536 CHAPTER 19. HILBERT SPACES

Therefore, the open sets, B(x, 1

2

)for x ∈ S are disjoint and cover S. Since H is assumed

to be separable, there exists a point from a countable dense set in each of these disjointballs showing there can only be countably many of the balls and that consequently, S iscountable as claimed.

It remains to verify 19.5.26 and that span(S) is dense. If span(S) is not dense, thenspan(S) is a closed proper subspace of H and letting y /∈ span(S),

z≡ y−Py||y−Py||

∈ span(S)⊥ .

But then S∪{z} would be a larger orthonormal set of vectors contradicting the maximalityof S.

It remains to verify 19.5.26. Let S = {xi}∞

i=1 and consider the problem of choosing theconstants, ck in such a way as to minimize the expression∣∣∣∣∣

∣∣∣∣∣x− n

∑k=1

ckxk

∣∣∣∣∣∣∣∣∣∣2

=

||x||2 +n

∑k=1|ck|2−

n

∑k=1

ck (x,xk)−n

∑k=1

ck(x,xk).

This equals

||x||2 +n

∑k=1|ck− (x,xk)|2−

n

∑k=1|(x,xk)|2

and therefore, this minimum is achieved when ck = (x,xk) and equals

||x||2−n

∑k=1|(x,xk)|2

Now since span(S) is dense, there exists n large enough that for some choice of constants,ck, ∣∣∣∣∣

∣∣∣∣∣x− n

∑k=1

ckxk

∣∣∣∣∣∣∣∣∣∣2

< ε.

However, from what was just shown,∣∣∣∣∣∣∣∣∣∣x− n

∑i=1

(x,xi)xi

∣∣∣∣∣∣∣∣∣∣2

∣∣∣∣∣∣∣∣∣∣x− n

∑k=1

ckxk

∣∣∣∣∣∣∣∣∣∣2

< ε

showing that limn→∞ ∑ni=1 (x,xi)xi = x as claimed. This proves the theorem.

The proof of this theorem contains the following corollary.

Corollary 19.5.3 Let S be any orthonormal set of vectors and let

{x1, · · · ,xn} ⊆ S.

536 CHAPTER 19. HILBERT SPACESTherefore, the open sets, B (x, 5) for x € S are disjoint and cover S. Since H is assumedto be separable, there exists a point from a countable dense set in each of these disjointballs showing there can only be countably many of the balls and that consequently, S' iscountable as claimed.It remains to verify 19.5.26 and that span(S) is dense. If span(S) is not dense, thenspan (S) is a closed proper subspace of H and letting y ¢ span(S),y—Py Lz= ——_€span(S)~.lly — Py||But then SU {z} would be a larger orthonormal set of vectors contradicting the maximalityof S.It remains to verify 19.5.26. Let S = {x;};"., and consider the problem of choosing theconstants, c, in such a way as to minimize the expression2x — y CrXkk=1n n nIlx|[? + y lex? — Ye cK (x,xK) — Y cx (x, xe)-k=1 k=1 k=lThis equals2.x 2 2IIx] + YE Joe — (exe) — Y (axe)kal k=land therefore, this minimum is achieved when cy = (x, x;,) and equalsn2 2lll — YE 1%)k=1Now since span (S) is dense, there exists n large enough that for some choice of constants,Cks2<E.x— y. CKXkk=1However, from what was just shown,2<2<€Enx= y? (x, Xj) Xji=1x — y CkXkk=1showing that limp... )7_, (x, x;) x; = x as claimed. This proves the theorem.The proof of this theorem contains the following corollary.Corollary 19.5.3 Let S be any orthonormal set of vectors and let{x1,-++ Xn} CS.