Open Mathematics Courses

English only for now

Compatible systems and global L-functions

Written by GPT-6.1 Sol (OpenAI) in Codex, Ultra setting, October 2026. Self-checked by the writing AI. Public domain (CC0).

An elliptic curve supplies a Tate module for every prime. These vector spaces have different coefficient fields, yet their Frobenius polynomials at a good prime are the same integer polynomial. The resulting Euler product is attached to the curve, rather than to a chosen Tate module. To include the bad primes, one must compare monodromy as well as Frobenius. To obtain a functional equation, one needs an additional analytic theorem.

1. The convention and the compatibility data

Let \(K\) be a number field. For a finite place \(v\), put \(q_v=\#k_v\) and let \(p(v)\) be its residue characteristic. Write \(\operatorname{Fr}_v\) for arithmetic Frobenius and \(\Phi_v=\operatorname{Fr}_v^{-1}\) for geometric Frobenius. The local norm and reciprocity convention remain \[ \|\Phi_v\|=q_v^{-1},\qquad \operatorname{Art}_v(\varpi_v)=\Phi_v\quad\text{in }W_{K_v}^{\mathrm{ab}}. \tag{1} \] In global Euler polynomials we use arithmetic Frobenius. Thus, for an unramified representation \(R\), \[ P_v(R,X)=\det(1-R(\operatorname{Fr}_v)X),\qquad L_v(s,R)=P_v(R,q_v^{-s})^{-1}. \tag{2} \] At such a place the local factor defined earlier using geometric Frobenius satisfies \[ L_v(s,R)=L\bigl(s,\operatorname{WD}_v(R^\vee)\bigr). \tag{3} \] Indeed, the eigenvalues of \(R^\vee(\Phi_v)\) are those of \(R(\operatorname{Fr}_v)\). The dual in (3) is essential.

Let \(E\) be a number field of coefficients and \(\lambda\) range over its finite places, with residue characteristic \(\ell(\lambda)\). Consider continuous representations \[ R_\lambda:G_K\longrightarrow\mathrm{GL}_d(E_\lambda). \] We use the following finite-place notions of compatibility. They make no assertion about a local representation at a place above its coefficient prime; that requires the theory introduced in the p-adic Hodge lesson.

Definition 1.1 (weak compatibility). There is a finite set \(S\) of finite places of \(K\), and for each \(v\notin S\) a polynomial \(P_v\in E[X]\), with constant term one and degree \(d\), such that whenever \(\ell(\lambda)\ne p(v)\), the representation \(R_\lambda\) is unramified at \(v\) and its polynomial (2) is \(P_v\).

Definition 1.2 (strict compatibility). In addition, for every finite \(v\) there is a Frobenius-semisimple Weil–Deligne isomorphism class \(D_v\) over \(\overline E\), invariant under \(\operatorname{Gal}(\overline E/E)\), whose scalar extensions give \[ \operatorname{WD}_v(R_\lambda)^{\mathrm{F\!ss}} \simeq D_v\otimes_{\overline E,\iota_\lambda}\overline E_\lambda \qquad\bigl(\ell(\lambda)\ne p(v)\bigr) \tag{4} \] for every embedding \(\iota_\lambda\) extending the coefficient embedding. At \(v\notin S\), this class is unramified and \(\det(1-D_v(\Phi_v^{-1})X)=P_v(X)\).

Invariance of an isomorphism class is weaker than the existence of matrices over \(E\). Here is a concrete proof of that distinction. The quaternion group has a two-dimensional complex representation with generators \[ I=\begin{pmatrix}i&0\\0&-i\end{pmatrix}, \qquad J=\begin{pmatrix}0&1\\-1&0\end{pmatrix}. \] Its character has values \(2,-2,0\), with corresponding characteristic polynomials \((T-1)^2,(T+1)^2,T^2+1\), all rational. The characteristic-polynomial Brauer–Nesbitt theorem proved in the first lesson therefore makes its class Galois-invariant. It has no real, hence no rational, two-dimensional realization. Indeed a real matrix with square \(-1\) has a basis in which it is \(\begin{psmallmatrix}0&-1\\1&0\end{psmallmatrix}\). A real matrix anticommuting with it has form \(\begin{psmallmatrix}a&b\\b&-a\end{psmallmatrix}\), whose square is \((a^2+b^2)1\), never \(-1\). This is an example of a Schur-index obstruction.

Definition 1.2 allows that obstruction. Isomorphic local pairs have the same invariant spaces, Frobenius determinants and nilpotent block lengths, so it supplies coefficient-independent local factors, conductors and monodromy lengths; the conductor integer is proved in the conductor lesson. Taylor uses “strongly compatible” for the comparison at every prime, within a definition that also includes de Rham and Hodge–Tate conditions [Taylor, §1, pp.81–83]. Those additional conditions are separate from our finite-place definitions.

Why is the stronger condition useful? The local pairs \[ (1\oplus\|\cdot\|,0),\qquad \mathrm{Sp}(2) \tag{5} \] have the same underlying Frobenius polynomial. Their local factors are respectively \[ \frac{1}{(1-q^{-s})(1-q^{-s-1})},\qquad \frac{1}{1-q^{-s-1}}, \] and their conductors are zero and one. Monodromy accounts for the difference. This is a local illustration of missing information, not a claimed example of two global semisimple systems with the same good-prime data.

Definition 1.3 (arithmetic weight). A weakly compatible system is pure of weight \(w\) if the inverse roots \(\alpha_{v,j}\) in \[ P_v(X)=\prod_{j=1}^d(1-\alpha_{v,j}X) \] satisfy \(|\sigma(\alpha_{v,j})|=q_v^{w/2}\) for every complex embedding \(\sigma\), at all \(v\notin S\). Enlarging \(S\) is allowed. Weights here refer to arithmetic Frobenius. The covariant elliptic Tate module has weight \(1\); its dual has arithmetic weight \(-1\), and geometric Frobenius on that dual has weight \(1\). The cyclotomic character \(\chi_\ell\) has arithmetic weight \(2\), since \(\chi_\ell(\operatorname{Fr}_v)=q_v\). Tensoring by \(\chi_\ell^m\) adds \(2m\) to an arithmetic weight. In the geometric convention it subtracts \(2m\).

2. The elliptic system and its exceptional set

Write \(A/K\) for an elliptic curve, reserving \(E\) for coefficients. The system \((V_\ell A)_\ell\) has coefficient field \(\mathbf Q\). Let \(S_A\) be the finite set of places of bad reduction. Its finiteness follows by choosing one global Weierstrass equation: outside the finitely many denominators of its coefficients and prime divisors of its nonzero discriminant, its integral projective cubic is smooth and gives a good model, as proved in the Tate-module lesson.

Theorem 2.1 (good-prime independence). At a good place \(v\), for every \(\ell\ne p(v)\), \[ P_v(V_\ell A,X)=1-a_vX+q_vX^2, \qquad a_v=q_v+1-\#\widetilde A(k_v). \tag{6} \] In particular \((V_\ell A)_\ell\) is weakly compatible and pure of arithmetic weight \(1\), with exceptional set \(S_A\).

Proof. Good reduction identifies the prime-to-\(p(v)\) torsion with that of \(\widetilde A\), equivariantly for arithmetic Frobenius. The Tate-module lesson proved that the Frobenius endomorphism has determinant \(q_v\), and that \(\deg(1-\operatorname{Fr}_v)=\#\widetilde A(k_v)\). Consequently its trace is \(q_v+1-\#\widetilde A(k_v)\), proving (6). Each quantity comes from the curve over its residue field and is independent of \(\ell\).

Here is the required Hasse inequality. For any integers \(m,n\), the degree–determinant theorem, valid also for inseparable endomorphisms, gives \[ 0\le\deg([m]+[n]\operatorname{Fr}_v) =m^2+a_vmn+q_vn^2. \tag{6A} \] If \(a_v^2>4q_v\), the real polynomial \(z^2+a_vz+q_v\) is negative on an open interval. Choose a rational number \(m/n\) in that interval and multiply by \(n^2\), contradicting (6A). Thus \(|a_v|\le2\sqrt{q_v}\). The roots of \(T^2-a_vT+q_v\) are consequently a complex-conjugate pair, or the repeated real root \(\pm\sqrt{q_v}\); their product is \(q_v\), so each has absolute value \(\sqrt{q_v}\). Every embedding gives the same two roots. This proves purity. ∎

We next prove the comparison at potentially good primes in arbitrary dimension. The extra geometry required at semistable primes is identified separately below.

Lemma 2.2 (endomorphisms have one characteristic polynomial). Let \(B/k\) be an abelian variety of dimension \(g\), over any field, and \(u\in\operatorname{End}(B)\). There is a monic \(P_u(T)\in\mathbf Q[T]\) of degree \(2g\) whose image in every \(\mathbf Q_\ell[T]\), \(\ell\ne\operatorname{char}k\), is \(\det(T-u\mid V_\ell B)\). The same conclusion holds for quasi-endomorphisms, with rational coefficients.

Proof. We use the actual earlier geometric proofs in Abelian varieties: the normalized cube, Proposition 4.4; the multivariable Euler polynomial, Lemma 5.2; Euler homogeneity, Proposition 5.5; finite-flat Euler comparison, Corollary 5.6; and isogeny kernels and prime-to-characteristic Tate torsion, §§6–7. Here are the remaining steps, including the coefficient comparison that a citation alone would not supply.

