26.12.2013 Views

A NULLSTELLENSATZ FOR AMOEBAS

A NULLSTELLENSATZ FOR AMOEBAS

A NULLSTELLENSATZ FOR AMOEBAS

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

534 BOUCKSOM, FAVRE, and JONSSON<br />

is represented by the effective zero divisor of the Jacobian determinant of the map<br />

X π ′ → X π induced by F , assuming that this is holomorphic. Now, for each blowup<br />

X π ,letC π be the set of nef classes α ∈ H 1,1<br />

R<br />

(X π) such that (α · K X ) ≤ 0.ThenC π is<br />

a closed convex cone with compact basis and is not reduced to zero since K X is not<br />

psef. It is, furthermore, invariant by S π . Indeed, if α ∈ H 1,1<br />

R<br />

(X π) is a nef class, we<br />

have<br />

(S π α · K X ) = (F ∗ α · K Xπ ) ≤ (F ∗ α · K X ) = (α · F ∗ K X ) ≤ (α · K X ).<br />

We can thus assume that the nonzero eigenclasses ϑ n in the proof of Theorem 3.2<br />

belong to C n , and we get (θ ∗ · K X ) ≤ 0.<br />

The same argument does not work for θ ∗ since F ∗ K X ≤ K X does not hold in<br />

general.<br />

3.2. Spectral properties<br />

Theorem 3.2 asserts the existence of eigenclasses for F ∗ and F ∗ with eigenvalue λ 1 .<br />

We now further analyze the spectral properties under the assumption that λ 2 1 >λ 2.<br />

THEOREM 3.5<br />

Assume that λ 2 1 >λ 2. Then the nonzero nef Weil classes θ ∗ ,θ ∗ ∈ L 2 (X) such that<br />

F ∗ θ ∗ = λ 1 θ ∗ and F ∗ θ ∗ = λ 1 θ ∗ are unique up to scaling. We have (θ ∗ · θ ∗ ) > 0 and<br />

(θ ∗2 ) = 0. We rescale them so that (θ ∗ · θ ∗ ) = 1. LetH ⊂ L 2 (X) be the orthogonal<br />

complement of θ ∗ and θ ∗ , so that we have the decomposition L 2 (X) = Rθ ∗ ⊕Rθ ∗ ⊕H.<br />

The intersection form is negative definite on H, and ‖α‖ 2 :=−(α 2 ) defines a Hilbert<br />

norm on H. The actions of F ∗ and F ∗ with respect to this decomposition are as<br />

follows.<br />

(i) The subspace H is F ∗ -invariant, and<br />

⎧<br />

F n∗ θ ∗ = λ n 1 θ ∗ ,<br />

( ⎪⎨ λ2<br />

) (<br />

nθ∗<br />

F n∗ θ ∗ = + (θ 2 ∗<br />

λ ) λn 1<br />

1 −<br />

1<br />

with h ⎪⎩<br />

n ∈ H, ‖h n ‖=O(λ n/2<br />

2 ),<br />

‖F n∗ h‖=λ n/2<br />

2 ‖h‖ for all h ∈ H.<br />

( λ2<br />

λ 2 1<br />

) n<br />

)<br />

θ ∗ + h n<br />

(ii) The subspace H is not F ∗ -invariant in general, but<br />

⎧<br />

F∗ ⎪⎨<br />

nθ ∗ = λ n 1 θ ∗,<br />

(<br />

F∗ nθ λ2<br />

) nθ ∗ =<br />

∗ ,<br />

λ 1<br />

⎪⎩ ‖F∗ nh‖≤Cλn/2<br />

2 ‖h‖ for some C>0 and all h ∈ H.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!