Suppose that
f
{\displaystyle f\!\,}
is a continuous real-valued function with domain
(
−
∞
,
∞
)
{\displaystyle (-\infty ,\infty )\!\,}
and that
f
{\displaystyle f\!\,}
is absolutely continuous on every finite interval
[
a
,
b
]
{\displaystyle [a,b]\!\,}
.
Prove: If
f
{\displaystyle f\!\,}
and
f
′
{\displaystyle f^{\prime }\!\,}
are both integrable on
(
−
∞
,
∞
)
{\displaystyle (-\infty ,\infty )\!\,}
, then
∫
−
∞
∞
f
′
=
0
{\displaystyle \int _{-\infty }^{\infty }f^{\prime }=0\!\,}
Since
f
{\displaystyle f\!\,}
is absolutely continuous for all
[
a
,
b
]
⊂
R
1
{\displaystyle [a,b]\subset R^{1}\!\,}
,
∫
a
b
f
′
(
x
)
d
x
=
f
(
b
)
−
f
(
a
)
{\displaystyle \int _{a}^{b}f^{\prime }(x)dx=f(b)-f(a)\!\,}
Hence
∫
−
∞
∞
f
′
(
x
)
d
x
=
lim
a
,
b
→
∞
∫
a
b
f
′
(
x
)
=
lim
a
,
b
→
∞
[
f
(
b
)
−
f
(
a
)
]
{\displaystyle \int _{-\infty }^{\infty }f^{\prime }(x)dx=\lim _{a,b\rightarrow \infty }\int _{a}^{b}f^{\prime }(x)=\lim _{a,b\rightarrow \infty }[f(b)-f(a)]\!\,}
Since
f
′
{\displaystyle f^{\prime }\!\,}
is integrable i.e.
f
′
∈
L
1
(
−
∞
,
∞
)
{\displaystyle f^{\prime }\in L^{1}(-\infty ,\infty )\!\,}
,
lim
b
→
∞
f
(
b
)
{\displaystyle \lim _{b\rightarrow \infty }f(b)\!\,}
and
lim
a
→
∞
f
(
a
)
{\displaystyle \lim _{a\rightarrow \infty }f(a)\!\,}
exist.
Assume for the sake of contradiction that
lim
b
→
∞
|
f
(
b
)
|
=
δ
>
0
{\displaystyle \lim _{b\rightarrow \infty }|f(b)|=\delta >0\!\,}
Then there exists
M
{\displaystyle M\!\,}
such that for all
x
>
M
{\displaystyle x>M\!\,}
|
f
(
x
)
|
>
δ
2
{\displaystyle |f(x)|>{\frac {\delta }{2}}\!\,}
since
f
{\displaystyle f\!\,}
is continuous. (At some point,
|
f
|
{\displaystyle |f|\!\,}
will either monotonically increase or decrease to
δ
{\displaystyle \delta \!\,}
.) This implies
∫
−
∞
∞
|
f
(
x
)
|
d
x
≥
∫
M
∞
|
f
(
x
)
|
d
x
≥
∫
M
∞
δ
2
d
x
=
∞
{\displaystyle {\begin{aligned}\int _{-\infty }^{\infty }|f(x)|dx&\geq \int _{M}^{\infty }|f(x)|dx\\&\geq \int _{M}^{\infty }{\frac {\delta }{2}}dx\\&=\infty \end{aligned}}}
which contradicts the hypothesis that
f
{\displaystyle f\!\,}
is integrable i.e.
f
∈
L
1
(
−
∞
,
∞
)
{\displaystyle f\in L^{1}(-\infty ,\infty )\!\,}
. Hence,
lim
b
→
∞
f
(
b
)
=
0
{\displaystyle \lim _{b\rightarrow \infty }f(b)=0\!\,}
Using the same reasoning as above,
lim
a
→
−
∞
f
(
a
)
=
0
{\displaystyle \lim _{a\rightarrow -\infty }f(a)=0\!\,}
Hence,
∫
−
∞
∞
f
′
(
x
)
d
x
=
lim
b
→
∞
f
(
b
)
−
lim
a
→
−
∞
f
(
a
)
=
0
{\displaystyle \int _{-\infty }^{\infty }f^{\prime }(x)dx=\lim _{b\rightarrow \infty }f(b)-\lim _{a\rightarrow -\infty }f(a)=0\!\,}
Suppose
∫
−
∞
∞
f
′
=
c
≠
0
{\displaystyle \int _{-\infty }^{\infty }f'=c\neq 0}
(without loss of generality,
c
>
0
{\displaystyle c>0}
). Then for small positive
ϵ
{\displaystyle \epsilon }
, there exists some real
M
{\displaystyle M}
such that for all
m
>
M
{\displaystyle m>M}
we have
c
−
ϵ
<
∫
−
m
m
f
′
<
c
+
ϵ
{\displaystyle c-\epsilon <\int _{-m}^{m}f'<c+\epsilon }
. By the fundamental theorem of calculus, this gives
f
(
−
m
)
+
c
−
ϵ
<
f
(
m
)
<
f
(
−
m
)
+
c
+
ϵ
{\displaystyle f(-m)+c-\epsilon <f(m)<f(-m)+c+\epsilon }
for all
m
>
M
{\displaystyle m>M}
.
