next up previous
Next: Optimal Quotients of Up: Statement and Proof of Previous: Statement and Proof of

Proof of the Main Theorem

The notation in this section is as in previous section.

Lemma 6.2   Let $ \pi_*:X_J \rightarrow X_{A^{\vee}}$ and $ \alpha: X_J \rightarrow \Hom(\pi^* X_A,\mathbf{Z})$ be as in previous section. Then

$\displaystyle \ker(\pi_*) = \ker(\alpha ).
$

Proof. Suppose $ x\in \ker(\pi_*)$, and let $ y=\pi^* z$ with $ z\in X_A$. Then

$\displaystyle \langle x, y \rangle = \langle x, \pi^* z \rangle
= \langle \pi_* x, z \rangle = 0,$

so $ x\in\ker(\alpha )$. Next let $ x\in\ker(\alpha )$. Then for all $ z\in X_A$,

$\displaystyle 0 = \langle x, \pi^* z\rangle = \langle \pi_* x, z\rangle,$

so $ \pi_* x$ is in the kernel of the monodromy map

$\displaystyle X_{A^{\vee}} \rightarrow \Hom(X_A,\mathbf{Z}).$

Since $ X_{A^{\vee}}$ and $ \Hom(X_A,\mathbf{Z})$ are free of the same finite rank and the cokernel is torsion, the monodromy map is injective. Thus $ \pi_* x=0$ and $ x\in \ker(\pi_*)$. $ \qedsymbol$

Let $ \pi^*:X_A \rightarrow X_J$ be as in previous section.

Lemma 6.3   The monodromy-pairing map $ X_J \rightarrow \Hom(X_J,\mathbf{Z})$ composed with restriction $ \Hom(X_J,\mathbf{Z})\rightarrow \Hom(\pi^* X_A,\mathbf{Z})$ gives rise to an exact sequence

$\displaystyle X_J \rightarrow \Hom(\pi^* X_A,\mathbf{Z}) \rightarrow \Phi_A \rightarrow 0.$

Proof. Lemma 6.2 gives the following commutative diagram with exact rows

$\displaystyle \xymatrix{0\ar[r]
& X_J/\ker(\alpha )\ar[d]^{\cong } \ar[r]
& {\...
...r[r] & X_{A^{\vee}}\ar[r] & {\Hom(X_A,\mathbf{Z})}\ar[r] & {\Phi_A}\ar[r] & 0.}$

By Lemma 6.2, the first vertical map is an isomorphism. The second is an isomorphism because it is induced by the isomorphism $ \pi^*:X_A\rightarrow \pi^* X_A$. It follows that $ \coker(\alpha )\cong \Phi_A$, as claimed. $ \qedsymbol$

Recall that $ \mathcal{L}$ denotes the saturation of $ \pi^* X_A$ in $ X_J$, and that $ L\subset \mathcal{L}$ denotes a subgroup of finite index.

Lemma 6.4   The rational number $ \displaystyle \frac{\char93 \Phi_L}{m_L}$ is independent of the choice of $ L$.

Proof. Suppose $ L'$ is another finite index subgroup of  $ \mathcal{L}$, and let $ n=[L:L']$. Here $ n$ is a rational number, the lattice index of $ L'$ in $ L$. Since $ \alpha $ is injective when restricted to $ \mathcal{L}$, it follows that

$\displaystyle m_{L'} = [\alpha (X_J):\alpha (L')]
= [\alpha (X_J):\alpha (L)]\cdot[\alpha (L):\alpha (L')] = m_L\cdot n.$

Similarly, $ \char93 \Phi_{L'} = \char93 \Phi_L\cdot n$. $ \qedsymbol$

Recall that $ m_A = \sqrt{\deg(\theta)}$ and

$\displaystyle \Phi_A \cong \coker(X_{A^{\vee}}\rightarrow \Hom(X_A,\mathbf{Z})),$

where $ m_A$ is the degree of $ A$ and $ \Phi_A$ is the component group of $ A$.

