riemann

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

Classical groups

The general linear, symplectic and unitary groups, their parabolic, Levi and compact subgroups, their adelic points and the quotients of finite volume, Hermitian lattices and the Siegel mass formula, and the Heisenberg and metaplectic groups.

Prerequisites: Algebra, Analysis, Number fields, Zeta and L-functions.

The proof uses GLn\mathrm{GL}_n over local fields, the symplectic group of the Weil representation, and unitary groups over a CM field. Here we define them, their parabolic and compact subgroups, their adelic points, and the quotients G(F)\G(𝔸F)G(F)\backslash G(\mathbb{A}_F) that carry automorphic forms.

General linear and symplectic groups

Let kk be a field and nβ‰₯1n \ge 1.

Definition. The general linear group GLn(k)\mathrm{GL}_n(k) is the group of invertible nΓ—nn \times n matrices over kk, and GL(V)\mathrm{GL}(V) is the group of invertible linear maps of a finite-dimensional kk-vector space VV. The special linear group SLn(k)\mathrm{SL}_n(k) is the kernel of det⁑\det.

Definition. A symplectic space over kk is a finite-dimensional vector space WW with a nondegenerate alternating bilinear form βŸ¨β‹…,β‹…βŸ©\langle \cdot, \cdot \rangle, that is, ⟨w,w⟩=0\langle w, w \rangle = 0 for each ww.

Theorem (Darboux basis). A symplectic space WW has even dimension 2n2n and a basis e1,…,en,f1,…,fne_1, \dots, e_n, f_1, \dots, f_n with ⟨ei,ej⟩=⟨fi,fj⟩=0\langle e_i, e_j \rangle = \langle f_i, f_j \rangle = 0 and ⟨ei,fj⟩=Ξ΄ij\langle e_i, f_j \rangle = \delta_{ij} (Lang, Algebra, XV.8).

Such a basis gives Darboux coordinates on WW, and W=XβŠ•YW = X \oplus Y with X=⟨ei⟩X = \langle e_i \rangle and Y=⟨fi⟩Y = \langle f_i \rangle Lagrangian: isotropic of dimension nn.

Definition. The symplectic group Sp(W)\mathrm{Sp}(W) is the group of g∈GL(W)g \in \mathrm{GL}(W) that preserve βŸ¨β‹…,β‹…βŸ©\langle \cdot, \cdot \rangle. In a Darboux basis it is Sp2n(k)={g∈GL2n(k):gTJg=J},J=(01nβˆ’1n0).\mathrm{Sp}_{2n}(k) = \{ g \in \mathrm{GL}_{2n}(k) : g^{T} J g = J \}, \qquad J = \begin{pmatrix} 0 & 1_n \\ -1_n & 0 \end{pmatrix}.

Example. Sp2=SL2\mathrm{Sp}_2 = \mathrm{SL}_2, since gTJg=(det⁑g)Jg^T J g = (\det g) J for g∈GL2(k)g \in \mathrm{GL}_2(k).

IsRecT is about irreducible representations of GLn(Ew)\mathrm{GL}_n(E_w), and the theta chapter fixes Darboux coordinates on XβŠ—WX \otimes W.

Unitary groups

Let E/FE/F be a separable quadratic extension with nontrivial automorphism cc; Chapter 1 defines Hermitian forms and signatures.

Definition. Let V=EnV = E^n with the Hermitian form h(x,y)=xTJc(y)h(x, y) = x^{T} J c(y), where J∈GLn(E)J \in \mathrm{GL}_n(E) and c(J)T=Jc(J)^{T} = J. The unitary group of VV is U(V)=UJ={g∈GLn(E):gTJ c(g)=J}.\mathrm{U}(V) = \mathrm{U}_J = \{ g \in \mathrm{GL}_n(E) : g^{T} J \, c(g) = J \}. With c(J)T=βˆ’Jc(J)^{T} = -J the same formula defines the unitary group of a skew-Hermitian space. If δ∈EΓ—\delta \in E^\times and c(Ξ΄)=βˆ’Ξ΄c(\delta) = -\delta, then Ξ΄h\delta h is skew-Hermitian with the same unitary group.