Extend the ground field to its algebraic closure. Choose a symmetric ample bundle \(H\). Define \(\delta(f)=\deg(f)\) for isogenies, and zero for other endomorphisms. Then \[ \delta(f)=\frac{\chi(B,f^*H)}{\chi(B,H)}. \tag{6B} \] For an isogeny this is Corollary 5.6. Otherwise factor \(f\) through its image, an abelian subvariety of dimension less than \(g\). Proper pushforward and the Euler-polynomial degree bound show that \(\chi(B,(f^*H)^a)\) has degree less than \(g\) in \(a\). Euler homogeneity says it is \(a^g\chi(B,f^*H)\), so its coefficient is zero. The denominator in (6B) is positive by Proposition 5.5.

For fixed \(f,h\), pull the cube back along \((nf+h,f,f)\). It says that the classes of \((nf+h)^*H\) have constant second difference; they are quadratic polynomials in \(n\) with values in the rational Picard group. The multivariable Euler polynomial has total degree at most \(g\). Formula (6B) therefore makes \(\delta(nf+h)\) a rational polynomial of degree at most \(2g\). Moreover \(\delta(nf)=n^{2g}\delta(f)\), by the degree of multiplication.

These assertions extend \(\delta\) to \(\operatorname{End}(B)\otimes\mathbf Q\) by \(\delta(f/n)=n^{-2g}\delta(f)\), independently of the denominator. They make it a homogeneous polynomial function of degree \(2g\) on every finite-dimensional subspace. To check the last assertion, choose coordinates on such a subspace and interpolate in each coordinate at \(2g+1\) distinct rational values. The line-degree bound produces a polynomial in all coordinates. Substituting arbitrary rational affine lines forces its total degree to be at most \(2g\); homogeneity forces its nonzero terms to have degree \(2g\). Multiplicativity \(\delta(fh)=\delta(f)\delta(h)\) holds: degrees multiply for isogenies, and a composition containing a map with smaller-dimensional image again has smaller-dimensional image. Polynomial interpolation extends this identity after any extension of the rational coefficient field.

Set \(P_u(T)=\delta(T-u)\). Its leading coefficient is \(\delta(1)=1\), and its degree is \(2g\). If \(a_i\) are its roots, multiplicativity after scalar extension gives, for any \(F\in\mathbf Q[T]\), \[ \delta(F(u))=\prod_{i=1}^{2g}F(a_i). \tag{6C} \] Indeed, factor \(F(T)=c\prod_j(T-b_j)\), apply multiplicativity and homogeneity to \(F(u)\), and interchange the two finite products. The even degree \(2g\) removes the possible sign.

For an isogeny \(h\), its kernel has \(\ell\)-primary order \(\ell^{v_\ell(\deg h)}\), for \(\ell\ne\operatorname{char}k\). Here are the scheme-length details. The actual isogeny theorem identifies its degree with the rank of its finite kernel, and its rank-annihilation lemma kills that kernel by this integer. Chinese-remainder multiplication idempotents split it into its primary factors. A factor killed by a power of \(\ell\) is a subgroup of the finite étale group \(B[\ell^r]\), so is étale. For a different primary factor, its geometric point group has order a power of its own prime. Its connected part is trivial in characteristic zero. In characteristic \(p>0\), its connected part is killed by a power of \(p\): multiplication by every prime-to-\(p\) integer is étale with trivial connected kernel, so is an automorphism on that finite connected part. Its rank is a power of \(p\), since it is a subgroup of \(B[p^r]\) and the finite subgroup quotient is a torsor whose ranks multiply. Thus no other factor contributes to the \(\ell\)-valuation of the rank. This proves the claimed \(\ell\)-primary order, including inseparable isogenies.

The \(\ell\)-primary point kernel is the kernel on \((\mathbf Q_\ell/\mathbf Z_\ell)^{2g}\). Diagonalizing the integral map \(T_\ell h\) by the DVR module theorem identifies this kernel with its finite lattice cokernel. Consequently \[ |\delta(h)|_\ell=|\det(T_\ell h)|_\ell. \tag{6D} \] For a nonisogeny both sides are zero: its positive-dimensional reduced kernel has nonzero rational Tate module. The Tate ranks and the finite-kernel length, including its inseparable characteristic-primary part, are exactly the earlier kernel results just specified. Thus (6D) applies to every \(F(u)\) with integral \(F\).

Let \(b_i\) be the eigenvalues of \(T_\ell u\). Equations (6C)–(6D) imply \[ \left|\prod_iF(a_i)\right|_\ell =\left|\prod_iF(b_i)\right|_\ell \quad(F\in\mathbf Z[T]). \tag{6E} \] Continuity, and then clearing a common denominator, give the same equality for \(F\in\mathbf Q_\ell[T]\). We verify that it forces equality of the two monic polynomials. For any algebraic \(z\) with minimal polynomial \(F_z\) over \(\mathbf Q_\ell\), Galois invariance of each multiset and uniqueness of the extended valuation give \[ \left|\prod_iF_z(a_i)\right|_\ell =|P_u(z)|_\ell^{[\mathbf Q_\ell(z):\mathbf Q_\ell]}, \] and the analogous identity for \(Q(T)=\prod_i(T-b_i)\). Hence \(|P_u(z)|_\ell=|Q(z)|_\ell\). Approach a root \(a\) by algebraic \(z=a+\ell^j c\) avoiding the finitely many roots. For large \(j\), all distances to roots unequal to \(a\) are constant. The valuation of \(P_u(z)\) has slope equal to the multiplicity of \(a\), and that of \(Q(z)\) has slope equal to its multiplicity there. Equality forces those multiplicities to agree. Applying this to every root proves \(P_u=Q\). Finally clear a denominator to obtain the quasi-endomorphism assertion. ∎

Lemma 2.3 (potentially good compatibility, in every dimension). If an abelian variety \(B/F\), over a characteristic-zero nonarchimedean local field, acquires good reduction over a finite extension, then all its \(V_\ell B\), \(\ell\ne p(F)\), have one Galois-invariant Frobenius-semisimple Weil–Deligne class over \(\overline{\mathbf Q}\); its monodromy is zero.

Proof. Choose a finite Galois extension \(F'/F\) of good reduction. Functoriality and uniqueness of the Néron model extend the descent action to its abelian scheme over \(\mathcal O_{F'}\). Reduction of invertible torsion is equivariant and identifies its Tate module with that of the special abelian variety \(B_0\). This is the written Néron models, Lemma 7.1, applied after strict henselization; properness identifies every generic torsion point with a model section. The common subgroup \(I_{F'}\) consequently acts trivially, so \(N_\ell=0\).

An element \(w\in W_F\) of nonnegative residue exponent acts on the special torsion through an algebraic endomorphism: compose the relative residue Frobenius with the descent isomorphism from the Frobenius-twisted special fibre back to \(B_0\). On geometric points this is precisely the semilinear descent action of \(w\). Relative Frobenius is an isogeny, and the descent map is an isomorphism. A negative exponent is handled by the inverse quasi-isogeny. Lemma 2.2 therefore makes the trace, and indeed the characteristic polynomial, of every \(w\) rational and independent of \(\ell\). This compares \(w\) with its inertia component present; it is more information than the trace of an unramified Frobenius power.

For completeness, this character comparison gives a common algebraic isomorphism class. The fixed finite group \(I_F/I_{F'}\) works for every coefficient prime, without requiring that its action be faithful. Some power \(\Phi^m\) centralizes it. The unipotent part of \(\Phi\) consequently commutes with inertia: its \(m\)-th power is the unipotent part of that centralizing power, and the unipotent \(m\)-th root is its finite binomial polynomial. It commutes with the semisimple part of \(\Phi\) as well. Removing it preserves all traces of \(i\Phi^a\): the product of its nilpotent part with any commuting operator has trace zero, as follows by the filtration by kernels of its powers. After Frobenius semisimplification, \(\Phi^m\) is semisimple, with algebraic eigenvalues by Lemma 2.2. On an eigenspace with eigenvalue \(c\), choose an algebraic \(m\)-th root \(b\) of \(c\), and replace \(\Phi\) there by \(b^{-1}\Phi\). Inertia together with this operator is a representation of a finite cyclic extension of its finite inertia group, because the new operator has \(m\)-th power one. Over an algebraically closed field of characteristic zero, the group algebra of a finite group splits into matrix blocks: averaging complements gives semisimplicity, and Schur's lemma identifies a block with the full matrix algebra on its simple module. The simple modules can all be defined over \(\overline{\mathbf Q}\), by splitting that finite group algebra there. Thus the original semisimple Weil representation has an algebraic model.

Equal traces on every group element force equal simple multiplicities. One can see this without choosing complex embeddings: for the finite-dimensional semisimple image algebra, its central idempotents and matrix units isolate each simple block, and equality of trace functionals on their spanning group elements gives its multiplicity. This is also the characteristic-zero case of the proved Brauer–Nesbitt theorem in the first lesson. The common rational traces are fixed by every automorphism of \(\overline{\mathbf Q}\), so the common isomorphism class is Galois-invariant. ∎

Theorem 2.4 (full local abelian compatibility; remaining geometric proof obligation). For every abelian variety over a nonarchimedean local field of characteristic zero, its prime-to-residue-characteristic rational Tate modules have a common Galois-invariant Frobenius-semisimple Weil–Deligne class over \(\overline{\mathbf Q}\).

The potentially good case has the full proof in Lemma 2.3. General semistable reduction is actually proved in Néron models, Theorem 8.19, with identity-component base change in Theorem 8.20. Those results alone do not identify the monodromy filtration. The remaining proof must identify its toric graded pieces and their monodromy pairing and compare the abelian graded piece, equivariantly for the finite descent group. No actual earlier proof of that general monodromy assertion has yet been bound here. Deligne's freely accessible Example 8.10, printed p.571, states the full theorem but supplies no proof of it; it does not discharge this obligation. The theorem is retained in full generality, and this edition does not claim its full proof complete.

