Bourbaki

Lemma 1

Tag 04VO, permanent. Cite it as [Bourbaki, Tag 04VO].

Commutative Algebra, chapter VI, § 7. Approximation theorem

Let viv_i (1≤i≤n1 \leq i \leq n) be ualuations on the jield KK and x∈K∗x \in K^*. Then there exists a polynomialf (X)(X) of theform

f(X)=1+n1X+⋯+nk−1Xk−1+Xk(k≥2, nj∈Z for 1≤j≤k−1) f(X) = 1 + n_1 X + \cdots + n_{k-1} X^{k-1} + X^k \quad (k \geq 2,\ n_j \in \mathbf{Z} \text{ for } 1 \leq j \leq k - 1)

such that f(x)≠0f(x) \neq 0 and the element z=f(x)−1z = f(x)^{-1} enjoys the following propertiesfor 1≤i≤n1 \leq i \leq n:

vi(z)=0ifvi(x)≥0vi(z)+vi(x)>0ifvi(x)<0. v_i(z) = 0 \qquad \text{if} \quad v_i(x) \geq 0 \\ v_i(z) + v_i(x) > 0 \qquad \text{if} \quad v_i(x) < 0.

Assuming this lemma for a moment, we show how it implies that A1⊂Bp1A_1 \subset B_{p_1}. Let xx be a non-zero element of AiA_i. We apply the lemma to xx and valuations viv_i associated with the Ai\mathbf{A}_i. Then vi(z)≥0v_i(z) \geq 0 and vi(zx)≥0v_i(zx) \geq 0 for all ii, hence z∈Bz \in B and zx∈Bzx \in B. As v1(x)≥0v_1(x) \geq 0, v1(z)=0v_1(z) = 0 and hence z∉p1z \notin p_1. Hence x=zx/z∈Bp1x = zx/z \in B_{p_1}. The field of fractions of BB then contains A1A_1 and hence is KK.

We now pass to the proof of the lemma. Let II be the set of indices ii such that vi(x)≥0v_i(x) \geq 0. For all i∈Ii \in I, let xˉi\bar{x}_i denote the canonical image of xx in κ(Ai)\kappa(\mathbf{A}_i). For all i∈Ii \in I we construct a polynomialf, as follows: if there exists a polynomial g(X)g(X) of the form (1) such that g(xˉi)=0g(\bar{x}_i) = 0 in κ(Ai)\kappa(\mathbf{A}_i), we takef, to be such a polynomial; otherwise we take fi=1f_i = 1. Then we write f(X)=1+X2∏i∈Ifi(X)f(X) = 1 + X^2 \prod_{i \in I} f_i(X). It is obviously a polynomial of the form (1). If i∈Ii \in I, then f(x)∈Aif(x) \in A_i and alsof (xˉi)≠0(\bar{x}_i) \neq 0 by construction; hence f(x)∉m(Ai)f(x) \notin m(\mathbf{A}_i), vi(f(x))=0v_i(f(x)) = 0 and v(z)=0v(z) = 0. If i∉Ii \notin I, then vi(x)<0v_i(x) < 0, whence vi(f(x))=kvi(x)v_i(f(x)) = kv_i(x) (§ 3, no. 1, Proposition 1) and

vi(x)+vi(z)=(1−k)vi(x)>0 v_i(x) + v_i(z) = (1 - k)v_i(x) > 0

(since k≥2k \geq 2). Whence the lemma.

Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.

Cited by

Cites

Transcribed from the printed volume. Not translated.