Compute
∫
0
∞
1
x
3
+
1
d
x
{\displaystyle \int _{0}^{\infty }{\frac {1}{x^{3}+1}}dx\!\,}
We will compute the general case:
∫
0
∞
1
x
n
+
1
d
x
{\displaystyle \int _{0}^{\infty }{\frac {1}{x^{n}+1}}dx\!\,}
The poles of
f
(
z
)
=
1
z
n
+
1
{\displaystyle f(z)={\frac {1}{z^{n}+1}}\!\,}
are just the zeros of
z
n
+
1
{\displaystyle z^{n}+1\!\,}
, so we can compute them in the following manner:
If
z
=
r
e
i
θ
{\displaystyle z=re^{i\theta }\!\,}
is a solution of
z
n
+
1
=
0
{\displaystyle z^{n}+1=0\!\,}
,
then
z
n
=
r
n
e
i
θ
n
=
−
1
{\displaystyle z^{n}=r^{n}e^{i\theta n}=-1\!\,}
⇒
r
=
1
{\displaystyle \Rightarrow r=1\!\,}
and
i
n
θ
=
i
(
π
+
2
π
k
)
{\displaystyle in\theta =i(\pi +2\pi k)\!\,}
⇒
θ
=
π
+
2
π
k
n
{\displaystyle \Rightarrow \theta ={\frac {\pi +2\pi k}{n}}\!\,}
, k=0,1,2,...,n-1.
Thus, the poles of
f
(
z
)
{\displaystyle f(z)\!\,}
are of the form
z
=
e
π
+
2
π
k
n
{\displaystyle z=e^{\frac {\pi +2\pi k}{n}}\!\,}
with
k
=
0
,
1
,
.
.
.
,
n
−
1
{\displaystyle k=0,1,...,n-1\!\,}
Choose Path of Contour Integral
edit
In order to get obtain the integral of
f
(
x
)
{\displaystyle f(x)\!\,}
from 0 to
∞
{\displaystyle \infty \!\,}
, let us consider the path
γ
{\displaystyle \gamma \!\,}
consisting in a line
A
{\displaystyle A\!\,}
going from 0 to
R
{\displaystyle R\!\,}
, then the arc
B
{\displaystyle B\!\,}
of radius
R
{\displaystyle R\!\,}
from the angle 0 to
2
π
n
{\displaystyle {\frac {2\pi }{n}}\!\,}
and then the line
C
{\displaystyle C\!\,}
joining the end point of
B
{\displaystyle B\!\,}
and the initial point of
A
{\displaystyle A\!\,}
,
where
R
{\displaystyle R\!\,}
is a fixed positive number such that
the pole
z
0
=
e
i
π
n
{\displaystyle z_{0}=e^{i{\frac {\pi }{n}}}\!\,}
is inside the curve
γ
{\displaystyle \gamma \!\,}
. Then , we need to estimate the integral
∫
γ
f
(
z
)
=
∫
0
R
f
(
z
)
⏟
A
+
∫
S
(
R
)
f
(
z
)
⏟
B
+
∫
R
e
i
2
π
n
0
f
(
z
)
⏟
C
{\displaystyle \int _{\gamma }f(z)=\underbrace {\int _{0}^{R}f(z)} _{A}+\underbrace {\int _{S(R)}f(z)} _{B}+\underbrace {\int _{Re^{i{\frac {2\pi }{n}}}}^{0}f(z)} _{C}\!\,}
Compute Residues of f at z0= exp{i\pi /n}
edit
R
e
s
(
f
,
z
0
)
=
1
(
z
n
+
1
)
′
|
z
=
z
0
=
1
n
z
0
n
−
1
=
1
n
e
i