For an elliptic curve the local classification gives every case without invoking the unproved higher-dimensional part. At good places use (6) and \(N=0\). At potentially good places use Lemma 2.3. At potentially multiplicative places use the Tate-curve proof and its quadratic descent in Elliptic curves over local fields and their Weil–Deligne representations, §§3–5: the common quadratic splitting character \(\eta_v\) and the common block are \[ D_v=\eta_v\otimes\mathrm{Sp}(2); \tag{7} \] The quadratic character depends on the curve, not on \(\ell\). These cases prove strict compatibility for the elliptic system. They preserve the higher-dimensional theorem and its outstanding proof; they do not replace it with an elliptic statement.

The cyclotomic system has \(S=\varnothing\), \(P_v(X)=1-q_vX\), and common local pair \((\|\cdot\|,0)\). It is therefore strictly compatible and has weight \(2\). The coefficient prime is excluded separately in (4); it is not added permanently to \(S\) for each member of the family.

For comparison, a finite-image Artin representation defined over a number field of coefficients gives a strictly compatible system by scalar extension. Its exceptional set consists of its ramified places, its local monodromy is zero, and its weight is zero, because every Frobenius eigenvalue is a root of unity.

3. Artin L-functions, including the missing places

Let \(\rho:G_K\to\mathrm{GL}_d(\mathbf C)\) have finite image. Its local factor at a finite place is \[ L_v(s,\rho)= \det\bigl(1-\rho(\operatorname{Fr}_v)q_v^{-s}\mid V^{I_v}\bigr)^{-1}. \tag{8} \] A lift of Frobenius acts unambiguously on the invariants. With our geometric local convention this is \(L(s,\rho_v^\vee|_{W_{K_v}})\). Finite inertia averaging identifies \((V^{I_v})^\vee\) with \((V^\vee)^{I_v}\), so this statement also holds at ramified places.

Proposition 3.1 (formalism at every prime). Direct sums multiply Artin L-functions, inflation preserves them, and for every finite extension \(M/K\), \[ L_K(s,\operatorname{Ind}_{G_M}^{G_K}\tau)=L_M(s,\tau). \tag{8A} \] All these products converge absolutely and locally uniformly, and are nonzero, on \(\Re s>1\).

Proof. Eigenvalues of finite-image operators are roots of unity. The absolute logarithmic series at \(v\) is bounded by \(d\sum_{m\ge1}q_v^{-m\sigma}/m\). If \(n=[K:\mathbf Q]\), at most \(n\) places lie above a rational prime \(p\), each with norm at least \(p\). Thus its global sum is bounded by \(n d\sum_p\sum_{m\ge1}p^{-m\sigma}/m\), finite for \(\sigma>1\). Fixing \(\sigma_0>1\) below the real parts on a compact parameter set gives uniform convergence. Exponentiating proves holomorphy and nonvanishing.

Direct sums give block determinants. In a quotient Galois group the decomposition and inertia groups map onto their corresponding groups and Frobenius maps to Frobenius; hence inflation preserves the invariant space and its operator.

For induction choose a finite Galois extension containing \(M\) and killing \(\tau\). Write its group as \(G\), the subgroup fixing \(M\) as \(H\), and local decomposition and inertia as \(D,I\). The coset model gives \[ \operatorname{Res}_D\operatorname{Ind}_H^G V =\bigoplus_{g\in D\backslash G/H}\operatorname{Ind}_{J_g}^{D}V_g, \qquad J_g=D\cap gHg^{-1}. \tag{8B} \] Indeed, partition the coset basis into \(D\)-orbits; the stabilizer and its fibre are \(J_g,V_g\). The double cosets are the places \(u\) of \(M\) above \(v\).

In one summand, an inertia-invariant vector on each \(I\)-orbit is determined by an initial vector fixed by \(I\cap J_g\). Hence the invariant space consists of \(f=[D:IJ_g]\) copies of \(V_g^{I\cap J_g}\). Arithmetic Frobenius cyclically permutes them, and after \(f\) steps acts by arithmetic Frobenius \(A\) at \(u\). For such a cyclic block operator \(\Psi\), \[ \det(1-T\Psi)=\det(1-T^f A). \tag{8C} \] Choose bases making its first \(f-1\) transition maps identities. Block elimination of \(1-T\Psi\) leaves identity diagonal blocks and the final block \(1-T^fA\). The calculation over \(\mathbf C(T)\) proves the polynomial identity. Since \(q_u=q_v^f\), the summand contributes exactly the factor at \(u\). Multiplying over double cosets proves (8A), even when \(I\ne1\). Absolute convergence justifies the global multiplication. ∎

The conductor ideal is \[ \mathfrak f(\rho)=\prod_{v<\infty}\mathfrak p_v^{a_v(\rho)}, \qquad C(\rho)=|D_K|^d\operatorname N_{K/\mathbf Q}\mathfrak f(\rho). \tag{9} \] Its integer exponents are proved in Conductors of Weil group representations, Theorem 2.6, before the ideal is defined. That lesson's Proposition 2.3 proves the finite induction formula and Lemma 1.4 proves different transitivity and the trace-dual discriminant formula. They imply \[ \begin{aligned} \mathfrak f_K(\operatorname{Ind}_M^K\tau) &=\mathfrak d_{M/K}^{\dim\tau} \operatorname N_{M/K}\mathfrak f_M(\tau),\\ |D_M|&=|D_K|^{[M:K]} \operatorname N_{K/\mathbf Q}\mathfrak d_{M/K},\\ C_K(\operatorname{Ind}_M^K\tau)&=C_M(\tau). \end{aligned} \tag{9A} \] Here is the global passage, including the residue factors. At \(v\), (8B) partitions induction into places \(u\mid v\). Its conductor exponent is \(\sum_{u\mid v}(\dim\tau\, f(u/v)d_{M_u/K_v}+f(u/v)a_u(\tau))\). The first sum is the valuation of the relative discriminant, and the second is the valuation of the ideal norm of \(\mathfrak f_M(\tau)\). This proves the first formula. For the second, complete the integral trace lattice at a rational prime. It is the product of the completed integer lattices above that prime; traces and trace determinants respect that product. The tower identity for their trace-dual differents, followed by their norm exponents \(f d\), gives the displayed valuation identity at each rational prime. Equality of these positive integers follows. Insert the first two identities into (9) to obtain the third.

The archimedean factors use \[ \Gamma_{\mathbf R}(s)=\pi^{-s/2}\Gamma(s/2),\qquad \Gamma_{\mathbf C}(s)=2(2\pi)^{-s}\Gamma(s). \] At a real place let \(d_v^+\) and \(d_v^-\) be the multiplicities of \(1\) and \(-1\) for complex conjugation. The factor is \[ \Gamma_{\mathbf R}(s)^{d_v^+}\Gamma_{\mathbf R}(s+1)^{d_v^-}. \tag{10} \] At a complex place it is \(\Gamma_{\mathbf C}(s)^d\). These are the archimedean Weil-group factors of the local-factor lesson. If \(\Gamma_\infty(s,\rho)\) denotes their product, the completed function \[ \Lambda(s,\rho)=C(\rho)^{s/2}\Gamma_\infty(s,\rho)L(s,\rho) \tag{11} \] has the functional equation \[ \Lambda(s,\rho)=W(\rho)\Lambda(1-s,\rho^\vee),\qquad |W(\rho)|=1. \tag{12} \] We now prove this assertion. The analytic prerequisite is an actual written proof, rather than a reference to a free theorem statement.

Lemma 3.2 (the Hecke theorem and its normalization). For any number field \(M\) and continuous unitary Hecke character \(\omega:\mathbb A_M^\times/M^\times\to S^1\), let \(\mathfrak f_\omega\) be its intrinsic finite conductor and \(Q_\omega=|D_M|N\mathfrak f_\omega\). At real places write \(\omega_v=\operatorname{sgn}^{e_v}|\cdot|^{it_v}\), and at complex places \(\omega_v=(z/|z|_{\mathrm{usual}})^{n_v}|\cdot|_v^{it_v}\). Set \[ \Gamma_\infty(s,\omega)= \prod_{v\text{ real}}\Gamma_{\mathbf R}(s+it_v+e_v) \prod_{v\text{ complex}}\Gamma_{\mathbf C}(s+it_v+|n_v|/2). \tag{12A} \] Then \(\Lambda_M(s,\omega)=Q_\omega^{s/2}\Gamma_\infty(s,\omega)L_M(s,\omega)\) continues meromorphically and satisfies \[ \Lambda_M(s,\omega)=W_M(\omega)\Lambda_M(1-s,\omega^{-1}), \qquad |W_M(\omega)|=1. \tag{12B} \] It is entire unless \(\omega=|\cdot|^{i\tau}\), in which case its only poles are simple poles at \(-i\tau,1-i\tau\). The constant is the product of its normalized rank-one local constants.

Proof with its earlier analytic locators. The written programme lessons Idèles and the idèle class group, Theorem 3.3, and Additive characters, self-dual measures and Poisson summation on the adèles, Theorems 5.4–5.5, prove compactness of \(C_M^1\), adelic self-duality, covolume one and Poisson summation. The full continuation argument is actually written in Tate's global theory: continuation and functional equation, Proposition 9.1, its Theta lemma, and Theorem 9.2, equations (15)–(24). Its proof unfolds a Schwartz integral \[ Z(f,\omega,s)=\int_{\mathbb A_M^\times} f(x)\omega(x)|x|^s\,d^\times x \] to the idèle class group. The large-norm part \(G(f,\omega,s)\) is entire: compact norm-one representatives reduce its theta sum to one fractional-ideal lattice, and Schwartz decay bounds the nonzero sum by \(O(|x|^{-B})\) for every \(B\). Poisson summation on the small-norm part gives the actual continuation \[ Z(f,\omega,s)=G(f,\omega,s) +G(\widehat f,\omega^{-1},1-s) +\begin{cases} \kappa\bigl(\widehat f(0)/(s+i\tau-1)-f(0)/(s+i\tau)\bigr), &\omega=|\cdot|^{i\tau},\\ 0,&\omega|_{C_M^1}\ne1. \end{cases} \tag{12C} \] Translation in the compact group kills its average unless the character is trivial there. Fourier inversion interchanges the two entire terms and preserves the correction, proving \(Z(f,\omega,s)=Z(\widehat f,\omega^{-1},1-s)\). This use of the earlier theta estimate includes arbitrary number fields and the whole Schwartz space, not just \(\mathbf Q\).

