For ζ∈Hm and k∈[m], ζk is the k-th component of ζ; by definition of the operations, (ζ+ζ′)k=ζk+ζk′ and (λζ)k=λζk for ζ,ζ′∈Hm and λ∈C, and two elements of Hm are equal when all their components are equal. For k∈[m] let πk:Hm→H be the map ζ↦ζk; it is linear by the preceding formulas.
1. (Hilbert structure) Vector space. Each condition of Vector Space over a Field for Hm, read in the k-th component, is the same condition for H; the tuple all of whose components are 0 is a zero vector of Hm, and (−ζ1,…,−ζm) is an additive inverse of ζ. So Hm is a complex vector space.
Inner product. Let ζ,ζ′,ζ′′∈Hm and λ∈C. By condition 1 of Complex Inner Product Space for H and claim 4 of Properties of Finite Sums,
⟨ζ,ζ′⟩Hm=k=1∑m⟨ζk′,ζk⟩H=k=1∑m⟨ζk′,ζk⟩H=⟨ζ′,ζ⟩Hm.
By conditions 2 and 3 of Complex Inner Product Space for H and claims 2 and 3 of Properties of Finite Sums,
⟨ζ,ζ′+ζ′′⟩Hm=k=1∑m(⟨ζk,ζk′⟩H+⟨ζk,ζk′′⟩H)=⟨ζ,ζ′⟩Hm+⟨ζ,ζ′′⟩Hm,⟨ζ,λζ′⟩Hm=k=1∑mλ⟨ζk,ζk′⟩H=λ⟨ζ,ζ′⟩Hm.
The summands ⟨ζk,ζk⟩H=∥ζk∥H2 of ⟨ζ,ζ⟩Hm are nonnegative real numbers. Since R⊆C is closed under addition, the map σ:[m]→C of Finite Sum Notation in a Field, formed in C for these summands, takes real values: the set of j∈N such that j∈/[m] or σ(j)∈R contains 1 and, with j, contains S(j) (if S(j)∈[m] then j∈[m] and σ(S(j))=σ(j)+∥ζS(j)∥H2), so it is N by Principle of Induction for the Natural Numbers; by the uniqueness in Existence and Uniqueness of Iterates of a Binary Operation it is then also the map formed in R, so ⟨ζ,ζ⟩Hm is the finite sum of the ∥ζk∥H2 in R. By claim 5 of Properties of Finite Sums it is nonnegative, and if it is 0 then ∥ζk∥H2=0 for every k∈[m], so ζk=0 by claim 4 of Elementary Properties of a Complex Inner Product, that is, ζ=0. Hence ⟨⋅,⋅⟩Hm is an inner product, and by Norm Induced by a Complex Inner Product
∥ζ∥Hm2=⟨ζ,ζ⟩Hm=k=1∑m∥ζk∥H2.
Consequently, by claim 6 of Properties of Finite Sums and claim 2 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field,
∥ζk∥H≤∥ζ∥Hmfor all ζ∈Hm and k∈[m].(1.1)
Completeness. Let (ζ(n))n∈N be a Cauchy sequence in Hm and let k∈[m]. The k-th component of ζ(n)−ζ(p) is ζk(n)−ζk(p), so by (1.1) ∥ζk(n)−ζk(p)∥H≤∥ζ(n)−ζ(p)∥Hm for all n,p∈N, and (ζk(n))n∈N is a Cauchy sequence in H. Since H is a complex Hilbert space, it is complete, so this sequence converges to some ζk∈H, which is unique by Uniqueness of Limits in a Metric Space. Let ζ=(ζ1,…,ζm)∈Hm.
