X \to \mathbb{C} with ∫∣f∣2 dμ<∞\int |f|^2 \, d\mu < \infty, modulo functions that are 00 almost everywhere. Its inner product is ⟨f,g⟩=∫fg‾ dμ\langle f, g \rangle = \int f \overline{g} \, d\mu, and its norm is ∥f∥2=⟨f,f⟩1/2\|f\|_2 = \langle f, f \rangle^{1/2}.

Definition. A Hilbert space is a complex vector space with an inner product whose norm is complete. A unitary representation of a topological group GG (defined below) on a Hilbert space HH is a homomorphism π\pi from GG to the group of linear bijections of HH that preserve the inner product, such that g↦π(g)vg \mapsto \pi(g) v is continuous for each vv.

Theorem (Cauchy-Schwarz inequality). In a space with an inner product, ∣⟨f,g⟩∣≤∥f∥ ∥g∥|\langle f, g \rangle| \le \|f\| \, \|g\|, with equality if and only if ff and gg are linearly dependent.

The space L2(X,μ)L^2(X, \mu) is a Hilbert space (Riesz-Fischer). For a unitary representation, Cauchy-Schwarz bounds each matrix coefficient: ∣⟨π(g)v,w⟩∣≤∥v∥ ∥w∥|\langle \pi(g) v, w \rangle| \le \|v\| \, \|w\|. The metaplectic representation of Step 8 of the Siegel-Weil chapter is unitary on an L2L^2 space, and RowsFinDim of HypsAlt is about automorphic L2L^2 spaces.

The Fourier transform

Definition. The Fourier transform of an integrable f:ℝn→ℂf : \mathbb{R}^n \to \mathbb{C} is f^(ξ)=∫ℝnf(x) e−2πix⋅ξ dx.\hat f(\xi) = \int_{\mathbb{R}^n} f(x) \, e^{-2\pi i x \cdot \xi} \, dx. It is a bounded continuous function, and f^(ξ)→0\hat f(\xi) \to 0 as ∣ξ∣→∞|\xi| \to \infty.

Example. The Gaussian φ0(x)=e−πx⋅x\varphi_0(x) = e^{-\pi x \cdot x} satisfies φ^0=φ0\hat\varphi_0 = \varphi_0 and ∫ℝnφ0=1\int_{\mathbb{R}^n} \varphi_0 = 1.

Definition. The Schwartz space 𝒮(ℝn)\mathcal{S}(\mathbb{R}^n) is the set of smooth ff such that each seminorm ∥f∥α,β=sup⁡x∈ℝn∣xα∂βf(x)∣\|f\|_{\alpha, \beta} = \sup_{x \in \mathbb{R}^n} |x^\alpha \partial^\beta f(x)| is finite. Its elements are Schwartz functions, for example the Gaussian.

Theorem (Fourier inversion and Plancherel). The Fourier transform is a linear bijection 𝒮(ℝn)→𝒮(ℝn)\mathcal{S}(\mathbb{R}^n) \to \mathcal{S}(\mathbb{R}^n), with inverse f(x)=∫f^(ξ)e2πix⋅ξ dξf(x) = \int \hat f(\xi) e^{2\pi i x \cdot \xi} \, d\xi. It satisfies ⟨f^,g^⟩=⟨f,g⟩\langle \hat f, \hat g \rangle = \langle f, g \rangle, so it extends to a unitary map L2(ℝn)→L2(ℝn)L^2(\mathbb{R}^n) \to L^2(\mathbb{R}^n).

Over a local field or the adeles the same definitions work with a nontrivial additive character ψ\psi in place of e2πixe^{2\pi i x} (Chapter 5). In the Siegel-Weil chapter the Weyl element acts by a Fourier transform, and the test functions of the theta matching are Schwartz functions.

Sobolev norms

Definition. Let GG be a Lie group with Lie algebra 𝔤\mathfrak{g} (both defined below), Haar measure on GG, and a basis X1,…,XkX_1, \dots, X_k of 𝔤\mathfrak{g}. Each X∈𝔤X \in \mathfrak{g} acts on smooth functions by (Xφ)(g)=ddt∣t=0φ(gexp⁡(tX))(X\varphi)(g) = \frac{d}{dt}\big|_{t=0} \varphi(g \exp(tX)). The Sobolev norm of order dd of φ\varphi is ∥φ∥S,d=∑∣α∣≤d∥Xαφ∥L2,Xα=X1α1⋯Xkαk.\|\varphi\|_{S, d} = \sum_{|\alpha| \le d} \|X^\alpha \varphi\|_{L^2}, \qquad X^\alpha = X_1^{\alpha_1} \cdots X_k^{\alpha_k}. Weighted norms such as ∑∣α∣≤dsup⁡g∣Xαφ(g)∣ ∥g∥−N\sum_{|\alpha| \le d} \sup_g |X^\alpha \varphi(g)| \, \|g\|^{-N} are of the same type.