As an algebraic group over FF, U(V)\mathrm{U}(V) becomes GLn\mathrm{GL}_n over EE. The determinant maps it onto U(1)={z∈EΓ—:z c(z)=1}\mathrm{U}(1) = \{ z \in E^\times : z \, c(z) = 1 \}, with kernel the special unitary group SU(V)\mathrm{SU}(V).

Definition. For p+q=np + q = n, U(p,q)\mathrm{U}(p, q) is the unitary group of β„‚n\mathbb{C}^n over ℝ\mathbb{R} for J=diag(1p,βˆ’1q)J = \mathrm{diag}(1_p, -1_q). It is compact if and only if pq=0pq = 0, and then it is the group U(n)\mathrm{U}(n) of unitary matrices.

Now let EE be a CM field with maximal totally real subfield FF (Chapter 4) and VV a Hermitian space over EE. At a real place vv of FF, hh has a signature (pv,qv)(p_v, q_v) and U(V)(Fv)β‰…U(pv,qv)\mathrm{U}(V)(F_v) \cong \mathrm{U}(p_v, q_v).

Definition. VV is definite if qv=0q_v = 0 at each real place vv (or pv=0p_v = 0 at each). Then U(V)\mathrm{U}(V) is a definite unitary group, and U(V)(FβŠ—β„šβ„)=∏vU(n)\mathrm{U}(V)(F \otimes_{\mathbb{Q}} \mathbb{R}) = \prod_v \mathrm{U}(n) is compact.

Proposition. Let VV be a Hermitian and WW a skew-Hermitian space over EE. Then Tr⁑E/F\operatorname{Tr}_{E/F} of the product form makes VβŠ—EWV \otimes_E W a symplectic space over FF of dimension 2dim⁑Vdim⁑W2 \dim V \dim W, and U(V)Γ—U(W)\mathrm{U}(V) \times \mathrm{U}(W) maps into Sp(VβŠ—EW)\mathrm{Sp}(V \otimes_E W).

The two groups are each other's centralizers in Sp(VβŠ—EW)\mathrm{Sp}(V \otimes_E W), a dual pair (Chapter 16). TLiftK takes JX=1J_X = 1 of signature (n,0)(n, 0) and a skew-Hermitian WW with iJWiJ_W of signature (d,0)(d, 0), and GeoOne a definite unitary group in an odd number of variables.

Parabolic subgroups

Let V=knV = k^n.

Definition. A flag in VV is a chain 0=V0⊊V1βŠŠβ‹―βŠŠVr=V0 = V_0 \subsetneq V_1 \subsetneq \cdots \subsetneq V_r = V. Its stabilizer PP in GL(V)\mathrm{GL}(V) is a parabolic subgroup: in an adapted basis, the block upper triangular matrices with diagonal blocks of sizes ni=dim⁑Viβˆ’dim⁑Viβˆ’1n_i = \dim V_i - \dim V_{i-1}. A full flag (r=nr = n) gives the Borel subgroup BB of upper triangular matrices.

Definition. For such PP, the Levi subgroup Mβ‰…GLn1Γ—β‹―Γ—GLnrM \cong \mathrm{GL}_{n_1} \times \cdots \times \mathrm{GL}_{n_r} is the block diagonal matrices, and the unipotent radical NN is the block upper triangular matrices with identity diagonal blocks. Then NN is normal in PP and P=M⋉NP = M \ltimes N (the Levi decomposition).

For a reductive group GG over kk (defined below), a parabolic subgroup is a closed subgroup PP with G/PG/P projective. It has a Levi decomposition P=M⋉NP = M \ltimes N, with NN the largest connected normal unipotent subgroup of PP (Borel and Tits, 1965). The parabolic subgroups of U(V)\mathrm{U}(V) and Sp(W)\mathrm{Sp}(W) are the stabilizers of flags of isotropic subspaces.

