TheoremBase

Proof

For u∈Vu\in V and k∈[n]k\in[n] write u^k=⟨ek,u⟩\hat u_{k}=\langle e_{k},u\rangle, so that Ξ¦(u)k=Re⁑u^k\Phi(u)_{k}=\operatorname{Re}\hat u_{k} and Ξ¦(u)n+k=Im⁑u^k\Phi(u)_{n+k}=\operatorname{Im}\hat u_{k}. Sums of vectors are finite sums in VV and sums of scalars are finite sums in a field; by the definition of a tuple a pp-tuple in a set XX is a map from [p][p] to XX, so those sums apply to tuples without change. By Modulus of a Complex Number, the modulus ∣z∣|z| of a complex number zz is the nonnegative real number with ∣z∣2=(Re⁑z)2+(Im⁑z)2|z|^{2}=(\operatorname{Re}z)^{2}+(\operatorname{Im}z)^{2}. By claim 3 of Canonical Form and Arithmetic of Complex Numbers a complex number is determined by its real and imaginary parts, and by claim 4 of that lemma these parts are additive, so that Re⁑(zβˆ’w)=Re⁑zβˆ’Re⁑w\operatorname{Re}(z-w)=\operatorname{Re}z-\operatorname{Re}w and Im⁑(zβˆ’w)=Im⁑zβˆ’Im⁑w\operatorname{Im}(z-w)=\operatorname{Im}z-\operatorname{Im}w. Since ee is an orthonormal basis, both claims of Orthonormal Expansion and Parseval's Identity in Finite Dimensions are available for it. Finally, let TdE\mathcal{T}_{d_{E}} be the collection of subsets of R2n\mathbb{R}^{2n} that are open in (R2n,dE)(\mathbb{R}^{2n},d_{E}), which is a topology on R2n\mathbb{R}^{2n} by Metric Open Sets Form a Topology.

Claim 1. Injectivity. Suppose Φ(u)=Φ(v)\Phi(u)=\Phi(v). Then Re⁑u^k=Re⁑v^k\operatorname{Re}\hat u_{k}=\operatorname{Re}\hat v_{k} and Im⁑u^k=Im⁑v^k\operatorname{Im}\hat u_{k}=\operatorname{Im}\hat v_{k} for every k∈[n]k\in[n], hence u^k=v^k\hat u_{k}=\hat v_{k} for every k∈[n]k\in[n]. By claim 1 of Orthonormal Expansion and Parseval's Identity in Finite Dimensions,

u=βˆ‘k=1nu^kek=βˆ‘k=1nv^kek=v.u=\sum_{k=1}^{n}\hat u_{k}e_{k}=\sum_{k=1}^{n}\hat v_{k}e_{k}=v .

Surjectivity. Let x∈R2nx\in\mathbb{R}^{2n} have coordinates xix_{i} for i∈[2n]i\in[2n]. For k∈[n]k\in[n] put ck=xk+xn+kic_{k}=x_{k}+x_{n+k}i, where ii is the imaginary unit of The Complex Numbers; by claim 3 of Canonical Form and Arithmetic of Complex Numbers and Real and Imaginary Parts of a Complex Number we have Re⁑ck=xk\operatorname{Re}c_{k}=x_{k} and Im⁑ck=xn+k\operatorname{Im}c_{k}=x_{n+k}. Put u=βˆ‘k=1nckeku=\sum_{k=1}^{n}c_{k}e_{k}. By claim 1 of Elementary Properties of an Orthonormal Family, u^j=cj\hat u_{j}=c_{j} for every j∈[n]j\in[n], so Ξ¦(u)k=xk\Phi(u)_{k}=x_{k} and Ξ¦(u)n+k=xn+k\Phi(u)_{n+k}=x_{n+k} for every k∈[n]k\in[n]. Since every index in [2n][2n] is of one of these two forms, Ξ¦(u)=x\Phi(u)=x.

