TheoremBase

Proof of Fatou's Lemma

The tail infima gkg_k of the sequence are measurable and increase to its limit inferior, so monotone convergence applies; each ∫g_k is at most the infimum of the later integrals.

Proof

Throughout, ≀\le on [βˆ’βˆž,∞][-\infty,\infty] is a total order (Order and Suprema in the Extended Real Line (i)), and every subset of [βˆ’βˆž,∞][-\infty,\infty] has a supremum and an infimum (Order and Suprema in the Extended Real Line (iii)). By Nonnegative Measurable Maps and Their Lebesgue Integral (i), (fm)(f_{m}) is a sequence in Meas(X,F;[0,∞],B([0,∞]))\mathrm{Meas}(X,\mathcal{F};[0,\infty],\mathcal{B}([0,\infty])), so Suprema, Infima, Limits Inferior and Limits Superior of Measurable [0,∞]-Valued Maps applies to it. Each fmf_{m} is a map Xβ†’[0,∞]X\to[0,\infty], so an element of [0,∞]X[0,\infty]^{X} and, as [0,∞]βŠ†[βˆ’βˆž,∞][0,\infty]\subseteq[-\infty,\infty] (The Extended Real Line, Its Order, and [0,∞] (ii)), of [βˆ’βˆž,∞]X[-\infty,\infty]^{X} (The Elements of the Class of Functions from a Set to a Set, and This Class Is a Set (i)); thus (fm)(f_{m}) is a sequence in [0,∞]X[0,\infty]^{X} and in [βˆ’βˆž,∞]X[-\infty,\infty]^{X}, as Maps Used to Build the Integral: Fibres of a Map, Pointwise Combinations with Finite Image, Tail Extrema of a Sequence of Maps, the Integer Part, and Indicators (iii) requires. For x∈Xx\in X let axa^{x} be the sequence in [0,∞][0,\infty] with amx=fm(x)a^{x}_{m}=f_{m}(x) for every m∈Nm\in\mathbb{N} (Evaluating a Sequence of Maps at a Point Gives a Sequence, and Its Range (i)); so lim inf⁑mfm(x)\liminf_{m}f_{m}(x) is lim inf⁑mamx\liminf_{m}a^{x}_{m}, and g(x)=lim inf⁑mamxg(x)=\liminf_{m}a^{x}_{m}.

Proof of (i). By Suprema, Infima, Limits Inferior and Limits Superior of Measurable [0,∞]-Valued Maps (iii), the map x↦sup⁑kinf⁑{fm(x):m∈N,Β mβ‰₯k}x\mapsto\sup_{k}\inf\{f_{m}(x):m\in\mathbb{N},\ m\ge k\} from XX to [0,∞][0,\infty] is measurable. It is gg, as both have the value lim inf⁑mamx=sup⁑kinf⁑{amx:m∈N,Β mβ‰₯k}\liminf_{m}a^{x}_{m}=\sup_{k}\inf\{a^{x}_{m}:m\in\mathbb{N},\ m\ge k\} at every x∈Xx\in X, with amx=fm(x)a^{x}_{m}=f_{m}(x) (Tail Extrema, Limit Inferior and Limit Superior of a Sequence in [βˆ’βˆž,∞] (ii), Tail Extrema, Limit Inferior and Limit Superior of a Sequence in [βˆ’βˆž,∞] (i)), and maps with equal values are equal (Values, Restriction and Equality of a Map, and the Map u↦t(u)u\mapsto t(u) as a Defined Set Symbol (iv)). So g∈M+g\in\mathcal{M}^{+} by Nonnegative Measurable Maps and Their Lebesgue Integral (i).

Proof of (ii). By Maps Used to Build the Integral: Fibres of a Map, Pointwise Combinations with Finite Image, Tail Extrema of a Sequence of Maps, the Integer Part, and Indicators (iii), there is a sequence (gk)(g_{k}) of maps Xβ†’[βˆ’βˆž,∞]X\to[-\infty,\infty] with gk(x)=inf⁑mβ‰₯kfm(x)=inf⁑{fm(x):m∈N,Β mβ‰₯k}g_{k}(x)=\inf_{m\ge k}f_{m}(x)=\inf\{f_{m}(x):m\in\mathbb{N},\ m\ge k\} for all k∈Nk\in\mathbb{N} and x∈Xx\in X. By Suprema, Infima, Limits Inferior and Limits Superior of Measurable [0,∞]-Valued Maps (ii), applied to J={m∈N:mβ‰₯k}J=\{m\in\mathbb{N}:m\ge k\}, each gkg_{k} takes values in [0,∞][0,\infty] and is measurable, so (gk)(g_{k}) is a sequence in M+\mathcal{M}^{+} (Nonnegative Measurable Maps and Their Lebesgue Integral (i)).

(gk)(g_{k}) is nondecreasing. Let k∈Nk\in\mathbb{N}. If m∈Nm\in\mathbb{N} and mβ‰₯k+1m\ge k+1, then mβ‰₯km\ge k: k<k+1k<k+1 by Arithmetic and Order of the Natural Numbers (xi), so k≀k+1k\le k+1 and k≀mk\le m by Arithmetic and Order of the Natural Numbers (vi). So for x∈Xx\in X, {fm(x):m∈N,Β mβ‰₯k+1}βŠ†{fm(x):m∈N,Β mβ‰₯k}\{f_{m}(x):m\in\mathbb{N},\ m\ge k+1\}\subseteq\{f_{m}(x):m\in\mathbb{N},\ m\ge k\}, and gk(x)≀gk+1(x)g_{k}(x)\le g_{k+1}(x) by Greatest Elements as Suprema, Suprema and Infima of Nested Sets in [βˆ’βˆž,∞], and a Scaling Criterion for Order in [0,∞] (ii).