Definition. Let TT be a maximal torus of a reductive group GG. The Weyl group is NG(T)/TN_G(T)/T; its elements are represented by Weyl elements in NG(T)N_G(T). For GLn\mathrm{GL}_n and the diagonal torus it is SnS_n, represented by the permutation matrices.

Theorem (Bruhat decomposition). GLn(k)\mathrm{GL}_n(k) is the disjoint union of the double cosets BwBBwB, with ww running over the permutation matrices (Borel, Linear Algebraic Groups, 14.12).

Definition (Siegel parabolic). Let W=E2nW = E^{2n} be skew-Hermitian with J=(01nβˆ’1n0)J = \begin{pmatrix} 0 & 1_n \\ -1_n & 0 \end{pmatrix}, so X=⟨e1,…,en⟩X = \langle e_1, \dots, e_n \rangle is maximal isotropic. The Siegel parabolic PP is the stabilizer of XX in U(W)\mathrm{U}(W), with Levi subgroup and unipotent radical M={m(a)=(a00c(a)βˆ’T):a∈GLn(E)},N={n(Ξ²)=(1nΞ²01n):Ξ²T=c(Ξ²)}.M = \left\{ m(a) = \begin{pmatrix} a & 0 \\ 0 & c(a)^{-T} \end{pmatrix} : a \in \mathrm{GL}_n(E) \right\}, \qquad N = \left\{ n(\beta) = \begin{pmatrix} 1_n & \beta \\ 0 & 1_n \end{pmatrix} : \beta^{T} = c(\beta) \right\}. So Mβ‰…GLn(E)M \cong \mathrm{GL}_n(E), and NN is the additive group of Hermitian matrices. The Weyl element w=Jw = J of PP lies in U(W)\mathrm{U}(W), and PwP=PwNPwP = PwN is open and dense in U(W)\mathrm{U}(W).

With cc trivial these matrices give the Siegel parabolic of Sp2n\mathrm{Sp}_{2n}. For Hermitian VV, the doubled space VβŠ•(βˆ’V)V \oplus (-V) has the maximal isotropic subspace {(x,x)}\{(x, x)\}, hence a Siegel parabolic. The Siegel-Weil chapter lets m(a)m(a), n(Ξ²)n(\beta) and ww act on the Weil representation in TPar and TWeyl; the determinants chapter and the Bessel periods use Levi determinants and unipotent radicals.

Reductive groups and maximal compact subgroups

Definition. A linear algebraic group GG over kk is reductive if it is smooth and connected and GkˉG_{\bar{k}} has no nontrivial connected normal unipotent subgroup.

Example. GLn\mathrm{GL}_n, SLn\mathrm{SL}_n, Sp2n\mathrm{Sp}_{2n} and U(V)\mathrm{U}(V) are reductive. The Borel subgroup of GLn\mathrm{GL}_n, nβ‰₯2n \ge 2, is not: its unipotent matrices form a connected normal subgroup.

Definition. A maximal compact subgroup of a locally compact group GG (Chapter 2) is a compact subgroup contained in no larger compact subgroup.

Definition (Cartan involution). Let GβŠ‚GLn(β„‚)G \subset \mathrm{GL}_n(\mathbb{C}) be the real points of a reductive group over ℝ\mathbb{R}, stable under g↦gΛ‰Tg \mapsto \bar{g}^{T}. The Cartan involution of GG is ΞΈ(g)=(gΛ‰T)βˆ’1\theta(g) = (\bar{g}^{T})^{-1}. Its fixed points form the compact group K=G∩U(n)K = G \cap \mathrm{U}(n).

Theorem (Cartan). Let 𝔭={X∈Lie⁑(G):XΛ‰T=X}\mathfrak{p} = \{ X \in \operatorname{Lie}(G) : \bar{X}^{T} = X \}. Then KK is a maximal compact subgroup of GG, (k,X)↦kexp⁑(X)(k, X) \mapsto k \exp(X) is a diffeomorphism K×𝔭→GK \times \mathfrak{p} \to G (the Cartan decomposition), and each compact subgroup of GG is conjugate into KK (Knapp, Lie Groups Beyond an Introduction, VI.2).

