Jump to content

Uniform integrability

From Wikipedia, the free encyclopedia

In mathematics, uniform integrability is an important concept in real analysis, functional analysis and measure theory, and plays a vital role in the theory of martingales.

Measure-theoretic definition

[edit]

Uniform integrability is an extension to the notion of a family of functions being dominated in L1 which is central in dominated convergence. Several textbooks on real analysis and measure theory use the following definition:[1]

Definition A: Let (X,𝔐,μ) be a positive measure space. A set Φ⊂L1(μ) is called uniformly integrable if supf∈Φ‖f‖L1(μ)<∞, and to each ε>0 there corresponds a δ>0 such that

∫E|f|dμ<ε

whenever f∈Φ and μ(E)<δ.

Definition A is rather restrictive for infinite measure spaces. A more general definition[2] of uniform integrability that works well in general measure spaces was introduced by G. A. Hunt.

Definition H: Let (X,𝔐,μ) be a positive measure space. A set Φ⊂L1(μ) is called uniformly integrable if and only if

infg∈L+1(μ)supf∈Φ∫{|f|>g}|f|dμ=0

where L+1(μ)={g∈L1(μ):g≥0}.


Since Hunt's definition is equivalent to Definition A when the underlying measure space is finite (see Theorem 2 below), Definition H is widely adopted in Mathematics.

The following result[3] provides another equivalent notion to Hunt's. This equivalency is sometimes given as definition for uniform integrability.

Theorem 1: If (X,𝔐,μ) is a (positive) finite measure space, then a set Φ⊂L1(μ) is uniformly integrable if and only if

infg∈L+1(μ)supf∈Φ∫(|f|−g)+dμ=0

If in addition μ(X)<∞, then uniform integrability is equivalent to either of the following conditions

1. infa>0supf∈Φ∫(|f|−a)+dμ=0.

2. infa>0supf∈Φ∫{|f|>a}|f|dμ=0

When the underlying space (X,𝔐,μ) is σ-finite, Hunt's definition is equivalent to the following:

Theorem 2: Let (X,𝔐,μ) be a σ-finite measure space, and h∈L1(μ) be such that h>0 almost everywhere. A set Φ⊂L1(μ) is uniformly integrable if and only if supf∈Φ‖f‖L1(μ)<∞, and for any ε>0, there exists δ>0 such that

supf∈Φ∫A|f|dμ<ε

whenever ∫Ahdμ<δ.

A consequence of Theorems 1 and 2 is that equivalence of Definitions A and H for finite measures follows. Indeed, the statement in Definition A is obtained by taking h≡1 in Theorem 2.

Tightness, boundedness, equi-integrability and uniform integrability

[edit]

Another concept associated with uniform integrability is that of tightness. In this article tightness is taken in a more general setting.

Definition: Suppose (X,𝔐,μ) is a measure space. Let 𝒦⊂𝔐 be a collection of sets of finite measure. A family Φ⊂L1(μ) is said to be tight with respect to 𝒦 if

infK∈𝒦supf∈Φ∫X∖K|f|dμ=0

When 𝒦=𝔐∩L1(μ), Φ is simply said to be tight.

When the measure space (X,𝔐,μ) is a metric space equipped with the Borel σ algebra, μ is a regular measure, and 𝒦 is the collection of all compact subsets of X, the notion of 𝒦-tightness discussed above coincides with the well known concept of tightness used in the analysis of regular measures in metric spaces

For σ-finite measure spaces, it can be shown that if a family Φ⊂L1(μ) is uniformly integrable, then Φ is tight. This is captured by the following result which is often used as definition of uniform integrability in the analysis literature:

Theorem 3: Suppose (X,𝔐,μ) is a σ-finite measure space. A family Φ⊂L1(μ) is uniformly integrable if and only if

  1. supf∈Φ‖f‖1<∞.
  2. infa>0supf∈Φ∫{|f|>a}|f|dμ=0
  3. Φ is tight.

When μ(X)<∞, condition 3 is redundant (see Theorem 1 above).