Monotone convergence. Let x∈Xx\in X. Let uu be the sequence in [βˆ’βˆž,∞][-\infty,\infty] with uk=gk(x)u_{k}=g_{k}(x) for every k∈Nk\in\mathbb{N} (Evaluating a Sequence of Maps at a Point Gives a Sequence, and Its Range (i)), (gk)(g_{k}) being a sequence in [βˆ’βˆž,∞]X[-\infty,\infty]^{X}. Its range is {gk(x):k∈N}\{g_{k}(x):k\in\mathbb{N}\} (Evaluating a Sequence of Maps at a Point Gives a Sequence, and Its Range (ii)), that is {inf⁑mβ‰₯kamx:k∈N}\{\inf_{m\ge k}a^{x}_{m}:k\in\mathbb{N}\} (Maps Used to Build the Integral: Fibres of a Map, Pointwise Combinations with Finite Image, Tail Extrema of a Sequence of Maps, the Integer Part, and Indicators (iii), Tail Extrema, Limit Inferior and Limit Superior of a Sequence in [βˆ’βˆž,∞] (i)); so g(x)=lim inf⁑mamxg(x)=\liminf_{m}a^{x}_{m} is the supremum of the range of uu (Tail Extrema, Limit Inferior and Limit Superior of a Sequence in [βˆ’βˆž,∞] (ii)), that is g(x)=sup⁑kgk(x)g(x)=\sup_{k}g_{k}(x). The sequence kβ†¦βˆ«Xgk dΞΌk\mapsto\int_{X}g_{k}\,d\mu is the composite of (gk)(g_{k}), a map Nβ†’M+\mathbb{N}\to\mathcal{M}^{+}, with the integral map M+β†’[0,∞]\mathcal{M}^{+}\to[0,\infty] (Nonnegative Measurable Maps and Their Lebesgue Integral (ii), Basic Properties of Functions: Equality, Composition, Identity, Inverse and Restriction (ii)), and likewise mβ†¦βˆ«Xfm dΞΌm\mapsto\int_{X}f_{m}\,d\mu is the composite of (fm)(f_{m}) with it. So Monotone Convergence Theorem (ii) applies to (gk)(g_{k}) and gg and gives

∫Xg dΞΌ=sup⁑k∫Xgk dΞΌ.\int_{X}g\,d\mu=\sup_{k}\int_{X}g_{k}\,d\mu.

Comparison. Let k∈Nk\in\mathbb{N} and Ξ²k=inf⁑{∫Xfm dΞΌ:m∈N,Β mβ‰₯k}\beta_{k}=\inf\{\int_{X}f_{m}\,d\mu:m\in\mathbb{N},\ m\ge k\}, so that lim inf⁑m∫Xfm dΞΌ=sup⁑kΞ²k\liminf_{m}\int_{X}f_{m}\,d\mu=\sup_{k}\beta_{k} by Tail Extrema, Limit Inferior and Limit Superior of a Sequence in [βˆ’βˆž,∞] (i) and Tail Extrema, Limit Inferior and Limit Superior of a Sequence in [βˆ’βˆž,∞] (ii), the integrals lying in [0,∞][0,\infty] by Nonnegative Measurable Maps and Their Lebesgue Integral (ii). For every m∈Nm\in\mathbb{N} with mβ‰₯km\ge k and every x∈Xx\in X, gk(x)≀fm(x)g_{k}(x)\le f_{m}(x), as gk(x)g_{k}(x) is a lower bound of {fj(x):jβ‰₯k}\{f_{j}(x):j\ge k\} (Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order (iv)); hence ∫Xgk dΞΌβ‰€βˆ«Xfm dΞΌ\int_{X}g_{k}\,d\mu\le\int_{X}f_{m}\,d\mu by Basic Properties of the Integral of Nonnegative Measurable Maps (iii). So ∫Xgk dΞΌ\int_{X}g_{k}\,d\mu is a lower bound of {∫Xfm dΞΌ:mβ‰₯k}\{\int_{X}f_{m}\,d\mu:m\ge k\}, and ∫Xgk dμ≀βk\int_{X}g_{k}\,d\mu\le\beta_{k}, Ξ²k\beta_{k} being its greatest lower bound. Since Ξ²k≀sup⁑jΞ²j\beta_{k}\le\sup_{j}\beta_{j}, the supremum being an upper bound, ∫Xgk dμ≀lim inf⁑m∫Xfm dΞΌ\int_{X}g_{k}\,d\mu\le\liminf_{m}\int_{X}f_{m}\,d\mu by transitivity (Order and Suprema in the Extended Real Line (i)). Thus lim inf⁑m∫Xfm dΞΌ\liminf_{m}\int_{X}f_{m}\,d\mu is an upper bound of {∫Xgk dΞΌ:k∈N}\{\int_{X}g_{k}\,d\mu:k\in\mathbb{N}\}, and

∫Xg dΞΌ=sup⁑k∫Xgk dμ≀lim inf⁑m∫Xfm dΞΌ,\int_{X}g\,d\mu=\sup_{k}\int_{X}g_{k}\,d\mu\le\liminf_{m}\int_{X}f_{m}\,d\mu,

as the supremum is the least upper bound (Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order (iv)).

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…