The uniform integrability , also called uniform integrability , is a reinforcement of the concept of integrability in mathematics . In contrast to integrability, it is always a property of a family of functions and not that of a single function. However, the family can also be one element. The equilibrium integrability is particularly important in probability theory and measure theory , as it enables, using Vitali's convergence theorem, to establish a connection from the convergence in the p-th mean to the convergence in probability of the probability theory and the convergence according to measure or the Beat convergence locally according to the measure of measure theory. A family of functions can clearly be integrated evenly if the integral over “small” quantities does not assume too large values.
definition
Let be a measure space with measure and be the set of functions that can be integrated with respect to this measure. The positive part of a function is denoted by.
(
Ω
,
A.
,
μ
)
{\ displaystyle (\ Omega, {\ mathcal {A}}, \ mu)}
μ
{\ displaystyle \ mu}
L.
1
(
μ
)
{\ displaystyle {\ mathcal {L}} ^ {1} (\ mu)}
f
+
: =
Max
{
0
,
f
}
{\ displaystyle f ^ {+}: = \ max \ {0, f \}}
For general measurements
A family of functions is said to be equally integrable if it fulfills one of the following equivalent definitions:
F.
⊂
L.
1
(
μ
)
{\ displaystyle {\ mathcal {F}} \ subset {\ mathcal {L}} ^ {1} (\ mu)}
The following applies to all positive functions
G
∈
L.
1
(
μ
)
{\ displaystyle g \ in {\ mathcal {L}} ^ {1} (\ mu)}
inf
G
≥
0
sup
f
∈
F.
∫
(
|
f
|
-
G
)
+
d
μ
=
inf
G
≥
0
sup
f
∈
F.
∫
{
|
f
|
>
G
}
|
f
|
d
μ
=
0
{\ displaystyle \ inf _ {g \ geq 0} \ sup _ {f \ in {\ mathcal {F}}} \ int (\ vert f \ vert -g) ^ {+} \ mathrm {d} \ mu = \ inf _ {g \ geq 0} \ sup _ {f \ in {\ mathcal {F}}} \ int _ {\ {| f |> g \}} \ vert f \ vert \ mathrm {d} \ mu = 0}
.
The following two conditions are met:
It is
sup
f
∈
F.
∫
|
f
|
d
μ
<
∞
{\ displaystyle \ sup _ {f \ in {\ mathcal {F}}} \ int \ vert f \ vert \ mathrm {d} \ mu <\ infty}
There is an integrable function , so that for any one there is one, so that for every quantity with
0
≤
H
∈
L.
1
(
μ
)
{\ displaystyle 0 \ leq h \ in {\ mathcal {L}} ^ {1} (\ mu)}
ε
>
0
{\ displaystyle \ varepsilon> 0}
δ
(
ε
)
>
0
{\ displaystyle \ delta (\ varepsilon)> 0}
A.
∈
A.
{\ displaystyle A \ in {\ mathcal {A}}}
∫
A.
H
d
μ
<
δ
(
ε
)
{\ displaystyle \ int _ {A} h \; \ mathrm {d} \ mu <\ delta (\ varepsilon)}
holds that
sup
f
∈
F.
∫
A.
|
f
|
d
μ
≤
ε
{\ displaystyle \ sup _ {f \ in {\ mathcal {F}}} \ int _ {A} | f | \; \ mathrm {d} \ mu \ leq \ varepsilon}
is.
For finite dimensions
If the measure is finite, so is , then the definitions simplify. The family is said to be equally integrable if one of the following equivalent definitions applies:
μ
(
Ω
)
<
∞
{\ displaystyle \ mu (\ Omega) <\ infty}
inf
a
∈
[
0
,
∞
)
sup
f
∈
F.
∫
(
|
f
|
-
a
)
+
d
μ
=
0
{\ displaystyle \ inf _ {a \ in [0, \ infty)} \ sup _ {f \ in {\ mathcal {F}}} \ int (| f | -a) ^ {+} \ mathrm {d} \ mu = 0}
inf
a
∈
[
0
,
∞
)
sup
f
∈
F.
∫
{
a
<
|
f
|
}
|
f
|
d
μ
=
0
{\ displaystyle \ inf _ {a \ in [0, \ infty)} \ sup _ {f \ in {\ mathcal {F}}} \ int _ {\ {a <| f | \}} | f | \ mathrm {d} \ mu = 0}
The following two conditions are met:
It is .
sup
f
∈
F.
