A Basis Theory Primer Expanded Edition Solutions Manual - Heil Problems

Christopher Heil
A Basis Theory Primer:

Solutions Manual
August 1, 2010
2010
c by Christopher Heil
Detailed Solutions to Exercises
These are my solutions to the exercises from “A Basis Theory Primer.” Of

course, many problems have solutions other than the ones I sketch. Please
send comments and corrections to “heil@math.gatech.edu”.
Detailed Solutions to Exercises from Chapter 1
1.1 Since k · k is a norm, we must have λ = k1k 6= 0. Then given any x ∈ F,

we have kxk = kx · 1k = |x| k1k = λ |x|.
1.2 (a) Suppose xn → x, and choose ε > 0. Then there exists N > 0 such that
kx − xn k < ε for all n > N. Hence, by the Triangle Inequality, if m, n > N
then kxm − xn k ≤ kxm − xk + kx − xn k < 2ε. Thus {xn }n∈N is Cauchy.
(b) Suppose that {xn }n∈N is Cauchy. Then there exists an N > 0 such
that kxm − xn k < 1 for all m, n ≥ N. Therefore, for n ≥ N we have
kxn k = kxn − xN + xN k ≤ kxn − xk + kxN k ≤ 1 + kxN k.
Hence for any n we have

kxn k ≤ max kx1 k, . . . , kxN −1 k, kxN k + 1 ,
so {xn } is bounded.
(c) Given x, y ∈ H, we have
kxk = k(x − y) + yk ≤ kx − yk + kyk,
so kxk − kyk ≤ kx − yk.

Reversing
the roles of x and y, we obtain kyk − kxk ≤
kx − yk, so we have kxk − kyk ≤ kx − yk.
(d) By the Reverse Triangle Inequality, if xn → y, then
2 Detailed Solutions

kxk − kxn k ≤ kx − xn k → 0.
(e) If xn → x and yn → y, then
k(x + y) − (xn + yn )k = k(x − xn ) + (y − yn )k ≤ kx − xn k + ky − yn k → 0.
(f) Suppose xn → x and cn → c. Then C = sup |cn | < ∞, so
kcx − cn xn k ≤ kcx − cn xk + kcn x − cn xn k

= |c − cn | kxk + |cn | kx − xn k
≤ |c − cn | kxk + C kx − xn k → 0.
1.3 (a) Suppose first that 1 ≤ p < ∞ and q = ∞. Given x = (x1 , . . . , xd ) ∈

Fd , we have |xk | ≤ kxk∞ for each k. Therefore
1/p 1/p
|x|p = |x1 |p + · · · + |xd |p ≤ d kxk∞ = d1/p kxk∞ .
Conversely, kxk∞ = |xk | for some particular k, so we have

1/p
kxk∞ = |xk | ≤ |x1 |p + · · · + |xd |p = kxkp .
Hence | · |p and | · |∞ are equivalent norms on Fd .

If we now choose any 1 ≤ p, q ≤ ∞, then | · |p is equivalent to | · |∞ , and
| · |∞ is equivalent to | · |q , so it follows that | · |p is equivalent to | · |q .
1.4 If kxk = 0 then ck (x) = 0 for each k, so x = 0. All of the other properties
of a norm follow easily. To show completeness, note that for each 1 ≤ k ≤ d
we have |ck (x)| ≤ kxk. Also, the ck are linear, so if {xn }n∈N is a Cauchy
sequence in X, then for each fixed k we have
|ck (xm ) − ck (xn )| = |ck (xm − xn )| ≤ kxm − xn k.
This implies that {ck (xn )}n∈N is a Cauchy sequence of scalars and hence
Pd
converges to some scalar ck . Define x = k=1 ck xk . The fact that xn → x
p
then follows just as in the proof that ℓ is complete.
1.5 We are given that kxn+1 − xn k < 2−n for every n. Choose any ε > 0,
and let N be large enough that 2−N +1 < ε. If n > m > N, then we have
n−1
X ∞
X 1 1 1
kxn − xm k ≤ kxk+1 − xk k ≤ k
= m−1 < N −1 < ε.
2 2 2
k=m k=m
Hence {xn } is Cauchy.

1.6 Suppose that every subsequence of {xn } has a subsequence that converges
to x, but the full sequence {xn } does not converge to x. Then there exists
an ε > 0 such that given any N we can find an n > N such that kx − xn k > ε.
Detailed Solutions 3
Then we can find a subsequence {xnk } such that kx − xnk k > ε for every k.
But then no subsequence of {xnk } can converge to x, which is a contradiction.
1.7 It is clear that k · kR is a norm on XR , so the issue is to show that XR is
complete. Suppose that {xn } is Cauchy in XR . Since kxm −xn k = kxm −xn kR ,
we have that {xn } is Cauchy in X and therefore converges to some x ∈ X.
But then kx − xn kR = kx − xn k → 0, so xn converges to x in XR . Hence XR
is complete.
1.8 (a) The Triangle Inequality follows from d(f, h) = k(f − g) + (g − h)k ≤
kf − gk + kg − hk = d(f, g) + d(g, h).
(b) A sequence {fn }n∈N converges to f ∈ X if limn→∞ d(fn , f ) = 0, i.e.,
if
∀ ε > 0, ∃ N > 0, ∀ n ≥ N, d(fn , f ) < ε.
A sequence {fn }n∈N is Cauchy if
∀ ε > 0, ∃ N > 0, ∀ m, n ≥ N, d(fm , fn ) < ε.
(c) For each n, let xn be a rational number such that π < xn < π + 1/n.
Then {xn } is Cauchy, but it does not converge in the space Q. It does converge
in the larger space R, but since the limit does not belong to Q, it is not
convergent in Q.
1.9 (a) Let δ1 , δ2 denote the first two standard basis vectors. These belong
to ℓp , but we have
kx + ykp = (1 + 1)1/p = 21/p
while
kxkp + kykp = 1 + 1 = 2.
Since p < 1 we have 21/p > 2, so the Triangle Inequality is not satisfied by
k · kp .
(b) Suppose that 0 < p < 1. Let f (t) = (1 + t)p and g(t) = 1 + tp for t > 0.
Then f (0) = 1 = g(0). Also,
1 1
f ′ (t) = p (1 + t)p−1 = p and g ′ (t) = ptp−1 = p .
(1 + t)1−p t1−p
Since 0 < 1 − p < 1, we have t1−p < (1 + t)1−p , and therefore f ′ (t) ≤ g ′ (t)
for t > 0. Hence g is increasing faster than f, and therefore f (t) ≤ g(t) for all
t ≥ 0. Next, given any a, b ≥ 0, we have
p p
b b
(a + b)p = ap 1 + ≤ ap 1 + = ap + b p .
a a
Hence, if x, y ∈ ℓp (I), then

X X
kx + ykpp = |xk + yk |p ≤ |xk |p + |yk |p = kxkpp + kykpp .
k∈I k∈I
This establishes the Triangle Inequality.

(c) To show that the unit ball is not convex, note that the the standard
basis vectors δ1 and δ2 both belong to the closed unit ball

D = x ∈ ℓp : kxkp ≤ 1 ,
but δ + δ p 1 p 1 p
1 2 2
= + = p = 21−p > 1,
2 p 2 2 2
so (δ1 + δ2 )/2 does not belong to the closed unit ball. Hence this set is not
convex in ℓp . This also shows that if ε > 0 is small, then the open unit ball
B1+ε (0) is not convex. By rescaling, the unit ball B1 (0) is not convex either.
1.10 (a) Set f (t) = tθ −θt−(1−θ). Then f ′ (t) = θtθ−1 −θ. We have f ′ (t) = 0
if and only if t = 1. Also, f is increasing for 0 < t < 1 and decreasing for
t > 1, and f (1) = 0, so f (t) ≤ 0 for all t > 0, with equality only for t = 1.
(b) Note that
1 1 p p′ 1 p′
+ ′ = 1, p′ = , = , p′ − = 1.
p p p−1 p p−1 p
′
With t = ap b−p and θ = 1/p, we have by part (a) that
1 1
′
−p′ /p

p −p′ 1/p p −p′ ap b−p 1
ab = a b ≤ a b + 1− = + ′.
p p p p
′
Multiplying through by bp and using the fact that p′ − (p′ /p) = 1, we obtain
′
′ ′ ap bp
ab = a bp −p /p ≤ + ′.
p p
′ ′
Equality holds if and only if ap b−p = 1. This is equivalent to bp = ap , or
′
b = ap/p = ap−1 .
p p′
1.11 Case 1 < p < ∞. By Exercise 1.10, equality holds in ab ≤ ap + bp′ if
and only if b = ap−1 . For the normalized case kxkp = kykp′ = 1, equality in
Hölder’s Inequality requires that we have equality in equation (1.5), and this
will happen if and only if |yk | = |xk |p−1 for each k. This is equivalent to
′
|yk |p = |yk |p/(p−1) = |xk |p .
For the nonnormalized case, if x, y 6= 0, equality holds in Hölder’s Inequality if

and only if it holds when we replace x and y by x/kxkp and y/kykp′ . Therefore,
we must have
′ p′ p
|yk |p |yk | |xk | |xk |p
= = = , k ∈ I.
p′
kykp′ kykp′ kxkp kxkpp
′ ′
Hence α |xk |p = β |yk |p with α = kykpp′ and β = kxkpp . On the other hand, if
either x = 0 or y = 0, then we have equality in Hölder’s Inequality, and we
′
also have α |xk |p = β |yk |p with α, β not both zero.
′
For the converse direction, suppose that α |xk |p = β |yk |p for each k ∈ I,
where α, β ∈ F are not both zero. If α = 0, then yk = 0 for every k, and hence
we trivially have kxyk1 = 0 = kxkp kykp′ . Likewise, equality holds trivially if
β = 0. Therefore, we can assume both α, β 6= 0, and by dividing both sides
′
by β, we may assume that β = 1 and α > 0. Then we have |yk |p = α |xk |p , so
′ X ′ X
kykpp′ = |yk |p = α |xk |p = α kxkpp .
k∈I k∈I
If either x = 0 or y = 0 then equality holds trivially in Hölder’s Inequality, so

let us assume both x, y 6= 0. Then we have
′
|yk |p α|xk |p |xk |p
= p = .
kxkpp
′
kykpp′ αkxkp
By the work above, this implies that equality holds in Hölder’s Inequality.
Case p = 1, p′ = ∞. Set M = supk |yk |. Suppose equality holds in Hölder’s
Inequality, i.e., X
X
|xk yk | = |xk | sup |yk | .
k
k∈I k∈I
Then X X
|xk yk | = M |xk |.
k∈I k∈I
Hence X
(M − |yk |) |xk | = 0,
k∈I
but 0 ≤ M − |yk | for every k, so we must have (M − |yk |) |xk | = 0 for every k.
Thus whenever xk 6= 0, we must have |yk | = M.
Conversely, if |yk | = M for all k such that xk 6= 0, equality holds in
Hölder’s Inequality.
1.12 We have ℓp ⊆ ℓ∞ for every p. Further, the constant sequence x =
(1, 1, 1, . . . ) belongs to ℓ∞ but not to any ℓp with p finite, so the inclusion is
proper.
Suppose 0 < p ≤ q < ∞ and x ∈ ℓp . If kxk∞ = 1, then
X
∞ 1/q X
∞ 1/q
q p q−p
kxkq = |xk | = |xk | |xk |
k=1 k=1
X∞ 1/q
≤ |xk |p = kxkp/q
p ≤ kxkp ,
k=1
the last inequality following from the fact that p/q ≤ 1 and kxkp ≥ kxk∞ = 1.
For the general case, apply this inequality to x/kxk∞ .
To show that the inclusion is strict, set xk = k −1/p . Then since q/p > 1,
we have
X∞
q 1
kxkq = q/p
< ∞,
k=1
k
while
∞
X 1
kxkpp = = ∞.
k
k=1
2
Another example is xk = (k log k)−1/q for k ≥ 2. The Integral Test shows
that
X∞
1
kxkqq = < ∞,
k=2
k log2 k
while ∞
X 1
kxkpp = 2 = ∞.
k=2
(k log k)p/q
1.13 We need the following lemma.
Lemma. If E ⊆ R is measurable and 0 < |E| < ∞, then there exists a