Example. The maximal compact subgroups of GLn(ℝ)\mathrm{GL}_n(\mathbb{R}), GLn(β„‚)\mathrm{GL}_n(\mathbb{C}), Sp2n(ℝ)\mathrm{Sp}_{2n}(\mathbb{R}) and U(p,q)\mathrm{U}(p, q) are O(n)\mathrm{O}(n), U(n)\mathrm{U}(n), U(n)\mathrm{U}(n) and U(p)Γ—U(q)\mathrm{U}(p) \times \mathrm{U}(q). Over a nonarchimedean local field FvF_v, each maximal compact subgroup of GLn(Fv)\mathrm{GL}_n(F_v) is conjugate to GLn(π’ͺv)\mathrm{GL}_n(\mathcal{O}_v).

TLiftK asks that ΞΈw\theta_w be a Cartan involution for JJ at each infinite place, and the local matching fixes the maximal compact subgroup KInf.

Adelic points and compact open subgroups

Let FF be a number field (Chapter 4 defines adeles).

Definition. Let GG be a linear algebraic group over FF with a model over π’ͺF\mathcal{O}_F. Its group of adelic points G(𝔸F)G(\mathbb{A}_F) is the restricted product of the G(Fv)G(F_v) with respect to the G(π’ͺv)G(\mathcal{O}_v). It is locally compact, with G(F)G(F) a discrete subgroup.

Definition. Let FvF_v be nonarchimedean with uniformizer Ο–\varpi and residue field 𝔽q\mathbb{F}_q. The groups Km=1n+Ο–mMn(π’ͺv),mβ‰₯1,K_m = 1_n + \varpi^m M_n(\mathcal{O}_v), \qquad m \ge 1, are compact open subgroups of GLn(Fv)\mathrm{GL}_n(F_v) and form a basis of neighbourhoods of 11. The first congruence subgroup K1K_1 is the kernel of reduction GLn(π’ͺv)β†’GLn(𝔽q)\mathrm{GL}_n(\mathcal{O}_v) \to \mathrm{GL}_n(\mathbb{F}_q), and the Iwahori subgroup II is the inverse image of B(𝔽q)B(\mathbb{F}_q). So K1βŠ‚IβŠ‚GLn(π’ͺv)K_1 \subset I \subset \mathrm{GL}_n(\mathcal{O}_v) and I/K1β‰…B(𝔽q)I/K_1 \cong B(\mathbb{F}_q).

Theorem (Iwahori and Matsumoto, 1965). GLn(Fv)\mathrm{GL}_n(F_v) is the disjoint union of the double cosets IwII w I, with ww running over the matrices diag(Ο–Ξ»1,…,Ο–Ξ»n) s\mathrm{diag}(\varpi^{\lambda_1}, \dots, \varpi^{\lambda_n}) \, s, Ξ»βˆˆβ„€n\lambda \in \mathbb{Z}^n, ss a permutation matrix.

The law depth of IsRecT is about representations with a nonzero KOne-fixed vector, and the Galois row proves local-global compatibility at places with Iwahori fixed vectors.

Strong approximation

Theorem (strong approximation). Let GG be a simply connected, absolutely almost simple group over a number field FF, and SS a finite set of places with ∏v∈SG(Fv)\prod_{v \in S} G(F_v) not compact. Then G(F)G(F) is dense in the restricted product G(𝔸FS)G(\mathbb{A}_F^S) over the places outside SS (Kneser; Platonov; see Platonov and Rapinchuk, 7.4).

Example. SLn\mathrm{SL}_n, and SU(V)\mathrm{SU}(V) for dim⁑Vβ‰₯2\dim V \ge 2, satisfy the hypotheses, but U(V)\mathrm{U}(V) does not, as it maps onto the torus U(1)\mathrm{U}(1). For definite VV, SS must contain a finite place, as SU(V)(Fv)\mathrm{SU}(V)(F_v) is compact at each real place.

Remark. If moreover G(FβŠ—β„)G(F \otimes \mathbb{R}) is not compact and KβŠ‚G(𝔸F,f)K \subset G(\mathbb{A}_{F,f}) is compact open, then G(𝔸F)=G(F) G(FβŠ—β„) KG(\mathbb{A}_F) = G(F) \, G(F \otimes \mathbb{R}) \, K.