Claim 2. Let u,v∈Vu,v\in V and put w=uβˆ’vw=u-v. For k∈[n]k\in[n], additivity and homogeneity in the second argument (conditions 2 and 3 of Complex Inner Product Space), together with βˆ’v=(βˆ’1)v-v=(-1)v from Elementary Identities in a Vector Space, give w^k=u^kβˆ’v^k\hat w_{k}=\hat u_{k}-\hat v_{k}, whence

Re⁑w^k=Ξ¦(u)kβˆ’Ξ¦(v)k,Im⁑w^k=Ξ¦(u)n+kβˆ’Ξ¦(v)n+k.\operatorname{Re}\hat w_{k}=\Phi(u)_{k}-\Phi(v)_{k},\qquad \operatorname{Im}\hat w_{k}=\Phi(u)_{n+k}-\Phi(v)_{n+k}.

Write x=Ξ¦(u)x=\Phi(u) and y=Ξ¦(v)y=\Phi(v), and let t:[2n]β†’Rt:[2n]\to\mathbb{R} be the map with ti=(xiβˆ’yi)2t_{i}=(x_{i}-y_{i})^{2}. By Euclidean Distance on Rn\mathbb{R}^n, dE(x,y)d_{E}(x,y) is the nonnegative real number whose square is βˆ‘i=12nti\sum_{i=1}^{2n}t_{i}. Let tβ€²t' be the restriction of tt to [n][n] and let tβ€²β€²:[n]β†’Rt'':[n]\to\mathbb{R} be given by tkβ€²β€²=tn+kt''_{k}=t_{n+k}. Since 2n=n+n2n=n+n, Concatenation of Finite Sums gives

βˆ‘i=12nti=(βˆ‘k=1ntkβ€²)+βˆ‘k=1ntkβ€²β€²=(βˆ‘k=1n(Re⁑w^k)2)+βˆ‘k=1n(Im⁑w^k)2.\sum_{i=1}^{2n}t_{i}=\Bigl(\sum_{k=1}^{n}t'_{k}\Bigr)+\sum_{k=1}^{n}t''_{k}=\Bigl(\sum_{k=1}^{n}(\operatorname{Re}\hat w_{k})^{2}\Bigr)+\sum_{k=1}^{n}(\operatorname{Im}\hat w_{k})^{2}.

By claim 2 of Properties of Finite Sums (additivity of finite sums) and the identity ∣z∣2=(Re⁑z)2+(Im⁑z)2|z|^{2}=(\operatorname{Re}z)^{2}+(\operatorname{Im}z)^{2}, the right-hand side equals

βˆ‘k=1n((Re⁑w^k)2+(Im⁑w^k)2)=βˆ‘k=1n∣w^k∣2=βˆ₯wβˆ₯2,\sum_{k=1}^{n}\bigl((\operatorname{Re}\hat w_{k})^{2}+(\operatorname{Im}\hat w_{k})^{2}\bigr)=\sum_{k=1}^{n}|\hat w_{k}|^{2}=\lVert w\rVert^{2},

the last equality by claim 2 of Orthonormal Expansion and Parseval's Identity in Finite Dimensions. Thus dE(x,y)d_{E}(x,y) and βˆ₯wβˆ₯=d(u,v)\lVert w\rVert=d(u,v) are nonnegative real numbers with the same square, so they are equal by Existence and Uniqueness of the Nonnegative Square Root.

Claim 3. SS is nonempty. Since ee is in particular orthonormal, ⟨e1,e1⟩=1\langle e_{1},e_{1}\rangle=1, and 11 is the nonnegative real number whose square is 11, so βˆ₯e1βˆ₯=1\lVert e_{1}\rVert=1 by Norm Induced by a Complex Inner Product and e1∈Se_{1}\in S.

Identification of Ξ¦(S)\Phi(S). For every k∈[n]k\in[n] we have ⟨ek,0V⟩=0\langle e_{k},0_{V}\rangle=0, by homogeneity in the second argument with the scalar 00 together with 0 ek=0V0\,e_{k}=0_{V} from Elementary Identities in a Vector Space; hence Ξ¦(0V)\Phi(0_{V}) is the point 0E0_{E} of R2n\mathbb{R}^{2n} all of whose coordinates are 00. By claim 2, dE(Ξ¦(u),0E)=d(u,0V)=βˆ₯uβˆ₯d_{E}(\Phi(u),0_{E})=d(u,0_{V})=\lVert u\rVert for every u∈Vu\in V. Consequently