measurable F ⊆ E such that |F | = |E|/2.
Proof. Let Et = E ∩ (−∞, t]. Then the sets Et are nested increasing
with t, their union is E, and their intersection is empty. Applying continuity
from both above and below, which is applicable since |E| < ∞, we conclude
that
lim |Et | = |E| and lim |Et | = 0.
t→∞ t→−∞
A similar argument shows that |Et | is a continuous function of t. Therefore

there must be some t such that |Et | = |E|/2. ⊓⊔
Now we return to the proof of the exercise. Assume that 1 ≤ p < q < ∞.
Taking E0 = E, by applying the Lemma, we can find a set E1 ⊆ E such
that |E1 | = |E|/2. Noting that |E\E1 | = |E|/2, we apply the lemma to
find E2 ⊆ E\E1 with |E2 | = |E|/4, and note that E2 is disjoint from E1 .
Continuing in this way we construct disjoint En ⊆ E such that |En | = 2−n |E|.
Consider
X
f = 2n/q χEn .
n
We have
Z X X
kf kqLq = |f |q = 2n |En | = 2n 2−n |E| = ∞,
E n n
/ Lq (E). On the other hand,

so f ∈
Z X
kf kpLp = |f |p = 2np/q |En |
E n
X
= 2np/q 2−n |E|
n
X p
≤ 2n( q −1) |E| < ∞,
n
since p/q < 1. Hence f ∈ Lp (E). Note that this f is unbounded, so this is
also an example of a function in Lp (E) that does not belong to L∞ (E).
1.14 Suppose that x ∈ ℓq (I) for some finite q. Since only countably many
components of x can be nonzero, it suffices to consider I = N.
If x = 0 then kxkp = 0 for every p, so we are done. Therefore, we may
assume x 6= 0, which implies kxk∞ 6= 0. By dividing through by kxk∞ , we
may assume that kxk∞ = 1. Then for every p we have 1 = kxk∞ ≤ kxkp . In
particular, |xk | ≤ 1 for every k. Therefore, for p ≥ q we have |xk |p ≤ |xk |q .
Hence x ∈ ℓp . Further, for p ≥ q,
X
∞ 1/p
p
kxk∞ = 1 ≤ kxkp = |xk |
k=1
X
∞ 1/p
q
≤ |xk |
k=1
= kxkq/p
q
→ 1 = kxk∞ as p → ∞,
where the limit exists because kxkq is finite and nonzero.

On the other hand, the vector x = (1, 1, 1, . . . ) satisfies kxk∞ = 1, but
kxkp = ∞ for every p < ∞.
1.15 The proof is very similar to the proof that ℓp (N) is a Banach space.
For example, if {xn }n∈N is a Cauchy sequence in ℓp (I) and we write xn =
(xn (i))i∈I , then for each fixed i we have that (xn (i))n∈N is a Cauchy sequence
of scalars, and hence converges to some scalar x(i). For a given n, at most
countable many components of xn can be nonzero. As a countable union of
countable sets is countable, at most countably many components of x can be

nonzero. An argument similar to the one used in the proof of Theorem 1.14
then shows that xn → x in the norm of ℓp (I), so ℓp (I) is complete.
1.16 (a) The Triangle Inequality follows from Hölder’s Inequality, and an
argument similar to the one used in the proof of Exercise 1.3 shows that the
norm k(x, y)kp is equivalent to the norm k(x, y)k∞ .

(b) Suppose that X and Y are complete. If (xn , yn ) is a Cauchy sequence
in X × Y, then {xn } is a Cauchy sequence in X since we always have kxkX ≤
k(x, y)kp . Hence xn → x for some x ∈ X, and similarly yn → y for some y ∈ Y.
It then follows that (xn , yn ) → (x, y) with respect to the norm on X × Y.
1.17 (a) ⇒. Suppose that E is closed. If x ∈/ E then since U = X\E is open,
there exists some r > 0 such that Br (x) ⊆ X\E. Consequently, every element
of E is at least a distance r from x, and therefore x cannot be a limit point
of E. Hence every limit point of E belongs to E.
⇐. Suppose that E is not closed. Then X\E is not open, so there exists
some x ∈/ E such that Br (x) is not contained in X\E for any r > 0. Therefore,
for each r = 1/n we can find an xn ∈ E ∩ B1/n (x). But then xn → x and
xn ∈ E. Since x ∈ / E while xn ∈ E, we must have xn 6= x. Therefore x is a
limit point of E, so E does not contain all of its limit points.
(b) Let F = E ∪ {x ∈ X : x is a limit point of E}. Our goal is to show
that F = E.
Suppose that x ∈ F C = X\F. Then, by definition, x ∈ / E and x is not a
limit point of E. If every open ball Br (x) contained an element of E (which
necessarily must not be x), then x would be a limit point of E. Therefore,
there exist some Br (x) that contains no points of E. Suppose that Br (x)
contained some limit point y of E. Then there would be points xn ∈ E such
that xn → y. But then for n large enough we would have xn ∈ Br (x), which
is a contradiction. Therefore Br (x) ⊆ F C . Hence F C is open, so F is closed.
Since E ⊆ F, this implies that E ⊆ F .
Now we’ll show that F ⊆ E. Since E ⊆ E, we simply have to show that
the limit points of E are contained in E. So, suppose that x ∈ / E. Since E is
closed, there exists some r > 0 such that Br (x) ⊆ X\E. Hence Br (x) contains
no points of E, and therefore x cannot be a limit point of E. Hence F ⊆ E.
(c) This follows by combining parts (a) and (b).
1.18 ⇒. Suppose that M is a Banach space with respect to the norm of X.
Suppose that xn ∈ M and xn → x ∈ X. Then {xn }n∈N is a Cauchy sequence
in M and hence must converge to some element y ∈ M. However, as a sequence
in X we then have that xn → x and xn → y, so by uniqueness of limits,
x = y ∈ M. Therefore M is closed.
⇐. Suppose that M is a closed subspace of X, and suppose that {xn }n∈N
is a Cauchy sequence in M. Then {xn }n∈N is Cauchy in X, so there exists
some x ∈ X such that xn → x. However, M is closed, so this implies that

x ∈ M. Therefore every Cauchy sequence in M converges to an element of M,
so M is complete.
1.19 Suppose that f ∈ Cb (R), and choose any M > 0. If |f | ≤ M everywhere,
then we certainly have |f | ≤ M a.e.
For the converse, choose any M > 0, and suppose that there is a point
where |f (x)| > M. Then since |f | is continuous, there must be an open inter-
val I containing x such that |f (y)| > M for y ∈ I. But then |f | > M on a set
with positive measure, i.e., it is not true that |f | ≤ M a.e. Hence this shows
by contrapositive argument that if |f | ≤ M a.e., then |f | ≤ M everywhere.
Consequently,
inf{M : f (x) ≤ M a.e.} = inf{M : f (x) ≤ M for every x} = sup |f (x)|,

x∈R
so the uniform and L∞ norms agree for functions in Cb (R).

1.20 (a) Suppose that
{xN }N ∈N is a sequence
in c and xN → x in ℓ∞ -norm.
Write xN = xN (k) k∈N and x = x(k) k∈N . Since ℓ∞ convergence implies
componentwise convergence, we have that x(k) = limN →∞ xN (k) for each
k ∈ N.
By hypothesis, yN = limk→∞ xN (k) exists for each N. We have
|yM −yN | = lim |xM (k)−xN (k)| ≤ sup |xM (k)−xN (k)| = kxM −xN kℓ∞ ,
k→∞ k
so {yN }N ∈N is a Cauchy sequence of scalars and therefore converges, say to y.

Fix any ε > 0. Then there exists an N such that kx − xN k∞ < ε and
|y − yN | < ε. Since |x(k) − xN (k)| < ε for every k, we have

lim |y − x(k)| ≤ lim sup |y − yN | + |yN − xN (k)| + |xN (k) − x(k)|
k→∞ k→∞
≤ ε + 0 + ε = 2ε.
Since ε is arbitrary, we conclude that y = limk→∞ x(k), so x ∈ c. Thus c is

closed in ℓ∞ .
Now assume in addition that xN ∈ c0 for each N. Then yN = 0 for every N,
so by the argument above we see that y = 0. Hence x ∈ c0 , so c0 is closed in
ℓ∞ as well.
(b) Choose any x = (x(1), x(2), . . . ) ∈ c0 . Define
xN = (x(1), . . . , x(N ), 0, 0, . . . ).
Then xN ∈ c00 , and
lim kx − xN kℓ∞ = lim sup |x(k)| = lim sup |x(k)| = 0.

N →∞ N →∞ k>N k→∞
Hence c00 is dense in c0 . However, c00 is not closed, since any x ∈ c0 with
infinitely many nonzero components is an accumulation point of c00 but does
not belong to c00 .
(c) Choose any x ∈ c0 . Write x = (x(1), x(2), . . . ), and set
N
X
xN = (x(1), . . . , x(N ), 0, 0, . . . ) = x(k) δk .
k=1
By part (b) we know thatP kx − xN kℓ∞ → 0 as N → ∞. Since P the xN is the

partial sums of the series x(k) δk , we conclude
P that x = x(k) δk .
On the other hand, if a series x = ck δk converges in ℓ∞ norm then
the partial sums must converge componentwise. The partial sums are xN =
(c1 , . . . , cN , 0, 0, . . . ), so the kth component of x is precisely ck .
1.21 (a) The fact that Cb R) is a vector space and k · k∞ is a norm on Cb (R)
is clear, so we only need to show completeness.
Suppose that {fn }n∈N is a Cauchy sequence in Cb (R) with respect to the
uniform norm. Then for each x, we have
|fm (x) − fn (x)| ≤ kfm − fn k∞ ,
so {fn (x)}n∈N is a Cauchy sequence of scalars, and hence converges. Define
f (x) = limn→∞ fn (x).
Now choose ε > 0. Then there exists an N such that kfm − fn k∞ < ε for
all m, n > N. Fix n > N. Then for every x we have
|f (x) − fn (x)| = lim |fm (x) − fn (x)| ≤ kfm − fn k∞ ≤ ε,
m→∞
so kf − fn k∞ ≤ ε for all n > N. Also, kf k∞ ≤ kf − fn k∞ + kfn k∞ , so f is

bounded. Finally, the uniform limit of continuous functions is continuous, so
f ∈ Cb (R) and fn → f uniformly. This shows that Cb (R) is complete.
(b) Suppose that fn ∈ C0 (R) and fn → f uniformly. By part (a) we have
f ∈ Cb (R). Given ε > 0, there exists some n such that kf − fn k∞ < ε. For
this n, there exists an R > 0 such that |fn (x)| < ε for all |x| > R. Hence for
|x| > R we have
|f (x)| ≤ |f (x) − fn (x)| + |fn (x)| ≤ kf − fn k∞ + ε ≤ 2ε.
Hence f (x) → 0 as |x| → ∞, so f ∈ C0 (R). Thus C0 (R) is a closed subspace
of Cb (R).
(c) Choose any g ∈ C0 (R). Then there exists an N > 0 such that |g(x)| < ε
for all |x| > N. Set

g(x),
 |x| ≤ N,
gN (x) = linear, N ≤ |x| ≤ N + 1,


0, |x| > N + 1.
Each gN belongs to Cc (R), and

kg − gN k∞ = sup |g(x) − gN (x)| ≤ sup |g(x)| + |gN (x)| ≤ 2ε,
|x|>N |x|>N
2
so gN → g uniformly. Hence Cc (R) is dense in C0 (R). However, if g(x) = e−x ,
then g belongs to C0 (R) but does not belong to Cc (R), so Cc (R) is not closed.
(d) Suppose that fn ∈ C(T) and fn → f uniformly. By part (a) we have
f ∈ Cb (R). Since uniform convergence implies pointwise convergence, for each
x ∈ R we have
f (x + 1) = lim fn (x + 1) = lim fn (x) = f (x).

n→∞ n→∞
Hence f is 1-periodic, so f ∈ C(T) and therefore C(T) is closed in Cb (R).

