Inequalities are an essential tool for proving central statements in function theory. Since
C
{\displaystyle \mathbb {C} }
does not have a complete/total order , one must rely on the magnitude of functions for estimations.
Inequality for the Sum of Real and Imaginary Parts - IRI
edit
Let
f
:
[
a
,
b
]
→
C
{\displaystyle f:[a,b]\rightarrow \mathbb {C} }
be a piecewise continuous function with
f
1
:
[
a
,
b
]
→
R
{\displaystyle f_{1}:[a,b]\rightarrow \mathbb {R} }
,
f
2
:
[
a
,
b
]
→
R
{\displaystyle f_{2}:[a,b]\rightarrow \mathbb {R} }
, and
f
=
f
1
+
i
⋅
f
2
{\displaystyle f=f_{1}+i\cdot f_{2}}
, then we have:
|
∫
a
b
f
(
t
)
d
t
|
≤
∫
a
b
|
f
1
(
t
)
|
d
t
+
∫
a
b
|
f
2
(
t
)
|
d
t
{\displaystyle \left|\int _{a}^{b}f(t)\,dt\right|\leq \int _{a}^{b}|f_{1}(t)|\,dt+\int _{a}^{b}|f_{2}(t)|\,dt}
Prove the IRI inequality. The proof is done by decomposing into real part function and imaginary part function, linearity of the integral, and applying the triangle inequality.
Inequality for the Absolute Value in the Integrand - AVI
edit
Let
f
:
[
a
,
b
]
→
C
{\displaystyle f:[a,b]\rightarrow \mathbb {C} }
be a piecewise continuous function, then we have:
|
∫
a
b
f
(
t
)
d
t
|
≤
∫
a
b
|
f
(
t
)
|
d
t
{\displaystyle \left|\int _{a}^{b}f(t)\,dt\right|\leq \int _{a}^{b}|f(t)|\,dt}
The proof is done by a case distinction with:
(AVI-1)
∫
a
b
f
(
t
)
d
t
=
0
{\displaystyle \int _{a}^{b}f(t)\,dt=0}
(AVI-2)
∫
a
b
f
(
t
)
d
t
≠
0
{\displaystyle \int _{a}^{b}f(t)\,dt\not =0}
Since
∫
a
b
f
(
t
)
d
t
=
0
{\displaystyle \int _{a}^{b}f(t)\,dt=0}
, we have
|
∫
a
b
f
(
t
)
d
t
|
=
0
{\displaystyle \left|\int _{a}^{b}f(t)\,dt\right|=0}
.
Since
|
f
(
t
)
|
≥
0
{\displaystyle |f(t)|\geq 0}
, we have
∫
a
b
|
f
(
t
)
|
d
t
≥
0
{\displaystyle \int _{a}^{b}|f(t)|\,dt\geq 0}
and we obtain:
|
∫
a
b
f
(
t
)
d
t
|
=
0
≤
∫
a
b
|
f
(
t
)
|
d
t
{\displaystyle \left|\int _{a}^{b}f(t)\,dt\right|=0\leq \int _{a}^{b}|f(t)|\,dt}
The integral
β
=
∫
a
b
f
(
t
)
d
t
∈
C
{\displaystyle \beta =\int _{a}^{b}f(t)\,dt\in \mathbb {C} }
is a complex number with
β
≠
0
{\displaystyle \beta \not =0}
, for which we have with
|
β
|
=
β
⋅
β
¯
{\displaystyle |\beta |={\sqrt {\beta \cdot {\overline {\beta }}}}}
:
|
β
|
=
|
β
|
2
|
β
|
=
β
⋅
β
¯
|
β
|
=
β
¯
|
β
|
⏟
α
:=
⋅
β
=
α
⋅
β
{\displaystyle |\beta |={\frac {|\beta |^{2}}{|\beta |}}={\frac {\beta \cdot {\overline {\beta }}}{|\beta |}}=\underbrace {\frac {\overline {\beta }}{|\beta |}} _{\alpha :=}\cdot \beta =\alpha \cdot \beta }
Case - (AVI-2) - Step 1
edit
Since
β
≠
0
{\displaystyle \beta \not =0}
, we have by the linearity of the integral:
|
β
|
=
α
⋅
β
=
α
⋅
∫
a
b
f
(
t
)
d
t
=
∫
a
b
α
⋅
f
(
t
)
d
t
{\displaystyle |\beta |=\alpha \cdot \beta =\alpha \cdot \int _{a}^{b}f(t)\,dt=\int _{a}^{b}\alpha \cdot f(t)\,dt}
Case - (AVI-2) - Step 3
edit
Let
g
:=
α
⋅
f
{\displaystyle g:=\alpha \cdot f}
and
g
:
[
a
,
b
]
→
C
{\displaystyle g:[a,b]\rightarrow \mathbb {C} }
be a piecewise continuous function with
g
1
:
[
a
,
b
]
→
R
{\displaystyle g_{1}:[a,b]\rightarrow \mathbb {R} }
,
g
2
:
[
a
,
b
]