π
n
(
n
−
1
)
=
e
i
π
n
n
e
i
π
=
−
1
n
e
i
π
n
{\displaystyle {\begin{aligned}Res(f,z_{0})&=\left.{\frac {1}{(z^{n}+1)'}}\right|_{z=z_{0}}\\&={\frac {1}{nz_{0}^{n-1}}}\\&={\frac {1}{ne^{{\frac {i\pi }{n}}(n-1)}}}\\&={\frac {e^{\frac {i\pi }{n}}}{ne^{i\pi }}}\\&={\frac {-1}{n}}e^{\frac {i\pi }{n}}\end{aligned}}\!\,}
Bound Arc Portion (B) of Integral
edit
|
B
|
=
|
∫
S
(
R
)
d
z
z
n
+
1
|
≤
∫
S
(
R
)
d
z
|
z
n
+
1
|
=
∫
S
(
R
)
d
z
|
R
n
+
1
|
≤
1
R
n
∫
S
(
R
)
d
z
=
1
R
n
2
π
n
R
=
2
π
n
R
n
−
1
{\displaystyle {\begin{aligned}|B|&=\left|\int _{S(R)}{\frac {dz}{z^{n}+1}}\right|\\&\leq \int _{S(R)}{\frac {dz}{|z^{n}+1|}}\\&=\int _{S(R)}{\frac {dz}{|R^{n}+1|}}\\&\leq {\frac {1}{R^{n}}}\int _{S(R)}dz\\&={\frac {1}{R^{n}}}{\frac {2\pi }{n}}R\\&={\frac {2\pi }{nR^{n-1}}}\end{aligned}}}
Hence as
R
→
∞
{\displaystyle R\rightarrow \infty \!\,}
,
|
B
|
→
0
{\displaystyle |B|\rightarrow 0\!\,}
Parametrize (C) in terms of (A)
edit
Let
z
=
r
e
i
2
π
n
{\displaystyle z=re^{i{\frac {2\pi }{n}}}\!\,}
where
r
{\displaystyle r\!\,}
is real number. Then
d
z
=
e
i
2
π
n
d
r
{\displaystyle dz=e^{i{\frac {2\pi }{n}}}dr\!\,}
C
=
∫
R
e
i
2
π
n
0
d
z
1
+
z
n
=
−
∫
0
R
e
i
2
π
n
d
z
1
+
z
n
=
−
∫
0
R
e
i
2
π
n
1
+
(
r
e
i
2
π
n
)
n
=
−
∫
0
R
e
i
2
π
n
1
+
r
n
=
−
e
i
2
π
n
∫
0
R
d
r
1
+
r
n
⏟
A
{\displaystyle {\begin{aligned}C&=\int _{Re^{i{\frac {2\pi }{n}}}}^{0}{\frac {dz}{1+z^{n}}}\\&=-\int _{0}^{Re^{i{\frac {2\pi }{n}}}}{\frac {dz}{1+z^{n}}}\\&=-\int _{0}^{R}{\frac {e^{i{\frac {2\pi }{n}}}}{1+(re^{i{\frac {2\pi }{n}}})^{n}}}\\&=-\int _{0}^{R}{\frac {e^{i{\frac {2\pi }{n}}}}{1+r^{n}}}\\&=-e^{i{\frac {2\pi }{n}}}\underbrace {\int _{0}^{R}{\frac {dr}{1+r^{n}}}} _{A}\\\end{aligned}}}
From Cauchy Integral Formula, we have,
A
+
B
+
C
=
2
π
i
−
e
i
π
n
n
{\displaystyle A+B+C=2\pi i{\frac {-e^{i{\frac {\pi }{n}}}}{n}}\!\,}
As
R
→
∞
{\displaystyle R\rightarrow \infty \!\,}
,
B
→
0
{\displaystyle B\rightarrow 0\!\,}
. Also
C
{\displaystyle C\!\,}
can be written in terms of
A
{\displaystyle A\!\,}
. Hence
A
+
B
+
C
=
A
+
C
=
(
1
−
e
i
2
π
n
)
A
{\displaystyle {\begin{aligned}A+B+C&=A+C\\&=(1-e^{i{\frac {2\pi }{n}}})A\end{aligned}}}
We then have,
A
=
2
π
i
n
e
i
π
n
e
i
2
π
n
−