Take \(f_v=1_{\mathcal O_v}\) at unramified finite places, \(f_v=\omega_v^{-1}1_{\mathcal O_v^\times}\) at ramified places, \(x^{e_v}e^{-\pi x^2}\) at real places, and the matching angular Gaussian at complex places. The actual local integral proofs are Tate's local theory at the finite places, Theorems 7.2–7.3, and Tate's local theory at the infinite places, Theorem 8.1 and Propositions 8.2–8.4. They make \(Z(f,\omega,s)\) exactly \(\Gamma_\infty(s,\omega)L_M(s,\omega)\), with no extra scalar.

For the trace character, let \(d_v\) be the different exponent and \(a_v\) the character conductor. The finite formulas give \[ \epsilon_v(s,\omega_v)=W_vq_v^{(a_v+d_v)(1/2-s)},\qquad |W_v|=1. \tag{12D} \] For a ramified character, \(v(b)=a_v+d_v\) and \[ W_v=\omega_v(b)q_v^{-a_v/2} \sum_{u\in(\mathcal O_v/\mathfrak p_v^{a_v})^\times} \omega_v(u)^{-1}\psi_v(u/b). \tag{12E} \] The finite Fourier calculation in Theorem 7.3 proves that the sum has absolute value \(q_v^{a_v/2}\); changing \(b\) by a unit cancels between its two factors. For an unramified character \(W_v=\omega_v(\delta_v)\), where \(\delta_v\) generates the different. Infinite constants for the negative trace character are \((-i)^{e_v}\) and \((-i)^{|n_v|}\), proved by Gaussian Fourier transforms in Proposition 8.3. Product trace-dual discriminants give \(\prod_{v<\infty}q_v^{d_v}=|D_M|\); hence the product of local constants is \(W_M(\omega)Q_\omega^{1/2-s}\). Apply the local equations to (12C), then multiply by \(Q_\omega^{s/2}\), to obtain (12B).

Formula (12C) gives no poles for characters nontrivial on \(C_M^1\). A character trivial there is a pure norm character, so its standard Gaussian is a shifted trivial-character test with both zero terms nonzero. Its two stated poles therefore have nonzero residues. This completes the same full Hecke proof written in Hecke L-functions and the Dedekind zeta function, Theorem 10.1 and equations (3)–(8). The freely readable Tate thesis, §4.4, Theorem 4.4.11 and its proof, is primary reading material; the proof obligation is discharged by the actual written arguments just identified. ∎

Corollary 3.2a (all continuous complex Hecke characters). Every continuous \(\theta:C_M\to\mathbb C^\times\) has the form \(\theta=\omega|\cdot|^a\) with \(\omega\) unitary and \(a\in\mathbb R\). Its completed primitive Hecke function continues meromorphically and satisfies \[ \Lambda_M(s,\theta)=Q_\theta^{-a}W_M(\omega) \Lambda_M(1-s,\theta^{-1}). \tag{12B'} \] Its analytic constant equals the product of its normalized local constants, and has absolute value \(Q_\theta^{-a}\). It is entire unless \(\theta=|\cdot|^{a+i\tau}\); in that case its poles are simple, at \(-a-i\tau\) and \(1-a-i\tau\).

Proof. The absolute value of a continuous character on the compact group \(C_M^1\) is one: its image is a compact subgroup of \(\mathbb R_{>0}\), whose logarithm is a compact additive subgroup of \(\mathbb R\), necessarily zero since the multiples of any nonzero member are unbounded. The norm splitting from the actual idèle-class proof identifies the remaining positive real character with \(t\mapsto t^a\). Indeed its logarithm is a continuous additive function on \(\mathbb R\); rational linearity follows by integer division and real linearity by density. Dividing \(\theta\) by that norm power produces \(\omega\).

Unit restrictions, and hence the conductor, are unchanged. Every Euler and Gamma factor is shifted from \(s\) to \(s+a\), so \(\Lambda(s,\theta)=Q^{-a/2}\Lambda(s+a,\omega)\). Similarly \(\Lambda(1-s,\theta^{-1})=Q^{a/2}\Lambda(1-s-a,\omega^{-1})\). Lemma 3.2 now gives (12B') and its poles. Its finite local exponent formula gives the central-product multiplier \(Q^{-a}\); infinite phases are unchanged by radial shifts. Thus the local product gives exactly the same constant. ∎

Theorem 3.3 (Artin meromorphy and functional equation). Equations (11)–(12) hold for every finite-image complex representation of \(G_K\), including reducible representations and all ramified places.

Proof. Take a finite Galois quotient \(G\) through which \(\rho\) factors. The actual integer line-induction theorem in the earlier Brauer's induction theorem, Theorem 5.1, gives \[ [\rho]=\sum_jm_j[\operatorname{Ind}_{H_j}^G\chi_j], \qquad m_j\in\mathbf Z,\quad\dim\chi_j=1. \tag{12F} \] Its proof in §§1–5 proves nilpotent monomiality, local induction ideals at every prime, integral descent and the final Bezout combination. Thus this is integer induction from lines, with negative coefficients permitted; rational Artin induction would not suffice. Put \(M_j=L^{H_j}\). The one-dimensional character gives the finite-order Hecke character \(\omega_j\) by arithmetic reciprocity. The actual local-to-global reciprocity and character construction are the preceding Weil lesson, §6 and Proposition 6.1, with its written class-field provider. In that provider's geometric convention apply the correspondence to \(\chi_j^\vee\): geometric reciprocity is the inverse of arithmetic reciprocity, so the two inversions give precisely \(\omega_j(x)=\chi_j(\operatorname{rec}_{\mathrm{arith}}(x))\). At a ramified place the Artin invariant line is zero exactly when the Hecke character is nontrivial on units. At an unramified place arithmetic reciprocity sends a uniformizer to arithmetic Frobenius. Hence their primitive Euler factors agree in both cases. The unit conductor equals the Artin conductor by the proved conductor lesson, Lemma 2.4.

Proposition 3.1 now gives \(L_K(s,\rho)=\prod_jL_{M_j}(s,\omega_j)^{m_j}\) on \(\Re s>1\). Equation (9A) gives equality of the conductor scales. At a real place, induction from a complex place gives one even and one odd line; induction from real places preserves their signs. At a complex place all dimensions simply add. Thus induction preserves the archimedean product, because \[ \Gamma_{\mathbf R}(s)\Gamma_{\mathbf R}(s+1)=\Gamma_{\mathbf C}(s). \tag{12G} \] This identity has an actual proof in the infinite Tate lesson, Proposition 8.4: the beta integral, the substitution \(t=(1+a)/2\), and then \(a^2=b\) give the Gamma duplication formula, and \(\Gamma(1/2)=\sqrt\pi\) follows from the Gaussian integral. Consequently \[ \Lambda_K(s,\rho)=\prod_j\Lambda_{M_j}(s,\omega_j)^{m_j}. \tag{12H} \] Lemma 3.2 continues this product meromorphically. The finite coset pairing identifies the dual of an induction with induction of the dual: pair corresponding coset fibres and sum their pairings. Apply (12B) to each factor in (12H). The resulting constant is \(W(\rho)=\prod_jW_{M_j}(\omega_j)^{m_j}\), of absolute value one, and gives exactly (12).

These continuations and constants are independent of the chosen Brauer expression. The Euler products agree on the common nonempty half-plane, so their meromorphic continuations agree by the identity theorem. On an open set where the dual completed function is finite and nonzero, their quotient determines the constant uniquely. Such an open set exists by the nonzero initial Euler product and meromorphic continuation. This proves the theorem, without asserting holomorphy of the negative-power product. ∎

The precise written analytic and character-theory scopes have thus been used. Section 4 proves the further local-product identity from the actual orthogonal and global Brauer proofs.

Artin's holomorphy conjecture. If \(\rho\) is nontrivial and irreducible, \(L(s,\rho)\) is entire. Meromorphic continuation is a theorem; removal of all possible poles in this generality is the conjecture. Brauer induction gives products with negative as well as positive powers, which explains why it yields meromorphy. A nontrivial irreducible representation induced from a one-dimensional character has an entire L-function, by induction and the Hecke theorem. This argument requires that induced-character hypothesis; solvability of a finite group by itself does not supply it.

Example 3.1 (Dedekind factorization). Let \(M/K\) be finite Galois with group \(G\). The regular representation decomposes as \[ \mathbf C[G]\simeq\bigoplus_{\tau\in\widehat G}\tau^{\oplus\dim\tau}. \] Induction of the trivial representation from \(G_M\) has L-function \(\zeta_M(s)\). The formalism therefore gives \[ \zeta_M(s)=\prod_{\tau\in\widehat G}L_K(s,\tau)^{\dim\tau}. \tag{13} \] This includes ramified Euler factors. For \(M=\mathbf Q(i)\), the two characters are \(1\) and \(\chi_{-4}\), so \(\zeta_M=\zeta L(\chi_{-4})\). For odd \(p\), the factor is \((1-p^{-s})^{-2}\) if \(p\equiv1\pmod4\), and \((1-p^{-2s})^{-1}\) if \(p\equiv3\pmod4\). At \(2\), inertia acts nontrivially on \(\chi_{-4}\), leaving only the trivial line: the factor is \((1-2^{-s})^{-1}\). The character at 2 is nontrivial on \(1+2\mathbf Z_2\), since it is minus one at \(-1\), and trivial on \(1+4\mathbf Z_2\); thus its unit conductor exponent is exactly two by Lemma 2.4 of the conductor lesson. Hence \(\mathfrak f(\chi_{-4})=(4)\), so \(C(1\oplus\chi_{-4})=4\), agreeing with \(|D_M|\). The gamma factors also agree, since the duplication formula gives \(\Gamma_{\mathbf R}(s)\Gamma_{\mathbf R}(s+1)=\Gamma_{\mathbf C}(s)\).