Proof. [Proof of Theorem 6.1] By Lemma 6.4 we may assume that $ L=\pi^*X_A$. With this choice of $ L$, Lemma 6.3 asserts that $ \Phi_L \cong \Phi_A$. By Lemma 6.2, properties of the index, and Lemma 5.2 we have
$\displaystyle m_L$ $\displaystyle =$ $\displaystyle [\alpha (X_J):\alpha (L)]$  
  $\displaystyle =$ $\displaystyle [\pi_*(X_J):\pi_*(L)]$  
  $\displaystyle =$ $\displaystyle [X_{A^{\vee}}:\pi_*(\pi^*X_A)]$  
  $\displaystyle =$ $\displaystyle [X_{A^{\vee}}:\theta^* X_A]$  
  $\displaystyle =$ $\displaystyle \char93 \coker(\theta^*)$  
  $\displaystyle =$ $\displaystyle \sqrt{\deg(\theta)} = m_A.$  

$ \qedsymbol$

Recall that $ \Phi_\mathcal{L}$ denotes the cokernel of the natural map $ X_J \rightarrow \Hom(\mathcal{L},\mathbf{Z})$ induced by composing the monodromy map $ X_J \rightarrow \Hom(X_J,\mathbf{Z})$ with the natural restriction map $ \Hom(X_J,\mathbf{Z})\rightarrow \Hom(\mathcal{L},\mathbf{Z})$.

Proposition 6.5   The group $ \Phi_\mathcal{L}$ is canonically isomorphic to the image of the map from $ \Phi_J$ to $ \Phi_A$ induced by $ \pi:J\rightarrow A$. Thus

$\displaystyle \image(\Phi_J\rightarrow \Phi_A) \cong \Phi_\mathcal{L}.$

Proof. Since $ \pi^*X_A\subset \mathcal{L}\subset X_J$, an application of Lemma 6.3 gives the following commutative diagram with exact rows:

$\displaystyle \xymatrix{
X_J\ar[r]\ar@{=}[d]& \Hom(X_J,\mathbf{Z})\ar[r]\ar[d]&...
...r]\ar[d] & 0\\
X_J\ar[r]& \Hom(\pi^*X_A,\mathbf{Z})\ar[r]& \Phi_A \ar[r]& 0.
}$

The map $ \Hom(\mathcal{L},\mathbf{Z})\rightarrow \Hom(\pi^*X_A,\mathbf{Z})$ is an isomorphism, so the map $ \Phi_\mathcal{L}\rightarrow \Phi_A$ is injective. Thus

$\displaystyle \image(\Phi_J\rightarrow \Phi_A) \cong \image(\Phi_J\rightarrow \Phi_\mathcal{L}).$

The cokernel of $ \Hom(X_J,\mathbf{Z})\rightarrow \Hom(\mathcal{L},\mathbf{Z})$ surjects onto the cokernel of $ \Phi_J\rightarrow \Phi_\mathcal{L}$. Using the exact sequence

$\displaystyle 0\rightarrow \mathcal{L}\rightarrow X_J \rightarrow X_J/\mathcal{L}\rightarrow 0,$

we find that

$\displaystyle \coker(\Hom(X_J,\mathbf{Z})\rightarrow \Hom(\mathcal{L},\mathbf{Z})) \subset \Ext^1(X_J/\mathcal{L},\mathbf{Z}).$

Because  $ \mathcal{L}$ is saturated, the quotient $ X_J/\mathcal{L}$ is torsion free, so the indicated $ \Ext^1$ group vanishes. Thus the map $ \Phi_J\rightarrow \Phi_\mathcal{L}$ is surjective, from which the proposition follows. $ \qedsymbol$

Corollary 6.6   The cokernel of the map from $ \Phi_J$ to $ \Phi_A$ induced by $ \pi:J\rightarrow A$ has order $ m_A / m_\mathcal{L}$. Thus

$\displaystyle \char93 \coker(\Phi_J\rightarrow \Phi_A) = \frac{m_A}{m_\mathcal{L}}.$

Proof. Combine Theorem 6.1 and Proposition 6.5. $ \qedsymbol$


next up previous
Next: Optimal Quotients of Up: Statement and Proof of Previous: Statement and Proof of
William A Stein 2001-12-09