Definition 7.3 :
Let
M
{\displaystyle M}
be a manifold of class
C
n
{\displaystyle {\mathcal {C}}^{n}}
. We define the vector field Lie bracket , denoted by
[
⋅
,
⋅
]
{\displaystyle [\cdot ,\cdot ]}
, as follows:
[
⋅
,
⋅
]
:
X
(
M
)
×
X
(
M
)
→
X
(
M
)
,
[
V
,
W
]
(
p
)
(
φ
)
:=
V
(
p
)
(
W
φ
)
−
W
(
p
)
(
V
φ
)
{\displaystyle [\cdot ,\cdot ]:{\mathfrak {X}}(M)\times {\mathfrak {X}}(M)\to {\mathfrak {X}}(M),[\mathbf {V} ,\mathbf {W} ](p)(\varphi ):=\mathbf {V} (p)(\mathbf {W} \varphi )-\mathbf {W} (p)(\mathbf {V} \varphi )}
Theorem 6.4 : If
V
,
W
{\displaystyle \mathbf {V} ,\mathbf {W} }
are vector fields of class
C
n
{\displaystyle {\mathcal {C}}^{n}}
on
M
{\displaystyle M}
, then
[
V
,
W
]
{\displaystyle [\mathbf {V} ,\mathbf {W} ]}
is a vector field of class
C
n
{\displaystyle {\mathcal {C}}^{n}}
on
M
{\displaystyle M}
(i. e.
[
⋅
,
⋅
]
{\displaystyle [\cdot ,\cdot ]}
really maps to
X
(
M
)
{\displaystyle {\mathfrak {X}}(M)}
)
Proof :
1. We show that for each
p
∈
M
{\displaystyle p\in M}
,
[
V
,
W
]
(
p
)
∈
T
(
p
)
M
{\displaystyle [\mathbf {V} ,\mathbf {W} ](p)\in T(p)M}
. Let
φ
,
ϑ
∈
C
∞
(
M
)
{\displaystyle \varphi ,\vartheta \in {\mathcal {C}}^{\infty }(M)}
and
c
∈
R
{\displaystyle c\in \mathbb {R} }
.
1.1 We prove linearity:
[
V
,
W
]
(
p
)
(
φ
+
c
ϑ
)
=
V
(
p
)
(
W
(
φ
+
c
ϑ
)
)
−
W
(
p
)
(
V
(
φ
+
c
ϑ
)
)
=
V
(
p
)
(
W
φ
+
c
W
ϑ
)
−
W
(
p
)
(
V
φ
+
c
V
ϑ
)
=
V
(
p
)
(
W
φ
)
−
W
(
p
)
(
V
φ
)
+
c
(
V
(
p
)
(
W
ϑ
)
−
W
(
p
)
(
V
ϑ
)
)
=
[
V
,
W
]
(
p
)
(
φ
)
+
c
[
V
,
W
]
(
p
)
(
ϑ
)
{\displaystyle {\begin{aligned}{[\mathbf {V} ,\mathbf {W} ]}(p)(\varphi +c\vartheta )&=\mathbf {V} (p)(\mathbf {W} (\varphi +c\vartheta ))-\mathbf {W} (p)(\mathbf {V} (\varphi +c\vartheta ))\\&=\mathbf {V} (p)(\mathbf {W} \varphi +c\mathbf {W} \vartheta )-\mathbf {W} (p)(\mathbf {V} \varphi +c\mathbf {V} \vartheta )\\&=\mathbf {V} (p)(\mathbf {W} \varphi )-\mathbf {W} (p)(\mathbf {V} \varphi )+c(\mathbf {V} (p)(\mathbf {W} \vartheta )-\mathbf {W} (p)(\mathbf {V} \vartheta ))\\&=[\mathbf {V} ,\mathbf {W} ](p)(\varphi )+c[\mathbf {V} ,\mathbf {W} ](p)(\vartheta )\end{aligned}}}
1.2 We prove the product rule:
[
V
,
W
]
(
p
)
(
φ
ϑ
)
=
V
(
p
)
(
W
(
φ
ϑ
)
)
−
W
(
p
)
(
V
(
φ
ϑ
)
)
=
V
(
p
)
(
φ
W
ϑ
+
ϑ
W
φ
)
−
W
(
p
)
(
φ
V
ϑ
+
ϑ
V
φ
)
=
V
(
p
)
(
φ
W
ϑ
)
+
V
(
p
)
(
ϑ
W
φ
)
−
W
(
p
)
(
φ
V
ϑ
)
−
W
(
p
)
(
ϑ
V
φ
)
=
φ
(
p
)
V
(
p
)
(
W
ϑ
)
+
(
Y
ϑ
)
(
p
)
⏞
=
Y
(
p
)
(
ϑ
)
V
(
p
)
(
φ
)
+
ϑ
(
p
)
V
(
p
)
(
W
φ
)
+
(
Y
φ
)
(
p
)
⏞
=
Y
(
p
)
(
φ
)
V
(
p
)
(
ϑ
)
−
φ
(
p
)
W
(
p
)
(
V
ϑ
)
−
(
X
ϑ
)
(
p
)
⏞
=
X
(
p
)
(
ϑ
)
W
(
p
)
(
φ
)
−
ϑ
(
p
)
W
(
p
)
(
V
φ
)
−
(
X
φ
)
(
p
)
⏞
=
X
(
p
)
(
φ
)
W
(
p
)
(
ϑ
)
=
φ
(
p
)
V
(
p
)
(
W
ϑ
)
−
φ
(
p
)
W
(
p
)
(
V
ϑ
)
+
ϑ
(
p
)
V
(
p
)
(
W
φ
)
−
ϑ
(
p
)
W
