1.Prove that
v
1
,
…
,
v
r
∈
V
∗
v_1,\dots,v_r\in V^*
v1,…,vr∈V∗ is linearly dependent if and only if
v
1
∧
⋯
∧
v
r
=
0
v_1\wedge\dots\wedge v_r=0
v1∧⋯∧vr=0
Proof:
If
v
1
,
…
.
v
r
v^1,\dots.v^r
v1,….vr is linearly dependent,
∃
v
r
=
a
1
v
1
+
⋯
+
a
r
−
1
v
r
−
1
\exists v^r=a_1v^1+\dots +a_{r-1}v^{r-1}
∃vr=a1v1+⋯+ar−1vr−1, thus
v
1
∧
⋯
∧
v
r
=
∑
λ
=
1
r
−
1
a
λ
v
1
∧
⋯
∧
v
r
−
1
∧
v
λ
=
0
v^1\wedge \dots \wedge v^r=\sum_{\lambda=1}^{r-1}a_{\lambda}v^1\wedge \dots\wedge v^{r-1}\wedge v^{\lambda}=0
v1∧⋯∧vr=λ=1∑r−1aλv1∧⋯∧vr−1∧vλ=0
If
v
1
,
…
,
v
r
v^1,\dots,v^r
v1,…,vr is linearly independent, we can enlarge them to be a basis of
V
∗
:
{
v
1
,
…
,
v
r
,
…
,
v
n
}
.
V^*:\{v^1,\dots,v^r,\dots,v^n\}.
V∗:{v1,…,vr,…,vn}.
{
e
i
}
\{e_i\}
{ei} is the conjugate basis of
{
v
i
}
\{v^i\}
{vi} in
V
V
V, thus
v
1
∧
⋯
∧
v
r
(
e
1
,
…
,
e
r
)
=
δ
i
1
…
i
r
1
…
r
v
i
1
(
e
1
,
…
,
e
r
)
=
(
v
1
(
e
1
)
…
v
1
(
e
r
)
⋮
⋮
v
r
(
e
1
)
…
v
r
(
e
r
)
)
=
1
\begin{aligned} v^1\wedge\dots\wedge v^r(e_1,\dots,e_r)&=\delta^{1\dots r}_{i_1\dots i_r}v^{i_1}(e_1,\dots ,e_r)\\ &=\begin{pmatrix} v^1(e_1)& \dots & v^1(e_r)\\ \vdots&&\vdots\\ v^r(e_1) & \dots & v^r(e_r) \end{pmatrix}\\ &=1 \end{aligned}
v1∧⋯∧vr(e1,…,er)=δi1…ir1…rvi1(e1,…,er)=⎝⎜⎛v1(e1)⋮vr(e1)……v1(er)⋮vr(er)⎠⎟⎞=1
thus, v 1 ∧ ⋯ ∧ v r ≠ 0. v^1\wedge \dots \wedge v^r\neq 0. v1∧⋯∧vr=0.
2.Let
F
:
V
→
W
F: V\rightarrow W
F:V→W be a linear map, prove that for
Φ
∈
A
r
(
W
)
\Phi\in A^r(W)
Φ∈Ar(W) and
Ψ
∈
A
s
(
W
)
\Psi\in A^s(W)
Ψ∈As(W)
F
∗
(
Φ
∧
Ψ
)
=
F
∗
Φ
∧
F
∗
Ψ
.
F^*(\Phi\wedge\Psi)=F^*\Phi\wedge F^*\Psi.
F∗(Φ∧Ψ)=F∗Φ∧F∗Ψ.
Proof:
∀
u
1
,
…
,
u
r
+
s
∈
V
\forall u_1,\dots,u_{r+s}\in V
∀u1,…,ur+s∈V, we have:
(
F
∗
(
Φ
∧
Ψ
)
)
(
u
1
,
…
,
u
r
+
s
)
=
(
Φ
∧
Ψ
)
(
F
(
u
1
)
,
…
,
F
(
u
r
+
s
)
)
=
1
r
!
s
!
δ
1
…
r
+
s
i
1
…
i
r
+
s
Φ
(
F
(
u
i
1
)
,
…
,
F
(
u
i
r
)
)
⋅
Ψ
(
F
(
u
i
r
+
1
)
,
…
,
F
(
u
i
r
+
s
)
)
=
(
F
∗
Φ
)
∧
(
F
∗
Ψ
)
(
u
1
,
…
,
u
r
+
s
)
\begin{aligned} &(F^*(\Phi\wedge\Psi))(u_1,\dots,u_{r+s})\\ &=(\Phi\wedge\Psi)(F(u_1),\dots,F(u_{r+s}))\\ &=\frac{1}{r!s!}\delta^{i_1\dots i_{r+s}}_{1\dots {r+s}}\Phi(F(u_{i_1}),\dots,F(u_{i_r}))\cdot \Psi(F(u_{i_{r+1}}),\dots,F(u_{i_{r+s}}))\\ &=(F^*\Phi)\wedge (F^*\Psi)(u_1,\dots,u_{r+s}) \end{aligned}
(F∗(Φ∧Ψ))(u1,…,ur+s)=(Φ∧Ψ)(F(u1),…,F(ur+s))=r!s!1δ1…r+si1…ir+sΦ(F(ui1),…,F(uir))⋅Ψ(F(uir+1),…,F(uir+s))=(F∗Φ)∧(F∗Ψ)(u1,…,ur+s)
3.Deduce the transformation formula of coefficients of a
(
r
,
s
)
(r,s)
(r,s)-tensor under the change
of local coordinates.
