The lemma of Fatou (after Pierre Fatou ) allows in mathematics to estimate the Lebesgue integral of the Limes inferior of a function sequence through the Limes inferior of the sequence of the associated Lebesgue integrals upwards. It thus provides a statement about the interchangeability of limit value processes .
Mathematical formulation
Be a measure space . For every sequence of non-negative, measurable functions, the following applies
(
S.
,
Σ
,
μ
)
{\ displaystyle (S, \ Sigma, \ mu)}
(
f
n
)
n
∈
N
{\ displaystyle (f_ {n}) _ {n \ in \ mathbb {N}}}
f
n
:
S.
→
R.
∪
{
∞
}
{\ displaystyle f_ {n} \ colon S \ to \ mathbb {R} \ cup \ {\ infty \}}
∫
S.
lim inf
n
→
∞
f
n
d
μ
≤
lim inf
n
→
∞
∫
S.
f
n
d
μ
,
{\ displaystyle \ int _ {S} \ liminf _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu \ leq \ liminf _ {n \ rightarrow \ infty} \ int _ {S} f_ {n} \ \ mathrm {d} \ mu,}
whereby on the left side the Limes inferior of the sequence is to be understood point by point .
(
f
n
)
n
∈
N
{\ displaystyle (f_ {n}) _ {n \ in \ mathbb {N}}}
Analogously, this set also for the Limes superior, unless a non-negative integrable function with are:
G
{\ displaystyle g}
f
n
≤
G
{\ displaystyle f_ {n} \ leq g}
∫
S.
lim sup
n
→
∞
f
n
d
μ
≥
lim sup
n
→
∞
∫
S.
f
n
d
μ
{\ displaystyle \ int _ {S} \ limsup _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu \ geq \ limsup _ {n \ rightarrow \ infty} \ int _ {S} f_ {n} \ \ mathrm {d} \ mu}
.
This can be summarized as the rule of thumb
∫
S.
lim inf
n
→
∞
f
n
d
μ
≤
lim inf
n
→
∞
∫
S.
f
n
d
μ
≤
lim sup
n
→
∞
∫
S.
f
n
d
μ
≤
∫
S.
lim sup
n
→
∞
f
n
d
μ
{\ displaystyle \ int _ {S} \ liminf _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu \ leq \ liminf _ {n \ rightarrow \ infty} \ int _ {S} f_ {n} \ \ mathrm {d} \ mu \ leq \ limsup _ {n \ rightarrow \ infty} \ int _ {S} f_ {n} \ \ mathrm {d} \ mu \ leq \ int _ {S} \ limsup _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu}
.
Proof idea
In order to prove Fatou's lemma for the Limes inferior, one turns to the monotonically increasing sequence of functions
G
n
: =
inf
k
≥
n
f
k
↗
lim inf
n
→
∞
f
n
{\ displaystyle g_ {n}: = \ inf _ {k \ geq n} f_ {k} \ nearrow \ liminf _ {n \ rightarrow \ infty} f_ {n}}
the theorem of monotonous convergence . With the resulting equation and the inequality based on the monotony of the integral
∫
S.
(
inf
k
≥
n
f
k
)
≤
∫
S.
f
n
{\ displaystyle \ int _ {S} \ left (\ inf _ {k \ geq n} f_ {k} \ right) \ leq \ int _ {S} f_ {n}}
one obtains from the calculation rules for the Limes:
∫
S.
lim inf
n
→
∞
f
n
=
lim
n
→
∞
∫
S.
(
inf
k
≥
n
f
k
)
≤
lim inf
n
→
∞
∫
S.
f
n
{\ displaystyle \ int _ {S} \ liminf _ {n \ rightarrow \ infty} f_ {n} = \ lim _ {n \ rightarrow \ infty} \ int _ {S} \ left (\ inf _ {k \ geq n} f_ {k} \ right) \ leq \ liminf _ {n \ rightarrow \ infty} \ int _ {S} f_ {n}}
.
For Fatou's lemma with limes superior analog can be moved, because, by assumption, is
to be integrated, that is integrated.
G
1
=
sup
k
≥
1
f
k
≤
G
{\ displaystyle g_ {1} = \ sup _ {k \ geq 1} f_ {k} \ leq g}
G
{\ displaystyle g}
G
1
{\ displaystyle g_ {1}}
Examples of strict inequality
The base space is provided with Borel's σ-algebra and the Lebesgue measure .
S.
{\ displaystyle S}
Example for a probability space : Let be the unit interval. Define for all and , where denotes the indicator function of the interval .
S.
=
[
0
,
1
]
{\ displaystyle S = [0,1]}
f
n
(
x
)
=
n
1
(
0
,
1
n
)
(
x
)
{\ displaystyle f_ {n} (x) = n \ mathbf {1} _ {(0, {\ frac {1} {n}})} (x)}
n
∈
N
{\ displaystyle n \ in \ mathbb {N}}
x
∈
S.