Φ(S)={x∈R2n:dE(x,0E)=1}:\Phi(S)=\{x\in\mathbb{R}^{2n}: d_{E}(x,0_{E})=1\}:

if u∈Su\in S then dE(Ξ¦(u),0E)=βˆ₯uβˆ₯=1d_{E}(\Phi(u),0_{E})=\lVert u\rVert=1; and if dE(x,0E)=1d_{E}(x,0_{E})=1 then, Ξ¦\Phi being surjective by claim 1, x=Ξ¦(u)x=\Phi(u) for some u∈Vu\in V, and βˆ₯uβˆ₯=dE(x,0E)=1\lVert u\rVert=d_{E}(x,0_{E})=1, so u∈Su\in S.

Ξ¦(S)\Phi(S) is bounded. Taking the point 0E0_{E} and the radius 11, every x∈Φ(S)x\in\Phi(S) satisfies dE(0E,x)=1≀1d_{E}(0_{E},x)=1\le1, using symmetry of the metric dEd_{E} (Euclidean Distance is a Metric on Rn\mathbb{R}^n); so Ξ¦(S)\Phi(S) is bounded in (R2n,dE)(\mathbb{R}^{2n},d_{E}).

Ξ¦(S)\Phi(S) is closed. Its complement relative to R2n\mathbb{R}^{2n} is

{x:dE(x,0E)<1}βˆͺ{x:1<dE(x,0E)},\{x: d_{E}(x,0_{E})<1\}\cup\{x: 1<d_{E}(x,0_{E})\},

by trichotomy for the total order of the ordered field R\mathbb{R} (Ordered Field). This complement is open in (R2n,dE)(\mathbb{R}^{2n},d_{E}). Indeed, if dE(x,0E)<1d_{E}(x,0_{E})<1, put r=1βˆ’dE(x,0E)>0r=1-d_{E}(x,0_{E})>0; every yy with dE(x,y)<rd_{E}(x,y)<r satisfies dE(y,0E)≀dE(y,x)+dE(x,0E)<r+dE(x,0E)=1d_{E}(y,0_{E})\le d_{E}(y,x)+d_{E}(x,0_{E})<r+d_{E}(x,0_{E})=1 by the triangle inequality. If instead 1<dE(x,0E)1<d_{E}(x,0_{E}), put r=dE(x,0E)βˆ’1>0r=d_{E}(x,0_{E})-1>0; every yy with dE(x,y)<rd_{E}(x,y)<r satisfies dE(x,0E)≀dE(x,y)+dE(y,0E)d_{E}(x,0_{E})\le d_{E}(x,y)+d_{E}(y,0_{E}), hence 1=dE(x,0E)βˆ’r<dE(x,0E)βˆ’dE(x,y)≀dE(y,0E)1=d_{E}(x,0_{E})-r<d_{E}(x,0_{E})-d_{E}(x,y)\le d_{E}(y,0_{E}). In both cases an open ball about xx is contained in the complement, so the complement belongs to TdE\mathcal{T}_{d_{E}} and Ξ¦(S)\Phi(S) is closed in (R2n,TdE)(\mathbb{R}^{2n},\mathcal{T}_{d_{E}}).

Conclusion. Since Ξ¦(S)\Phi(S) is closed in (R2n,TdE)(\mathbb{R}^{2n},\mathcal{T}_{d_{E}}) and bounded in (R2n,dE)(\mathbb{R}^{2n},d_{E}), the implication from statement 2 to statement 1 of Heine-Borel Theorem in Rn\mathbb{R}^n shows that Ξ¦(S)\Phi(S) is compact in (R2n,TdE)(\mathbb{R}^{2n},\mathcal{T}_{d_{E}}). By claims 1 and 2, Ξ¦\Phi is a distance-preserving bijection from the metric space (V,d)(V,d) onto the metric space (R2n,dE)(\mathbb{R}^{2n},d_{E}), so claim 3 of A Distance-Preserving Bijection is a Homeomorphism shows that SS is compact in (V,Td)(V,\mathcal{T}_{d}).

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…