1
=
π
n
2
i
e
i
π
n
e
i
π
n
(
e
i
π
n
−
e
−
i
π
n
)
=
π
n
1
sin
(
π
n
)
{\displaystyle {\begin{aligned}A&={\frac {2\pi i}{n}}{\frac {e^{i{\frac {\pi }{n}}}}{e^{i{\frac {2\pi }{n}}}-1}}\\&={\frac {\pi }{n}}{\frac {2ie^{i{\frac {\pi }{n}}}}{e^{i{\frac {\pi }{n}}}(e^{i{\frac {\pi }{n}}}-e^{-i{\frac {\pi }{n}}})}}\\&={\frac {\pi }{n}}{\frac {1}{\sin({\frac {\pi }{n}})}}\end{aligned}}}
Lemma: Two fixed points imply identity
edit
Lemma. Let
f
{\displaystyle f\!\,}
be analytic on the unit
D
{\displaystyle D\!\,}
, and assume that
|
f
(
z
)
|
<
1
{\displaystyle |f(z)|<1\!\,}
on the disc. Prove that if there exist two distinct points
a
{\displaystyle a\!\,}
and
b
{\displaystyle b\!\,}
in the disc which are fixed points, that is,
f
(
a
)
=
a
{\displaystyle f(a)=a\!\,}
and
f
(
b
)
=
b
{\displaystyle f(b)=b\!\,}
, then
f
(
z
)
=
z
{\displaystyle f(z)=z\!\,}
.
Proof Let
h
:
D
→
D
{\displaystyle h:D\rightarrow D\!\,}
be the automorphism defined as
h
(
z
)
=
a
−
z
1
−
a
¯
z
{\displaystyle h(z)={\frac {a-z}{1-{\overline {a}}z}}\!\,}
Consider now
F
(
z
)
=
h
∘
f
∘
h
−
1
(
z
)
{\displaystyle F(z)=h\circ f\circ h^{-1}(z)\!\,}
. Then, F has two fixed points, namely
F
(
0
)
=
h
∘
f
∘
h
−
1
(
0
)
=
h
∘
f
(
a
)
=
h
(
a
)
=
0
{\displaystyle F(0)=h\circ f\circ h^{-1}(0)=h\circ f(a)=h(a)=0\!\,}
F
(
a
−
b
1
−
a
¯
b
)
=
h
∘
f
∘
h
−
1
(
a
−
b
1
−
a
¯
b
)
=
h
∘
f
(
b
)
=
h
(
b
)
=
a
−
b
1
−
a
¯
b
{\displaystyle F\left({\frac {a-b}{1-{\overline {a}}b}}\right)=h\circ f\circ h^{-1}\left({\frac {a-b}{1-{\overline {a}}b}}\right)=h\circ f(b)=h(b)={\frac {a-b}{1-{\overline {a}}b}}\!\,}
.
Since
F
(
0
)
=
0
{\displaystyle F(0)=0\!\,}
,
a
−
b
1
−
a
¯
b
≠
0
{\displaystyle {\frac {a-b}{1-{\overline {a}}b}}\neq 0\!\,}
(since
a
{\displaystyle a\!\,}
is different to
b
{\displaystyle b\!\,}
), and
|
F
(
a
−
b
1
−
a
¯
b
)
|
=
|
a
−
b
1
−
a
¯
b
|
{\displaystyle \left|F\left({\frac {a-b}{1-{\overline {a}}b}}\right)\right|=\left|{\frac {a-b}{1-{\overline {a}}b}}\right|\!\,}
,
by Schwarz Lemma,
F
(
z
)
=
α
z
{\displaystyle F(z)=\alpha z\!\,}
.
But, replacing
a
−
b
1
−
a
¯
b
{\displaystyle {\frac {a-b}{1-{\overline {a}}b}}\!\,}
into the last formula, we get
α
=
1
{\displaystyle \alpha =1\!\,}
.