1.22 (a) Let us show that Cb1 (R) is complete. Suppose that {fn }n∈N is
a Cauchy sequence in Cb1 (R). Then {fn }n∈N is Cauchy in Cb (R), so there
exists an f ∈ Cb (R) such that fn → f uniformly. Additionally, by definition
of Cb1 (R), we know that
′
kfm − fn′ k∞ ≤ kfm − fn k∞ + kfm
′
− fn′ k∞ = kfm − fn kCb1 ,
so {fn′ }n∈N is Cauchy with respect to the uniform norm. That is, {fn′ }n∈N is a
Cauchy sequence in Cb (R). Since Cb (R) is complete, there exists a g ∈ Cb (R)
such that fn′ → g uniformly. So, the remaining point is to show that g = f ′ ,
for then we will have that fn → f in the norm of Cb1 (R).
′
To see this, fix ε > 0. Then there exists an N > 0 such that kfm −fn′ k∞ < ε
whenever m, n > N. Fix x, y ∈ R and m, n > N. Applying the Mean-Value
Theorem to the function fm − fn , there exists a point c (depending on m, n,
x, and y) between x and y such that
′
(fm − fn )(y) − (fm (x) − fn )(x) = (y − x) (fm − fn′ )(c).
Consequently,

fm (y) − fm (x) fn (y) − fn (x)
− = |f ′ (c) − f ′ (c)| ≤ kf ′ − f ′ k∞ < ε.
y−x y−x m n m n
Letting m → ∞, we conclude that

f (y) − f (x) fn (y) − fn (x)
− ≤ ε.
y−x y−x
This is valid for x, y ∈ R as long as n > N.

Now, since fn is differentiable, there exists a δ > 0 such that

′ fn (y) − fn (x)

|x − y| < δ =⇒ fn (x) −
y−x < ε.
Further, since fn′ → g uniformly, there exists an M such that kfn′ − gk∞ < ε
whenever n > M. Fix x, and suppose that |x − y| < δ. Then for n > M, N
we have

g(x) − f (y) − f (x) < |g(x) − f ′ (x)| + f ′ (x) − fn (y) − fn (x)
y−x n n
y−x

fn (y) − fn (x) f (y) − f (x)
+ −
y−x y−x
< ε + ε + ε = 3ε.
Hence
f (y) − f (x)
g(x) = lim ,
y→x y−x
so f is differentiable at x, and f ′ (x) = g(x). Thus fn → f in the norm of
Cb1 (R), so this space is complete.
A proof by induction shows that Cbm (R) is complete for each m.
(b) If we replace the norm on Cb1 (R) by the uniform norm, then it is no
longer complete. Let w(x) = max{1 − |x|, 0} be the hat function on [−1, 1].
Then we can find differentiable functions fn ∈ Cb1 (R) such that kw − fn k∞ →
0. For example, we just need to “smooth out” the corners of the graph of w
to find fn . Therefore {fn } is a Cauchy sequence in the uniform norm, but it
does not converge within Cb1 (R) because w ∈ / Cb1 (R).
1.23 (a) If f is Hölder continuous with α > 0 then

f (x) − f (y) α

lim ≤ lim C |x − y| = lim C |x − y|1−α = 0.
y→x x−y y→x |x − y| y→x
Therefore f is differentiable and f ′ (x) = 0 for every x, so f is constant.

(b) By the Mean-Value Theorem, given x and y there exists some c be-
tween x and y such that f (x) − f (y) = f ′ (c) (x − y), so
|f (x) − f (y)| = |f ′ (c)| |x − y| ≤ kf ′ k∞ |x − y|.
If f ′ is bounded, then it follows that f is Lipschitz.

The function f (x) = |x| is Lipschitz, but is not differentiable at every
point.
(c) By definition, 0 ≤ kf kC α < ∞ for each f ∈ C α (R).
Suppose that kf kC α = 0. Then f (0) = 0 and
|f (x) − f (y)
= 0, all x 6= y.
|x − y|α
Consequently, f (x) = f (y) for all x 6= y. Hence f (x) = 0 for every x, i.e.,
f = 0.
Homogeneity and the Triangle Inequality follow directly, so k · kC α is a

norm.
We will use the technique from Exercise 3.3 to show that that C α (R) is
complete, i.e., we will show that every absolutely convergent series in C α (R)
is convergent.
Suppose that {fn }n∈N is absolutely convergent in C α (R). For each n let
|fn (x) − fn (y)|

Cn = sup .
x6=y |x − y|α
Then we have both

X X
S0 = |fn (0)| < ∞ and Cn < ∞.
Define ∞ ∞
X X
f (0) = fn (0) and C = Cn .
n=1 n=1
Given x 6= 0, we have
∞
X ∞
X ∞
X
|fn (x)| ≤ |fn (x) − fn (0)| + |fn (0)|
n=1 n=1 n=1
∞
X
≤ Cn |x − 0|α + S0 < ∞.
n=1
P
Therefore, the series f (x) = |fn (x)| converges pointwise absolutely. Then
∞
X ∞
X
|f (x) − f (y)| ≤ |fn (x) − fn (y)| ≤ Cn |x − y|α = C |x − y|α ,
n=1 n=1
so f ∈ Cα (R). Also,
N N
X X
f (x) − fn (x) − f (y) − fn (y)

n=1 n=1
X
∞ ∞
X

= fn (x) − fn (y)
n=N +1 n=N +1
∞
X
≤ |fn (x) − fn (y)|
n=N +1
∞
X
≤ Cn |x − y|α
n=N +1
Therefore,
N
X

f − fn

n=1 Cα
N
X N
X

f (x) − fn (x) − f (y) − fn (y)
N
X

= f (0) − fn (0) + sup n=1 n=1
n=1 x6=y |x − y|α

∞
X ∞
X
≤ |fn (0)| + Cn
n=N +1 n=N +1
→ 0 as N → ∞.
P
Therefore the series f = fn converges in C α -norm, so we conclude that
C α (R) is complete.
1.24 (b) Since kϕk+1 − ϕk k∞ < 2−k , we have that ϕk is Cauchy in C[0, 1] by
Exercise 1.5. Consequently ϕk converges uniformly on [0, 1], and since each
ϕk is continuous it follows that the limit ϕ is continuous.
Since
f (3−k ) − f (0) = 2−k − 0 = 2−k = (3−k )log3 2 ,
we have
f (3−k ) − f (0)
= (3−k )−1+log3 2 = 3 k(1−log3 2) .
3−k
However, 1 − log3 2 > 0, so this is not bounded independently of k. Therefore
f is not Lipschitz.
(c) Let ϕ be the Cantor–Lebesgue function. If |x − y| ≤ 3−k , then we have
|ϕ(x) − ϕ(y)| ≤ 2−k . Let k ≥ 0 be the integer such that
1 1
≤ |x − y| ≤ .
3k+1 3k
Then we have
2 1 log3 2
|ϕ(x) − ϕ(y)| ≤ = 2 ≤ 2 |x − y|log3 2 .
2k+1 3k+1
Hence ϕ is Hölder continuous for α = log3 2.
On the other hand, we have
|ϕ(3−k ) − ϕ(0)| = |2−k − 0| = 2−k = (3−k )log3 2 .
Therefore ϕ is not Hölder continuous for any exponent α > − log3 2.
Since the sum of the lengths of the intervals on which ϕ is constant is 1,
we have ϕ′ = 0 a.e.
1.25 We are given a normed linear space X that is not complete, and a a
relation ∼ on the set C of all Cauchy sequences, defined by {xn } ∼ {yn } if
limn→∞ kxn − yn k = 0.
(a) Given {xn } ∈ C, we have lim kxn − xn k = 0, so {xn } ∼ {xn }.

If {xn } ∼ {yn } then {yn } ∼ {xn } since kxn − yn k = kyn − xn k.
Suppose {xn } ∼ {yn } and {yn } ∼ {zn }. Then since
kxn − zn k ≤ kxn − yn k + kyn − zn k,
we have {xn } ∼ {zn }.

Therefore ∼ is an equivalence relation on C.

(b) We let [xn ] = {yn } : {yn } ∼ {xn } denote the equivalence class of
e to be the set of all equivalence classes under ∼ .
{xn }, and define X
Given {xn } ∈ C, we have by the Reverse Triangle Inequality that

kxn k − kxm k ≤ kxm − xn k.

Since {xn } is Cauchy, this implies that kxn k is a Cauchy sequence of real
numbers, and therefore lim kxn k exists. If {yn } ∼ {xn } then the Reverse
Triangle Inequality again implies that

kxn k − kyn k ≤ kxn − yn k → 0,

so lim kxn k = lim kyn k. Hence [xn ] Xe = lim kxn k is a well-defined nonnega-
tive real
number.

If [xn ] Xe = lim kxn k = 0 then {xn } ∼ {0, 0, 0, . . . }, and therefore [xn ] =
[0], the zero element of X.e

We clearly have c[xn ] Xe = lim kcxn k = |c| [xn ] Xe .

The Triangle Inequality follows from

[xn ] + [yn ] e = lim kxn + yn k
X n→∞

≤ lim kxn k + kyn k = [xn ] Xe + [yn ] Xe .
n→∞
e is a normed space.
Therefore X
(c) Given x ∈ X, let [x] denote the equivalence class of the Cauchy se-
quence {x, x, x, . . . }. Then the map T : x → [x] is clearly linear, and

T (x) e = lim kxk = kxk,
X n→∞
so T is an isometry.
We must show that the range of T is a dense subspace of X. e Given A =
e
[xn ] ∈ X, for each m ∈ N let Ym = T (xm ) = [(xm , xm , xm , . . . )]. Then since
{xn } is Cauchy in X, we have
lim sup kA − Ym kXe = lim sup lim kxn − xm k = 0.

m→∞ m→∞ n→∞
e so T (X) is dense in X.
Hence Ym → A in X, e
(d) For this part, it will be important to distinguish between the equiv-
alence class of a sequence {xn }n∈N = {x1 , x2 , . . . } and the equivalence class
determined by a constant sequence {xn , xn , . . . }. Therefore, for this part we
will write [{xn }n∈N ] instead of [xn ].
Suppose that {XM }M∈N is a Cauchy sequence in X. e Each XM is an
equivalence class of a sequence, say
XM = [{xM (n)}n∈N ].
For each M, by part (c) there exists some yM ∈ X such that the constant
sequence determined by yM is close to XM :
1
kXM − T (yM )kXe < .
M
Call this constant sequence
YM = T (yM ) = [{yM , yM , . . . }].
Then, rewriting a line from above, we have

1
kXM − YM kXe < .
M
Since T is an isometry we have kyM k = kT (yM )kXe = kYM kXe . Therefore

kyM − yN k = YM − YN Xe

≤ YM − XM Xe + kXM − XN kXe + XN − YN Xe
1 1
≤ kxM − xN k + + ,
M N
Hence {yM }M∈N is a Cauchy sequence in X. Its equivalence class therefore
e
belongs to X:
e
Y = [{yM }M∈N ] = [{y1 , y2 , . . . }] ∈ X.
Fix ε > 0. Since {yM }M∈N is Cauchy, there exists an N0 > 0 such that
M, N ≥ N0 =⇒ kyM − yN k < ε.
Hence for N ≥ N0 we have

kY − YN kXe = {y1 − yN , y2 − yN , . . . } Xe = lim kyM − yN k ≤ ε.
M→∞
e Also, if N ≥ N0 and N ≥ 1/ε, then

This says that YN → Y in X.
1
kY − XN kXe ≤ kY − YN kXe + kYN − XN kXe ≤ ε + ≤ 2ε.
N
e so X
Therefore XN → Y in X, e is complete.
(e) Suppose that Y is a Banach space and U : X → Y is a linear isometry

such that U (X) is dense in Y. Choose any f ∈ X. e Since range(T ) is dense
e there exist vectors xn ∈ X such that T (xn ) → f. Hence {T (xn )}
in X,
is Cauchy in range(T ). Since isometries map Cauchy sequences to Cauchy
sequences, by applying U T −1 to this sequence we see that {U (xn )} is Cauchy
in range(U ) ⊆ Y. Since Y is complete, there exists g ∈ Y such that U (xn ) → g.
Define V (f ) = g. We will show that V is a isometric map of X e onto Y.
To see that V is well defined, suppose that we had both T (xn ) → f and
T (x′n ) → f. Then T (xn − x′n ) → f − f = 0, and since U T −1 is an isometry,
this implies U (xn − x′n ) → 0. Since U (xn ) → g, we conclude that U (x′n ) → g
as well. Therefore V is well defined.
Suppose that f, h ∈ X are given. Let xn , hn ∈ X be such that T (xn ) → f
and T (hn ) → h. Then T (xn + hn ) → f + h, so
V (f + h) = lim U (xn + hn ) = lim U (xn ) + U (hn ) = V (f ) + V (h).

n→∞ n→∞
Similarly, V (cf ) = cV (f ), so V is linear.