Proof:
Suppose
(
U
,
φ
;
x
i
)
(U,\varphi;x^i)
(U,φ;xi) is a chart of
M
M
M,
(
r
,
s
)
(r,s)
(r,s)-tensor
Φ
\Phi
Φ can be locally represented:
Φ
(
x
)
=
Φ
j
1
…
j
s
i
1
…
i
r
(
x
)
∂
∂
x
i
1
⊗
⋯
⊗
∂
∂
x
i
r
⊗
d
x
j
1
⊗
⋯
⊗
d
x
j
\Phi(x)=\Phi^{i_1\dots i_{r}}_{j_1\dots j_s}(x)\frac{\partial}{\partial x^{i_1}}\otimes\dots\otimes\frac{\partial}{\partial x^{i_r}}\otimes dx^{j_1}\otimes\dots\otimes dx^j
Φ(x)=Φj1…jsi1…ir(x)∂xi1∂⊗⋯⊗∂xir∂⊗dxj1⊗⋯⊗dxj
Thus for another chart ( V , ϕ ; x ′ i ) (V,\phi;x'^i) (V,ϕ;x′i), we have the transformation formula on the intersection:
Φ j 1 … j s i 1 … i r ( x ′ ) = Φ k 1 … k s h 1 … h r ( x ) ∂ x ′ i 1 ∂ x h 1 … ∂ x ′ i r ∂ x h r ∂ x ′ k 1 ∂ x ′ j 1 … ∂ x ′ k s ∂ x ′ j s \Phi^{i_1\dots i_{r}}_{j_1\dots j_s}(x')=\Phi^{h_1\dots h_{r}}_{k_1\dots k_s}(x)\frac{\partial x'^{i_1}}{\partial x^{h_1}}\dots\frac{\partial x'^{i_r}}{\partial x^{h_r}}\frac{\partial x'^{k_1}}{\partial x'^{j_1}}\dots \frac{\partial x'^{k_s}}{\partial x'^{j_s}} Φj1…jsi1…ir(x′)=Φk1…ksh1…hr(x)∂xh1∂x′i1…∂xhr∂x′ir∂x′j1∂x′k1…∂x′js∂x′ks
4.Show that any
C
∞
(
M
)
C^{\infty}(M)
C∞(M)-linear map
τ
:
Λ
1
(
M
)
×
⋯
×
Λ
1
(
M
)
⏟
s
×
X
(
M
)
×
…
X
(
M
)
⏟
r
→
C
∞
(
M
)
\tau:\underbrace{\Lambda^1(M)\times\dots\times\Lambda^1(M)}_{s}\times\underbrace{\mathscr{X}(M)\times\dots\mathscr{X}(M)}_r\rightarrow C^{\infty}(M)
τ:s
Λ1(M)×⋯×Λ1(M)×r
X(M)×…X(M)→C∞(M)
can be identified with a
(
s
,
r
)
(s,r)
(s,r)-tensor field.
Proof:
We only show the result of
(
0
,
1
)
(0,1)
(0,1)-tensor field, for the situation of
(
r
,
s
)
(r,s)
(r,s)-tensor can be proved by Cartesian product of every vector field, and it’s too long to write completely.
Suppose we have a linear map
τ
:
X
(
M
)
→
C
∞
(
M
)
\tau: \mathscr{X}(M)\rightarrow C^{\infty}(M)
τ:X(M)→C∞(M), s.t.
∀
v
∈
X
(
M
)
\forall v\in\mathscr{X}(M)
∀v∈X(M) and
μ
∈
C
∞
(
M
)
\mu\in C^{\infty}(M)
μ∈C∞(M) we have:
τ
(
μ
⋅
v
)
=
μ
⋅
τ
(
v
)
\tau(\mu\cdot v)=\mu\cdot\tau(v)
τ(μ⋅v)=μ⋅τ(v)
We need to find out a
(
0
,
1
)
(0,1)
(0,1)-tensor field
τ
~
\tilde{\tau}
τ~ on
M
M
M, s.t.
∀
x
0
∈
M
\forall x_0\in M
∀x0∈M:
τ
~
(
x
0
)
(
X
(
x
0
)
)
=
(
τ
(
X
)
)
(
x
0
)
,
∀
X
∈
X
(
M
)
\tilde{\tau}(x_0)(X(x_0))=(\tau(X))(x_0), \forall X\in \mathscr{X}(M)
τ~(x0)(X(x0))=(τ(X))(x0),∀X∈X(M)
First, we hvae τ : X ( M ) → C ∞ ( M ) \tau:\mathscr{X}(M)\rightarrow C^{\infty}(M) τ:X(M)→C∞(M) is local defined, X , Y ∈ X ( M ) X,Y\in\mathscr{X}(M) X,Y∈X(M), and for open subset U ⊂ M U\subset M U⊂M, X ∣ U = Y ∣ U X|_U=Y|_U X∣U=Y∣U, we have τ ( X ) ∣ U = τ ( Y ) ∣ U ∈ C ∞ ( U ) \tau(X)|_U=\tau(Y)|_U\in C^{\infty}(U) τ(X)∣U=τ(Y)∣U∈C∞(U):
By the partition of unity, f ⋅ ( X − Y ) f\cdot (X-Y) f⋅(X−Y) is a zero vector field on M M M.