→
R
{\displaystyle g_{2}:[a,b]\rightarrow \mathbb {R} }
, and
g
=
g
1
+
i
⋅
g
2
{\displaystyle g=g_{1}+i\cdot g_{2}}
, then we have by the linearity of the integral:
R
e
(
∫
a
b
g
(
t
)
d
t
)
=
R
e
(
∫
a
b
g
1
(
t
)
d
t
⏟
∈
R
+
i
⋅
∫
a
b
g
2
(
t
)
d
t
⏟
∈
R
)
=
R
e
(
∫
a
b
g
1
(
t
)
d
t
⏟
∈
R
)
=
∫
a
b
g
1
(
t
)
d
t
=
∫
a
b
R
e
(
g
1
(
t
)
)
d
t
{\displaystyle {\begin{array}{rcl}{\mathfrak {Re}}{\bigg (}\int _{a}^{b}g(t)\,dt{\bigg )}&=&{\mathfrak {Re}}{\bigg (}\underbrace {\int _{a}^{b}g_{1}(t)\,dt} _{\in \mathbb {R} }+i\cdot \underbrace {\int _{a}^{b}g_{2}(t)\,dt} _{\in \mathbb {R} }{\bigg )}\\&=&{\mathfrak {Re}}{\bigg (}\underbrace {\int _{a}^{b}g_{1}(t)\,dt} _{\in \mathbb {R} }{\bigg )}=\int _{a}^{b}g_{1}(t)\,dt\\&=&\int _{a}^{b}{\mathfrak {Re}}(g_{1}(t))\,dt\\\end{array}}}
Case - (AVI-2) - Step 4
edit
Since
|
β
|
=
∫
a
b
α
⋅
f
(
t
)
d
t
∈
R
{\displaystyle |\beta |=\int _{a}^{b}\alpha \cdot f(t)\,dt\in \mathbb {R} }
holds, we have by the above calculation from Step 3 for the real part:
|
β
|
=
R
e
(
∫
a
b
α
⋅
f
(
t
)
d
t
⏟
∈
R
)
=
∫
a
b
R
e
(
α
⋅
f
(
t
)
)
⏟
∈
R
d
t
{\displaystyle |\beta |={\mathfrak {Re}}{\bigg (}\underbrace {\int _{a}^{b}\alpha \cdot f(t)\,dt} _{\in \mathbb {R} }{\bigg )}=\int _{a}^{b}\underbrace {{\mathfrak {Re}}\left(\alpha \cdot f(t)\right)} _{\in \mathbb {R} }\,dt}
Case - (AVI-2) - Step 5
edit
The following real part estimate against the absolute value of a complex number
z
{\displaystyle z}
R
e
(
z
)
=
z
1
≤
|
z
1
|
=
z
1
2
≤
z
1
2
+
z
2
2
=
|
z
|
{\displaystyle {\mathfrak {Re}}(z)=z_{1}\leq |z_{1}|={\sqrt {z_{1}^{2}}}\leq {\sqrt {z_{1}^{2}+z_{2}^{2}}}=|z|}
for
z
=
z
1
+
i
⋅
z
2
{\displaystyle z=z_{1}+i\cdot z_{2}}
is now applied to the integrand of the above integral
R
e
(
α
⋅
f
(
t
)
)
⏟
∈
R
{\displaystyle \underbrace {{\mathfrak {Re}}\left(\alpha \cdot f(t)\right)} _{\in \mathbb {R} }}
.
Case - (AVI-2) - Step 6
edit
The following estimate is obtained analogously to Step 5 by the linearity of the integral
|
β
|
=
∫
a
b
R
e
(
α
⋅
f
(
t
)
)
d
t
≤
∫
a
b
|
α
⋅
f
(
t
)
|
d
t
=
|
α
|
⋅
∫
a
b
|
f
(
t
)
|
d
t
{\displaystyle |\beta |=\int _{a}^{b}{\mathfrak {Re}}\left(\alpha \cdot f(t)\right)\,dt\leq \int _{a}^{b}\left|\alpha \cdot f(t)\right|\,dt=|\alpha |\cdot \int _{a}^{b}\left|f(t)\right|\,dt}
Case - (AVI-2) - Step 7
edit
Since
|
α
|
=
|
β
¯
|
β
|
|
=
|
β
¯
|
|
β
|
=
1
{\displaystyle |\alpha |=\left|{\frac {\overline {\beta }}{|\beta |}}\right|={\frac {|{\overline {\beta }}|}{|\beta |}}=1}
holds, we have in total the desired estimate:
|
∫
a
b
f
(
t
)
d
t
|
=
α
⋅
∫
a
b
|
f
(
t
)
|
d
t
≤
|
α
|
⋅
∫
a
b
|
f
(
t
)
|
d
t
=
∫
a
b
|
f
(
t
)
|
d
t
{\displaystyle \left|\int _{a}^{b}f(t)\,dt\right|=\alpha \cdot \int _{a}^{b}|f(t)|\,dt\leq |\alpha |\cdot \int _{a}^{b}|f(t)|\,dt=\int _{a}^{b}|f(t)|\,dt}
Inequality - Length of Integration Path - LIP
edit
Let
γ
:
[
a
,
b
]
→
C
{\displaystyle \gamma :[a,b]\rightarrow \mathbb {C} }
be an integration path and
f
:
U
→
C
{\displaystyle f:U\rightarrow \mathbb {C} }
be a function on the trace of
γ
{\displaystyle \gamma }
(i.e.