Also, if f ∈ X and T (xn ) → f, then U (xn ) → V (f ). Since U T −1 is an
isometry,
kf k = lim kT (xn )k = lim kU (xn )k = kV (f )k.

n→∞ n→∞
Thus V is an isometry.
Finally, to show that V is onto, suppose that g is any element of Y. Then
since range(U ) is dense, there exist vectors xn ∈ X such that U (xn ) → g.
e and
Since T U −1 is an isometry, it follows that {T (xn )}n∈N is Cauchy in X,
e
hence T (xn ) → f for some f ∈ X. But then, by definition, V (f ) = g. Hence
V is surjective.
Thus V : X e → Y is an isometric isomorphism.
P P P
1.26 Assume that s = xn and t = yn converge. Let sN = N n=1 xn and
PN
tN = n=1 yn . Then

(s + t) − (sN + tN ) ≤ ks − sN k + kt − tN k → 0 as N → ∞.
P
Hence (xn + yn ) exists and equals s + t.
1.27 Let ek (x) = xk . Let θM be 1 on [−M, M ], zero outside [−M − 1, M + 1],
and linear on [−M − 1, −M ] and [M, M + 1]. The set
X
N
S = rk ek θM : M, N ∈ N, rk rational
k=0
is countable. Suppose f ∈ C0 (R) and ε > 0 is fixed. Then there exists an M

PN
such that |f (x)| < ε for all |x| ≥ M. Let p = k=0 ck ek be a polynomial such
that k(f − p) χ[−M−1,M+1] k∞ < ε. Then kf − pθM k∞ < 2ε. Choose rational
rk with
ε
|ck − rk | < , k = 0, . . . , N.
N (M + 1)k
PN PN
Set q = k=0 rk ek . Then kqθM − pθM k∞ ≤ k=0 |ck − rk | (M + 1)k < ε.
Hence kf − qθM k∞ < 3ε, and qθM ∈ S. Therefore S is a countable dense
subset of C0 (R), so this space is separable.
1.28 (a) Suppose that X is separable, and let {xn } be a countable dense
subset. For each y ∈ S, there exists some ny such that ky − xny k < 1/2. By
the Triangle Inequality, ny is uniquely determined by y. But then y 7→ ny is
an injective mapping of the uncountable set S into the countable set N, which
is a contradiction. Hence no such set S can exist.
(b) There are uncountably many sequences in ℓ∞ that contain only 0
and 1 components. Any two distinct such vectors are unit distance apart in
ℓ∞ -norm. Hence ℓ∞ (I) cannot be separable by part (c).

(c) S = χ[t,t+1] t∈R is an uncountable subset of L∞ (R), and if s 6= t
then
kχ[s,s+1] − χ[t,t+1] kL∞ = 1.
Hence part (b) implies that L∞ (R) is not separable.
P∞ There
1.29
k
is nothing to prove if y = 0, so we may assume y 6= 0. Since
k=0 kc y converges, we must have ck y k → 0 as k → ∞. Therefore M =
k
sup |cy y | < ∞. If |x| < |y| then |x/y| < 1, so
∞ ∞ k ∞ k
X X x X x
k
|ck x | = k
|ck y | ≤ M < ∞.
y y
k=0 k=0 k=0
P∞
Thus the series f (x) = k=0 ck xk converges absolutely for all |x| <
P|y|. k
Now suppose that |x| < |y|, and fix r with |x| < r < |y|. Then |ck |r <
∞ by our previous work. Let
X k
x
M = k < ∞.
r
k∈N
Then
∞
X ∞
X k ∞
|ck | rk x M X
k |ck xk−1
| ≤ k ≤ |ck | rk < ∞.
r r r
k=1 k=1 k=1
Now, applying the Mean Value Theorem to the function xk , given h ∈ R

and k ∈ N, there exists ξh,k between x and x + h such that
(x + h)k − xk k−1
= k ξh,k .
h
Assuming that the discrete Fubini’s theorem is justified (this is demonstrated
below), we have for those h with |x + h| ≤ r < |y| that
∞
f (x + h) − f (x) X
lim − k ck xk−1
h→0 h
k=1
X
∞ (x + h)k − xk
∞
X

= lim ck − k ck xk−1
h→0 h
k=1 k=1
∞
X
≤ lim kck (ξ k−1 − xk−1 )
h→0 h,k
k=1
∞
X k−1
= lim k |ck | ξh,k − xk−1
h→0
k=1
∞
X
= 0 = 0.
k=1
Therefore f is differentiable at x, and we have

∞
X
f ′ (x) = k ck xk−1 .
k=1
To justify the interchange of limit and series by Fubini’s theorem, note

that ξh,k lies between x and x + h, and |x|, |x + h| ≤ r < |y|, so we have
∞
X ∞
X
k−1
k |ck | ξh,k − xk−1 ≤ k |ck | rk−1 + |x|k−1 < ∞,
k=1 k=1
where the convergence follows since r, |x| < |y|. Since the summands in the
final series are independent of h, Fubini is justified.
1.30 Suppose that F = C (the same proofs work for F = R).
(b) Let A = [hei , ej i]dj,i=1 . Then for any x, y ∈ Cd we have
     
he1 , e1 i · · · hed , e1 i x1 y1
 .. .. ..   ..   .. 
Ax · y =  . . .   .  ·  . 
he1 , ed i · · · hed , ed i xd yd
   
x1 he1 , e1 i + · · · + xd hed , e1 i y1
 .
.   .. 
=  .  · . 
x1 he1 , ed i + · · · + xd hed , ed i yd
   
hx, e1 i y1
   
=  ...  ·  ... 
hx, ed i yd
= hx, e1 i ȳ1 + · · · + hx, ed i ȳd = hx, yi.
Finally, A is positive definite since Ax · x = hx, xi > 0 for all x 6= 0.

(c) Let A be a positive definite matrix, and let |||x||| = hx, xi1/2 be the
corresponding induced norm.
Since A is diagonalizable, there exists an orthonormal basis {v1 , . . . , vn }
for Cn with corresponding positive eigenvalues λ1 ≤ · · · ≤ λn . Given x =
c1 v1 + · · · + cn vn , we therefore have
X
n n
X
2
|||x||| = Ax · x = cj λj vj , ck vk
j=1 k=1
n
X n
X
= |ck |2 λk kvk k2 ≤ λn |ck |2 = λn |x|2 ,
k=1 k=1
2
Similarly |||x||| ≥ λ1 |x|2 , so ||| · ||| is equivalent to the Euclidean norm on Cn .
If F = R then we regard A as a complex matrix and follow the same proof.
1.31 (a) Given f, g ∈ H, we have
kf + gk2 = hf + g, f + gi
= hf, f i + hf, gi + hg, f i + hg, gi
= kf k2 + hf, gi + hf, gi + kgk2

= kf k2 + 2 Rehf, gi + kgk2 .
(b) If f ⊥ g, then the Polar Identity implies that kf + gk2 = kf k2 + kgk2 .

Since f is also orthogonal to −g, we have kf − gk2 = kf k2 + kgk2 as well.
(c) Given f, g ∈ H, we have
kf + gk2 + kf − gk2
= kf k2 + hf, gi + hg, f i + kgk2 + kf k2 − hf, gi − hg, f i + kgk2
= 2kf k2 + 2kgk2 .
1.32 If p 6= 2 and δ1 , δ2 are the first two standard basis vectors in ℓp , then
kδ1 + δ2 k2p + kδ1 − δ2 k2p = (1 + 1)2/p + (1 + 1)2/p = 2 · 22/p = 21+(2/p) ,
while

2 kδ1 k2p + kδ2 kp
2
= 2 (12 + 12 ) = 4.
These are not equal when p 6= 2, so the Parallelogram Law does not hold in
ℓp .
1.33 ⇒. The proof of the Cauchy–Bunyakovski–Schwarz Inequality shows
that for every real t we have
0 ≤ kf − αtgk2 = kf k2 − 2t |hf, gi| + t2 kgk2 ,
where |α| = 1 satisfies hf, gi = α |hf, gi|. This is a quadratic polynomial in

the real variable t, and since it is everywhere nonnegative, it can have at most
one real root. This requires that the discriminant be at most zero, so
(−2 |hf, gi|)2 − 4 kf k2 kgk2 ≤ 0,
which yields the CBS inequality. Suppose now that equality holds in CBS,
i.e.,
|hf, gi| = kf k kgk
for some f and g. Then we have
(−2 |hf, gi|)2 − 4 kf k2 kgk2 = 0,
which means that the discriminant above is zero. This implies that the poly-
nomial above does have a real root, i.e., there is be a t such that
kf − αtgk2 = kf k2 − 2t |hf, gi| + t2 kgk2 = 0.
⇐. Suppose that kaf + bgk = 0 for some scalars a, b ∈ F, not both zero.
Suppose that a = 0. Then b 6= 0 but |b| kgk = kaf + bgk = 0, so kgk = 0.
Hence 0 ≤ |hf, gi| ≤ kf k kgk = 0, so equality holds in CBS. Similarly, equality
holds if b = 0, so we can assume a, b 6= 0.
By the Polar Identity,
0 ≤ kaf + tbgk2 = kaf k2 + 2t Rehaf, bgi + t2 kbgk2.
If |b| kgk = kbgk = 0, then we have for t = 1 that
0 = kaf + bgk2 = kaf k2 + 2 Rehaf, bgi.
Hence
0 ≤ kaf k2 = −2 Rehaf, bgi ≤ 2 |haf, bgi| ≤ 2 kaf k kbgk = 0.
Thus in this case we have that
|ab| |hf, gi| = 0 = |ab| kf k kgk.
Since ab 6= 0, it follows that equality holds in CBS.