{\ displaystyle x \ in S}
1
(
0
,
1
n
)
{\ displaystyle \ mathbf {1} _ {(0, {\ frac {1} {n}})}}
(
0
,
1
n
)
{\ displaystyle (0, {\ tfrac {1} {n}})}
Example with uniform convergence : Let be the set of real numbers . Define for everyone and . (Note that there is no integrable majorante in this example and therefore the sup part of Fatou's lemma is not applicable.)
S.
{\ displaystyle S}
f
n
(
x
)
=
1
n
1
[
0
,
n
]
(
x
)
{\ displaystyle f_ {n} (x) = {\ tfrac {1} {n}} \ mathbf {1} _ {[0, n]} (x)}
n
∈
N
{\ displaystyle n \ in \ mathbb {N}}
x
∈
S.
{\ displaystyle x \ in S}
Each has an integral one,
f
n
{\ displaystyle f_ {n}}
∫
S.
f
n
d
μ
=
1
{\ displaystyle \ int _ {S} f_ {n} \ \ mathrm {d} \ mu = 1}
therefore applies
1
=
lim
n
→
∞
∫
S.
f
n
d
μ
=
lim inf
n
→
∞
∫
S.
f
n
d
μ
=
lim sup
n
→
∞
∫
S.
f
n
d
μ
{\ displaystyle 1 = \ lim _ {n \ rightarrow \ infty} \ int _ {S} f_ {n} \ \ mathrm {d} \ mu = \ liminf _ {n \ rightarrow \ infty} \ int _ {S} f_ {n} \ \ mathrm {d} \ mu = \ limsup _ {n \ rightarrow \ infty} \ int _ {S} f_ {n} \ \ mathrm {d} \ mu}
The sequence converges point by point to the null function
(
f
n
)
n
∈
N
{\ displaystyle (f_ {n}) _ {n \ in \ mathbb {N}}}
S.
{\ displaystyle S}
0
=
lim
n
→
∞
f
n
=
lim inf
n
→
∞
f
n
=
lim sup
n
→
∞
f
n
,
{\ displaystyle 0 = \ lim _ {n \ rightarrow \ infty} f_ {n} = \ liminf _ {n \ rightarrow \ infty} f_ {n} = \ limsup _ {n \ rightarrow \ infty} f_ {n}, }
therefore the integral is also zero
0
=
∫
S.
lim inf
n
→
∞
f
n
d
μ
=
∫
S.
lim sup
n
→
∞
f
n
d
μ
,
{\ displaystyle 0 = \ int _ {S} \ liminf _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu = \ int _ {S} \ limsup _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu,}
therefore the strict inequalities apply here
∫
S.
lim inf
n
→
∞
f
n
d
μ
<
lim inf
n
→
∞
∫
S.
f
n
d
μ
,
{\ displaystyle \ int _ {S} \ liminf _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu <\ liminf _ {n \ rightarrow \ infty} \ int _ {S} f_ { n} \ \ mathrm {d} \ mu,}
∫
S.
lim sup
n
→
∞
f
n
d
μ
<
lim sup
n
→
∞
∫
S.
f
n
d
μ
{\ displaystyle \ int _ {S} \ limsup _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu <\ limsup _ {n \ rightarrow \ infty} \ int _ {S} f_ { n} \ \ mathrm {d} \ mu}
Discussion of the requirements
The precondition of the nonnegativity of the individual functions cannot be dispensed with, as the following example shows: Let the half-open interval with Borel's σ-algebra and the Lebesgue measure. Define for everyone . The sequence converges on (even uniformly) to the null function (with integral 0), but each has an integral −1. thats why
S.
{\ displaystyle S}
[
0
,
∞
)
{\ displaystyle [0, \ infty)}
n
∈
N
{\ displaystyle n \ in \ mathbb {N}}
f
n
(
x
)
: =
-
1
n
1
[
0
,
n
]
(
x
)
{\ displaystyle f_ {n} (x): = - {\ tfrac {1} {n}} {\ mathfrak {1}} _ {[0, n]} (x)}
(
f
n
)
n
∈
N
{\ displaystyle (f_ {n}) _ {n \ in \ mathbb {N}}}
S.
{\ displaystyle S}
f
n
{\ displaystyle f_ {n}}
0
=
∫
S.
lim
n
→
∞
f
n
d
μ
>
lim
n
→
∞
∫
S.
f
n
d
μ
=
-
1
{\ displaystyle 0 = \ int _ {S} \ lim _ {n \ rightarrow \ infty} f_ {n} \ \ mathrm {d} \ mu> \ lim _ {n \ rightarrow \ infty} \ int _ {S} f_ {n} \ mathrm {d} \ mu = -1}
.
See also
literature
<img src="https://de.wikipedia.org/wiki/Special:CentralAutoLogin/start?type=1x1" alt="" title="" width="1" height="1" style="border: none; position: absolute;">