S
p
u
r
(
γ
)
:=
{
γ
(
t
)
:
t
∈
[
a
,
b
]
}
⊂
U
{\displaystyle Spur(\gamma ):=\{\gamma (t)\,:\,t\in [a,b]\}\subset U}
). Then we have:
|
∫
γ
f
(
z
)
d
z
|
≤
max
z
∈
S
p
u
r
(
γ
)
|
f
(
z
)
|
⋅
L
(
γ
)
{\displaystyle \left|\int _{\gamma }f(z)\,dz\right|\leq \max _{z\in Spur(\gamma )}|f(z)|\cdot {\mathcal {L}}(\gamma )}
where
L
(
γ
)
=
∫
a
b
|
γ
′
(
t
)
|
d
t
{\displaystyle {\mathcal {L}}(\gamma )=\int _{a}^{b}|\gamma '(t)|\,dt}
is the length of the integral.
By using the above estimate for the absolute value of the integrand
|
f
(
z
)
d
z
|
≤
max
z
∈
S
p
u
r
(
γ
)
|
f
(
z
)
|
{\displaystyle \left|f(z)\,dz\right|\leq \max _{z\in Spur(\gamma )}|f(z)|}
and the UG-BI inequality, we obtain:
|
∫
γ
f
(
z
)
d
z
|
≤
A
V
I
∫
a
b
|
f
(
γ
(
t
)
)
⋅
γ
′
(
t
)
|
d
t
=
∫
a
b
|
f
(
γ
(
t
)
)
|
⋅
|
γ
′
(
t
)
|
d
t
≤
∫
a
b
max
z
∈
S
p
u
r
(
γ
)
|
f
(
z
)
|
⏟
M
:=
⋅
|
γ
′
(
t
)
|
d
z
=
M
⋅
∫
γ
|
γ
′
(
t
)
|
d
z
=
M
⋅
L
(
γ
)
{\displaystyle {\begin{array}{rcl}\displaystyle \left|\int _{\gamma }f(z)\,dz\right|&{\stackrel {AVI}{\leq }}&\displaystyle \int _{a}^{b}\left|f(\gamma (t))\cdot \gamma {\,}'(t)\right|\,dt=\int _{a}^{b}\left|f(\gamma (t))\right|\cdot \left|\gamma {\,}'(t)\right|\,dt\\&\leq &\displaystyle \int _{a}^{b}\underbrace {\max _{z\in Spur(\gamma )}|f(z)|} _{M:=}\cdot |\gamma {\,}'(t)|\,dz=M\cdot \int _{\gamma }|\gamma {\,}'(t)|\,dz\\&=&\displaystyle M\cdot {\mathcal {L}}(\gamma )\\\end{array}}}
Inequality for Estimation Over Integration Paths
edit
Let
γ
:
[
a
,
b
]
→
C
{\displaystyle \gamma :[a,b]\rightarrow \mathbb {C} }
be an Integration path and
f
:
U
→
C
{\displaystyle f:U\rightarrow \mathbb {C} }
a continuous function on the trace of
γ
{\displaystyle \gamma }
(
T
r
a
c
e
(
γ
)
:=
γ
(
t
)
,
:
,
t
∈
[
a
,
b
]
⊂
U
{\displaystyle Trace(\gamma ):={\gamma (t),:,t\in [a,b]}\subset U}
). Then, the following holds:
|
∫
γ
f
(
z
)
,
d
z
|
≤
max
z
∈
T
r
a
c
e
(
γ
)
|
f
(
z
)
|
⋅
L
(
γ
)
{\displaystyle \left|\int _{\gamma }f(z),dz\right|\leq \max _{z\in Trace(\gamma )}|f(z)|\cdot L(\gamma )}
Here,
L
(
γ
)
=
∫
a
b
|
γ
′
(
t
)
|
,
d
t
{\displaystyle L(\gamma )=\int _{a}^{b}|\gamma '(t)|,dt}
is the length of the integral.