τ ( 0 ) = τ ( 0 ) + τ ( 0 ) ⇒ τ ( 0 ) = 0 \tau(0)=\tau(0)+\tau(0)\Rightarrow \tau(0)=0 τ(0)=τ(0)+τ(0)⇒τ(0)=0 on zero vector field.
thus, 0 = τ ( f ⋅ ( X − Y ) ) = f ⋅ ( τ ( X ) − τ ( Y ) ) 0=\tau(f\cdot (X-Y))=f\cdot (\tau(X)-\tau(Y)) 0=τ(f⋅(X−Y))=f⋅(τ(X)−τ(Y)), thus we do have τ ( X ) ∣ U = τ ( Y ) ∣ U \tau(X)|_U=\tau(Y)|_U τ(X)∣U=τ(Y)∣U.
Suppose
x
0
∈
(
U
;
x
i
)
x_0\in(U;x^i)
x0∈(U;xi) is a chart, and
v
∈
X
(
U
)
v\in\mathscr{X}(U)
v∈X(U), thus
v
=
∑
i
v
i
∂
∂
x
i
v=\sum_iv^i\frac{\partial}{\partial x^i}
v=i∑vi∂xi∂
thus, τ ( v ) = ∑ i v i ⋅ τ ( ∂ ∂ x i ) \tau(v)=\sum_iv^i\cdot\tau(\frac{\partial}{\partial x^i}) τ(v)=∑ivi⋅τ(∂xi∂)
Then, we can define τ ~ ( x 0 ) ∈ T x 0 ∗ M \tilde{\tau}(x_0)\in T^*_{x_0}M τ~(x0)∈Tx0∗M for x 0 ∈ M x_0\in M x0∈M:
∀
v
0
∈
T
x
0
M
\forall v_0\in T_{x_0}M
∀v0∈Tx0M,
v
∈
X
(
U
)
v\in\mathscr{X}(U)
v∈X(U) is a extension of
v
0
v_0
v0 at
U
U
U whcih contains
x
0
x_0
x0 , s.t.
v
(
x
0
)
=
v
0
v(x_0)=v_0
v(x0)=v0, and
(
τ
~
(
x
0
)
)
(
v
0
)
=
(
τ
(
v
)
)
(
x
0
)
(\tilde{\tau}(x_0))(v_0)=(\tau(v))(x_0)
(τ~(x0))(v0)=(τ(v))(x0)
and τ ~ ( x 0 ) : T x 0 M → R \tilde{\tau}(x_0):T_{x_0}M\rightarrow \mathbb{R} τ~(x0):Tx0M→R is linear, thus τ ~ ( x 0 ) ∈ T x 0 ∗ M \tilde{\tau}(x_0)\in T^*_{x_0}M τ~(x0)∈Tx0∗M.
In
(
U
;
x
i
)
(U;x^i)
(U;xi), we have:
τ
~
∣
U
=
∑
i
=
1
m
τ
~
(
∂
∂
x
i
)
d
x
i
\tilde{\tau}|_U=\sum^m_{i=1}\tilde{\tau}(\frac{\partial}{\partial x^i})dx^i
τ~∣U=i=1∑mτ~(∂xi∂)dxi
τ ~ ( ∂ ∂ x i ) ( x 0 ) = τ ~ ( x 0 ) ( ∂ ∂ x i ) = τ ( ∂ ∂ x i ) ( x 0 ) , ∀ x 0 ∈ U \tilde{\tau}(\frac{\partial}{\partial x^i})(x_0)=\tilde{\tau}(x_0)(\frac{\partial}{\partial x^i})=\tau(\frac{\partial}{\partial x^i})(x_0), \forall x_0\in U τ~(∂xi∂)(x0)=τ~(x0)(∂xi∂)=τ(∂xi∂)(x0),∀x0∈U
Thus,
τ
~
(
∂
∂
x
i
)
=
τ
(
∂
∂
x
i
)
∈
C
∞
(
U
)
\tilde{\tau}(\frac{\partial}{\partial x^i})=\tau(\frac{\partial}{\partial x^i})\in C^{\infty}(U)
τ~(∂xi∂)=τ(∂xi∂)∈C∞(U)
thus τ ~ \tilde{\tau} τ~ is cotangent vector field, and ∀ v ∈ X ( M ) \forall v\in \mathscr{X}(M) ∀v∈X(M), we have τ ~ ( v ) = τ ( v ) \tilde{\tau}(v)=\tau(v) τ~(v)=τ(v)