Therefore,
h
∘
f
∘
h
−
1
(
z
)
=
z
{\displaystyle h\circ f\circ h^{-1}(z)=z\!\,}
,
which implies
f
(
z
)
=
z
{\displaystyle f(z)=z\!\,}
Shift Points to Create Fixed Points
edit
Let
f
(
z
)
=
g
(
z
)
−
1
{\displaystyle f(z)=g(z)-1\!\,}
. Then
f
(
0
)
=
0
{\displaystyle f(0)=0\!\,}
and
f
(
−
1
)
=
−
1
{\displaystyle f(-1)=-1\!\,}
.
Notice that
S
{\displaystyle S\!\,}
is an infinite horizontal strip centered around the real axis with height
π
{\displaystyle \pi \!\,}
. Since
f
(
z
)
{\displaystyle f(z)\!\,}
is a unit horizontal shift left,
f
(
S
)
⊂
S
{\displaystyle f(S)\subset S\!\,}
.
Use Riemann Mapping Theorem
edit
From the Riemann mapping theorem, there exists a biholomorphic (bijective and holomorphic) mapping
h
{\displaystyle h\!\,}
, from the open unit disk
D
{\displaystyle D\!\,}
to
S
{\displaystyle S\!\,}
.
Define Composition Function
edit
Let
F
=
h
−
1
∘
f
∘
h
{\displaystyle F=h^{-1}\circ f\circ h\!\,}
. Then
F
{\displaystyle F\!\,}
maps
D
{\displaystyle D\!\,}
to
D
{\displaystyle D\!\,}
.
From the lemma, since
F
(
z
)
{\displaystyle F(z)\!\,}
has two fixed points,
F
(
z
)
=
z
{\displaystyle F(z)=z\!\,}
which implies
f
(
z
)
=
z
{\displaystyle f(z)=z\!\,}
which implies
g
=
z
+
1
{\displaystyle g=z+1\!\,}
.
Choose any compact set K in D
edit
Choose any compact set
K
{\displaystyle K\!\,}
in the open unit disk
D
{\displaystyle D\!\,}
. Since
K
{\displaystyle K\!\,}
is compact, it is also closed and bounded.
We want to show that for all
f
∈
F
{\displaystyle f\in {\mathcal {F}}\!\,}
and all
z
∈
K
{\displaystyle z\in K\!\,}
,
|
f
(
z
)
|
{\displaystyle |f(z)|\!\,}
is bounded i.e.
|
f
(
z
)
|
<
B
K
{\displaystyle |f(z)|<B_{K}\!\,}
where
B
K
{\displaystyle B_{K}\!\,}
is some constant dependent on the choice of
K
{\displaystyle K\!\,}
.
Apply Maximum Modulus Principle to find |f(z0)|
edit
Choose
z
0
{\displaystyle z_{0}\!\,}
that is the shortest distance from the boundary of the unit disk
D
{\displaystyle D\!\,}
. From the maximum modulus principle,
|
f
(
z
0
)
|
=
max
z
∈
K
|
f
(
z
|
{\displaystyle |f(z_{0})|=\max _{z\in K}|f(z|\!\,}
.
Note that
z
0
{\displaystyle z_{0}\!\,}
is independent of the choice of
f
∈
F
{\displaystyle f\in {\mathcal {F}}\!\,}
.
We will apply Cauchy's Integral formula to
f
2
(
z
0
)
{\displaystyle f^{2}(z_{0})\!\,}
(instead of
f
(
z
0
)
{\displaystyle f(z_{0})\!\,}
) to take advantage of the hypothesis.
