TheoremBase

The spanned subspaces are handled by the lemma on projections onto the span of an orthonormal tuple; the equivalence with the exhausting property uses the orthonormal expansion in one direction and the nearest-point property of the projection in the other.

Proof

For each n∈Nn\in\mathbb{N} the tuple e(n)e^{(n)} is orthonormal, since its components are components of the orthonormal sequence (ek)(e_{k}) and distinct indices in [n][n] are distinct in N\mathbb{N}. Write P(n)P^{(n)} for the projection onto Hn=span⁡(e(n))H_{n}=\operatorname{span}(e^{(n)}) of Projection onto the Span of an Orthonormal Tuple, and Coordinates on a Finite-Dimensional Subspace, so that P(n)x=∑k=1n⟨x,ek⟩ekP^{(n)}x=\sum_{k=1}^{n}\langle x,e_{k}\rangle e_{k}.

Claim 1. By The Span of a Finite Family is the Smallest Subspace Containing It the set HnH_{n} is a linear subspace of HH, and it is closed in HH by claim 4 of Projection onto the Span of an Orthonormal Tuple, and Coordinates on a Finite-Dimensional Subspace, hence a closed linear subspace. Since ∣e1∣=1|e_{1}|=1 we have e1≠0He_{1}\ne0_{H}, so Hn≠{0H}H_{n}\ne\{0_{H}\}; apply claim 1 of Gram-Schmidt Orthonormalisation in a Real Inner Product Space to the tuple e(n)e^{(n)}, whose span is HnH_{n}. Since e(n)e^{(n)} is itself an orthonormal nn-tuple with span⁡(e(n))=Hn\operatorname{span}(e^{(n)})=H_{n}, it is a tuple of the kind to which the second sentence of that claim refers, so the corresponding tuple in HnH_{n} is a basis of HnH_{n}; hence HnH_{n} is finite-dimensional.

For the inclusion, let u∈Hnu\in H_{n}, say u=∑k=1nckeku=\sum_{k=1}^{n}c_{k}e_{k} with c∈Rnc\in\mathbb{R}^{n}. Let c′∈Rn+1c'\in\mathbb{R}^{n+1} have components ck′=ckc'_{k}=c_{k} for k∈[n]k\in[n] and cn+1′=0c'_{n+1}=0. By the recursion of Finite Sum Notation in a Vector Space and claim 1 of Properties of Finite Sums of Vectors,

∑k=1n+1ck′ek=(∑k=1nckek)+0 en+1=u+0H=u,\sum_{k=1}^{n+1}c'_{k}e_{k}=\Bigl(\sum_{k=1}^{n}c_{k}e_{k}\Bigr)+0\,e_{n+1}=u+0_{H}=u,

using claim 3 of Elementary Identities in a Vector Space for 0 en+1=0H0\,e_{n+1}=0_{H}. Hence u∈Hn+1u\in H_{n+1} and Hn⊆Hn+1H_{n}\subseteq H_{n+1}.

Finally, HH is a real Hilbert space and HnH_{n} is closed, so by claim 5 of Projection onto the Span of an Orthonormal Tuple, and Coordinates on a Finite-Dimensional Subspace the map P(n)P^{(n)} is the orthogonal projection PHnP_{H_{n}}, which is the asserted formula.

Claim 2. Suppose first that (ek)k∈N(e_{k})_{k\in\mathbb{N}} is an orthonormal basis of HH. Conditions 1 and 2 of Exhausting Sequence of Finite-Dimensional Subspaces of a Real Hilbert Space hold by claim 1. For condition 3, let x∈Hx\in H and let ε\varepsilon be positive. By claim 3 of Orthonormal Expansions in a Real Hilbert Space the series ∑k=1∞⟨x,ek⟩ek\sum_{k=1}^{\infty}\langle x,e_{k}\rangle e_{k} converges to xx, so there is n∈Nn\in\mathbb{N} with ∣x−PHnx∣<ε|x-P_{H_{n}}x|<\varepsilon, and PHnx∈HnP_{H_{n}}x\in H_{n}. Hence (Hn)n∈N(H_{n})_{n\in\mathbb{N}} is an exhausting sequence for HH.

Conversely, suppose (Hn)n∈N(H_{n})_{n\in\mathbb{N}} is an exhausting sequence for HH, and let x∈Hx\in H satisfy ⟨x,ek⟩=0\langle x,e_{k}\rangle=0 for every k∈Nk\in\mathbb{N}. By claim 3 of Properties of Finite Sums of Vectors, applied with the factor 00, and claim 1 above,

PHnx=∑k=1n0 ek=0Hfor every n∈N.P_{H_{n}}x=\sum_{k=1}^{n}0\,e_{k}=0_{H}\qquad\text{for every }n\in\mathbb{N}.

Let ε\varepsilon be positive. By condition 3 of Exhausting Sequence of Finite-Dimensional Subspaces of a Real Hilbert Space there are n∈Nn\in\mathbb{N} and y∈Hny\in H_{n} with ∣x−y∣<ε|x-y|<\varepsilon. By claim 5 of Projection onto the Span of an Orthonormal Tuple, and Coordinates on a Finite-Dimensional Subspace the point PHnxP_{H_{n}}x is a nearest point of HnH_{n} to xx, so

∣x∣=∣x−0H∣=∣x−PHnx∣≤∣x−y∣<ε.|x|=|x-0_{H}|=|x-P_{H_{n}}x|\le|x-y|<\varepsilon .

As ε\varepsilon was an arbitrary positive real number, Comparison of Real Numbers with Arbitrary Positive Slack gives ∣x∣≤0|x|\le0; since 0≤∣x∣0\le|x|, we get ∣x∣=0|x|=0 and hence ⟨x,x⟩=0\langle x,x\rangle=0, so x=0Hx=0_{H} by condition (d) of Real Inner Product Spaces (i). Thus (ek)k∈N(e_{k})_{k\in\mathbb{N}} is an orthonormal basis of HH.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…