Let A be a noetherian ring, q an ideal of A contained in the radical of A and M a finitely generated A-module such that M/qM is of finite length. Let x1,...,xm be a finite sequence of elements of q. Put x=Ax1+⋯+Axm⊂q.
a) One has dimA(M/xM)≥dimA(M)−m.
b) If dimA(M/xM)=dimA(M)−m, then eq(M/xM)≥eq(M).
c) If m<dimA(M), and if for i=1,...,m, the element xi of A is superficial for M/(x1M+⋯+xi−1M) relative to q, then one has
dimA(M/xM)=dimA(M)−mandeq(M/xM)=eq(M).
d) If m=dimA(M), and if, for i=1,...,m, the element xi of A is superficial for M/(x1M+⋯+xi−1M) relative to q, then one has
eq(M)=ex(M)≤long(M/xM)<+∞.
Parts a),b), and c) follow, for m=1, from Prop. 8, and the general case is deduced by induction. Suppose the hypotheses of d) are satisfied and set x′=Ax1+⋯+Axm−1 and M′=M/x′M, so that M/xM is identified with M′/xmM′. Then, by c), one has dimA(M′)=1 and eq(M)=eq(M′). By Prop. 8, M/xM is of finite length and one has eq(M′)=exmA(M′)≤long(M/xM). But, since xmnM′=xnM′ for every n, one has exmA(M′)=ex(M′). On the other hand, one has ex(M′)≥ex(M): this follows from b), where m is replaced by m−1, x by x′, and q by x. Consequently, one has
ex(M)≤ex(M′)=exmA(M′)=eq(M′)=eq(M).
Since x is contained in q, this implies ex(M)=eq(M) (No. 1, Remark 1), and completes the proof.
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.
Nothing.
Nothing.