Example 3.2 (a character that detects the dual). Let \(\zeta_3=e^{2\pi i/3}\). Define the primitive character modulo \(7\) by \(\chi(3)=\zeta_3\), since \(3\) generates \((\mathbf Z/7\mathbf Z)^\times\). Then \(2\equiv3^2\pmod7\), so \(\chi(2)=\zeta_3^2\). The corresponding Artin representation acts on arithmetic Frobenius at \(2\) by \(\zeta_3^2\), and its Euler factor is \[ L_2(s,\chi)=\frac{1}{1-\zeta_3^2\,2^{-s}}. \tag{14} \] It is even, because \(-1\equiv3^3\), so its infinite factor is \(\Gamma_{\mathbf R}(s)\). Its conductor is \(7\), its factor at \(7\) is one, and \(L(s,\chi)=\sum_{n\ge1}\chi(n)n^{-s}\), with \(\chi(n)=0\) when \(7\mid n\). If one uses this representation itself in the geometric local-factor definition, its eigenvalue at \(\Phi_2\) is \(\zeta_3\), giving the factor for \(\chi^{-1}\). Applying that definition to the dual gives (14). A quadratic character would not expose the error.

4. Root numbers and the global additive character

Choose a nontrivial continuous unitary additive character \(\psi:\mathbf A_K/K\to\mathbf C^\times\), with local components \(\psi_v\), and at each place use the self-dual Haar measure \(dx_{\psi_v}\). At a complex place the absolute value means the squared usual absolute value. This convention makes \(\prod_v|a|_v=1\) for \(a\in K^\times\).

The actual written adelic self-duality and Poisson proofs are Additive characters, self-dual measures and Poisson summation on the adèles, Theorems 5.4–5.5. They prove that the product self-dual measure has covolume one. The first theorem identifies the annihilator of \(K\) in \(\mathbb A_K\) with \(K\) by a rational trace-dual basis, so every nontrivial additive character on \(\mathbb A_K/K\) is \(\psi_a(x)=\psi(ax)\) for a unique \(a\in K^\times\).

Put \(r_v=\rho_v^\vee|_{W_{K_v}}\). At infinity use the analogous finite-image Weil representation. Define \[ \epsilon(s,\rho,\psi)= \prod_v\epsilon_v(s,r_v,\psi_v,dx_{\psi_v}),\qquad W_{\mathrm{loc}}(\rho)=\epsilon(1/2,\rho,\psi). \tag{15} \] Almost every finite place is unramified both for \(r_v\) and for the additive character, with \(\mathcal O_v\) of self-dual volume one. Its constant is one, so the product has only finitely many nontrivial terms. Finite-image representations are unitary: average any positive Hermitian form over their finite image. The proved local-factor lesson, Theorem 3.0 and §§4–5, therefore gives absolute value one for every central local factor.

Proposition 4.0 (the complete conductor power). The product in (15) has the form \[ \epsilon(s,\rho,\psi) =W_{\mathrm{loc}}(\rho)C(\rho)^{1/2-s}, \qquad |W_{\mathrm{loc}}(\rho)|=1. \tag{15A} \]

Proof. First take the standard trace character. The local exponent formula in the actually written local-factor lesson gives \[ \epsilon_v(s,r_v,\psi_v) =\epsilon_v(1/2,r_v,\psi_v) q_v^{(a_v(\rho)+d\,d_v)(1/2-s)}. \] Here \(d_v\) is the local different exponent of \(K_v/\mathbf Q_p\); the character conductor is \(d_v\) by that lesson's trace-dual Lemma 2.2. Duality preserves every invariant codimension in the conductor formula, hence \(a(r_v)=a(\rho_v)\). At infinity the finite-image real parameter is a sum of trivial and sign lines, and the complex parameter is trivial; their Gaussian epsilon factors do not depend on \(s\). Taking the product and using \(\prod_{v<\infty}q_v^{d_v}=|D_K|\) gives exactly (15A). Local unitary normalization gives the absolute value assertion. Proposition 4.1 below extends the formula to every global additive character. ∎

Theorem 4.2 (global factorization). The analytic constant in Theorem 3.3 equals \(W_{\mathrm{loc}}(\rho)\). Equivalently, \[ \epsilon(s,\rho,\psi)=W(\rho)C(\rho)^{1/2-s}. \tag{16} \] Proof. The rank-one case has the complete proof in Lemma 3.2. Local induction, proved in the local-factor lesson for virtual dimension zero, gives for an actual representation \(\tau\) of a local extension \(M_u/K_v\) \[ W_{K_v}(\operatorname{Ind}\tau) =\lambda(M_u/K_v)^{\dim\tau}W_{M_u}(\tau), \quad \lambda(M_u/K_v)= \frac{W_{K_v}(\operatorname{Ind}1)}{W_{M_u}(1)}. \tag{16A} \] This follows by applying dimension-zero induction to \(\tau-(\dim\tau)1\) and then restoring the trivial factors. Over all places it reduces Theorem 4.2, by (12F), to the exact assertion \[ \prod_v\prod_{u\mid v}\lambda(M_u/K_v)=1 \quad\text{for every finite number-field extension }M/K. \tag{16B} \] Here are the complete steps proving (16B). Write \(P=\operatorname{Ind}_{G_M}^{G_K}1\), \(n=[M:K]\), \(\eta=\det P\), and \[ A=P-\eta-(n-1)1. \tag{16C} \] The permutation matrices give \(P\) a real orthogonal realization; \(\eta\) is a real line. Thus \(A\) is a finite-image virtual real representation with dimension zero and determinant one. The preceding local-factor lesson actually proves its orthogonal theorem in Theorem 7.4, with the real induction, Clifford cocycle and quadratic Fourier calculations in Lemmas 7.5–7.8. It gives at every completion \[ W_{K_v}(A_v)= \exp\bigl(2\pi i\operatorname{inv}_v(b_v)\bigr), \qquad b=\operatorname{cl}(w_2(A))\in\operatorname{Br}(K). \tag{16D} \] The class is global: pull back the degree-two sign cocycle of the finite real representation along \(G_K\) and embed its signs in \(\bar K^\times\). Restriction of that cocycle gives each \(b_v\). The actual earlier Brauer groups of local and global fields, §4 and Theorem 24.4, proves finite support and \(\sum_v\operatorname{inv}_v(b_v)=0\). Its written proof first proves injectivity by cyclic norms and Sylow restriction, constructs a cyclic splitting extension for the finite local support, and then obtains the sum law from the principal reciprocity law. Hence (16D) gives \(W_{\mathrm{loc}}(A)=1\).

The trivial character has global root one directly from the Gaussian and finite unramified calculations in Lemma 3.2: its central finite factors for the trace character are one, as are its infinite factors. If \(\eta\ne1\), let \(M'/K\) be its quadratic field. Proposition 3.1, (9A) and the real/complex Gamma induction in (12G) identify its completed function with the quotient of the completed zeta functions of \(M'\) and \(K\). The trivial-character case of Lemma 3.2 gives each zeta function root one. Consequently its quotient has analytic root one, and the proved rank-one factorization gives \(W_{\mathrm{loc}}(\eta)=1\). This is also true when \(\eta=1\). Multiplicativity in (16C) now gives \(W_{\mathrm{loc}}(P)=1\).

At \(v\), restricting the global permutation module decomposes it as the sum of \(\operatorname{Ind}_{W_{M_u}}^{W_{K_v}}1\) over \(u\mid v\), by the same coset decomposition as (8B), also at the real and complex places. Therefore the numerator in the product of (16A)'s constants is \(W_{\mathrm{loc}}(P)=1\), while its denominator is the trivial-character product over \(M\), also one. This proves (16B) for every finite extension, including non-Galois extensions and ramified completions.

For a finite-order line \(\chi\) over \(M\), equations (16A)–(16B) give \(W_{\mathrm{loc}}(\operatorname{Ind}_M^K\chi)=W_{\mathrm{loc}}(\chi)=W(\chi)\). Apply the integer Brauer identity (12F). Multiplicativity of the actual local epsilon family gives the product of these constants with the signed integers \(m_j\); the analytic construction (12H) gives the identical product. Thus \(W_{\mathrm{loc}}(\rho)=W(\rho)\) for every finite-image \(\rho\). Proposition 4.0 proves (16). All the higher-dimensional inputs used here are actual earlier written proofs; their free-source chain and publication checks remain the programme's separate integration obligations. ∎

The product of constants at \(s=0\) has absolute value \(C(\rho)^{1/2}\), by (15A). For instance its absolute value for \(\chi_{-4}\) is \(2\), whereas its central product has absolute value one. Taking constants at zero does not give the normalized root.

Proposition 4.1 (independence of \(\psi\)). Replacing the global additive character, with its corresponding self-dual local measures, leaves \(\epsilon(s,\rho,\psi)\) and \(W_{\mathrm{loc}}(\rho)\) unchanged. The analytic \(W(\rho)\) is uniquely fixed by (12).