(
p
)
(
V
φ
)
=
φ
(
p
)
[
V
,
W
]
(
p
)
(
ϑ
)
+
ϑ
(
p
)
[
V
,
W
]
(
p
)
(
φ
)
{\displaystyle {\begin{aligned}{[\mathbf {V} ,\mathbf {W} ]}(p)(\varphi \vartheta )&=\mathbf {V} (p)(\mathbf {W} (\varphi \vartheta ))-\mathbf {W} (p)(\mathbf {V} (\varphi \vartheta ))\\&=\mathbf {V} (p)(\varphi \mathbf {W} \vartheta +\vartheta \mathbf {W} \varphi )-\mathbf {W} (p)(\varphi \mathbf {V} \vartheta +\vartheta \mathbf {V} \varphi )\\&=\mathbf {V} (p)(\varphi \mathbf {W} \vartheta )+\mathbf {V} (p)(\vartheta \mathbf {W} \varphi )-\mathbf {W} (p)(\varphi \mathbf {V} \vartheta )-\mathbf {W} (p)(\vartheta \mathbf {V} \varphi )\\&=\varphi (p)\mathbf {V} (p)(\mathbf {W} \vartheta )+\overbrace {\mathbf {(} Y\vartheta )(p)} ^{=Y(p)(\vartheta )}\mathbf {V} (p)(\varphi )+\vartheta (p)\mathbf {V} (p)(\mathbf {W} \varphi )+\overbrace {\mathbf {(} Y\varphi )(p)} ^{=Y(p)(\varphi )}\mathbf {V} (p)(\vartheta )\\&~~~~-\varphi (p)\mathbf {W} (p)(\mathbf {V} \vartheta )-\overbrace {\mathbf {(} X\vartheta )(p)} ^{=X(p)(\vartheta )}\mathbf {W} (p)(\varphi )-\vartheta (p)\mathbf {W} (p)(\mathbf {V} \varphi )-\overbrace {\mathbf {(} X\varphi )(p)} ^{=X(p)(\varphi )}\mathbf {W} (p)(\vartheta )\\&=\varphi (p)\mathbf {V} (p)(\mathbf {W} \vartheta )-\varphi (p)\mathbf {W} (p)(\mathbf {V} \vartheta )+\vartheta (p)\mathbf {V} (p)(\mathbf {W} \varphi )-\vartheta (p)\mathbf {W} (p)(\mathbf {V} \varphi )\\&=\varphi (p)[\mathbf {V} ,\mathbf {W} ](p)(\vartheta )+\vartheta (p)[\mathbf {V} ,\mathbf {W} ](p)(\varphi )\end{aligned}}}
2. We show that
[
V
,
W
]
{\displaystyle [\mathbf {V} ,\mathbf {W} ]}
is differentiable of class
C
n
{\displaystyle {\mathcal {C}}^{n}}
.
Let
φ
∈
C
n
(
M
)
{\displaystyle \varphi \in {\mathcal {C}}^{n}(M)}
be arbitrary. As
V
,
W
{\displaystyle \mathbf {V} ,\mathbf {W} }
are vector fields of class
C
n
{\displaystyle {\mathcal {C}}^{n}}
,
V
φ
{\displaystyle \mathbf {V} \varphi }
and
W
φ
{\displaystyle \mathbf {W} \varphi }
are contained in
C
n
(
M
)
{\displaystyle {\mathcal {C}}^{n}(M)}
. But since
V
,
W
{\displaystyle \mathbf {V} ,\mathbf {W} }
are vector fields of class
C
n
{\displaystyle {\mathcal {C}}^{n}}
,
V
(
W
φ
)
{\displaystyle \mathbf {V} (\mathbf {W} \varphi )}
and
W
(
V
φ
)
{\displaystyle \mathbf {W} (\mathbf {V} \varphi )}
are contained in
C
n
(
M
)
{\displaystyle {\mathcal {C}}^{n}(M)}
. But the sum of two differentiable functions is again differentiable (this is what theorem 2.? says), and thus
[
V
,
W
]
φ
{\displaystyle [\mathbf {V} ,\mathbf {W} ]\varphi }
is in
C
n
(
M
)
{\displaystyle {\mathcal {C}}^{n}(M)}
, and since
φ
{\displaystyle \varphi }
was arbitrary,
[
V
,
W
]
{\displaystyle [\mathbf {V} ,\mathbf {W} ]}
is differentiable of class
C
n
{\displaystyle {\mathcal {C}}^{n}}
.