On the other hand, if kbgk 6= 0, then we have a true quadratic polynomial

with at most one real root. Further, it does in fact have a root, when t = 1.
Therefore the discriminant is zero,
(2 Rehaf, bgi)2 − 4 kaf k2 kbgk2 = 0,
or
kaf k kbgk = |Rehaf, bgi|.
|ab| kf k kgk = kaf k kbgk ≤ |haf, bgi| = |ab| |hf, gi| ≤ |ab| kf k kgk.
Since ab 6= 0, equality in CBS holds.
1.34 Let h·, ·i be a semi-inner product on H. Then we proceed just as in
the proof of Cauchy–Bunyakovski–Schwarz. If x = 0 or y = 0 then there is
nothing to prove, so suppose that both are nonzero. Write hx, yi = α |hx, yi|
where α ∈ F and |α| = 1. The Polar Identity holds for semi-inner products,
so for t ∈ R we have
0 ≤ kx − αtyk2 = kxk2 − 2 Re(ᾱt hx, yi) + t2 kyk2

= kxk2 − 2t |hx, yi| + t2 kyk2 .
This is a real-valued quadratic polynomial in the variable t. In order for it

to be nonnegative, it can have at most one real root. This requires that the
2
discriminant be at most zero, so −2 |hx, yi| − 4 kxk2 kyk2 ≤ 0. The desired
inequality then follows upon rearranging.
1.35 Suppose that p and q are elements of M that are both closest to x, say
with d = kx − pk = kx − qk. The point m = (p + q)/2 belongs to M so we
have kx − mk ≥ d. Therefore, by the Parallelogram Law,
4d2 = 2 kx − pk2 + 2 kx − qk2

= k(x − p) + (x − q)k2 + k(x − p) − (x − q)k2
= k2x − (p + q)k2 + kp − qk2
= 4 kx − mk2 + kp − qk2
≥ 4 d2 + kp − qk2 .
But then kp − qk = 0, so p = q.
1.36 If y, z ∈ A⊥ then for any x ∈ A we have
hay + bz, xi = ahy, xi + bhz, xi = 0,
so ay + bz ∈ A⊥ . Therefore A⊥ is a subspace of H.
Suppose that yn ∈ A⊥ and yn → y in H. Then for every x ∈ A we have
by the continuity of the inner product that
hx, yi = lim hx, yn i = 0.

n→∞
Therefore y ∈ A⊥ , so A⊥ is closed.
1.37 (a) ⇒ (b). Fix F = C in this part (the proof for F = R is similar).
Let p be the (unique) point in M closest to h, and let e = p − h. Choose any
f ∈ M. We must show that hf, ei = 0. Since M is a subspace, p + λf ∈ M for
any scalar λ ∈ C. Hence,
kh − pk2 ≤ kh − (p + λf )k2 = k(h − p) − λf k2

= kh − pk2 − 2 Re(hλf, h − pi) + |λ|2 kf k2
= kh − pk2 − 2 Re(λhf, ei) + |λ|2 kf k2 .
Therefore,
∀ λ ∈ C, 2 Re(λhf, ei) ≤ |λ|2 kf k2 .
If we consider λ = t > 0, then we can divide through by t to get
∀ t > 0, 2 Re(hf, ei) ≤ t kf k2 .
Letting t → 0+ , we conclude that Re(hf, ei) ≤ 0. Similarly, taking λ = t < 0

and letting t → 0− , we obtain Re(hf, ei) ≥ 0, so Re(hf, ei) = 0.
Finally, by taking λ = it with t > 0 and then λ = it with t < 0, it follows
similarly that Im(hf, ei) = 0.
(b) ⇒ (a). Suppose that h = p + e where p ∈ M and e ⊥ M. Choose any
f ∈ M. Then p − f ∈ M, so h − p = e ⊥ p − f. Therefore, by the Pythagorean
Theorem,
kh − f k2 = k(h − p) + (p − f )k2 = kh − pk2 + kp − f k2 ≥ kh − pk2 .
Hence p is the point in M that is closest to h.

(b) ⇒ (c). Suppose that x = p + e where p ∈ M and e ∈ M ⊥ . If y ∈ (M ⊥ )
then hp, yi = 0, so p ∈ (M ⊥ )⊥ . Therefore x = e + p where e ∈ M ⊥ and
p ∈ (M ⊥ )⊥ . The equivalence of statements (a) and (b) therefore implies that e
is the orthogonal projection of x onto M ⊥ .
(c) ⇒ (b). Suppose that x = p + e where e is the orthogonal projection
of x onto M ⊥ . Then by the equivalence of statements (a) and (c) we have
that p ∈ (M ⊥ )⊥ . We must show that p ∈ M.
Write p = q + f where q ∈ M and f ∈ M ⊥ . Since p ∈ (M ⊥ )⊥ we have
hp, f i = 0. Since we also have hq, f i = 0, it follows that
0 = hp, f i = hq, f i + hf, f i = 0 + kf k2 .
Hence f = 0 for p = q ∈ M.
1.38 (a) If f ∈ M then hf, yi = 0 for every y ∈ M ⊥ , so f ⊥ M ⊥ and therefore
f ∈ (M ⊥ )⊥ .
Conversely, suppose that f ∈ (M ⊥ )⊥ . Since M is a closed subspace, we

can write f = p + e with p ∈ M and e ∈ M ⊥ . Then e = f − p ∈ (M ⊥ )⊥ and
e ∈ M ⊥ , so e = 0. Hence f = p ∈ M.
(b) Let M = span(A). We must show that A⊥ = M ⊥ . Since A ⊆ M, we
have M ⊥ ⊆ A⊥ .
Suppose that f ∈ A⊥ . Then f ⊥ A. By forming linear combinations, it
follows that f ⊥ span(A). Then by taking limits, it follows that f ⊥ span(A) =
M. Hence f ∈ M ⊥ and therefore A⊥ ⊆ M ⊥ .
(c) {xn } is complete if and only if {xn }⊥ = span{xn } = H.
1.39 (a) Suppose that zn ∈ M ⊕ N and zn → z in H. Then we can write
zn = xn + yn where xn ∈ M and yn ∈ N. Then xm − xn ∈ M is orthogonal
to ym − yn ∈ N so, by the Pythagorean Theorem,
kzm − zn k2 = k(xm − xn ) + (ym − yn )k2 = kxm − xn k2 + kym − yn k2 .
Hence {xn }n∈N and {yn }n∈N are each Cauchy, so must converge in H. There-
fore xn → x and yn → y for some x, y ∈ H, and since M and N are closed we
have x ∈ M and y ∈ N. Thus, we have both xn + yn → x+ y and xn + yn → h,
so z = x + y ∈ M ⊕ N. Therefore M ⊕ N is closed.
(b) This follows from Theorem 1.43.

1.40 Let H, K be Hilbert spaces. It is clear that (h1 , k1 ), (h2 , k2 ) =
hh1 , h2 iH + hk1 , k2 iK is an inner product on H × K. The induced norm is

(h, k) 2 = khk2H + kkk2K ,
and it follows from Exercise 1.16 that H × K is a Banach space with respect
to this norm. Therefore H × K is a Hilbert space.
1.41 Let H be an inner product space. Then by Exercise 1.25, its completion
e is a Banach space.
H
e there exist xn , yn ∈ H such that xn → x and yn → y. Let
Given x, y ∈ H,
C = max{kxn k, kyn k}n∈N . Given ε > 0, there exists some N > 0 such that
m, n > N =⇒ kxm − xn k, ky − ym k < ε.
Therefore, for m, n > N we have
|hxm , ym i − hxn , yn i|
≤ |hxm , ym i − hxm , yi| + |hxm , yi − hxn , yi| + |hxn , yi − hxn , yn i|
≤ kxm k kym − yk + kxm − xn k kyk + kxn k ky − yn k
≤ C kym − yk + C kyk + C ky − yn k ≤ 3Cε.

Hence hxn , yn i is a Cauchy sequence in F, so we can define
hx, yi = lim hxn , yn i.

n→∞
If we also have x′n → x and yn′ → y, then
|hxn , yn i − hx′n , yn′ i| ≤ |hxn , yn − yn′ i| + |hxn − x′n , yn′ i|

≤ kxn k kyn − yn′ k + kxn − x′n k kyn′ k
→ 0,
e Considering constant sequences, we see that h·, ·i

so h·, ·i is well defined on H.
extends the inner product on H.
Suppose that hx, xi = 0 for some x ∈ H.e Let xn ∈ H be such that xn → x
2
in H. Then kxn k = hxn , xn i → hx, xi = 0, so xn → 0 in H.e Therefore x = 0.
e is a Hilbert space.
The other properties of an inner product are clear, so H
1.42 First we prove the remaining parts of Theorem 1.49.
P
(b) If x = cn xn converges, then for each fixed m we have by continuity
of the inner product that
X X
hx, xm i = cn hxn , xm i = cn δmn = cm .
n n
P∞
(c) ⇐. Suppose that n=1 |cn |2 < ∞. Set
N
X N
X
sN = cn xn , tN = |cn |2 .
n=1 n=1
We know that {tN }N ∈N is a convergent (hence Cauchy) sequence of scalars,

and we must show that {sN }N ∈N is a convergent sequence of vectors. We
have for N > M that
X
N 2

ksN − sM k2 = cn xn
n=M+1
N
X
= kcn xn k2
n=M+1
N
X
= |cn |2 = |tN − tM |.
n=M+1
Since {tN }N ∈N is Cauchy, we conclude that {sN }N ∈N is Cauchy and hence

converges.
P
⇒.P If cn xn converges, then we have by part (b) that cn = hx, xn i, and
hence |cn | < ∞ by Bessel’s Inequality.
P∞(e) By Bessel’s Inequality and part (c), we know that the series p =
n=1 hf, xn i xn converges, so we just
have to show that it is the orthogonal
projection of f onto span {xn }n∈N . Given any k we have by the linearity
and continuity of the inner product that
∞
X
hf − p, xk i = hf, xk i − hf, xn i hxn , xk i = hf, xk i − hf, xk i = 0.
n=1

By linearity of the inner product, this implies that f −p ⊥ span {xn }n∈N . By
continuity of the inner product, this extends to f − p ⊥ span {xn } . Hence p
is indeed the orthogonal projection of f onto span {xn }n∈N .
P
(e) ⇐. Suppose x = hx, xn i xn . Then, by part (e), x = p, the orthogonal
projection of x onto span{xn }, so x ∈ span{xn }.
Now we prove the remaining implications in Theorem 1.50.
(a) ⇒ (b). If {xn } is complete,
P∞ then its closed span is all of H, so by
Exercise 1.42(e) we have f = n=1 hf, xn i xn for every f ∈ H.
(b) ⇒ (c). If statement (b) holds, then we must have cn = hx, xn i by
Theorem 1.49.
(c) ⇒ (e). Suppose that (c) holds, and choose any f, g ∈ H. Then
DX
∞ E ∞
X
∞
X
hf, gi = hf, xn i xn , g = hf, xn i xn , g = hf, xn i hxn , gi,
n=1 n=1 n=1
where we have the continuity of the inner product.

(e) ⇒ (d). This follows by taking x = y.
(d) ⇒ (a). Suppose that Plancherel’s
P Equality holds, and that hx, xn i = 0
for every n ∈ N. Then kxk = |hx, xn i|2 = 0, so x = 0. Hence {xn } is
complete.
1.43 (a) Let M be a proper, closed subspace of a Hilbert space H. Given
x ∈ H\M, let p be the orthogonal projection of x onto M. Since x 6= p we can
define y = (x − p)/kx − pk, and we have that kyk = 1. Suppose that m ∈ M.
By Theorem 1.43 we have e = x − p ∈ M ⊥ , so since y is a scalar multiple of e
we have y ∈ M ⊥ as well. Since 0 ∈ M, we have dist(y, M ) ≤ ky − 0k = 1. On
the other hand, if m ∈ M, then y ⊥ m, so
ky − mk2 = kyk2 + kmk2 ≥ kyk2 = 1.
Therefore dist(y, M ) = 1.
(b) Let H be an infinite-dimensional Hilbert space. Choose any nonzero
vector in H, and normalize it to obtain a vector e1 with norm 1. Since H is
infinite dimensional, H 6= span{e1 }. Therefore by part (a) we can find a vector
e2 ⊥ e1 with ke2 k = 1. Then H 6= span{e1 , e2 }, so we apply part (a) again.

