Section II.6.1

The beginnings of tensor calculus

[7/57%]
Show
LaTeX
Problem II.6.1.1

The Kronecker symbol. Verify that δij\delta_{i}^{j} is an element of T11(V)\mathbb {T}_{1}^{1}(V), representing the identity map of the space VV into itself.

?
Problem II.6.1.2

The metric tensor. Let V=⟨e1,…,en⟩RV= \left\langle \mathbf{e}_{1}, \ldots , \mathbf{e}_{n}\right\rangle_{\mathbb {R}} be a Euclidean vector space with inner product given by a positive definite bilinear form

(x∣y)=g(x,y)=∑i,jgijxixj,gij=g(ei,ej) (\mathbf{x} \mid \mathbf{y})=g(\mathbf{x}, \mathbf{y})=\sum _{i, j} g_{i j} x^{i} x^{j}, \quad g_{i j}=g\left(\mathbf{e}_{i}, \mathbf{e}_{j}\right)

(following the tradition adopted by geometers, we use here the letter gg and the symbols gijg_{i j}). It is customary to call G0=(gij)G_{0}=(g_{i j}) the metric tensor (of type (2,0)(2,0)) of the space VV. Thus, the components of the metric tensor are the entries of the Gram matrix of the original basis in VV with respect to the corresponding inner product.

Now we recall that a Euclidean space is canonically isomorphic to its dual space V∗V^{*}, and the basis (ei)(\mathbf{e}_{i}) is identified with its dual basis (ei)(\mathbf{e}^{i}) of the space V∗:(ei∣∗)=eiV^{*}:\left(\mathbf{e}_{i} \mid *\right)=\mathbf{e}^{i} (see (13) from § 1 Ch. 3). If gij=g(ei,ej)g^{i j}=g\left(\mathbf{e}^{i}, \mathbf{e}^{j}\right), then G0=(gij)G^{0}=\left(g^{i j}\right) is also called the metric tensor (now of type (0,2)(0,2)) of the space VV. Here gijg_{i j} and gijg^{i j} are the covariant and contravariant coordinates of the metric tensor GG.

Verify that G0G0=EG^{0} G_{0}=E, i.e.,

∑jgijfjk=δji. \sum _{j} g^{i j} f_{j k}=\delta _{j}^{i}.
?
Problem II.6.1.3

Extension of the scalar field. Let VV be a real vector space with basis (e1,…,en)(\mathbf{e}_{1}, \ldots , \mathbf{e}_{n}), VCV^{\mathbb {C}} its complexification (see item 3§43 \S 4 Ch. 3). Since the field C\mathbb {C} is a vector space over R\mathbb {R} with basis (1,i)(1, i), we can construct the vector space

C⊗V=⟨1⊗e1,…,1⊗en,i⊗e1,…,i⊗en⟩ \mathbb {C} \otimes V= \left\langle 1 \otimes \mathbf{e}_{1}, \ldots , 1 \otimes \mathbf{e}_{n}, i \otimes \mathbf{e}_{1}, \ldots , i \otimes \mathbf{e}_{n}\right\rangle

Verify that the R\mathbb {R}-linear map

R⊗V→VC:1⊗ek↦ek,i⊗ek↦iek \mathbb {R} \otimes V \rightarrow V^{\mathbb {C}}: 1 \otimes \mathbf{e}_{k} \mapsto \mathbf{e}_{k}, \quad i \otimes \mathbf{e}_{k} \mapsto i \mathbf{e}_{k}

defines an isomorphism of C⊗V\mathbb {C} \otimes V with VCV^{\mathbb {C}}.

More generally: let K\mathfrak {K} be a subfield of a field L,V\mathfrak {L}, V a vector space over K\mathfrak {K}. First regarding L\mathfrak {L} as a vector space over K\mathfrak {K}, let us construct the tensor product L⊗KV\mathfrak {L} \otimes_{\mathfrak {K}} V. Then let us introduce on it the structure of a vector space over L\mathfrak {L}, defining multiplication by scalars a∈La \in \mathfrak {L} by the formula

a(b⊗x)=(ab)⊗x,a,b∈L,x∈V a(b \otimes \mathbf{x})=(a b) \otimes \mathbf{x}, \quad a, b \in \mathfrak {L}, \quad \mathbf{x} \in V

Verify the correctness of this definition.

?
Problem II.6.1.4

Let dim⁡V>1,dim⁡W>1\operatorname {dim} V>1, \operatorname {dim} W>1. Show that in the tensor product V⊗WV \otimes W there exist elements not representable in the form v⊗w\mathbf{v} \otimes \mathbf{w}.

?
Problem II.6.1.5

Let A,BA, B be invertible square matrices. Give the matrix inverse to A⊗BA \otimes B.

?
Problem II.6.1.6

Compare

∥β11Aβ12A⋯β1mAβ21Aβ22A⋯β2mA⋯⋯⋯⋯βm1Aβm2A⋯βmmA∥ \left\Vert \begin{smallmatrix} \beta _{11}A & \beta _{12}A & \cdots & \beta _{1m}A \\ \beta _{21}A & \beta _{22}A & \cdots & \beta _{2m}A \\ \cdots & \cdots & \cdots & \cdots \\ \beta _{m1}A & \beta _{m2}A & \cdots & \beta _{mm}A \end{smallmatrix}\right\Vert

with the block matrix (20).

?
Problem II.6.1.7

Prove formula (22) for square matrices A,BA, B of orders nn and mm, respectively, with complex coefficients, using the possibility of reducing them to triangular form.

?