Bourbaki

Definition 1

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

Integration, chapter IV, § 6. Convexity inequalities

— Let X be a locally compact space, μ\mu a measure on X. Given a numerical function f (finite or not), defined locally almost everywhere in X, we call maximum in measure, or μ\mu-maximum (resp. minimum in measure, or μ\mu-minimum) of the function f, and denote by M∞(f)M_\infty(f) (resp. m∞(f)m_\infty(f)), the infimum (resp. supremum) of the set of numbers α\alpha such that f(x)≤αf(x) \leq \alpha (resp. f(x)≥αf(x) \geq \alpha) locally almost everywhere (for μ\mu).

It follows at once from the definition that m∞(f)=−M∞(−f)m_\infty(f) = -M_\infty(-f), thus from every property of the maximum in measure one deduces a corresponding property of the minimum in measure.

For every α>M∞(f)\alpha > M_\infty(f), the set of x∈Xx \in X such that f(x)>αf(x) > \alpha is locally negligible; now, the set of x∈Xx \in X such that f(x)>M∞(f)f(x) > M_\infty(f) is the union of the sets where f(x)>rnf(x) > r_n, with rnr_n running over the set of rational numbers >M∞(f)> M_\infty(f); therefore f(x)≤M∞(f)f(x) \leq M_\infty(f) locally almost everywhere (§ 5, No. 2). Similarly f(x)≥m∞(f)f(x) \geq m_\infty(f) locally almost everywhere; it follows that m∞(f)≤M∞(f)m_\infty(f) \leq M_\infty(f) if the measure μ\mu is nonzero; moreover, the relation m∞(f)=M∞(f)m_\infty(f) = M_\infty(f) is equivalent to saying that ff is equal to a constant locally almost everywhere. It is clear that if the measure μ\mu is nonzero, then

inf⁡x∈Xf(x)≤m∞(f)≤M∞(f)≤sup⁡x∈Xf(x). \inf_{x \in X} f(x) \leq m_\infty(f) \leq M_\infty(f) \leq \sup_{x \in X} f(x).

If two functions f,gf, g are equal locally almost everywhere, then m∞(f)=m∞(g)m_\infty(f) = m_\infty(g) and M∞(f)=M∞(g)M_\infty(f) = M_\infty(g).

Finally, if ff and gg are two functions such that f+gf + g is defined locally almost everywhere, then

(1)

M∞(f+g)≤M∞(f)+M∞(g) M_\infty(f + g) \leq M_\infty(f) + M_\infty(g)
provided the second member is defined, as follows at once from Def. 1; similarly, if ff and gg are both ≥0\geq 0 and are such that fgfg is defined locally almost everywhere, then

(2)

M∞(fg)≤M∞(f)M∞(g) M_\infty(fg) \leq M_\infty(f) M_\infty(g)
provided the second member is defined.

If M∞(f)<+∞M_\infty(f) < +\infty, then f(x)<+∞f(x) < +\infty locally almost everywhere, but not necessarily almost everywhere. A numerical function ff is said to be bounded in measure (for the measure μ\mu) if it is defined and finite almost everywhere and if, moreover, the numbers m∞(f)m_\infty(f) and M∞(f)M_\infty(f) are both finite (the latter condition amounts to saying that M∞(∣f∣)<+∞M_\infty(|f|) < +\infty).

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

Cited by

Nothing.

Cites

Transcribed from the printed volume. Not translated.