Continuing in this way we obtain an infinite orthonormal sequence {en }.
1.44 (a) Choose any u ∈ X \ M. Since M is closed, we have
a = dist(u, M ) = inf ku − f k > 0.

f ∈M
a
Fix δ > 0 small enough that a+δ > 1 − ε. By the definition of infimum, there
exists some v ∈ M such that a ≤ ku − vk < a + δ. Set
u−v
g = ,
ku − vk
and note that kgk = 1. Given f ∈ M we have h = v + ku − vk f ∈ M, so

u − v − ku − vk f
kg − f k = = ku − hk > a
> 1 − ε.
ku − vk ku − vk a+δ
(b) Suppose that X is infinite dimensional. Choose any nonzero vector

e1 ∈ X with ke1 k ≤ 1. Applying part (a) to M = span{e1 }, we can find a
vector e2 ∈ X \span{e1 } with ke2 k ≤ 1 such that ke2 −e1 k > 12 . Then we apply
part (a) to M = span{e1 , e2 } to find e3 ∈ X \ span{e1 , e2 } with ke3 k ≤ 1 such
that ke3 − e1 k > 21 and ke3 − e2 k > 12 . Continuing in this way, we construct
vectors en such that {en }n∈N contains no convergent subsequences. Hence the
closed unit ball B is not compact.
1.46
P (a) Suppose that {xn }n∈NPis a basis and there exist scalars cn such that
cn xn = 0. Since we also have 0 · xn = 0, we conclude by uniqueness that
cn = 0 for every n. Hence {xn }n∈N is ω-independent.
(b) Let {δk }∞ 2
k=1 be the standard basis for ℓ . We are given that x0 = δ1
and xk = αδk + βδk+1 for k > 0.
To show that {xk }k≥0 is complete we just have to show that z = 0 is the
only vector orthogonal to every xk . So, suppose that z ⊥ xk for all k ≥ 0.
Then z1 = z · x0 = 0. Also for k > 0 we have
0 = z · xk = α(z · xk ) + β(z · xk+1 ) = αzk + βzk+1 .
Hence zk+1 = − α β zk . Since z1 = 0 and α, β 6= 0, it follows by induction that

zk = 0 for all k. Hence {xk }k≥0 is complete.
PN
To show finite linear independence, suppose that k=0 ck xk = 0 for
some N and scalars ck . Then
N
X
0 = ck xk
k=0
= (c0 + c1 α, c1 β + c2 α, c2 β + c3 α, . . . , cN −1 β + cN α, cN β, 0, 0, . . . ),
where the cN β term occurs in the (N + 1)st component. Since β 6= 0, we

conclude that cN = 0. Then we have cN −1 β = 0 and hence cN −1 = 0. By
induction we obtain cN = cN −1 = · · · = c1 = 0, and finally this implies c0 = 0
as well.
Next, to show that {xk }k≥0 is not ω-independent, let c0 = −α and ck =
β k−1
(− α ) for k > 0. Then c0 + c1 α = 0 and ck β + ck+1 α = 0 for k > 0. Set
N
X
sN = ck xk
k=0
= (c0 + c1 α, c1 β + c2 α, c2 β + c3 α, . . . , cN −1 β + cN α, cN β, 0, 0, . . . ),
= (0, 0, . . . , 0, cN β, 0, 0, . . . ).
β
Since | α | < 1,
2N −2
β
ksN k2 = |cN β|2 = β 2 → 0 as N → ∞.
α
P∞
Thus sN → 0, which says that k=0 ck xk = 0.
Since {xk }k≥0 is not ω-independent, part (a) therefore implies that it
cannot be a Schauder basis.
1.47 (a) ⇒ (b). For each m, let Em be the closed subspace Em =
span{xn }n6=m . Then we can write xm uniquely as xm = pm + qm where
⊥
pm ∈ Em and qm ∈ Em . In particular, we have qm ⊥ xn for every n 6= m.
If we had hxm , qm i = 0, then we would have
0 = hxm , qm i = hpm , qm i + hqm , qm i = 0 + kqm k.
But this implies xm = pm ∈ Em , which is a contradiction. Hence hxm , qm i =

6 0.
Define
qm
ym = .
hqm , xm i
Since ym is a multiple of qm we have ym ⊥ xn for every n 6= m. And since
hxm , ym i = 1, the sequence {ym }m∈N has the required properties.
(b) ⇒ (a). Suppose that a biorthogonal sequence {yn }n∈N exists. With
Em as above, we have by linearity and continuity of the inner product that
ym ⊥ Em . Since hxm , ym i = 1 6= 0, the vector xm cannot belong to Em .
(a) + completeness ⇒ (b) + uniquness. Suppose that (a) holds and that
{xn }n∈N is complete. Suppose that {yn }n∈N and {zn }n∈N are both biorthog-
onal to {xn }n∈N . Fix any n. Then for any m we have
hxm , yn − zn i = hxm , yn i − hxm , zn i = δmn − δmn = 0.

Since {xm }m∈N is complete, this implies that yn − zn = 0. Hence we have

uniqueness.
(b) + uniquness ⇒ (a) + completeness. Suppose that a biorthogonal se-
quence exists and is unique. Suppose that z ∈ H is orthogonal to every xn .
Then
hxn , ym + zi = hxn , ym i + hxn , zi = δmn + 0 = δmn .
Thus {ym +z}m∈N is also birthogonal to {xn }n∈N . By uniqueness, we therefore
have z = 0. Hence {xn }n∈N is complete.
1.48 Let f = χ[0,1/2) − χ[1/2,1) . For n = 0 we have
Z 1 Z 1
fb(0) = f (x) e −2πi0x
dx = f (x) dx = 0.
0 0
For n 6= 0 we have
Z 1 Z 1/2 Z 1
fb(n) = f (x) e −2πinx
dx = e −2πinx
dx − e−2πinx dx
0 0 1/2
1/2 1
e−2πinx e−2πinx
= −
−2πin 0 −2πin 1/2
e−πin − 1 1 − e−πin
= −
−2πin −2πin
(
2(−1)n − 1 0, n even,
= = 2i
−2πin − πn , n odd.
Therefore,
2
X X 2i X 4
1 = kf k22 = |fb(n)|2 = = 2
πn .
π 2 n2
n∈Z n∈Z, n odd n∈N, n odd
Rearranging, we see that

∞
X
π2 1
= .
8 (2k − 1)2
k=1
√ √
1.49 Set E = {1} ∪ { 2 sin 2πnt}n∈N ∪ { 2 cos 2πnt}n∈N .
We have eiθ = cos θ + i sin θ. Therefore e−iθ = cos θ − i sin θ. Write
2 cos 2πnt = e2πint + e−2πint = en (t) + e−n (t).
Then for m, n ≥ 0 we have


4 cos 2πmt, cos 2πnt = em + e−m , en + e−n
= hem , en i + hem , e−n i + he−m , en i + he−m , e−n i
= δmn + δm,−n + δ−m,n + δ−m,−n
= 2δmn + 2δm,−n = 2δmn .
A similar argument applies to the other inner products, so we conclude that E

is orthogonal. Since each element is normalized, E is orthonormal.
Since span(E) = span{en }n∈Z , it follows that E is an orthonormal basis
for complex L2 (T). Hence every vector f in real L2 (T) can be represented in
the orthonormal basis E. Since each element of E is real-valued, the scalars in
this representation, which are inner products of f with the elements of E, will
be real. Hence f has a real-valued series representation, and therefore E is an
orthonormal basis for real L2 (T).
1.50 (a) Given any two distinct elements f, g of the Haar system, we either
have f g = 0, f g = f, or f g = g. It follows from this that the Haar system is
orthonormal.
2
(b) Suppose
that f ∈ L [0, 1] and f is orthogonal to each element of
H = χ ∪ ψn,k n≥0, k=0,...,2n −1 . If we extend f by zero to all of R, then

this extended f is orthogonal to every element of ψn,k n≥0, k∈Z , which we
have shown is an orthonormal basis for L2 (R). Hence f = 0, so H is complete
and therefore is an orthonormal basis for L2 [0, 1].
1.51 P(a) We are given an orthonormal basis {xn } and a sequence (yn ) such
that kxn − yn k < 1. Suppose that hx, yn i = 0 for every n. If x 6= 0 then we
have by the Parseval Equality that
X X
kxk2 = |hx, xn i|2 = |hx, xn − yn i + hx, yn i|2
n n
X
= |hx, xn − yn i|2
n
X
≤ kxk2 kxn − yn k2
n
< kxk2 ,
which is a contradiction. Therefore x = 0, so {yn } is complete.

P
(b) If we take y1 = x2 and yn = xn for n > 1, then kxn − yn k2 = 1 but
{yn } is not complete.
1.52 Let {xn } be an orthonormal sequence in a Hilbert space H.
P
⇒. If x ∈ span{xn } then x = hx, xn i xn by Theorem 1.49. Hence kxk2 =
P
hx, xi = |hx, xn i|2 .
P
⇐. Suppose that kxk2 = |hx, xn i|2 . By Theorem 1.49, the orthogonal
P
projection of x onto span{xn } is p = hx, xn i xn . Since x−p ⊥ p, we therefore
have that
X
kxk2 = kx − pk2 + kpk2 = kx − pk2 + |hx, xn i|2 = kx − pk2 + kxk2 .
Hence kx − pk = 0, so x = p ∈ span{xn }.
1.53 Let {en } be any orthonormal basis for H. Let {fn } be an orthonormal
basis for H that contains an orthonormal basis for M, i.e., {fn }n∈I is an
orthonormal basis for M and {fn }n∈J is an orthonormal basis for M ⊥ , where
I ∪ J is a partition of N. Then
X XX
kP en k2 = |hP en , fm i|2
n n m
XX
= |hen , P fm i|2
m n
X
= kP fm k2
m
X
= 12 = |I| = dim(M ).
m∈I
1.54 ⇒. If {fn } is complete, then we have by Plancherel’s Equality that

∞ Z
X x 2 ∞
X

fn (t) dt = |hχ[a,x] , fn i|2

n=1 a n=1
Z b
= kχ[a,x] k22 = |χ[a,x] (t)|2 dt = x − a.
a
⇐. Suppose that
∞ Z
X x 2

fn (t) dt = x − a, x ∈ [a, b].

n=1 a
Then,
∞
X
|hχ[a,x] , fn i|2 = x − a = kχ[a,x] k22 .
n=1
Thus, the Plancherel Equality holds for χ[a,x] , so by Exercise 1.52 we have
that χ[a,x] ∈ span{fn } for every x ∈ [a, b]. Hence χ[x,y] = χ[a,y] − χ[a,x] ∈
span{fn } for every x < y. But span{χ[x,y] : a ≤ x < y ≤ b} is the set of
“really simple functions,” which is dense in L2 [a, b] by Lemma A.28. Hence
span{fn } = L2 [a, b], as desired.
1.55 We apply Vitali’s criterion with [a, b] = [0, 1]. Note that
|e2πinx − 1|2 = (e2πinx − 1)(e−2πinx − 1)

= 1 − e2πinx − e−2πinx + 1
= 2 − 2 cos 2πnx.
Therefore
2 Z 2
X Z x x X e2πinx − 1 2
e 2πint
dt =

1 dt +
2πin
n∈Z 0 0 n6=0
X 2 − 2 cos 2πnx
= x2 +
4π 2 n2
n6=0
X∞
1 − cos 2πnx
= x2 + . (A)
n=1
π 2 n2
Vitali’s criterion tells us that {e2πinx }n∈Z is complete in L2 [0, 1] if and only
if 2
X Z x
e 2πint
dt = x, x ∈ [0, 1]. (B)

n∈Z 0
Substituting equation (A), equation (B) reduces to
X∞
1 − cos 2πnx
x2 + = x.
n=1
π 2 n2
1.56 ⇒. If {fn } is complete, then Exercise 1.54 implies that

∞ Z Z 2 Z
X b x b
(b − a)2
fn (t) dt = (x − a) dx = .
2
n=1 a a a
⇐. Suppose that
∞ Z ∞ Z Z 2
X b X b x (b − a)2
|hχ[a,x] , fn i|2 dx = fn (t) dt = .
2
n=1 a n=1 a a
Since {fn } is orthonormal, Bessel’s Inequality implies that

∞
X
|hχ[a,x] , fn i|2 ≤ kχ[a,x] k2L2 = x − a.
n=1
Therefore
Z b ∞
X
χ 2 χ 2
0 ≤ k [a,x] kL2 − |h [a,x] , fn i| dx
a n=1
Z b Z b ∞
X
= kχ[a,x] k2L2 − |hχ[a,x , fn i|2 dx
a a n=1
Z b
(b − a)2
= (x − a) dx − = 0.
a 2
Consequently,
∞
X
x − a = kχ[a,x] k2L2 = |hχ[a,x] , fn i|2 a.e. x.
n=1
Now, kχ[a,x] k2L2 = x − a is a continuous function of x. By Bessel’s Inequality

and the Triangle Inequality on ℓ2 ,
X
∞ 1/2
2
|||f ||| = |hf, fn i|
n=1
is a seminorm on L2 [a, b] and therefore is continuous with respect to L2 -norm.

