Bourbaki

Theorem 3

Tag 08PN, permanent. Cite it as [Bourbaki, Tag 08PN].

Integration, chapter IV, § 1. Upper integral of a positive function

— *For every increasing sequence (fn)(f_n) of numerical functions ⩾0\geqslant 0 defined on X,

μ∗(sup⁡nfn)=sup⁡nμ∗(fn). \mu^*\left( \sup_n f_n \right) = \sup_n \mu^*(f_n).

Since each of the functions fnf_n is less than or equal to sup⁡nfn\sup_n f_n, everything comes down to proving that μ∗(sup⁡nfn)⩽sup⁡nμ∗(fn)\mu^*(\sup_n f_n) \leqslant \sup_n \mu^*(f_n); this is obvious if the second member of the inequality is +∞+\infty. In the contrary case, μ∗(fn)<+∞\mu^*(f_n) < +\infty for all nn; we are going to show that, for every ε>0\varepsilon > 0, there exists an increasing sequence (gn)(g_n) of functions in I+\mathscr{I}_+ such that fn⩽gnf_n \leqslant g_n and μ∗(gn)⩽μ∗(fn)+ε\mu^*(g_n) \leqslant \mu^*(f_n) + \varepsilon. If gg is the upper envelope of the sequence (gn)(g_n), we will then have μ∗(g)=sup⁡nμ∗(gn)\mu^*(g) = \sup_n \mu^*(g_n) (No. 1, Th. 1), whence μ∗(g)⩽sup⁡μ∗(fn)+ε\mu^*(g) \leqslant \sup \mu^*(f_n) + \varepsilon; since sup⁡fn⩽g\sup f_n \leqslant g and ε\varepsilon is arbitrary, the theorem will then have been proved.

By hypothesis, there exists a function hn∈J+h_n \in \mathscr{J}_+ such that fn≤hnf_n \leq h_n and μ∗(fn)≤μ∗(hn)≤μ∗(fn)+ε/2n\mu^*(f_n) \leq \mu^*(h_n) \leq \mu^*(f_n) + \varepsilon/2^n; let us show that the functions gn=sup⁡(h1,h2,…,hn)g_n = \sup(h_1, h_2, \ldots, h_n) meet the requirements. They belong to J+\mathscr{J}_+, form an increasing sequence, and satisfy fn≤gnf_n \leq g_n for all nn; we shall prove that

μ∗(gn)≤μ∗(fn)+ε(1−12n). \mu^*(g_n) \leq \mu^*(f_n) + \varepsilon \left( 1 - \frac{1}{2^n} \right).

Let us argue by induction on nn; the case n=1n = 1 is trivial. On the other hand gn+1=sup⁡(gn,hn+1)g_{n+1} = \sup(g_n, h_{n+1}), gn≥fng_n \geq f_n and hn+1≥fn+1≥fnh_{n+1} \geq f_{n+1} \geq f_n, whence inf⁡(gn,hn+1)≥fn\inf(g_n, h_{n+1}) \geq f_n; since

inf⁡(gn,hn+1)+sup⁡(gn,hn+1)=gn+hn+1, \inf(g_n, h_{n+1}) + \sup(g_n, h_{n+1}) = g_n + h_{n+1},

it follows from Th. 2 of No. 1 that

μ∗(gn+1)=μ∗(gn)+μ∗(hn+1)−μ∗(inf⁡(gn,hn+1))≤μ∗(gn)+μ∗(hn+1)−μ∗(fn)≤μ∗(fn+1)+ε(1−12n)+ε2n+1=μ∗(fn+1)+ε(1−12n+1). \begin{align*} \mu^*(g_{n+1}) &= \mu^*(g_n) + \mu^*(h_{n+1}) - \mu^*(\inf(g_n, h_{n+1})) \\ &\leq \mu^*(g_n) + \mu^*(h_{n+1}) - \mu^*(f_n) \\ &\leq \mu^*(f_{n+1}) + \varepsilon \left( 1 - \frac{1}{2^n} \right) + \frac{\varepsilon}{2^{n+1}} \\ &= \mu^*(f_{n+1}) + \varepsilon \left( 1 - \frac{1}{2^{n+1}} \right). \end{align*}

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.