Since
f
{\displaystyle f}
is integrable, this means that for any small positive
δ
{\displaystyle \delta }
, there exists an
N
{\displaystyle N}
such that for all
n
>
N
{\displaystyle n>N}
, we have
∫
−
∞
−
n
+
∫
n
∞
f
<
δ
{\displaystyle \int _{-\infty }^{-n}+\int _{n}^{\infty }f<\delta }
. But by the above estimate,
∫
−
∞
−
n
f
+
∫
n
∞
f
>
∫
−
∞
−
n
f
+
∫
−
∞
−
n
f
+
(
c
−
ϵ
)
=
∫
−
∞
−
n
2
f
+
∫
−
∞
−
n
(
c
−
ϵ
)
=
∞
{\displaystyle \int _{-\infty }^{-n}f+\int _{n}^{\infty }f>\int _{-\infty }^{-n}f+\int _{-\infty }^{-n}f+(c-\epsilon )=\int _{-\infty }^{-n}2f+\int _{-\infty }^{-n}(c-\epsilon )=\infty }
This contradicts the integrability of
f
{\displaystyle f}
. Therefore, we must have
c
=
0
{\displaystyle c=0}
.
Suppose that
{
f
n
}
{\displaystyle \{f_{n}\}\!\,}
is a sequence of real valued measurable functions defined on the interval
[
0
,
1
]
{\displaystyle [0,1]\!\,}
and suppose that
f
n
(
x
)
→
f
(
x
)
{\displaystyle f_{n}(x)\rightarrow f(x)\!\,}
for almost every
x
∈
[
0
,
1
]
{\displaystyle x\in [0,1]\!\,}
. Let
p
>
1
{\displaystyle p>1\!\,}
and
M
>
0
{\displaystyle M>0\!\,}
and suppose that
‖
f
n
‖
p
≤
M
{\displaystyle \|f_{n}\|_{p}\leq M\!\,}
for all
n
{\displaystyle n\!\,}
(a) Prove that
‖
f
‖
p
≤
M
{\displaystyle \|f\|_{p}\leq M\!\,}
.
(b)Prove that
‖
f
−
f
n
‖
1
→
0
{\displaystyle \|f-f_{n}\|_{1}\rightarrow 0\!\,}
as
n
→
∞
{\displaystyle n\rightarrow \infty \!\,}
By definition of norm,
‖
f
‖
p
=
(
∫
|
f
(
x
)
|
p
d
x
)
1
p
{\displaystyle \|f\|_{p}=\left(\int |f(x)|^{p}\,\mathrm {d} x\right)^{\frac {1}{p}}\!\,}
Since
‖
f
n
‖
p
≤
M
{\displaystyle \|f_{n}\|_{p}\leq M\!\,}
,
‖
f
n
‖
p
p
≤
M
p
{\displaystyle \|f_{n}\|_{p}^{p}\leq M^{p}\!\,}
By Fatou's Lemma ,
‖
f
‖
p
p
=
∫
0
1
|
f
(
x
)
|
p
d
x
=
∫
0
1
lim
inf
n
|
f
n
(
x
)
|
p
d
x
≤
lim
inf
n
∫
0
1
|
f
n
(
x
)
|
p
d
x
≤
M
p
{\displaystyle {\begin{aligned}\|f\|_{p}^{p}&=\int _{0}^{1}|f(x)|^{p}dx\\&=\int _{0}^{1}{\underset {n}{\lim \inf }}|f_{n}(x)|^{p}dx\\&\leq {\underset {n}{\lim \inf }}\int _{0}^{1}|f_{n}(x)|^{p}dx\\&\leq M^{p}\end{aligned}}}
which implies, by taking the
p
{\displaystyle p\!\,}
th root,
‖
f
‖
p
≤
M
{\displaystyle \|f\|_{p}\leq M\!\,}