If y → x, then χ[a,y] → χ[a,x] in L2 -norm, so
∞
X
F (x) = |hχ[a,x] , fn i|2
n=1
is a continuous function of x. Since we have shown that F (x) = x − a a.e.,

it follows that F (x) = x − a for every x. Exercise 1.54 therefore implies that
{fn } is complete.
1.57 We are given an orthonormal basis {fn }n∈N for L2 [a, b]. The function
f1 is not the zero element of L2 [a, b], so
S
{|f1 | > 1/n} = {|f1 | > 0}
n
has positive measure. Therefore, there must exist some n such that the set
E = {|f1 | > 1/n} has positive measure. Write E = ∪Fk disjointly where each
Fk has positive measure. Define
(
|Fk |1/2
m(t) = |f1 (t)| , t ∈ Fk ,
1, otherwise.
Then for t ∈ Fk we have
|Fk |1/2 n
|m(t)| ≤ ≤ n (b − a)1/2 ,
k
so m ∈ L∞ [a, b]. Note that f1 /m is defined a.e., and

Z b
X Z f1 (t) 2

XZ
f1 (t) 2 1 X
dt ≥ dt = dt = 1 = ∞,
m(t) m(t)
a k∈N Fk Fk |Fk |
k∈N k∈N
so f1 /m ∈/ L2 [a, b].
Since m ∈ L∞ [a, b], we have mfn ∈ L2 [a, b] for every n. Now suppose
that g ∈ L2 [a, b] satisfies hg, mfn i = 0 for every n ≥ 2. Then (since m is
real-valued)
Z b
0 = hg, mfn i = g(t) m(t) fn (t) dt = hgm, fn i, n ≥ 2.
a
Since gm ∈ L [a, b] and {fn }n∈N is an orthonormal basis for L2 [a, b], we
2
conclude that
∞
X
gm = hgm, fn i fn = hgm, f1 i f1 .
n=1
Thus, gm = cf1 where c = hgm, f1 i. Now, if c 6= 0 then since m(t) 6= 0 a.e.

/ L2 [a, b], which is a contradiction. Hence we must have
we have g = cf1 /m ∈
c = 0. But then hgm, fn i = 0 for every n, so gm = 0 a.e. Since m(t) 6= 0 a.e.,
we conclude that g = 0 a.e., and therefore {mfn }n≥2 is complete.
1.58 (a) If {xi }i∈I is an uncountable
√ orthonormal set, then the Pythagorean
Theorem implies that kxi − xj k = 2 for each i 6= j. Hence H is nonseparable
by Exercise 1.28.
1.59 First note that if ξ 6= η then, since |eξ | ≤ 1,
Z T
1
heξ , eη i = lim e2πiξt e−2πiηt dt
T →∞ 2T −T
T
1 e2πi(ξ−η)t
= lim
T →∞ 2T 2πi(ξ − η) −T
e2πi(ξ−η)T − e−2πi(ξ−η)T
= lim = 0,
T →∞ 4T πi(ξ − η)
and Z T
1
heξ , eη i = lim e2πiξt e−2πiξt dt = 1.
T →∞ 2T −T
Hence, once we show that h·, ·i is an inner product, we can conclude that
{eξ }ξ∈R is an orthonormal system.
Suppose that f, g, h ∈ H and the limits defining hf, hi and hg, hi exist.
Then
Z T
1
hf + g, hi = lim (f (t) + g(t)) h(t) dt = hf, hi + hg, hi
T →∞ 2T −T
exists as well, and similarly hcf, hi = chf, hi and hh, f i = hf, hi.
PM
In particular, given f, g ∈ H we can write f = m=1 cm eξm and g =
PN
n=1 dn eξn , so we have that
M X
X N M X
X N
hf, gi = cm dn heξ , eη i = cm dn δξη
m=1 n=1 m=1 n=1
exists. Note we have that hf, f i ≥ 0 for every f ∈ H, as

Z T
1
hf, f i = lim |f (t)|2 dt.
T →∞ 2T −T
Further, if hf, f i = 0 then it follows from the preceding calculation that

f = 0 a.e. on each interval [−T, T ], and hence f = 0 a.e.
Therefore we conclude that h·, ·i is an inner product on H, and {eξ }ξ∈R is
an uncountable orthonormal system in H.
1.60 ⇒. Suppose that f : X → Y is continuous, and let V ⊆ Y be open.
Suppose that f −1 (V ) was not open. Then there exists some x ∈ f −1 (V ),
which means y = f (x) ∈ V. Since V is open, there exists some ε > 0 such
that Bε (y) ⊆ V. Since f −1 (V ) is not open, for each n ∈ N the ball B1/n (x)
is not contained in f −1 (V ). Hence there exists some xn ∈ B1/n (x) such that
xn ∈/ f −1 (V ). Thus we have kx−xn k < 1/n, so xn → x. Since f is continuous,
this implies f (xn ) → f (x). Then there must exist some N > 0 such that
kf (x) − f (xn )k < ε for all n > N. But this implies that f (xn ) ∈ Bε (y) ⊆ V,
so xn ∈ f −1 (V ) for all n > N. This is a contradiction. Hence f −1 (V ) must be
open.
⇐. Suppose that f −1 (V ) is open for each open V ⊆ Y. Suppose that
xn → x in X, and fix ε > 0. Let y = f (x). Then B = Bε (y) is open, so f −1 (B)
is open. Since x ∈ f −1 (B), there exists a δ > 0 such that Bδ (x) ⊆ f −1 (B).
There exists some N > 0 such that kx − xn k < δ for all n > N. Hence for
n > N we have xn ∈ Bδ (x) ⊆ f −1 (B), so f (xn ) ∈ B by definition of inverse
image. Therefore for n > N we have f (xn ) ∈ Bε (y), so kf (x) − f (xn )k =
ky − f (xn )k < ε. Thus f (xn ) → f (x), so f is continuous.
1.61 (a) If L is an isometry and Lx = 0, then kxk = kLxk = 0 so x = 0.
x
(c), (d) Consider any x ∈ X. If x 6= 0 then define y = kxk , so kyk = 1.
Therefore kLk, which is the supremum of kLzk over ALL elements z of norm 1,
must exceed kLyk for this specific vector y, i.e., kLyk ≤ kLk for this y. Hence
x 1 1

kLyk = L = Lx = kLxk.
kxk kxk kxk
Rearranging, we obtain
kLxk = kLyk kxk ≤ kLk kyk kxk = kLk kxk.

This is also true when x = 0, so it is valid for every vector x ∈ X.

Now we show that kLk is the smallest number satisfying kLxk ≤ Kkxk
for every x ∈ X. Suppose that K is any number for which this is true, and
let y be any vector with kyk = 1. Then
kLyk ≤ K kyk = K.
Hence K is an upper bound for {kLyk : kyk = 1}. But, by definition, kLk is
the least upper bound for that set, so we must have kLk ≤ K.
(e) We certainly have
sup kLxk ≤ sup kLxk.

kxk=1 kxk≤1
On the other hand, if kxk ≤ 1 and x 6= 0, then x = αy where kyk = 1 and

|α| < 1. Since L is linear, we therefore have kLxk = |α| kLyk ≤ kLyk. And if
x = 0 then we still have kLxk ≤ kLyk for any unit vector y. Consequently,
sup kLxk ≤ sup kLxk.

kxk≤1 kxk=1
Additionally, if x 6= 0 then x/kxk is a unit vector, so
kLxk
sup kLxk = sup .
kxk=1 x6=0 kxk
1.62 Suppose that fn ∈ ker(L) and fn → f ∈ X. Then L(fn ) = 0 for every n,

and, since L is continuous, L(fn ) → L(f ). Hence L(f ) = 0, so f ∈ ker(L).
Therefore ker(L) is closed.
1.63 If λ is an eigenvalue of L, then there exists a nonzero vector x such that
Lx = λx. By multiplying x by a nonzero scalar, we may assume that kxk = 1.
Then
kLxk = kλxk = |λ| kxk,
so we must have kLk ≥ |λ|.
1.64 (a) Linear: L is linear because
L(αx + βy) = L(αx1 + βy1 , αx2 + βy2 , . . . )

= (αx2 + βy2 , αx3 + βy3 , . . . )
= α (x2 , x3 , . . . ) + β (y2 , y3 , . . . ) = α L(x) + βL(y).
Continuous: Since
∞
X ∞
X
kLxk22 = k(x2 , x3 , . . . )k22 = |xk |2 ≤ |xk |2 = kxk22 ,
k=2 k=1
we have kLk ≤ 1. On the other hand, if x = (0, 1, 0, 0, 0, . . . ), then Lx =

(1, 0, 0, 0, . . . ), so kxk2 = 1 and kLxk2 = 1. Therefore kLk = 1.
Injective: L is not injective because L(1, 0, 0, . . . ) = 0 = L(0, 0, 0, . . . ).
Surjective: L is surjective because if y = (y1 , y2 , . . . ) ∈ ℓ2 and then Lx = y
where x = (0, y1 , y2 , . . . ).
Norm-preserving: L is not norm-preserving since k(1, 0, 0, . . . )k1 = 1 but
kL(1, 0, 0, . . . )k2 = 0. Hence L is not an isometry.
Eigenvalues: Suppose that λ is an eigenvalue of L, and let x 6= 0 be a
corresponding eigenvector. Then Lx = λ x, so
(x2 , x3 , . . . ) = Lx = λ x = (λ x1 , λ x2 , . . . ).
Therefore:
x2 = λ x1 , x3 = λ x2 = λ2 x1 , ....
n−1
Thus xn = λ x1 for every n. But x must be an element of ℓ2 , so kxk2 must
be finite. Therefore
∞
X ∞
X ∞
X
kxk22 = |xn |2 = |λn−1 x1 |2 = |x1 |2 |λ|2n−2 < ∞.
n=1 n=1 n=1
If x1 = 0 then this implies that kxk2 = 0 and therefore that x = 0, a con-

tradiction. Hence we must have x1 6= 0, so the above inequality implies that
|λ| < 1. These are the only possible eigenvalues.
On the other hand, if |λ| < 1, then let x be any vector such that xn =
λn−1 x1 for every n and x1 6= 0. For example, set x = (1, λ, λ2 , . . . ). Then
Lx = (λ, λ2 , λ3 , . . . ) = λ (1, λ, λ2 , . . . ) = λ x.
Thus every scalar λ with |λ| < 1 is an eigenvalue of L.