Example. On ℝ\mathbb{R} with X=d/dxX = d/dx, the norm of order 11 is ∥φ∥2+∥φ′∥2\|\varphi\|_2 + \|\varphi'\|_2.

The bound CuspDecaySobolev of Step 10 of the Siegel-Weil chapter says that a smooth cusp function decays faster than each power of the height, with a constant controlled by a weighted Sobolev norm.

Topological groups and covering maps

Definition. A topological group is a group GG with a topology for which (x,y)↦xy(x, y) \mapsto xy and x↦x−1x \mapsto x^{-1} are continuous. It is a locally compact group if its topology is locally compact and Hausdorff.

Example. ℝn\mathbb{R}^n, ℂ×\mathbb{C}^\times, GLn(ℝ)\mathrm{GL}_n(\mathbb{R}), ℚp\mathbb{Q}_p, each discrete group, and the adele group GLn(𝔸F)\mathrm{GL}_n(\mathbb{A}_F) of Chapter 4 are locally compact.

Definition. A continuous surjection p:X~→Xp : \tilde X \to X is a covering map if each x∈Xx \in X has an open neighbourhood UU such that p−1(U)p^{-1}(U) is a disjoint union of open sets, each mapped homeomorphically onto UU by pp. It is a double cover if each fibre has two points. A covering homomorphism p:G~→Gp : \tilde G \to G of topological groups is a covering map that is a homomorphism; its kernel is a discrete normal subgroup.

Example. z↦z2z \mapsto z^2 is a double cover ℂ×→ℂ×\mathbb{C}^\times \to \mathbb{C}^\times with kernel {±1}\{\pm 1\}, and ℝ→ℝ/ℤ\mathbb{R} \to \mathbb{R}/\mathbb{Z} is a covering with kernel ℤ\mathbb{Z}.

Proposition. Let n≥1n \ge 1. The fundamental group of Sp2n(ℝ)\mathrm{Sp}_{2n}(\mathbb{R}) is ℤ\mathbb{Z}. So Sp2n(ℝ)\mathrm{Sp}_{2n}(\mathbb{R}) has, up to isomorphism, exactly one connected double cover, the metaplectic group Mp2n(ℝ)\mathrm{Mp}_{2n}(\mathbb{R}), and it is not a matrix group.

MetaplecticRepresentation, in Step 8 of the Siegel-Weil chapter, asks for a covering homomorphism onto Sp(ℝ)\mathrm{Sp}(\mathbb{R}) with kernel of order 22 on which the Weil representation is a true representation.

Haar measure

Theorem (Haar, Weil). Let GG be a locally compact group. There is a nonzero regular Borel measure μ\mu on GG that is finite on compact sets, positive on nonempty open sets, and left invariant: μ(gA)=μ(A)\mu(gA) = \mu(A) for each g∈Gg \in G and each Borel set AA. Any two such measures differ by a positive constant factor.

Definition. Such a μ\mu is a Haar measure. For g∈Gg \in G, the measure A↦μ(Ag)A \mapsto \mu(Ag) is again left invariant, so it is Δ(g)μ\Delta(g) \mu for a number Δ(g)>0\Delta(g) > 0. The continuous homomorphism Δ:G→ℝ>0\Delta : G \to \mathbb{R}_{>0} is the modular function, and GG is unimodular if Δ=1\Delta = 1, that is, if μ\mu is also right invariant.

Example. A Haar measure is dxdx on ℝ\mathbb{R}, dx/∣x∣dx/|x| on ℝ×\mathbb{R}^\times, and ∣det⁡g∣−n∏i,jdgij|\det g|^{-n} \prod_{i,j} dg_{ij} on GLn(ℝ)\mathrm{GL}_n(\mathbb{R}). Abelian groups, compact groups, GLn\mathrm{GL}_n, Sp2n\mathrm{Sp}_{2n} and U(p,q)\mathrm{U}(p, q) are unimodular.