In many books in analysis [4][5][6][7], condition 2 in Theorem 3 is often replaced by another condition called equi-integrability:

Definition: A family 𝒞 of complex or real valued measurable functions is equi-integrable (or uniformly absolutely continuous with respect to a measure μ) if for any ε>0 there is δ>0 such that supf∈𝒞∫A|f|dμ<εwheneverμ(A)<δ

Theorem 3 then says that equi-integrability together with L1 boundedness and tightness (conditions (1) and (3) in Theorem 3) is equivalent to uniform integrability.

Relevant theorems

[edit]

The following theorems describe very useful criteria for uniform integrability which have many applications in Analysis and Probability.

de la Vallée-Poussin theorem[8][9]

Suppose (X,𝔐,μ) is a finite measure space. The family ℱ⊂L1(μ) is uniformly integrable if and only if there exists a function G:[0,∞)→[0,∞) such that limt→∞G(t)t=∞ and supf∈ℱ∫XG(|f|)dμ<∞. The function G can be chosen to be monotone increasing and convex.

Uniform integrability gives a characterization of weak compactness in L1.

Dunford–Pettis theorem[10][11]

Suppose (X,𝔐,μ) is a σ-finite measure. A family ℱ⊂L1(μ) has compact closure in the weak topology σ(L1,L∞) if and only if ℱ is uniformly integrable.

Probability definition

[edit]

In probability theory, Definition A or the statement of Theorem 1 are often presented as definitions of uniform integrability using the notation expectation of random variables.,[12][13][14] that is,

1. A class 𝒞 of random variables is called uniformly integrable if:

  • There exists a finite M such that, for every X in 𝒞, E⁡(|X|)≤M and
  • For every ε>0 there exists δ>0 such that, for every measurable A such that P(A)≤δ and every X in 𝒞, E⁡(|X|IA)≤ε.

or alternatively

2. A class 𝒞 of random variables is called uniformly integrable (UI) if for every ε>0 there exists K∈[0,∞) such that E⁡(|X|I|X|≥K)≤ε  for all X∈𝒞, where I|X|≥K is the indicator function I|X|≥K={1if |X|≥K,0if |X|<K..

[edit]

The following results apply to the probabilistic definition.[15]

  • Definition 1 could be rewritten by taking the limits as limK→∞supX∈𝒞E⁡(|X|I|X|≥K)=0.
  • A non-UI sequence. Let Ω=[0,1]⊂ℝ, and define Xn(ω)={n,ω∈(0,1/n),0,otherwise. Clearly Xn∈L1, and indeed E⁡(|Xn|)=1 , for all n. However, E⁡(|Xn|I{|Xn|≥K})=1  for all n≥K, and comparing with definition 1, it is seen that the sequence is not uniformly integrable.
Non-UI sequence of RVs. The area under the strip is always equal to 1, but Xn→0 pointwise.
  • By using Definition 2 in the above example, it can be seen that the first clause is satisfied as L1 norm of all Xns are 1 i.e., bounded. But the second clause does not hold as given any δ positive, there is an interval (0,1/n) with measure less than δ and E[|Xm|:(0,1/n)]=1 for all m≥n.
  • If X is a UI random variable, by splitting E⁡(|X|)=E⁡(|X|I{|X|≥K})+E⁡(|X|I{|X|<K}) and bounding each of the two, it can be seen that a uniformly integrable random variable is always bounded in L1.
  • If any sequence of random variables Xn is dominated by an integrable, non-negative Y: that is, for all ω and n, |Xn(ω)|≤Y(ω), Y(ω)≥0, E⁡(Y)<∞, then the class 𝒞 of random variables {Xn} is uniformly integrable.
  • A class of random variables bounded in Lp (p>1) is uniformly integrable.

Uniform integrability and stochastic ordering

[edit]

A family of random variables {Xi}i∈I is uniformly integrable if and only if[16][17] there exists a random variable X such that EX<∞ and |Xi|≤icxX for all i∈I, where ≤icx denotes the increasing convex stochastic order defined by A≤icxB if Eϕ(A)≤Eϕ(B) for all nondecreasing convex real functions ϕ.