Proof. Write \(\psi_{v,a}(x)=\psi_v(ax)\). Self-duality changes the measure by \[ dx_{\psi_{v,a}}=|a|_v^{1/2}dx_{\psi_v}. \] The measure and character laws proved in the local-factor lesson, valid also at infinity by Tate's one-dimensional formulas and induction, give \[ \epsilon_v(s,r_v,\psi_{v,a},dx_{\psi_{v,a}}) =\det r_v(\operatorname{Art}_v(a)) |a|_v^{d(s-1/2)} \epsilon_v(s,r_v,\psi_v,dx_{\psi_v}). \tag{17} \] For clarity the product formula itself follows here from norms: the product of the real and squared complex absolute values is \(|N_{K/\mathbf Q}(a)|\), while the finite product is \(\prod_p p^{-v_p(N_{K/\mathbf Q}a)}=|N_{K/\mathbf Q}(a)|^{-1}\). The local norm valuation \(f v\) was proved in the first lesson; factoring the completed trace lattice gives this global factorization.

The determinant character corresponds, by the actual preceding Weil lesson's §6 and Proposition 6.1, to a character of \(C_K=\mathbb A_K^\times/K^\times\), with these local restrictions. Evaluating it on the principal idèle \(a\) gives one, since that idèle represents the identity of \(C_K\). Thus both products of multipliers in (17) are one. All but finitely many factors are one, so their multiplication is legitimate. Every other nontrivial global additive character is \(\psi_a\), by the actual self-duality theorem identified above. This proves the full assertion. ∎

At \(s=1/2\) the norm multiplier is already one at each place. The determinant multipliers still require the global principal-idèle cancellation. Thus the product is independent even when individual local roots vary. Theorem 4.2 identifies that invariant product with the analytic root number.

5. The Hasse–Weil product and its convergence

For an elliptic curve \(A/\mathbf Q\), write \(a_p=p+1-\#\widetilde A(\mathbf F_p)\) at good primes. At bad primes define \(a_p=1\) for split multiplicative reduction, \(a_p=-1\) for nonsplit multiplicative reduction, and \(a_p=0\) for additive reduction. The local elliptic lesson proved the factors \[ L_p(A,s)= \begin{cases} (1-a_pp^{-s}+p^{1-2s})^{-1},&p\text{ good},\\ (1-a_pp^{-s})^{-1},&p\text{ bad}. \end{cases} \tag{18} \] The second line is one in the additive case. These factors are \(L(s,\operatorname{WD}_p(V_\ell A^\vee))\), for any \(\ell\ne p\). Set \[ L(A,s)=\prod_p L_p(A,s),\qquad N(A)=\prod_p p^{a(\operatorname{WD}_p(V_\ell A))}. \tag{19} \] Local compatibility makes both definitions independent of coefficient choices. The conductor is also unchanged by duality, as proved in the local elliptic lesson.

Theorem 5.1 (absolute convergence). The Euler product in (19) converges absolutely, locally uniformly, to a nonzero holomorphic function on \(\operatorname{Re}s>3/2\).

Proof. For a good prime, factor \[ 1-a_pX+pX^2=(1-\alpha_pX)(1-\beta_pX), \qquad |\alpha_p|=|\beta_p|=\sqrt p \] by Theorem 2.1. If \(\sigma=\operatorname{Re}s>3/2\), then \(|\alpha_pp^{-s}|=|\beta_pp^{-s}|=p^{1/2-\sigma}<1\). The logarithm defined by its power series gives \[ \log L_p(A,s) =\sum_{m\ge1}\frac{\alpha_p^m+\beta_p^m}{m\,p^{ms}}. \tag{20} \] Its absolute value, and indeed the sum of the absolute values of its terms, is at most \[ 2\sum_{m\ge1}\frac{p^{-m(\sigma-1/2)}}m \le \frac{2p^{-(\sigma-1/2)}}{1-p^{-(\sigma-1/2)}}. \tag{21} \] On a compact subset of the half-plane, choose \(\sigma_0>3/2\) below all real parts there and put \(b=\sigma_0-1/2>1\). The bound is at most \(2p^{-b}/(1-2^{-b})\), uniformly on that compact set. Its sum over primes converges, since \(\sum_p p^{-b}\le\sum_{n\ge2}n^{-b}<\infty\). Thus the double logarithmic series converges absolutely and locally uniformly. Its exponential is holomorphic and nonzero, and is the limit of the products over good primes.

There are only finitely many bad primes. Their nontrivial factors have \(a_p=\pm1\), so their denominators cannot vanish when \(\operatorname{Re}s>3/2\). Multiplication by this finite product proves the assertion. ∎

This is a statement in a right half-plane. It does not construct the continuation through that half-plane's boundary.

Example 5.2 (the conductor-11 curve). Take \[ A\colon\quad y^2+y=x^3-x^2. \] Its discriminant is \(-11\). The previous lesson proved that at \(11\) it has split multiplicative reduction of type \(I_1\), conductor exponent one, and \(a_{11}=1\). At every other prime it has good reduction. Hence \[ N(A)=11,\qquad L(A,s)=\frac{1}{1-11^{-s}} \prod_{p\ne11}\frac{1}{1-a_pp^{-s}+p^{1-2s}}. \tag{22} \] Here are direct good-prime counts. The middle column lists the numbers of \(y\)-solutions as \(x\) runs from \(0\) to \(p-1\); the projective point at infinity contributes one more.

\(p\) Numbers of \(y\)-solutions for successive \(x\) \(\#A(\mathbf F_p)\) \(a_p\)
\(2\) \(2,2\) \(5\) \(-2\)
\(3\) \(2,2,0\) \(5\) \(-1\)
\(5\) \(2,2,0,0,0\) \(5\) \(1\)
\(7\) \(2,2,0,0,2,2,1\) \(10\) \(-2\)

For example the factors at \(2\) and \(5\) are \[ (1+2\,2^{-s}+2^{1-2s})^{-1},\qquad (1-5^{-s}+5^{1-2s})^{-1}. \] One must use the linear bad factor at \(11\), not a degree-two factor obtained from a singular cubic's total point count.

The elliptic completed function, in its usual normalization, is \[ \Lambda(A,s)=N(A)^{s/2}(2\pi)^{-s}\Gamma(s)L(A,s). \tag{23} \] The local archimedean factor is \(\Gamma_{\mathbf C}(s)\); dividing it by the constant two gives the conventional factor in (23), without changing the functional equation.

Theorem 5.3 (modularity and analytic continuation; remaining modularity proof). Every elliptic curve over \(\mathbf Q\) has the same L-function as a weight-two cuspidal newform of level \(N(A)\). The function (23) is entire and \[ \Lambda(A,s)=w(A)\Lambda(A,2-s),\qquad w(A)\in\{1,-1\}. \tag{24} \] The modularity assertion is retained for every elliptic curve over \(\mathbf Q\). The freely readable author manuscript of Breuil–Conrad–Diamond–Taylor states it as Theorem A and discusses the conductor-level equivalences in its introduction. Reading that statement does not prove it in this programme. No actual earlier proof of general modularity, of the exact conductor-level comparison, or of the complete local newform/elliptic comparison has yet been bound here. These remain core proof obligations.

Here is the full analytic deduction once those modularity data are supplied. It replaces the former appeal to an unspecified Mellin-transform theorem.

Lemma 5.4 (the Mellin deduction). Suppose \(f(z)=\sum_{n\ge1}b_ne^{2\pi inz}\) is a weight-two cusp form with the same Euler coefficients as \(A\), and is a Fricke eigenfunction \[ \frac1{Nz^2}f\left(-\frac1{Nz}\right)=\eta f(z), \qquad \eta\in\{1,-1\},\quad N=N(A). \tag{24A} \] Then (23) is entire and satisfies (24) with \(w(A)=-\eta\).

Proof. At a good prime the coefficients of the reciprocal quadratic factor satisfy \(b_{p^r}=\sum_{j=0}^r\alpha_p^j\beta_p^{r-j}\), so \(|b_{p^r}|\le(r+1)p^{r/2}\). The multiplicative factors have coefficients bounded by one, and the additive factors have all positive coefficients zero. Multiplying prime-power bounds gives \(|b_n|\le d(n)n^{1/2}\). Since \(\sum_n d(n)n^{-b}=\sum_{u,v}(uv)^{-b}<\infty\) for \(b>1\), the Dirichlet series and its termwise Mellin integral converge absolutely on \(\Re s>3/2\). Integrating \(e^{-2\pi ny}y^s\,dy/y\) gives \[ (2\pi)^{-s}\Gamma(s)L(A,s)=\int_0^\infty f(iy)y^s\,\frac{dy}{y}. \tag{24B} \] Set \(F(u)=f(iu/\sqrt N)\). Cusp holomorphy gives exponential decay as \(u\to\infty\): its convergent Fourier series has zero constant term. Equation (24A) says \(F(1/u)=-\eta u^2F(u)\). Therefore \(F(u)\) has \(O(u^{-2}e^{-c/u})\) decay at zero. On any compact set of complex \(s\), these two bounds dominate the Mellin integrand and every parameter derivative, so its integral is entire. Equivalently, put \[ A(s)=\int_1^\infty F(u)u^s\,\frac{du}{u}. \] This is entire, and substitution \(u\mapsto1/u\) in the small part gives \[ \Lambda(A,s)=A(s)-\eta A(2-s). \tag{24C} \] The same formula at \(2-s\), and \(\eta^2=1\), give \(\Lambda(A,s)=-\eta\Lambda(A,2-s)\). Initially (24B) identifies this entire function with (23); the identity theorem supplies its continuation. ∎

This proof establishes the analytic implication completely. It does not construct the cusp form or establish its exact level or Fricke data; those are the identified remaining parts of Theorem 5.3.