Choose sufficiently small
r
0
>
0
{\displaystyle r_{0}>0\!\,}
so that
D
(
z
0
,
r
0
)
∈
D
{\displaystyle D(z_{0},r_{0})\!\,\in D}
f
2
(
z
0
)
=
1
2
π
i
∫
|
z
−
z
0
|
=
r
0
f
2
(
z
)
z
−
z
0
d
z
=
1
2
π
i
∫
0
2
π
f
2
(
z
0
+
r
0
e
i
θ
)
i
r
0
e
i
θ
(
z
0
+
r
0
e
i
θ
)
−
z
0
d
θ
=
1
2
π
∫
0
2
π
f
2
(
z
0
+
r
0
e
i
θ
)
d
θ
{\displaystyle {\begin{aligned}f^{2}(z_{0})&={\frac {1}{2\pi i}}\int _{|z-z_{0}|=r_{0}}{\frac {f^{2}(z)}{z-z_{0}}}dz\\&={\frac {1}{2\pi i}}\int _{0}^{2\pi }{\frac {f^{2}(z_{0}+r_{0}e^{i\theta })ir_{0}e^{i\theta }}{(z_{0}+r_{0}e^{i\theta })-z_{0}}}d\theta \\&={\frac {1}{2\pi }}\int _{0}^{2\pi }f^{2}(z_{0}+r_{0}e^{i\theta })d\theta \end{aligned}}}
Integrate with respect to r
edit
∫
0
r
0
r
f
2
(
z
0
)
d
r
=
∫
0
r
0
∫
0
2
π
f
2
(
z
0
+
r
e
i
θ
)
r
d
r
d
θ
=
1
2
π
∫
∫
D
(
z
0
,
r
)
f
2
(
x
+
i
y
)
d
x
d
y
{\displaystyle {\begin{aligned}\int _{0}^{r_{0}}rf^{2}(z_{0})dr&=\int _{0}^{r_{0}}\int _{0}^{2\pi }f^{2}(z_{0}+re^{i\theta })rdrd\theta \\&={\frac {1}{2\pi }}\int \int _{D(z_{0},r)}f^{2}(x+iy)dxdy\end{aligned}}}
Integrating the left hand side, we have
∫
0
r
0
r
f
2
(
z
0
)
d
r
=
r
0
2
2
f
2
(
z
0
)
{\displaystyle \int _{0}^{r_{0}}rf^{2}(z_{0})dr={\frac {r_{0}^{2}}{2}}f^{2}(z_{0})\!\,}
Hence,
r
0
2
2
f
2
(
z
0
)
=
1
2
π
∫
∫
D
(
z
0
,
r
)
f
2
(
x
+
i
y
)
d
x
d
y
{\displaystyle {\frac {r_{0}^{2}}{2}}f^{2}(z_{0})={\frac {1}{2\pi }}\int \int _{D(z_{0},r)}f^{2}(x+iy)dxdy\!\,}
Bound |f(z0)| by using hypothesis
edit
|
r
0
2
2
f
2
(
z
0
)
|
=
|
1
2
π
∫
∫
D
(
z
0
,
r
0
)
f
2
(
x
+
i
y
)
d
x
d
y
|
≤
1
2
π
∫
∫
D
(
z
0
,
r
0
)
|
f
2
(
x
+
i
y
)
|
d
x
d
y
≤
1
2
π
∫
∫
D
|
f
2
(
x
+
i
y
)
|
d
x
d
y
≤
1
2
π
Then
|
r
0
2
2
f
2
(
z
0
)
|
≤
1
2
π
This implies
|
f
(
z
0
)
|
≤
1
r
0
π
{\displaystyle {\begin{aligned}\left|{\frac {r_{0}^{2}}{2}}f^{2}(z_{0})\right|&=\left|{\frac {1}{2\pi }}\int \int _{D(z_{0},r_{0})}f^{2}(x+iy)dxdy\right|\\&\leq {\frac {1}{2\pi }}\int \int _{D(z_{0},r_{0})}|f^{2}(x+iy)|dxdy\\&\leq {\frac {1}{2\pi }}\int \int _{D}|f^{2}(x+iy)|dxdy\\&\leq {\frac {1}{2\pi }}\\\\{\mbox{Then }}\\\left|{\frac {r_{0}^{2}}{2}}f^{2}(z_{0})\right|&\leq {\frac {1}{2\pi }}\\\\{\mbox{This implies}}\\\\|f(z_{0})|&\leq {\frac {1}{r_{0}{\sqrt {\pi }}}}\end{aligned}}}
Apply Montel's Theorem
edit
Then, since any
f
∈
F
{\displaystyle f\in {\mathcal {F}}\!\,}
is uniformly bounded in every compact set, by Montel's Theorem, it follows that
F
{\displaystyle {\mathcal {F}}\!\,}
is normal