Relation to convergence of random variables

[edit]

A sequence {Xn} converges to X in the L1 norm if and only if it converges in measure to X and it is uniformly integrable. In probability terms, a sequence of random variables converging in probability also converge in the mean if and only if they are uniformly integrable.[18] This is a generalization of Lebesgue's dominated convergence theorem, see Vitali convergence theorem.

Citations

[edit]
  1. ↑ Royden, H.L. & Fitzpatrick, P.M. (2010). Real Analysis (4 ed.). Boston: Prentice Hall. p. 93. ISBN 978-0-13-143747-0.
  2. ↑ Hunt, G. A. (1966). Martingales et Processus de Markov. Paris: Dunod. p. 33.
  3. ↑ Klenke, A. (2008). Probability Theory: A Comprehensive Course. Berlin: Springer Verlag. pp. 134–137. ISBN 978-1-84800-047-6.
  4. ↑ Fonseca, Irene; Leoni, Giovanni (2007). Modern Methods in the Calculus of Variations: Lp Spaces. New York, NY: Springer New York Springer e-books. ISBN 978-0387690063.
  5. ↑ Benedetto, J. J. (1976). Real Variable and Integration. Stuttgart: B. G. Teubner. p. 89. ISBN 3-519-02209-5.
  6. ↑ Burrill, C. W. (1972). Measure, Integration, and Probability. McGraw-Hill. p. 180. ISBN 0-07-009223-0.
  7. ↑ Bass, Richard F. (2011). Stochastic Processes. Cambridge: Cambridge University Press. pp. 356–357. ISBN 978-1-107-00800-7.
  8. ↑ Meyer, P.A. (1966). Probability and Potentials, Blaisdell Publishing Co, N. Y. (p.19, Theorem T22).
  9. ↑ De La Vallée Poussin, C. (1915). "Sur L'Integrale de Lebesgue". Transactions of the American Mathematical Society. 16 (4): 435–501. doi:10.2307/1988879. hdl:10338.dmlcz/127627. JSTOR 1988879.
  10. ↑ Dunford, Nelson (1938). "Uniformity in linear spaces". Transactions of the American Mathematical Society. 44 (2): 305–356. doi:10.1090/S0002-9947-1938-1501971-X. ISSN 0002-9947.
  11. ↑ Dunford, Nelson (1939). "A mean ergodic theorem". Duke Mathematical Journal. 5 (3): 635–646. doi:10.1215/S0012-7094-39-00552-1. ISSN 0012-7094.
  12. ↑ Williams, David (1997). Probability with Martingales (Repr. ed.). Cambridge: Cambridge Univ. Press. pp. 126–132. ISBN 978-0-521-40605-5.
  13. ↑ Gut, Allan (2005). Probability: A Graduate Course. Springer. pp. 214–218. ISBN 0-387-22833-0.
  14. ↑ Bass, Richard F. (2011). Stochastic Processes. Cambridge: Cambridge University Press. pp. 356–357. ISBN 978-1-107-00800-7.
  15. ↑ Gut 2005, pp. 215–216.
  16. ↑ Chong, K. M. (1974). "Spectral orders, uniform integrability and Lebesgue's Dominated Convergence Theorem". Transactions of the American Mathematical Society. 191: 395–404 (Thm 4.2). doi:10.1090/S0002-9947-1974-0369646-2.
  17. ↑ Leskelä, L.; Vihola, M. (2013). "Stochastic order characterization of uniform integrability and tightness". Statistics and Probability Letters. 83 (1): 382–389. arXiv:1106.0607. doi:10.1016/j.spl.2012.09.023.
  18. ↑ Bogachev, Vladimir I. (2007). "The spaces Lp and spaces of measures". Measure Theory Volume I. Berlin Heidelberg: Springer-Verlag. p. 268. doi:10.1007/978-3-540-34514-5_4. ISBN 978-3-540-34513-8.

References

[edit]