The full root-number comparison in Theorem 5.3 asserts that \(w(A)\) is the product of the local constants for the dual elliptic parameters at \(s=1\). Its proof additionally requires the complete local comparison with the conductor-level newform; an Artin finite-image theorem alone would not prove this infinite-image assertion. The following local normalization calculation is independent of that remaining global comparison. At a finite place the dual elliptic determinant is \(\|\cdot\|^{-1}\). Consequently the scaling law (17), with \(d=2\) and \(s=1\), has multiplier \[ |a|_v^{-1}|a|_v^{2(1-1/2)}=1. \] This also holds for the determinant of the elliptic archimedean parameter. Thus in this case each normalized local root is independent of the local additive character.

For the curve in Example 5.2, use the standard global character of \(\mathbf A_{\mathbf Q}/\mathbf Q\), with \(x\mapsto e^{2\pi ix}\) at infinity and conductor zero at every finite place. The good-prime constants are one. At \(11\), the dual parameter is \(\|\cdot\|^{-1}\mathrm{Sp}(2)\), and the block calculation in the monodromy lesson gives \[ \epsilon_{11}(s)=-11^{1-s},\qquad w_{11}(A)=-1. \] The real parameter is \(\operatorname{Ind}_{W_{\mathbf C}}^{W_{\mathbf R}}(z/|z|_{\mathrm{usual}})\otimes\|\cdot\|^{-1/2}\). Its epsilon factor is \(-1\). The actual infinite Tate Gaussian proof gives, for the positive character, \(\epsilon_{\mathbf R}(1)=1\), \(\epsilon_{\mathbf R}(\mathrm{sgn})=i\), and the angular complex character's constant \(i\). Dimension-zero induction in the local-factor lesson gives the induction constant \(i\), since \(\operatorname{Ind}1=1\oplus\mathrm{sgn}\). Thus the induced angular constant is \(i\cdot i=-1\). The Gaussian formula is constant under radial twists, so the factor \(\|\cdot\|^{-1/2}\) does not change it. The local product is consequently one; under the root-number comparison asserted in Theorem 5.3 this gives \[ w(A)=w_\infty(A)w_{11}(A)=(-1)(-1)=1. \tag{25} \] This calculation fixes the sign in (24); it does not determine the order of vanishing at the central point.

6. Artin, Hecke and geometric functions in one framework

Three sources of Euler products now fit together. An Artin representation has finite image, weight zero, and arithmetic Euler factors (8). An algebraic Hecke character has compatible one-dimensional \(\ell\)-adic realizations. We prove that construction rather than treating the words “through class field theory” as a proof.

Proposition 6.1 (algebraic Hecke realizations). Let \(\theta:C_K\to\mathbb C^\times\) be a continuous Hecke character whose infinite component is \[ \theta_\infty(x_\infty)=\epsilon_\infty(x_\infty) \prod_{\tau:K\hookrightarrow\mathbb C}\tau(x_\infty)^{-n_\tau}, \qquad n_\tau\in\mathbb Z, \tag{26} \] where \(\epsilon_\infty\) is a product of real sign characters. There is a number field \(E\) containing all its finite-component values and a continuous character \(\rho_\lambda:G_K\to E_\lambda^\times\) for every finite place \(\lambda\) of \(E\). At every finite \(v\nmid\ell\), the finite-inertia local characters and Frobenius values are independent of \(\lambda\), after the specified embeddings. In particular they form a strictly compatible system away from the residue characteristic; at a place where \(\theta_v\) is unramified their arithmetic polynomial is \(1-\theta_v(\varpi_v)X\).

Proof. The finite character \(\theta_f\) has finite image on \(U_f=\prod_{v<\infty}\mathcal O_v^\times\). To prove this directly, choose an arc about one in the unit circle containing no nontrivial subgroup: every nontrivial unit complex number has a power outside an arc of length less than a semicircle. The absolute value of the image of a compact group is one. Continuity and the basis of open subgroups in \(U_f\) now give an open subgroup mapped into that arc, hence into one. Thus the character factors through a finite quotient on \(U_f\); it is unramified outside finitely many places.

The actual earlier Idèles and the idèle class group, Theorem 3.3 and Corollary 3.4, proves finiteness of the ideal class group using norm-one compactness. Choose finitely many finite idèles \(t_j\) representing its classes. Every finite idèle has the form \(a t_j u\), with \(a\in K^\times\) and \(u\in U_f\), by removing its principal ideal. Triviality on principal idèles gives \(\theta_f(a)=\epsilon_\infty(a)^{-1}\prod_\tau\tau(a)^{n_\tau}\). These values lie in a fixed normal closure of \(K\). If the ideal class of \(t_j\) has order \(h_j\), then \(t_j^{h_j}=a_j u_j\), so \(\theta_f(t_j)\) is an \(h_j\)-th root of an algebraic number. Adjoining the finitely many such values and the finite unit-image roots of unity gives a number field \(E\) containing every finite value. Enlarge it to contain the normal closure of \(K\).

Fix \(i_\lambda:E\hookrightarrow E_\lambda\subset\overline{\mathbb Q}_\ell\). For each \(\tau\), let \(u_\tau\mid\ell\) be the place of \(K\) defined by \(i_\lambda\tau\); that embedding extends continuously from \(K\) to \(K_{u_\tau}\). Define on idèles \[ R_\lambda(x)=i_\lambda(\theta_f(x_f))\epsilon_\infty(x_\infty) \prod_\tau (i_\lambda\tau)(x_{u_\tau})^{-n_\tau}. \tag{27} \] All factors are continuous: the finite character is locally constant on finite units, its valuation coordinates are discrete in an idèle chart, and the additional factors are continuous powers at the finitely many places above \(\ell\). Its values lie in \(E_\lambda\). Formula (26) makes (27) one on \(K^\times\). It is also one on the connected archimedean component, because the only remaining infinite factors are signs.

The actual earlier The global reciprocity isomorphism, Proposition 17.3 and its preceding compact-group lemma, proves that the arithmetic reciprocity map identifies \(G_K^{\mathrm{ab}}\) with \(C_K/D_K\), where \(D_K\) is the closure of that connected archimedean image. Its proof passes through all finite reciprocity quotients and the compact norm-one group. Continuity makes (27) kill that closure, so it descends to the required \(\rho_\lambda\). Here arithmetic reciprocity is the inverse of the geometric convention in that provider. This change of convention does not change its kernel or topology.

For \(v\nmid\ell\), insert a local idèle at \(v\) in (27). Every correction factor above \(\ell\) is one. The resulting local character is exactly \(i_\lambda\theta_v\) under arithmetic reciprocity, including its finite unit character and its value on a uniformizer. Its geometric Weil parameter therefore sends geometric Frobenius to \(\theta_v(\varpi_v)^{-1}\), with the common finite inertia character obtained by inversion of reciprocity. It has \(N=0\), and this common one-dimensional pair proves strict compatibility. Arithmetic Frobenius has the asserted inverse-root value. This proves every claim. ∎

The Euler product of this system is the primitive Hecke product after the same Frobenius translation. General continuous complex Hecke characters need not have the integer exponents (26), and the construction above does not assert that they all have such realizations.

For a smooth projective variety \(X/K\), the general cohomological assertions are that its étale cohomology supplies geometric Galois representations, its good-prime characteristic polynomials are independent of \(\ell\), and its geometric-Frobenius eigenvalues on \(H^j_{\mathrm{et}}\) have absolute value \(q_v^{j/2}\). These assertions retain their full scope here, but their construction, smooth proper comparison, trace formula and general purity proofs have not been supplied in an actual earlier lesson inspected for this edition. They remain core proof obligations. The elliptic \(H^1_{\mathrm{et}}=V_\ell A^\vee\) comparison is available from the actual torsion and Kummer proof in the Tate-module lesson. With the arithmetic convention (2), dualizing a geometric cohomology representation makes arithmetic Frobenius have the same eigenvalues as geometric Frobenius on the original space: if a matrix is \(B\), that dual arithmetic matrix is \((B^{-1})^{-t}=B^t\).

An algebraic correspondence which has an idempotent Galois-equivariant action \(e\) separates a representation into \(\operatorname{im}e\oplus\ker e\): write \(v=ev+(v-ev)\), and the intersection is zero since \(e^2=e\). Every Frobenius matrix is block diagonal in that decomposition, so its characteristic polynomial and Euler factor split as the product of those of the pieces. Purity passes to a piece because its eigenvalues form a submultiset. Establishing the cohomological action, finding the required correspondences, and comparing their factors for all coefficient primes require further geometric proofs; the linear-algebra argument alone does not construct those correspondences. In dimension zero the permutation calculation in Proposition 3.1 gives the Artin and Dedekind factors. In dimension one the proved elliptic Tate modules supply a concrete geometric example.

A pure arithmetic weight \(w\) suggests a functional equation with center \((w+1)/2\): weight zero gives \(1/2\), and the covariant elliptic system of weight one gives \(1\). Purity and weak compatibility alone do not prove such an equation, nor do they supply all bad-prime monodromy data. General motivic continuation and functional equations over number fields require additional theorems or remain conjectural. The unconditional elliptic statement (24) is specifically over \(\mathbf Q\).

There is a useful distinction in Deligne's §9: its global field is a function field over a finite field. Theorem 9.8 compares ordinary local semisimplifications and explicitly warns on p.578 that it does not retain \(N\). It must not be read as a theorem that good-prime compatibility automatically supplies full bad-prime Weil–Deligne compatibility for an arbitrary number-field system.