ThetaLocalInputs takes strong approximation for SU\mathrm{SU} from the literature as its field strongApproxSU.

Quotients of finite volume

Let GG be reductive over a number field FF. Then G(F)\G(𝔸F)G(F)\backslash G(\mathbb{A}_F) carries a G(𝔸F)G(\mathbb{A}_F)-invariant measure.

Definition. A fundamental domain for G(F)G(F) in G(𝔸F)G(\mathbb{A}_F) is a measurable set DD whose translates Ξ³D\gamma D, γ∈G(F)\gamma \in G(F), cover G(𝔸F)G(\mathbb{A}_F) and meet only in sets of measure zero. Then ∫G(F)\G(𝔸F)f=∫Df\int_{G(F)\backslash G(\mathbb{A}_F)} f = \int_D f for integrable G(F)G(F)-invariant ff.

Definition (Siegel set). Let P0=M0N0P_0 = M_0 N_0 be a minimal parabolic subgroup of GG over FF, A0A_0 the identity component of the real points of the largest split central torus of M0M_0, Ξ”0\Delta_0 the simple roots of A0A_0 in N0N_0, and KK a maximal compact subgroup of G(𝔸F)G(\mathbb{A}_F) with G(𝔸F)=P0(𝔸F)KG(\mathbb{A}_F) = P_0(\mathbb{A}_F) K. A Siegel set is 𝔖=ω A0(t) K,A0(t)={a∈A0:Ξ±(a)>tΒ forΒ eachΒ Ξ±βˆˆΞ”0},\mathfrak{S} = \omega \, A_0(t) \, K, \qquad A_0(t) = \{ a \in A_0 : \alpha(a) > t \text{ for each } \alpha \in \Delta_0 \}, with Ο‰βŠ‚N0(𝔸F)M0(𝔸F)1\omega \subset N_0(\mathbb{A}_F) M_0(\mathbb{A}_F)^1 compact and t>0t > 0.

Theorem (reduction theory). For tt small and Ο‰\omega large enough, G(𝔸F)=G(F) 𝔖G(\mathbb{A}_F) = G(F) \, \mathfrak{S} (Borel; Godement; see Moeglin and Waldspurger, I.2).

Theorem (Borel and Harish-Chandra, 1962). If GG has no nontrivial character G→𝔾mG \to \mathbb{G}_m over FF, then G(F)\G(𝔸F)G(F)\backslash G(\mathbb{A}_F) has finite volume, and it is compact if and only if GG has no proper parabolic subgroup over FF.

Example. Each character of U(V)\mathrm{U}(V) factors through det⁑\det onto U(1)\mathrm{U}(1), which has no nontrivial character over FF, so U(V)(F)\U(V)(𝔸F)\mathrm{U}(V)(F)\backslash \mathrm{U}(V)(\mathbb{A}_F) has finite volume. It is compact if VV is definite, since then VV has no isotropic vector and U(V)\mathrm{U}(V) no proper parabolic subgroup.

Definition (height). Let P=MNP = MN be a parabolic subgroup and KK a maximal compact subgroup of G(𝔸F)G(\mathbb{A}_F) with G(𝔸F)=P(𝔸F)KG(\mathbb{A}_F) = P(\mathbb{A}_F) K (an Iwasawa decomposition). For the Siegel parabolic of U(W)\mathrm{U}(W), the height of g=n(Ξ²)m(a)kg = n(\beta) m(a) k is a(g)=∣det⁑aβˆ£π”ΈEa(g) = |\det a|_{\mathbb{A}_E}. It is independent of the decomposition, a(pk)=a(p)a(pk) = a(p), and a(Ξ³g)=a(g)a(\gamma g) = a(g) for γ∈P(F)\gamma \in P(F) by the product formula.