◻
{\displaystyle \Box }
Theorem 6.5 :
If
M
{\displaystyle M}
is a manifold, and
[
⋅
,
⋅
]
{\displaystyle [\cdot ,\cdot ]}
is the vector field Lie bracket, then
X
(
M
)
{\displaystyle {\mathfrak {X}}(M)}
and
[
⋅
,
⋅
]
{\displaystyle [\cdot ,\cdot ]}
form a Lie algebra together.
Proof :
1. First we note that
X
(
M
)
{\displaystyle {\mathfrak {X}}(M)}
as defined in definition 5.? is a vector space (this was covered by exercise 5.?).
2. Second, we prove that for the vector Lie bracket, the three calculation rules of definition 6.1 are satisfied. Let
V
,
W
,
U
∈
X
(
M
)
{\displaystyle \mathbf {V} ,\mathbf {W} ,\mathbf {U} \in {\mathfrak {X}}(M)}
and
c
∈
R
{\displaystyle c\in \mathbb {R} }
.
2.1 We prove bilinearity. For all
p
∈
M
{\displaystyle p\in M}
and
φ
∈
C
n
(
M
)
{\displaystyle \varphi \in {\mathcal {C}}^{n}(M)}
, we have
[
V
,
W
+
c
U
]
(
p
)
(
φ
)
=
V
(
p
)
(
(
W
+
c
U
)
φ
)
−
(
W
+
c
U
)
(
p
)
(
V
φ
)
=
V
(
p
)
(
W
φ
+
c
U
φ
)
−
W
(
p
)
(
V
φ
)
−
c
U
(
p
)
(
V
φ
)
=
V
(
p
)
(
W
φ
)
−
W
(
p
)
(
V
φ
)
+
c
V
(
p
)
(
U
φ
)
−
c
U
(
p
)
(
V
φ
)
=
[
V
,
W
]
(
p
)
(
φ
)
+
c
[
V
,
U
]
(
p
)
(
φ
)
{\displaystyle {\begin{aligned}{[\mathbf {V} ,\mathbf {W} +c\mathbf {U} ]}(p)(\varphi )&=\mathbf {V} (p)((\mathbf {W} +c\mathbf {U} )\varphi )-(\mathbf {W} +c\mathbf {U} )(p)(\mathbf {V} \varphi )\\&=\mathbf {V} (p)(\mathbf {W} \varphi +c\mathbf {U} \varphi )-\mathbf {W} (p)(\mathbf {V} \varphi )-c\mathbf {U} (p)(\mathbf {V} \varphi )\\&=\mathbf {V} (p)(\mathbf {W} \varphi )-\mathbf {W} (p)(\mathbf {V} \varphi )+c\mathbf {V} (p)(\mathbf {U} \varphi )-c\mathbf {U} (p)(\mathbf {V} \varphi )\\&=[\mathbf {V} ,\mathbf {W} ](p)(\varphi )+c[\mathbf {V} ,\mathbf {U} ](p)(\varphi )\end{aligned}}}
and hence, since
p
∈
M
{\displaystyle p\in M}
and
φ
∈
C
n
(
M
)
{\displaystyle \varphi \in {\mathcal {C}}^{n}(M)}
were arbitrary,
[
V
,
W
+
c
U
]
=
[
V
,
W
]
+
c
[
V
,
U
]
{\displaystyle [\mathbf {V} ,\mathbf {W} +c\mathbf {U} ]=[\mathbf {V} ,\mathbf {W} ]+c[\mathbf {V} ,\mathbf {U} ]}
Analogously (see exercise 1), it can be proven that
[
V
+
c
W
,
U
]
=
[
V
,
U
]
+
c
[
W
,
U
]
{\displaystyle [\mathbf {V} +c\mathbf {W} ,\mathbf {U} ]=[\mathbf {V} ,\mathbf {U} ]+c[\mathbf {W} ,\mathbf {U} ]}
2.2 We prove skew-symmetry. We have for all
p
∈
M
{\displaystyle p\in M}
and
φ
∈
C
n
(
M
)
{\displaystyle \varphi \in {\mathcal {C}}^{n}(M)}
:
[
V
,
W
]
(