∫
|
f
|
d
μ
<
∞
{\ displaystyle \ sup _ {f \ in {\ mathcal {F}}} \ int \ vert f \ vert \ mathrm {d} \ mu <\ infty}
For anything one exists , so that for everyone with
ε
>
0
{\ displaystyle \ varepsilon> 0}
δ
(
ε
)
>
0
{\ displaystyle \ delta (\ varepsilon)> 0}
A.
∈
A.
{\ displaystyle A \ in {\ mathcal {A}}}
μ
(
A.
)
<
δ
(
ε
)
{\ displaystyle \ mu (A) <\ delta (\ varepsilon)}
sup
f
∈
F.
∫
A.
|
f
|
d
μ
≤
ε
{\ displaystyle \ sup _ {f \ in {\ mathcal {F}}} \ int _ {A} | f | \ mathrm {d} \ mu \ leq \ varepsilon}
applies.
Equally integrable in the p-th mean
A family of functions is said to be equally integrable in the p-th mean , if the family is equally integrable.
(
f
i
)
i
∈
I.
{\ displaystyle (f_ {i}) _ {i \ in I}}
(
|
f
i
|
p
)
i
∈
I.
{\ displaystyle (| f_ {i} | ^ {p}) _ {i \ in I}}
properties
Every finite set can be integrated equally.
F.
⊂
L.
1
(
μ
)
{\ displaystyle {\ mathcal {F}} \ subset {\ mathcal {L}} ^ {1} (\ mu)}
Be families of functions and be equally integrable. Exists for each one , so that , is also integrable equally.
F.
,
G
{\ displaystyle {\ mathcal {F}}, {\ mathcal {G}}}
F.
{\ displaystyle {\ mathcal {F}}}
G
∈
G
{\ displaystyle g \ in {\ mathcal {G}}}
f
∈
F.
{\ displaystyle f \ in {\ mathcal {F}}}
|
G
|
≤
|
f
|
{\ displaystyle \ vert g \ vert \ leq \ vert f \ vert}
G
{\ displaystyle {\ mathcal {G}}}
If one exists , so that for all , it is equally integrable. This is a direct special case of the two above properties
G
∈
L.
1
(
μ
)
{\ displaystyle g \ in L ^ {1} (\ mu)}
|
f
|
≤
|
G
|
{\ displaystyle | f | \ leq | g |}
f
∈
F.
{\ displaystyle f \ in {\ mathcal {F}}}
F.
{\ displaystyle {\ mathcal {F}}}
A sequence of measurable functions converges on the mean , i.e. with respect to the norm against a function , when it converges in terms of measure and is equally integrable. This follows from Vitali's convergence theorem .
f
n
{\ displaystyle f_ {n}}
L.
1
{\ displaystyle L ^ {1}}
f
{\ displaystyle f}
General converge a sequence of measurable functions if and in pth mean , that with respect to the norm to a function when the dimensions to be uniformly integrable converges in pth mean. This statement also follows from Vitali's convergence theorem.
f
n
{\ displaystyle f_ {n}}
L.
p
{\ displaystyle L ^ {p}}
f
{\ displaystyle f}
Are equi-integrable families, so too , for , , and equi-integrable families. The operations are always to be understood element by element, if not clear.
F.
,
G
{\ displaystyle {\ mathcal {F}}, {\ mathcal {G}}}
α
F.
+
β
G
{\ displaystyle \ alpha {\ mathcal {F}} + \ beta {\ mathcal {G}}}
α
,
β
∈
R.
{\ displaystyle \ alpha, \ beta \ in \ mathbb {R}}
F.
∪
G
{\ displaystyle {\ mathcal {F}} \ cup {\ mathcal {G}}}
F.
∩
G
{\ displaystyle {\ mathcal {F}} \ cap {\ mathcal {G}}}
|
F.
|
{\ displaystyle \ vert {\ mathcal {F}} \ vert}
literature
Jürgen Elstrodt : Measure and integration theory . 6th, corrected edition. Springer-Verlag, Berlin Heidelberg 2009, ISBN 978-3-540-89727-9 , doi : 10.1007 / 978-3-540-89728-6 .
Achim Klenke : Probability Theory . 3. Edition. Springer-Verlag, Berlin Heidelberg 2013, ISBN 978-3-642-36017-6 , doi : 10.1007 / 978-3-642-36018-3 .
Norbert Kusolitsch: Measure and probability theory . An introduction. 2nd, revised and expanded edition. Springer-Verlag, Berlin Heidelberg 2014, ISBN 978-3-642-45386-1 , doi : 10.1007 / 978-3-642-45387-8 .
<img src="https://de.wikipedia.org/wiki/Special:CentralAutoLogin/start?type=1x1" alt="" title="" width="1" height="1" style="border: none; position: absolute;">