Let ε>0 and put δ=ε/(4m). First choose N∈N with ∥ζ(n)−ζ(p)∥Hm<δ for all n,p≥N. Let n≥N and k∈[m]. Next choose Nk∈N with ∥ζk(p)−ζk∥H<δ for all p≥Nk, and let p be the larger of N and Nk (claim 3 of Properties of the Order on the Natural Numbers). By the triangle inequality (claim 2 of The Induced Norm is a Norm, and Induces a Metric) and (1.1),
∥ζk(n)−ζk∥H≤∥ζk(n)−ζk(p)∥H+∥ζk(p)−ζk∥H<2δ,
so ∥ζk(n)−ζk∥H2<4δ2 by claim 1 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field. The k-th component of ζ(n)−ζ is ζk(n)−ζk; by the norm formula, claims 2, 3 and 5 of Properties of Finite Sums (applied to the nonnegative differences 4δ2−∥ζk(n)−ζk∥H2) and the recursion in claim 1 of that lemma (which gives ∑k=1m4δ2=4mδ2),
∥ζ(n)−ζ∥Hm2=k=1∑m∥ζk(n)−ζk∥H2≤4mδ2=4mε2<ε2,
so ∥ζ(n)−ζ∥Hm<ε by claim 1 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field. As n≥N was arbitrary, (ζ(n)) converges to ζ. Hence Hm is complete, and it is a complex Hilbert space.
Two remarks. From now on we use the following facts. (R1) For a map j↦ξ(j) from [n] to Hm and k∈[m], the k-th component of ∑j=1nξ(j) is ∑j=1nξk(j), by claim 4 of Properties of Finite Sums of Vectors applied to the linear map πk. (R2) Let V,W,U be among H and Hm. For fixed v∈V the map R↦Rv from L(V,W) to W is linear, because sums and scalar multiples in L(V,W) are formed pointwise; and for fixed P∈L(W,U) and P′∈L(U,V) the maps R↦PR and R↦RP′ are linear, because composition distributes over sums and commutes with scalar multiples (Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations). So, by claim 4 of Properties of Finite Sums of Vectors, finite sums in L(V,W) may be evaluated termwise at v and composed termwise with P and P′.
2. (Coordinate inclusions) Let k,l∈[m]. For v,w∈H and λ∈C, the tuples ιk(v+λw) and ιkv+λιkw both have k-th component v+λw and all other components 0, so ιk is linear. By the norm formula of claim 1 and claim 7 of Properties of Finite Sums (the summands with index j=k are ∥0∥H2=0), ∥ιkv∥Hm2=∥v∥H2, so ∥ιkv∥Hm=∥v∥H by claim 3 of Monotonicity of Squaring on the Nonnegative Elements of an Ordered Field and 1 is a bound for ιk; thus ιk∈L(H,Hm).
For ζ∈Hm and v∈H, claim 3 of Elementary Properties of a Complex Inner Product makes the summands with index j=k of ⟨ζ,ιkv⟩Hm equal to ⟨ζj,0⟩H=0, so by claim 7 of Properties of Finite Sums
⟨ζ,ιkv⟩Hm=⟨ζk,v⟩H=⟨πkζ,v⟩H.
Thus the linear map πk is an adjoint of ιk, and by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-unique ιk∗=πk, that is, ιk∗ζ=ζk; also ιk∗∈L(Hm,H) by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint, as H is a complex Hilbert space. Hence ιk∗ιlv=(ιlv)k is v if k=l and 0 if k=l, so ιk∗ιl=IH if k=l and ιk∗ιl=0 if k=l. Finally let ζ∈Hm and j∈[m]. By (R2) and (R1), the j-th component of (∑k=1mιkιk∗)ζ=∑k=1mιkζk is ∑k=1m(ιkζk)j, whose summands with k=j are 0; by claim 7 of Properties of Finite Sums of Vectors it equals ζj. Hence ∑k=1mιkιk∗=IHm.
3. (Block entries) By claim 1, Hm is a complex Hilbert space, so every T∈L(Hm) has an adjoint T∗∈L(Hm) by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint; and Tkl=ιk∗Tιl∈L(H) by claim 2 and Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations. Let S,T∈L(Hm), ζ,ζ′∈Hm and k,l∈[m].
Components. By claim 2 and (R2),
(Tζ)k=ιk∗T(j=1∑mιjιj∗)ζ=j=1∑mιk∗Tιjιj∗ζ=j=1∑mTkjζj.