p
)
(
φ
)
=
V
(
p
)
(
W
φ
)
−
W
(
p
)
(
V
φ
)
=
−
(
W
(
p
)
(
V
φ
)
−
V
(
p
)
(
W
φ
)
)
=
−
[
W
,
V
]
(
p
)
(
φ
)
{\displaystyle [\mathbf {V} ,\mathbf {W} ](p)(\varphi )=\mathbf {V} (p)(\mathbf {W} \varphi )-\mathbf {W} (p)(\mathbf {V} \varphi )=-(\mathbf {W} (p)(\mathbf {V} \varphi )-\mathbf {V} (p)(\mathbf {W} \varphi ))=-[\mathbf {W} ,\mathbf {V} ](p)(\varphi )}
2.3 We prove Jacobi's identity. We have for all
p
∈
M
{\displaystyle p\in M}
and
φ
∈
C
n
(
M
)
{\displaystyle \varphi \in {\mathcal {C}}^{n}(M)}
:
[
V
,
[
W
,
U
]
]
(
p
)
(
φ
)
+
[
U
,
[
V
,
W
]
]
(
p
)
(
φ
)
+
[
W
,
[
U
,
V
]
]
(
p
)
(
φ
)
=
V
(
p
)
(
[
W
,
U
]
φ
)
−
[
W
,
U
]
(
p
)
(
V
φ
)
+
U
(
p
)
(
[
V
,
W
]
φ
)
−
[
V
,
W
]
(
p
)
(
U
φ
)
+
W
(
p
)
(
[
U
,
V
]
φ
)
−
[
U
,
V
]
(
p
)
(
W
φ
)
=
V
(
p
)
(
W
(
U
φ
)
−
U
(
W
φ
)
)
−
W
(
p
)
(
U
(
V
φ
)
)
+
U
(
p
)
(
W
(
V
φ
)
)
+
U
(
p
)
(
V
(
W
φ
)
−
W
(
V
φ
)
)
−
V
(
p
)
(
W
(
U
φ
)
)
+
W
(
p
)
(
V
(
U
φ
)
)
+
W
(
p
)
(
U
(
V
φ
)
−
V
(
U
φ
)
)
−
U
(
p
)
(
V
(
W
φ
)
)
+
V
(
p
)
(
U
(
W
φ
)
)
=
0
{\displaystyle {\begin{aligned}{[\mathbf {V} ,[\mathbf {W} ,\mathbf {U} ]]}(p)(\varphi )+[\mathbf {U} ,[\mathbf {V} ,\mathbf {W} ]](p)(\varphi )+[\mathbf {W} ,[\mathbf {U} ,\mathbf {V} ]](p)(\varphi )&=\mathbf {V} (p)([\mathbf {W} ,\mathbf {U} ]\varphi )-[\mathbf {W} ,\mathbf {U} ](p)(\mathbf {V} \varphi )\\&~~~~+\mathbf {U} (p)([\mathbf {V} ,\mathbf {W} ]\varphi )-[\mathbf {V} ,\mathbf {W} ](p)(\mathbf {U} \varphi )\\&~~~~+\mathbf {W} (p)([\mathbf {U} ,\mathbf {V} ]\varphi )-[\mathbf {U} ,\mathbf {V} ](p)(\mathbf {W} \varphi )\\&=\mathbf {V} (p)(\mathbf {W} (\mathbf {U} \varphi )-\mathbf {U} (\mathbf {W} \varphi ))-\mathbf {W} (p)(\mathbf {U} (\mathbf {V} \varphi ))+\mathbf {U} (p)(\mathbf {W} (\mathbf {V} \varphi ))\\&~~~~+\mathbf {U} (p)(\mathbf {V} (\mathbf {W} \varphi )-\mathbf {W} (\mathbf {V} \varphi ))-\mathbf {V} (p)(\mathbf {W} (\mathbf {U} \varphi ))+\mathbf {W} (p)(\mathbf {V} (\mathbf {U} \varphi ))\\&~~~~+\mathbf {W} (p)(\mathbf {U} (\mathbf {V} \varphi )-\mathbf {V} (\mathbf {U} \varphi ))-\mathbf {U} (p)(\mathbf {V} (\mathbf {W} \varphi ))+\mathbf {V} (p)(\mathbf {U} (\mathbf {W} \varphi ))\\&=0\end{aligned}}}
, where the last equality follows from the linearity of
V
(
p
)
,
W
(
p
)
{\displaystyle \mathbf {V} (p),\mathbf {W} (p)}
and
U
(
p
)
{\displaystyle \mathbf {U} (p)}
.
◻
{\displaystyle \Box }