On GLn(𝔸F)\mathrm{GL}_n(\mathbb{A}_F) the norm is βˆ₯gβˆ₯=∏vmax⁑i,j(∣gij∣v,∣(gβˆ’1)ij∣v)\|g\| = \prod_v \max_{i,j}(|g_{ij}|_v, |(g^{-1})_{ij}|_v), and a function on G(F)\G(𝔸F)G(F)\backslash G(\mathbb{A}_F) has moderate growth if it is bounded by a constant times a power of βˆ₯gβˆ₯\|g\|.

The Siegel-Weil chapter fixes a fundamental domain, the standard KK and a height, and uses that U(F)\U(𝔸)\mathrm{U}(F)\backslash \mathrm{U}(\mathbb{A}) has finite volume; the determinants chapter proves that a(g,1)ta(g, 1)^t is integrable, and HypsAlt is nonempty reaches Siegel sets through siegelSetUnitary.

Hermitian lattices and the mass formula

Let E/FE/F be a CM extension and VV a positive definite Hermitian space over EE of dimension nn, with form hh.

Definition. A Hermitian lattice is a finitely generated π’ͺE\mathcal{O}_E-submodule LβŠ‚VL \subset V that spans VV. Its dual is L#={x∈V:h(x,L)βŠ‚π’ͺE}L^\# = \{ x \in V : h(x, L) \subset \mathcal{O}_E \}, and LL is self-dual if L#=LL^\# = L.

Definition. Lattices L,Lβ€²L, L' are in the same class if Lβ€²=Ξ³LL' = \gamma L for some γ∈U(V)(F)\gamma \in \mathrm{U}(V)(F), and in the same genus if Lvβ€²=gvLvL'_v = g_v L_v for some gv∈U(V)(Fv)g_v \in \mathrm{U}(V)(F_v) at each finite place vv, where Lv=LβŠ—π’ͺFπ’ͺvL_v = L \otimes_{\mathcal{O}_F} \mathcal{O}_v.

With KLβŠ‚U(V)(𝔸F,f)K_L \subset \mathrm{U}(V)(\mathbb{A}_{F,f}) the stabilizer of ∏vLv\prod_v L_v, the classes in the genus of LL are in bijection with the finite set U(V)(F)\U(V)(𝔸F,f)/KL\mathrm{U}(V)(F)\backslash \mathrm{U}(V)(\mathbb{A}_{F,f}) / K_L (Shimura, 1964).

Definition. As VV is definite, each U(Lβ€²)={g∈U(V)(F):gLβ€²=Lβ€²}\mathrm{U}(L') = \{ g \in \mathrm{U}(V)(F) : g L' = L' \} is finite. The mass of the genus of LL is mass⁑(L)=βˆ‘[Lβ€²]1∣U(Lβ€²)∣,\operatorname{mass}(L) = \sum_{[L']} \frac{1}{|\mathrm{U}(L')|}, over the classes [Lβ€²][L'] in the genus. It is the volume of U(V)(F)\U(V)(𝔸F)\mathrm{U}(V)(F)\backslash \mathrm{U}(V)(\mathbb{A}_F) for the Haar measure with vol⁑(U(V)(FβŠ—β„) KL)=1\operatorname{vol}(\mathrm{U}(V)(F \otimes \mathbb{R}) \, K_L) = 1.

Definition (local density). Let vv be a finite place of FF with uniformizer Ο–\varpi and residue field of order qvq_v, let SS and TT be Hermitian matrices over π’ͺEv\mathcal{O}_{E_v} of sizes mm and nn, and let NkN_k be the number of X∈MmΓ—n(π’ͺEv/Ο–k)X \in M_{m \times n}(\mathcal{O}_{E_v}/\varpi^k) with XTS c(X)≑T(modΟ–k)X^{T} S \, c(X) \equiv T \pmod{\varpi^k}. The local density is Ξ±v(S,T)=lim⁑kβ†’βˆžqvk(n2βˆ’2mn)Nk.\alpha_v(S, T) = \lim_{k \to \infty} q_v^{k(n^2 - 2mn)} N_k.