Inner products. By the definition of ⟨⋅,⋅⟩Hm, the formula just proved and claim 5 of Properties of Finite Sums of Vectors,
⟨ζ,Tζ′⟩Hm=k=1∑m⟨ζk,j=1∑mTkjζj′⟩H=k=1∑mj=1∑m⟨ζk,Tkjζj′⟩H.
Adjoints. By Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-calculus, ιl is the adjoint of ιl∗, and, applying the rule for composites twice, Tlk=ιl∗(Tιk) has the adjoint (Tιk)∗ιl=ιk∗T∗ιl. By Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-unique, (Tlk)∗=ιk∗T∗ιl=(T∗)kl.
Products. By claim 2 and (R2),
(ST)kl=ιk∗S(j=1∑mιjιj∗)Tιl=j=1∑m(ιk∗Sιj)(ιj∗Tιl)=j=1∑mSkjTjl.
Converse. Let (Akl)k,l∈[m] be a family in L(H) and T=∑k=1m∑l=1mιkAklιl∗, an element of the complex vector space L(Hm) by Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations. Let p,q∈[m]. By (R2), Tpq=∑k=1m∑l=1mιp∗ιkAklιl∗ιq. For fixed k, the summands with l=q vanish since ιl∗ιq=0 by claim 2, so by claim 7 of Properties of Finite Sums of Vectors the inner sum is ιp∗ιkAkq; in the same way, since ιp∗ιk=0 for k=p and ιp∗ιp=IH, the outer sum is Apq. So Tpq=Apq. If also T′∈L(Hm) satisfies Tkl′=Akl for all k,l∈[m], then by the component formula (T′ζ)k=∑j=1mAkjζj=(Tζ)k for all ζ∈Hm and k∈[m], so T′=T.
4. (Diagonal operators) Let A∈L(H), ζ∈Hm and j∈[m]. By (R2), (R1) and claim 2, the j-th component of (∑k=1mιkAιk∗)ζ is ∑k=1m(ιkAζk)j, whose summands with k=j are 0, so by claim 7 of Properties of Finite Sums of Vectors it equals Aζj=(A(m)ζ)j. Hence
A(m)=k=1∑mιkAιk∗,
which lies in L(Hm) by claim 2 and Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §operations. Let Dkl=A if k=l and Dkl=0 if k=l. For fixed k the summands ιkDklιl∗ with l=k are 0, so by claim 7 of Properties of Finite Sums of Vectors ∑k=1m∑l=1mιkDklιl∗=A(m), and the converse part of claim 3 gives (A(m))kl=Dkl for all k,l∈[m].
Adjoint. By Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint, A∗∈L(H), so by what was just shown (A∗)(m)∈L(Hm) has the entry A∗ at (k,k) and 0 at (k,l) with k=l. The zero map is its own adjoint, since ⟨0w,v⟩=0=⟨w,0v⟩ by claim 3 of Elementary Properties of a Complex Inner Product, so by the adjoint formula of claim 3 and Bounded Linear Maps between Complex Inner Product Spaces: the Least Bound, Operations, the Underlying Real Structure, Adjoints, Completeness and the Quadratic-Form Bound §adjoint-unique, ((A(m))∗)kl=((A(m))lk)∗ is A∗ if k=l and 0 if k=l. By the uniqueness in the converse part of claim 3, (A(m))∗=(A∗)(m).
Commutation. Let T∈L(Hm) and k,l∈[m]. By the product formula of claim 3, (TA(m))kl=∑j=1mTkj(A(m))jl and (A(m)T)kl=∑j=1m(A(m))kjTjl; the summands with j=l, respectively j=k, are composites with the zero map and hence 0, so by claim 7 of Properties of Finite Sums of Vectors
(TA(m))kl=TklA,(A(m)T)kl=ATkl.
If TA(m)=A(m)T, then these entries agree, so Tkl commutes with A for all k,l∈[m]. Conversely, if TklA=ATkl for all k,l∈[m], then TA(m) and A(m)T have the same entries and are equal by the uniqueness in the converse part of claim 3.