Bourbaki

Proposition 3

Tag 09QP, permanent. Cite it as [Bourbaki, Tag 09QP].

Integration, IX, p. 51, § 4. Inverse limits of measures

Let there be given an inverse system of topological spaces T=(Ti,pij)\mathscr{T} = (T_i, p_{ij}) indexed by II, a topological space TT, a coherent and separating family of continuous mappings pi:T→Tip_i : T \to T_i (for i∈Ii \in I) and a sub-inverse system (μi)i∈I(\mu_i)_{i \in I} of measures on T\mathscr{T}. For every compact subset K of TT, set

J(K)=inf⁡i∈Iμi∙(pi(K)). J(K) = \inf_{i \in I} \mu_i^\bullet(p_i(K)).

Then there exists a bounded measure π\pi on TT, and only one, such that π∙(K)=J(K)\pi^\bullet(K) = J(K) for every compact subset KK of TT. One has μi≥pi(π)\mu_i \geq p_i(\pi) for all i∈Ii \in I, and π\pi is the largest measure on TT satisfying this condition.

Let us first prove that J(K)J(K) is the limit of μi∙((pi(K)))\mu_i^\bullet((p_i(K))) with respect to the section filter F\mathfrak{F} of the directed preordered set II: for this, it suffices (GT, IV, §5, No. 2, Th. 2) to show that μi∙(pi(K))≥μj∙(pj(K))\mu_i^\bullet(p_i(K)) \geq \mu_j^\bullet(p_j(K)) for i≤ji \leq j; now, setting μij′=pij(μj)\mu_{ij}' = p_{ij}(\mu_j), we have μij′≤μi\mu_{ij}' \leq \mu_i and pj(K)⊂pij‾−1(pi(K))p_j(K) \subset \overline{p_{ij}}^{-1}(p_i(K)), whence

μj∙(pj(K))≤μj∙(pij‾−1(pi(K)))=(μij′)∙(pi(K))≤μi∙(pi(K)). \mu_j^\bullet(p_j(K)) \leq \mu_j^\bullet(\overline{p_{ij}}^{-1}(p_i(K))) = (\mu_{ij}')^\bullet(p_i(K)) \leq \mu_i^\bullet(p_i(K)).

Let us now pass to the study of the properties of the function JJ: 1) It is clear that J(K)≤J(L)J(K) \leq J(L) when K⊂LK \subset L. 2) Let KK and LL be two compact subsets of TT. For every i∈Ii \in I, we have pi(K∪L)=pi(K)∪pi(L)p_i(K \cup L) = p_i(K) \cup p_i(L), whence

μi∙(pi(K∪L))≤μi∙(pi(K))+μi∙(pi(L)); \mu_i^\bullet(p_i(K \cup L)) \leq \mu_i^\bullet(p_i(K)) + \mu_i^\bullet(p_i(L));

passing to the limit with respect to the filter F\mathfrak{F}, we obtain J(K∪L)≤J(K)+J(L)J(K \cup L) \leq J(K) + J(L). 3) Suppose that the compact sets KK and LL are disjoint. By Prop. 2 of No. 1, there exists an i∈Ii \in I such that pj(K)∩pj(L)=∅p_j(K) \cap p_j(L) = \varnothing for j≥ij \geq i. For j≥ij \geq i we therefore have

μj∙(pj(K∪L))=μj∙(pj(K))+μj∙(pj(L)), \mu_j^\bullet(p_j(K \cup L)) = \mu_j^\bullet(p_j(K)) + \mu_j^\bullet(p_j(L)),

whence J(K∪L)=J(K)+J(L)J(K \cup L) = J(K) + J(L) on passing to the limit with respect to the filter F\mathfrak{F}. 4) Let (Kα)α∈A(K_\alpha)_{\alpha \in A} be a decreasing directed family of compact subsets of TT, with intersection KK. By Prop. 1 of No. 1, pi(K)=⋂α∈Api(Kα)p_i(K) = \bigcap_{\alpha \in A} p_i(K_\alpha) and therefore μi∙(pi(K))=inf⁡α∈Aμi∙(pi(Kα))\mu_i^\bullet(p_i(K)) = \inf_{\alpha \in A} \mu_i^\bullet(p_i(K_\alpha)) for all i∈Ii \in I (§ 1, No. 6, Cor. of Prop. 5). From this, one deduces

J(K)=inf⁡i∈Iμi∙(pi(K))=inf⁡i∈Iinf⁡α∈Aμi∙(pi(Kα))=inf⁡α∈Ainf⁡i∈Iμi∙(pi(Kα))=inf⁡α∈AJ(Kα). J(K) = \inf_{i \in I} \mu_i^\bullet(p_i(K)) = \inf_{i \in I} \inf_{\alpha \in A} \mu_i^\bullet(p_i(K_\alpha)) = \inf_{\alpha \in A} \inf_{i \in I} \mu_i^\bullet(p_i(K_\alpha)) = \inf_{\alpha \in A} J(K_\alpha).
5) Let us choose an i∈Ii \in I and set c=μi∙(Ti)c = \mu_i^\bullet(T_i). Then cc is finite and J(K)≤μi∙(pi(K))≤μi∙(Ti)J(K) \leq \mu_i^\bullet(p_i(K)) \leq \mu_i^\bullet(T_i), thus J(K)≤cJ(K) \leq c for every compact set KK in TT.

The preceding properties permit applying Th. 1 of § 3, No. 1; we conclude that there exists one and only one bounded measure π\pi on TT such that π∙(K)=J(K)\pi^\bullet(K) = J(K) for every compact subset KK of TT. For every i∈Ii \in I, let us denote by νi\nu_i the measure on TiT_i that is the image of π\pi under pip_i. Let i∈Ii \in I, AA a compact subset of TiT_i, and L\mathscr{L} the set of compact subsets of pi−1(A)‾\overline{p_i^{-1}(A)}. By Remark 3 of § 1, No. 2, we have π∙(pi−1(A)‾)=sup⁡K∈Lπ∙(K)\pi^\bullet(\overline{p_i^{-1}(A)}) = \sup_{K \in \mathscr{L}} \pi^\bullet(K); moreover, νi∙(A)=π∙(pi−1(A)‾)\nu_i^\bullet(A) = \pi^\bullet(\overline{p_i^{-1}(A)}) and J(K)=π∙(K)J(K) = \pi^\bullet(K) for K∈LK \in \mathscr{L}, whence νi∙(A)=sup⁡K∈LJ(K)\nu_i^\bullet(A) = \sup_{K \in \mathscr{L}} J(K).

For K∈LK \in \mathscr{L}, we have pi(K)⊂Ap_i(K) \subset A, whence J(K)≤μi∙(pi(K))≤μi∙(A)J(K) \leq \mu_i^\bullet(p_i(K)) \leq \mu_i^\bullet(A) and finally νi∙(A)≤μi∙(A)\nu_i^\bullet(A) \leq \mu_i^\bullet(A). Since AA is an arbitrary compact set in TiT_i, we conclude that νi≤μi\nu_i \leq \mu_i. The last assertion of the proposition is obvious.

Q.E.D.

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.