Theorem (Siegel mass formula). Let LβŠ‚VL \subset V have Gram matrix SS, and let Ο‡\chi be the quadratic character of E/FE/F (Chapter 6). Then mass⁑(L)\operatorname{mass}(L) is an explicit constant times ∏vΞ±v(S,S)βˆ’1\prod_v \alpha_v(S, S)^{-1} over the finite places, and outside a finite set of places Ξ±v(S,S)βˆ’1\alpha_v(S, S)^{-1} is the vv-factor of ∏i=1nL(i,Ο‡i)\prod_{i=1}^{n} L(i, \chi^i) (Siegel, 1935, for quadratic forms; Gan, Hanke and Yu, 2001, for Hermitian lattices).

In its general form, Siegel's theorem writes a weighted average over the genus of the number of representations of TT by the Lβ€²L' as an explicit constant times a product of local densities over all places; the Siegel-Weil formula of Chapter 16 is its analytic form. GenusMassAt in HypsAlt is the mass of each positive definite Hermitian space of rank 22, TLiftK uses a self-dual Hermitian lattice, and SiegelWeilUnitaryConvergent in the Siegel-Weil chapter contains local densities.

The Heisenberg and metaplectic groups

Let WW be a symplectic space over a local field FF of characteristic 00.

Definition. The Heisenberg group of WW is H(W)=WΓ—FH(W) = W \times F with the product (w,t)(wβ€²,tβ€²)=(w+wβ€², t+tβ€²+12⟨w,wβ€²βŸ©).(w, t)(w', t') = \left(w + w', \, t + t' + \tfrac{1}{2} \langle w, w' \rangle \right). Its center is {0}Γ—F\{0\} \times F, and Sp(W)\mathrm{Sp}(W) acts on it by gβ‹…(w,t)=(gw,t)g \cdot (w, t) = (gw, t), trivially on the center.

For a nontrivial unitary character ψ\psi of FF, the Stone-von Neumann theorem (Chapter 16) gives one irreducible unitary representation ρψ\rho_\psi of H(W)H(W) with central character ψ\psi, up to isomorphism. As hβ†¦ΟΟˆ(gβ‹…h)h \mapsto \rho_\psi(g \cdot h) has the same central character, there is a unitary Ο‰(g)\omega(g), unique up to a scalar of absolute value 11, with Ο‰(g)ρψ(h)Ο‰(g)βˆ’1=ρψ(gβ‹…h)\omega(g) \rho_\psi(h) \omega(g)^{-1} = \rho_\psi(g \cdot h), a projective representation of Sp(W)\mathrm{Sp}(W).

Theorem (Weil, 1964). Let F≠ℂF \ne \mathbb{C} and W≠0W \ne 0. There is a nonsplit central extension 1→{±1}→Mp(W)→Sp(W)→11 \to \{\pm 1\} \to \mathrm{Mp}(W) \to \mathrm{Sp}(W) \to 1 such that the ω(g)\omega(g) can be chosen to form a true representation of Mp(W)\mathrm{Mp}(W).

Definition. Mp(W)\mathrm{Mp}(W) is the metaplectic group. For F=ℝF = \mathbb{R} it is the unique connected double cover of Sp2n(ℝ)\mathrm{Sp}_{2n}(\mathbb{R}) (Chapter 2). Chapter 16 defines its Weil representation Ο‰Οˆ\omega_\psi.

The Siegel-Weil chapter proves that the metaplectic representation intertwines the Heisenberg action, and the Bessel periods use the Heisenberg group.

Further reading

  • Platonov and Rapinchuk, Algebraic Groups and Number Theory: Ch. 4 (reduction theory, finite volume), Ch. 5 (adele groups), Ch. 7 (strong approximation), Ch. 8 (class numbers, genera).
  • Borel, Linear Algebraic Groups: Ch. IV and V (Borel and parabolic subgroups, Bruhat decomposition).
  • Knapp, Lie Groups Beyond an Introduction: Ch. VI (Cartan and Iwasawa decompositions, maximal compact subgroups).
  • Moeglin and Waldspurger, Spectral Decomposition and Eisenstein Series: I.1 and I.2 (adelic groups, heights, Siegel sets).

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