7. Proof scopes and remaining obligations

The local elliptic classification, the Tate-curve and dual factors, and the conductor calculations are imported from Elliptic curves over local fields and their Weil–Deligne representations, Theorems 3.1, 4.1 and 5.1. The good-reduction torsion comparison, determinant–degree identity and Hasse bound are imported from The ℓ-adic Tate module of an elliptic curve, Theorems 3.1, 2.2 and 4.1 respectively. Their combination proving common good-prime polynomials is Theorem 2.1 here.

Lemma 2.2 proves the common characteristic polynomial for every abelian-variety endomorphism; its exact geometric prerequisites are the written cube and Euler-characteristic proofs in Abelian varieties, Proposition 4.4, Lemma 5.2, Proposition 5.5 and Corollary 5.6, and the finite kernels and Tate modules in §§6–7. Lemma 2.3 proves strict local compatibility for potentially good abelian varieties of every dimension, using the actual torsion-specialization proof in Néron models, Lemma 7.1. For elliptic curves, potentially multiplicative descent and the proved Tate curve complete the local comparison. The full arbitrary-abelian-variety theorem, Theorem 2.4, still needs the semistable monodromy filtration, its toric pairing and equivariant descent comparison. Semistable reduction alone does not prove that representation theorem. No assertion at \(\ell=p(v)\) is part of strict compatibility here.

Proposition 3.1 proves all ramified Euler induction and convergence. The conductor lesson's Theorems 2.6 and 4.1 prove integrality and induction, yielding (9A). Lemma 3.2 gives the rank-one analytic proof with the actual earlier adelic Poisson and Tate proofs: NT-ADL-03, Theorem 3.3; NT-ADL-05, Theorems 5.4–5.5; NT-ADL-07, Theorems 7.2–7.3; NT-ADL-08, Theorem 8.1 and Propositions 8.2–8.4; NT-ADL-09, Theorem 9.2; and NT-ADL-10, Theorem 10.1. Theorem 3.3 then proves the full Artin meromorphic functional equation through the actually written integer induction theorem RT-FIN-11, Theorem 5.1. General Artin holomorphy remains a conjecture; meromorphic quotients in (12H) do not prove it. Proposition 6.1 proves algebraic Hecke realizations using NT-CFT-17, Proposition 17.3.

The earlier local-factor lesson's Theorem 3.0 and §§3A–3D supply the actual local epsilon existence proof and all Brauer relations. Its Lemma 2.2 proves the trace conductor, §§4–5 prove scaling and normalization, and Theorem 7.4 with Lemmas 7.5–7.8 proves the local orthogonal formula. Theorem 4.2 here proves the global product of induction constants, hence the full Artin root-number identity, using the actual global invariant-sum proof NT-CFT-24, §4 and Theorem 24.4. Proposition 4.1 proves additive-character independence with the actual adelic annihilator theorem NT-ADL-05, Theorem 5.4, and the preceding Weil lesson's Proposition 6.1.

The full modularity, exact conductor level and local newform/elliptic comparison of Theorem 5.3 remain without an actual programme proof. Lemma 5.4 proves the complete Mellin analytic deduction once those data are available, including entireness and the exact Fricke sign. The elliptic global root-number comparison also needs that local automorphic comparison. The general cohomological construction, comparison and purity assertions in §6 remain without the corresponding written proofs. These are genuine retained mathematical obligations, so this edition is not certified as a fully proved lesson.

8. Exercises with complete solutions

Exercise 8.1 (easy). Check strict finite-place compatibility for the cyclotomic system and the elliptic Tate-module system. Give exceptional sets and arithmetic weights.

Solution. Fix a finite place \(v\) and a coefficient prime \(\ell\ne p(v)\). Prime-to-\(p(v)\) roots of unity lie in the maximal unramified local extension. Arithmetic Frobenius raises them to the \(q_v\)-th power, so the cyclotomic character is unramified and has polynomial \(1-q_vX\). Its geometric local pair is \((\|\cdot\|,0)\), independently of \(\ell\). This applies to every finite \(v\): take \(S=\varnothing\). The sole inverse root is \(q_v\), of absolute value \(q_v^{2/2}\), giving weight two.

For \(V_\ell A\), take \(S=S_A\). At every \(v\notin S_A\), Theorem 2.1 supplies the same integral polynomial \(1-a_vX+q_vX^2\), and its inverse roots have absolute value \(\sqrt{q_v}\). Hence weak compatibility and weight one hold. At a potentially good place, Lemma 2.3 proves a common finite-inertia parameter with \(N=0\), including all its Frobenius and inertia character values. At a potentially multiplicative place the Tate-curve calculation and quadratic descent in the preceding local elliptic lesson give explicitly \(\eta_v\mathrm{Sp}(2)\), with the fixed splitting character \(\eta_v\). The integral-\(j\) potential-good proof and the nonintegral-\(j\) Tate descent in that lesson show these alternatives exhaust the elliptic case. Thus every finite \(v\), including \(v\in S_A\), has the common class \(D_v\) required by (4); no appeal to the still incomplete general semistable-abelian Theorem 2.4 is needed in this elliptic solution. A different coefficient prime creates no new permanent exceptional place, because the comparison excludes that prime's residue characteristic locally. These statements include no assertion at \(\ell=p(v)\).

Exercise 8.2 (medium). Give a Dirichlet-series proof of absolute convergence on \(\operatorname{Re}s>3/2\). Define its coefficients from the Euler factors and bound them.

Solution. At a good prime let \(b_{p,0}=1\) and write \[ \frac1{1-a_pX+pX^2}=\sum_{r\ge0}b_{p,r}X^r. \] Multiplication gives \(b_{p,1}=a_p\) and \(b_{p,r}=a_pb_{p,r-1}-pb_{p,r-2}\) for \(r\ge2\). Using the two roots, including the repeated-root case by the same product of geometric series, gives \[ b_{p,r}=\sum_{j=0}^r\alpha_p^j\beta_p^{r-j},\qquad |b_{p,r}|\le(r+1)p^{r/2}. \] At a multiplicative prime the coefficients are \(b_{p,r}=a_p^r\), bounded by one. At an additive prime, \(b_{p,0}=1\) and \(b_{p,r}=0\) for \(r>0\). Define \(b_n=\prod_{p^r\parallel n}b_{p,r}\). The uniform bound is \[ |b_n|\le d(n)n^{1/2}, \] where \(d(n)=\prod_{p^r\parallel n}(r+1)\) is the number of positive divisors. For \(\sigma>3/2\), put \(b=\sigma-1/2>1\). Then \[ \sum_{n\ge1}|b_n|n^{-\sigma} \le\sum_{n\ge1}d(n)n^{-b} =\sum_{u,v\ge1}(uv)^{-b} =\zeta(b)^2<\infty. \] Absolute convergence permits multiplication and regrouping of the prime-power series, proving equality with the Euler product. The same bound with a fixed \(\sigma_0>3/2\) gives local uniform convergence. Nonvanishing follows from the convergent logarithmic series (20), whose positive sign is the expansion of \(-\log(1-z)\).

Exercise 8.3 (medium). Let \(\rho=\operatorname{Ind}_{G_{\mathbf Q(i)}}^{G_{\mathbf Q}}1\). Write its Artin L-function and compute its Euler factor at \(2\).

Solution. The representation is the permutation action on the two cosets of the index-two subgroup. The sum of the two basis vectors spans a trivial line, and their difference spans the quadratic line \(\chi_{-4}\). Therefore \(\rho=1\oplus\chi_{-4}\), and multiplicativity gives \(L(s,\rho)=\zeta(s)L(s,\chi_{-4})=\zeta_{\mathbf Q(i)}(s)\).

The prime \(2\) ramifies: \(2=-i(1+i)^2\), up to the displayed unit, and the residue field at \(1+i\) has two elements. Thus inertia is the whole group of order two. It fixes the sum line and acts by minus one on the difference line. The invariant space is one-dimensional, and arithmetic Frobenius on it is the identity. Formula (8) gives \(L_2(s,\rho)=(1-2^{-s})^{-1}\). Equivalently the trivial character supplies this factor and the ramified quadratic character supplies one. It would be wrong to omit the trivial line merely because the extension ramifies.

Exercise 8.4 (hard). Prove that the global Artin root number is independent of the global additive character, with the self-dual measure and the geometric/arithmetic translation explicit.

Solution. The local parameter to use for the arithmetic Artin L-function is \(r_v=\rho_v^\vee\), not \(\rho_v\). For \(\psi_a(x)=\psi(ax)\), the self-dual measure is \(|a|_v^{1/2}dx_{\psi_v}\). The character law at \(s=0\) contributes \(\det r_v(\operatorname{Art}_v(a))|a|_v^{-d}\); twisting by \(\|\cdot\|^s\) contributes an additional \(|a|_v^{sd}\) to this determinant, and rescaling the measure contributes \(|a|_v^{d/2}\). Their combined multiplier is \[ \det r_v(\operatorname{Art}_v(a))|a|_v^{d(s-1/2)}. \] At the root-number point \(s=1/2\), the absolute-value multiplier is one at every place. The remaining product is \(\det\rho^\vee\) evaluated on the global Artin image of the principal idele \(a\), so it is one by reciprocity. More generally the product formula cancels the absolute values for every \(s\). The finite exceptional set may grow under scaling, but all factors outside its union remain one, so the multiplication is legitimate. Additive adelic self-duality says that every other nontrivial global character has this form with \(a\in K^\times\). Hence the normalized product (15), identified with the analytic root by Theorem 4.2, is independent of the character. Dropping the measure change would lose the \(|a|_v^{d/2}\) contribution; taking constants at \(s=0\) would also fail to give the normalized local roots.

Freely readable primary materials

The following verified free texts were read for the specified mathematical scopes. They provide materials and conventions; they do not replace the proofs or settle the remaining obligations in §7.

Editable Markdown source of this lesson