Definition. Let Γ⊆G\Gamma \subseteq G be a discrete subgroup. A fundamental domain for Γ\Gamma is a measurable set D⊆GD \subseteq G that meets each coset Γg\Gamma g in exactly one point. If GG is unimodular, integration over DD defines a GG-invariant measure on Γ\G\Gamma \backslash G, and its total mass μ(D)\mu(D) is the volume of Γ\G\Gamma \backslash G.

Model.IsGood, in the theta matching, asks for a Haar measure on UX(𝔸)U_X(\mathbb{A}) and a fundamental domain for UX(F)U_X(F) on which each theta function is integrable. Step 9 of the Siegel-Weil chapter proves that the domain of UJ(F)U_J(F) has finite volume.

Lie groups and Lie algebras

Definition. A Lie group is a group that is a smooth manifold, with smooth multiplication and inverse. A closed subgroup of GLn(ℝ)\mathrm{GL}_n(\mathbb{R}) is a Lie group (Cartan).

Definition. For X∈Mn(ℝ)X \in M_n(\mathbb{R}) the matrix exponential is exp⁡(X)=∑k≥0Xk/k!\exp(X) = \sum_{k \ge 0} X^k / k!. The series converges for each XX, exp⁡(X+Y)=exp⁡(X)exp⁡(Y)\exp(X + Y) = \exp(X) \exp(Y) if XY=YXXY = YX, and det⁡exp⁡(X)=etr⁡X\det \exp(X) = e^{\operatorname{tr} X}.

Definition. Let G⊆GLn(ℝ)G \subseteq \mathrm{GL}_n(\mathbb{R}) be a closed subgroup. Its Lie algebra is 𝔤={X∈Mn(ℝ):exp⁡(tX)∈G for each t∈ℝ}.\mathfrak{g} = \{X \in M_n(\mathbb{R}) : \exp(tX) \in G \text{ for each } t \in \mathbb{R}\}. It is a real vector space, closed under the bracket [X,Y]=XY−YX[X, Y] = XY - YX.

Definition. A one-parameter subgroup of a topological group GG is a continuous homomorphism γ:ℝ→G\gamma : \mathbb{R} \to G. For a closed subgroup GG of GLn(ℝ)\mathrm{GL}_n(\mathbb{R}), each one-parameter subgroup is t↦exp⁡(tX)t \mapsto \exp(tX) for exactly one X∈𝔤X \in \mathfrak{g}.

Example. Let J=(01n−1n0)J = \begin{pmatrix} 0 & 1_n \\ -1_n & 0 \end{pmatrix} and Sp2n(ℝ)={g:gTJg=J}\mathrm{Sp}_{2n}(\mathbb{R}) = \{g : g^T J g = J\}. Its Lie algebra is 𝔰𝔭2n={X:XTJ+JX=0}\mathfrak{sp}_{2n} = \{X : X^T J + J X = 0\}.

Definition. Let π\pi be a representation of GG on a Banach space VV. A vector vv is smooth if g↦π(g)vg \mapsto \pi(g) v is smooth. On smooth vectors, 𝔤\mathfrak{g} acts by the derived representation dπ(X)v=ddt∣t=0π(exp⁡(tX))vd\pi(X) v = \frac{d}{dt}\big|_{t=0} \pi(\exp(tX)) v, and dπ([X,Y])=[dπ(X),dπ(Y)]d\pi([X, Y]) = [d\pi(X), d\pi(Y)].

In the Siegel-Weil chapter, RealMetaplecticRepresentation lifts each t↦exp⁡(tX)t \mapsto \exp(tX), X∈𝔰𝔭X \in \mathfrak{sp}, to a one-parameter subgroup of the metaplectic cover, and IsRealMaps of LocalMatching uses the Lie algebra action at the infinite places.

Further reading

  • W. Rudin, Real and Complex Analysis, Chapters 1, 2, 4, 7 and 8 (measures, L2L^2, Fubini).
  • G. Folland, A Course in Abstract Harmonic Analysis, Chapters 2 and 3 (topological groups, Haar measure, unitary representations).
  • E. Stein and R. Shakarchi, Fourier Analysis, Chapters 5 and 6 (the Fourier transform and Schwartz functions).
  • B. Hall, Lie Groups, Lie Algebras, and Representations, Chapters 2 and 3.
  • A. Hatcher, Algebraic Topology, Section 1.3 (covering spaces).

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