By Holder's Inequality , for all
A
⊂
[
0
,
1
]
{\displaystyle A\subset [0,1]\!\,}
that are measurable,
∫
A
|
(
f
(
x
)
−
f
n
(
x
)
)
⋅
1
|
d
x
≤
(
∫
A
|
f
(
x
)
−
f
n
(
x
)
|
p
d
x
)
1
p
⋅
(
∫
A
1
q
d
x
)
1
q
≤
(
∫
A
|
2
f
(
x
)
|
p
d
x
)
1
p
⋅
(
∫
A
1
q
d
x
)
1
q
≤
2
(
∫
A
|
f
(
x
)
|
p
d
x
)
1
p
⋅
(
∫
A
1
q
d
x
)
1
q
≤
2
M
⋅
(
m
(
A
)
)
1
q
{\displaystyle {\begin{aligned}\int _{A}|(f(x)-f_{n}(x))\cdot 1|dx&\leq \left(\int _{A}|f(x)-f_{n}(x)|^{p}dx\right)^{\frac {1}{p}}\cdot \left(\int _{A}1^{q}dx\right)^{\frac {1}{q}}\\&\leq \left(\int _{A}|2f(x)|^{p}dx\right)^{\frac {1}{p}}\cdot \left(\int _{A}1^{q}dx\right)^{\frac {1}{q}}\\&\leq 2\left(\int _{A}|f(x)|^{p}dx\right)^{\frac {1}{p}}\cdot \left(\int _{A}1^{q}dx\right)^{\frac {1}{q}}\\&\leq 2M\cdot (m(A))^{\frac {1}{q}}\end{aligned}}}
where
1
p
+
1
q
=
1
{\displaystyle {\frac {1}{p}}+{\frac {1}{q}}=1\!\,}
Hence,
|
f
(
x
)
−
f
n
(
x
)
|
≤
2
M
(
m
(
A
)
)
1
q
{\displaystyle |f(x)-f_{n}(x)|\leq 2M(m(A))^{\frac {1}{q}}\!\,}
The Vitali Convergence Theorem then implies
lim
n
→
∞
∫
0
1
|
f
(
x
)
−
f
n
(
x
)
|
d
x
=
∫
0
1
lim
n
→
∞
|
f
(
x
)
−
f
n
(
x
)
|
d
x
=
0
{\displaystyle \lim _{n\rightarrow \infty }\int _{0}^{1}|f(x)-f_{n}(x)|dx=\int _{0}^{1}\lim _{n\rightarrow \infty }|f(x)-f_{n}(x)|dx=0\!\,}
Suppose
f
(
x
)
,
x
f
(
x
)
∈
L
2
(
R
)
{\displaystyle f(x),xf(x)\in L^{2}(R)\!\,}
. Prove that
f
(
x
)
∈
L
1
(
R
)
{\displaystyle f(x)\in L^{1}(R)\!\,}
and that
‖
f
‖
1
≤
2
(
‖
f
‖
2
+
‖
x
f
‖
2
)
{\displaystyle \|f\|_{1}\leq {\sqrt {2}}(\|f\|_{2}+\|xf\|_{2})\!\,}
∫
R
|
f
(
x
)
|
d
x
=
∫
|
x
|
>
1
|
x
f
(
x
|
⋅
1
|
x
|
d
x
+
∫
|
x
|
<
1
|
f
(
x
)
|
⋅
1
d
x
≤
(
∫
|
x
|
>
1
|
x
f
(
x
)
|
2
d
x
)
1
2
⋅
(
∫
|
x
|
>
1
1
|
x
|
2
d
x
)
1
2
+
(
∫
|
x
|
<
1
|
f
(
x
)
|
2
d
x
)
1
2
⋅
(
∫
|
x
|
<
1
1
2
d
x
)
1
2
{\displaystyle {\begin{aligned}\int _{R}|f(x)|dx&=\int _{|x|>1}|xf(x|\cdot {\frac {1}{|x|}}dx+\int _{|x|<1}|f(x)|\cdot 1dx\\&\leq \left(\int _{|x|>1}|xf(x)|^{2}dx\right)^{\frac {1}{2}}\cdot \left(\int _{|x|>1}{\frac {1}{|x|^{2}}}dx\right)^{\frac {1}{2}}+\left(\int _{|x|<1}|f(x)|^{2}dx\right)^{\frac {1}{2}}\cdot \left(\int _{|x|<1}1^{2}dx\right)^{\frac {1}{2}}\end{aligned}}}