Closed

Independence with no extra support

Proposition. (Independence with no extra support). The operation (C.1.50) is independent of the signed presentation on the same prescribed ZZ. It is additive under tensor product with union of supports, commutes with other such operations on their intersection of supports, and satisfies (C.1.48) for pulled-back presentations. These assertions remain meaningful when a pulled-back factor section vanishes identically. Its image in Cr−1(X;T)C_{r-1}(\mathcal X;T) is the first Chern operation of L∣T\mathcal L|_T.

Every rational Cartier section on a locally factorial integral Noetherian X\mathcal X has a signed presentation on exactly the support of its divisor. Thus the result applies, in particular, to arbitrary rational Cartier sections on locally factorial proper DVR models. On other integral models the existence of the stated presentation is a hypothesis, not a conclusion obtained by enlarging ZZ.

Proof

Proof. Suppose (A,a)/(B,b)(A,a)/(B,b) and (A′,a′)/(B′,b′)(A',a')/(B',b') present the same (L,σ)(\mathcal L,\sigma). The specified line identifications give A⊗B′≃A′⊗BA\otimes B'\simeq A'\otimes B, and the global sections a⊗b′a\otimes b' and a′⊗ba'\otimes b agree: they agree at the generic point, and an invertible sheaf on an integral scheme has no section supported off that point. Equation (C.1.49), with all targets pushed to T∩ZT\cap Z, gives Ga+Gb′=Ga′+GbG_a+G_{b'}=G_{a'}+G_b. Subtraction proves independence. The equality of these global sections persists under every pullback, even if both pull back to zero. Consequently the same calculation proves independence for pulled-back presentations on arbitrary source schemes.

Tensor additivity follows from (C.1.49). Proper compatibility follows by subtracting the two identities (C.1.48). For commutation, expand into four compositions and use Theorem C.1.101(3), then push to the common intersection of the prescribed supports. Naturality for the intermediate closed immersions justifies this last step. The Chern assertion follows by subtraction and Stacks, Tag 02SP.

For existence in the locally factorial case, write the Cartier divisor as the finite Weil sum D=∑nPPD=\sum n_P P. Let D+=∑max⁡(nP,0)PD^+=\sum\max(n_P,0)P and D−=∑max⁡(−nP,0)PD^-=\sum\max(-n_P,0)P. At each point, unique factorization makes each of these effective sums principal. These local equations glue as effective Cartier divisors: on an overlap their quotient has zero exponent at every prime factor of each local UFD, and so is a unit. Their canonical sections of O(D+)\mathcal O(D^+) and O(D−)\mathcal O(D^-) have union of zero supports ∣D∣|D|. The usual local identification of O(D)\mathcal O(D) with L\mathcal L carries their ratio to σ\sigma. These are the required data. This argument does not replace a prescribed support by the support of an auxiliary ample divisor. ◻

Citation

Non-Archimedean Geometry Foundations, Tag NAG-003116.

Proof in chapter · Fixed-edition record