(b) Linearity of R is similar.
Continuous: Given any x ∈ ℓ2 , we have
∞
X
kRxk22 = k(0, x1 , x2 , . . . )k22 = |xk |2 = kxk22 .
k=1
Therefore R is bounded and kRk = 1. Also, R is an isometry and is therefore

injective.
Surjective: R is not surjective since Rx 6= (1, 0, 0, . . . ) for any x. Therefore
R is not an isomorphism.
Eigenvalues: Assume that λ is an eigenvalue of R, with corresponding
eigenvector x. Then
(0, x1 , x2 , x3 , . . . ) = Rx = λ x = (λ x1 , λ x2 , . . . ).
Hence,
λ x1 = 0, λ x2 = x1 , λ x3 = x2 , etc.
That is, λ x1 = 0, and λ xn+1 = xn for every n. If λ 6= 0 then this implies
that xn = 0 for every n, a contradiction. Thus no λ 6= 0 can be an eigenvalue.
On the other hand, if λ = 0, then xn = λ xn+1 = 0 xn+1 = 0 for every n,
also a contradiction. Thus λ = 0 can’t be an eigenvalue either. Thus R has no
eigenvalues.
(c) We have
LR(x) = L(0, x1 , x2 , . . . ) = (x1 , x2 , . . . ) = x.
Therefore LR = I, the identity mapping. This says that R has a left-inverse,

and its left-inverse is L.
On the other hand,
RL(x) = R(x2 , x3 , . . . ) = (0, x2 , x3 , . . . ).
Therefore RL is the operator that zeros out the first component of x.

Consequently, L and R do not commute.
1.65 (a) Given f ∈ Cb1 (R), we have
kDf kCb = kf ′ k∞ ≤ kf k∞ + kf ′ k∞ = kf kCb1 .
Hence D is bounded and kDk ≤ 1.

(b) Let fn (x) = einx . Then kfn k∞ = 1, but
kDfn kCb = kfn′ k∞ = kineinx k∞ = n.
Hence kDk = ∞, so D is not bounded.

If F = R, we can take fn (x) = sin nx.
1.66 (a) We are given λ ∈ ℓ∞ . For any x ∈ H we have
∞
X ∞
X
|λn hx, en i|2 ≤ kλk2∞ hx, en i|2 = kλk2∞ kxk2 < ∞,
n=1 n=1
so the series defining Mλ x converges (because {en } is an orthonormal se-

quence). Moreover, the preceding calculation also shows that
∞
X
kMλ xk2 = |λn hx, en i|2 ≤ kλk2∞ kxk2 ,
n=1
so we see that kMλ k ≤ kλkℓ∞ . On the other hand, by orthonormality we have

Mλ en = λn en , i.e., each en is an eigenvector for Mλ with eigenvalue λn . Since
ken k = 1 and kMλ en k = |λn | ken k = |λn | we conclude that
kMλ k = sup kMλ xk ≥ sup kMλ en k = sup |λn | = kλk∞ .

kxk=1 n∈N n∈N
(b) Follows from Mλ (en ) = λn en .

(c) ⇒. Suppose that Mλ is injective. Then the fact that en 6= 0 implies
that λn en = Mλ (en ) 6= 0, so λn 6= 0.
⇐. Suppose λn 6= 0 for every n, and suppose that Mλ x = 0. Then
∞
X
0 = kMλ xk2 = |λn |2 |hx, en i|2 ,
n=1
so hx, en i = 0 for every n. This implies that x = 0, so Mλ is injective.

(d) ⇐. Suppose that |λn | ≥ δ > 0 for every n. If g ∈ H, then
X∞
hg, en i
f = en ∈ H,
n=1
λn
so Mλ f = g and therefore Mλ is surjective.

⇒. If λn = 0 for some n then en ∈
/ range(Mλ ), so Mλ is not surjective in
this case.
Suppose that λn 6= 0 for every n but inf |λn | = 0. Then we can find nk
such that |λnk | < 2−k . Hence
X
y = λnk enk
k
converges. However, if T x = y then

X
λnk = hT x, enk i = λn hx, en i en , enk
n
X
= λn hx, en i hen , enk i = λnk hx, enk i,
n
so hx, enk i = 1 for every k. This is impossible, so no such x can exist. There-
fore Mλ is not surjective. On the other hand, Mλ en = λn en and λn 6= 0, so
range(Mλ ) contains span{en }n∈N , which is dense.
1.67 If f ∈ Lp (E) with p finite, then
Z 1/p
kTm f kLp = |f (t) m(t)|p dt
E
Z 1/p
p
≤ kmk L∞ |f (t)| dt = kmkL∞ kf kLp .
E
Therefore Tm is bounded and kTm k ≤ kmkL∞ . A similar proof shows that the
same inequality holds if p = ∞.
If kmkL∞ = 0 then m = 0 a.e. and Tm is the zero operator, so there is
nothing to prove in this case. Suppose M = kmkL∞ > 0 and fix ε > 0. Then
A = {x ∈ E : |m(x)| > M − ε} has positive measure. Since Lebesgue measure
is σ-finite, some set Ak = A∩[k, k+1] must have positive (and finite) measure.
If p < ∞, consider f = |Ak |−1/p χAk . We have
Z
p
kf kp = |Ak |−1 dx = 1,
Ak
but Z
kTm f kpp = |Ek |−1 |m(x)|p dx ≥ (kmk∞ − ε)p .
Ak
Hence kTm k ≥ kmk∞ −ε, and ε is arbitrary, so we conclude that kTm k−kmk∞.
On the other hand, if p = ∞, then consider f = χAk . We have kf k∞ = 1,
and
kTm f k∞ = kmχAk k∞ ≥ kmk∞ − ε,
and the inequality again follows.
1.68 Let B = {e1 , . . . , ed } be a basis for X, and let k · k1 be the ℓ1 -norm on X
Pd
with respect to that basis. Given x ∈ X, write x = k=1 xk ek . Then
X
n n
X

kLxk = L(xk ek ) ≤ |xk | kLek k

k=1 k=1
Xn
≤ |xk | max kLek k = C kxk1 ,
k
k=1
where C = maxk kLek k < ∞. Hence L is bounded. Since all norms on X are
equivalent, it follows that L is bounded with respect to any norm on X.
1.69 (a) The norm properties follow directly from the definition. For example,
if kLk = 0, then kLf k = 0 for every unit vector f. Hence Lf = 0 for every unit
vector f. Since every vector is a scalar multiple of a unit vector, we conclude
that Lf = 0 for every f ∈ X.
Given any f ∈ X we have

k(L + K)f k = kLf + Kf k ≤ kLf k + kKf k = kLk + kKk kf k.
Taking the supremum over all unit vectors, we have kL + Kk ≤ kLk + kKk.
(b) Let X be a normed space and Y a Banach space. Assume that {An }n∈N
is a sequence of operators in B(X, Y ) that is Cauchy in operator norm. For
any given f ∈ X, we have
kAm f − An f k ≤ kAm − An k kf k,
so we conclude that {An f }n∈N is a Cauchy sequence of vectors in Y. Since Y

is complete, this sequence must converge, say Afn → g ∈ Y. Define Af = g.
This gives us a candidate limit operator A, and we leave as an exercise the
task of showing that A defined in this way is linear.
To show that A is bounded, recall that all Cauchy sequences are bounded,
so C = sup kAn k < ∞. If f ∈ X, then since An f → Af,
kAf k = lim kAn f k ≤ sup kAn f k ≤ sup kAn k kf k = C kf k.

n→∞ n∈N n∈N
Therefore A is bounded, with kAk ≤ C.

Finally, we must show that An → A in operator norm. Fix any ε > 0.
Then there exists an N such that
ε
m, n > N =⇒ kAm − An k < .
2
Choose any f ∈ X with kf k = 1. Then since Am f → Af, there exists an
m > N such that
ε
kAf − Am f k < .
2
Hence for any n > N we have
kAf − An f k ≤ kAf − Am f k + kAm f − An f k

ε ε
≤ kAf − Am f k + kAm − An k kf k < + = ε.
2 2
Taking the supremum over all unit vectors, we conclude that kA − An k ≤ ε
for all n > N, so An → A. Therefore B(X, Y ) is complete.
(c) If f ∈ X, then we have
k(AB)f k = kA(Bf )k ≤ kAk kBf k ≤ kAk kBk kf k.
Hence kABk ≤ kAk kBk.

1.70 (a) Let v1 , . . . , vn be the columns of A. Given x ∈ Fn , we then have
that
kAxk1 = kx1 v1 + · · · + xn vn k1
≤ |x1 | kv1 k1 + · · · + |xn | kvn k1

n o
≤ max kvj k1 |x1 | + · · · + |xn |
j
n o
= max kvj k1 kxk1 .
j
Hence n o
kAkℓ1 →ℓ1 ≤ max kvj k1 .
j
On the other hand, kej k1 = 1 and kAej k1 = kvj k1 , so we actually have

equality:
n o X m
kAkℓ1 →ℓ1 = max kvj k1 = max |aij | .
j j
i=1
(b) Given x ∈ Fn , we have

X
n
kAxk∞ = max aij xi
i
j=1
X
n
≤ max |aij xi |
i
j=1
X
n
≤ kxk∞ max |aij | .
i
j=1
Hence X
n
kAkℓ∞ →ℓ∞ ≤ max |aij | .
i
j=1
On the other hand, fix any particular i, and set x = (x1 , . . . , xn ) where xj aij =
|aij |. Then kxk∞ = 1 and
n n
X X
kAxk∞ ≥ (Ax)i = aij xi = |aij |.
j=1 j=1
Hence we actually have

X
n
kAkℓ∞ →ℓ∞ = max |aij | .
i
j=1
1.71 Choose any orthonormal basis {en }n∈N for P H any orthonormal basis
{fn }n∈N for K. Define L : H → K by Lg = hg, fn i fn . The Plancherel
Equality implies that L is an isometry, and it also follows that the range of L
is closed. Since fn ∈ range(L) for each n, we conclude that range(L) = K,
so L is surjective.
1.72 Fix any f ∈ X. Since Y is dense in X, there exist gn ∈ Y such that
gn → f. Since L is bounded, we have kLgm − Lgn k ≤ kLk kgm − gn k. But
{gn }n∈N is Cauchy in X, so this implies that {Lgn }n∈N is Cauchy in Z.
Since Z is a Banach space, we conclude that there exists an h ∈ Z such that
e = h.
Lgn → h. Define Lf
e
To see that L is well defined, suppose that we also had gn′ → f for some
gn ∈ Y. Then kLgn′ − Lgn k ≤ kLk kgn′ − gn k → 0. Since Lgn → h, it follows
′
that Lgn′ = Lgn + (Lgn′ − Lgn ) → h + 0 = h. Thus Le is well defined.

To see that L e is linear, suppose that f, g ∈ X are given and c ∈ F. Then

there exist fn , gn ∈ Y such that fn → f and gn → g. Since cfn + gn → cf + g
and
L(cfn + gn ) = cLfn + Lgn → cLf e + Lg,e
e
by definition we have that L(cf e + Lg.
+ g) = cLf e
e
To see that L is an extension of L, suppose that g ∈ Y is fixed. If we set
e = Lg. Hence
gn = g, then gn → g and Lgn → Lg, so by definition we have Lg
e
the restriction of L to Y is L. Consequently,
e =
kLk sup e k ≥
kLf sup e k =
kLf sup kLf k = kLk.
f ∈X, kf k=1 f ∈Y, kf k=1 f ∈Y, kf k=1
Now suppose that f ∈ X. Then there exist gn ∈ Y such that gn → f and

e so
Lgn → Lf,
e k = lim kLgn k ≤ lim kLk kgnk = kLk kf k.
kLf
n→∞ n→∞
Hence kLke ≤ kLk. Combining this with the opposite inequality derived above,
we conclude that kLke = kLk.
Finally, we must show that Le is unique. Suppose that A ∈ B(X, Y ) also
satisfied A|Y = L. Then Af = Lfe for all f ∈ Y. Since Y is dense, this extends
e
by continuity to all f ∈ X, which implies that A = L.
′
1.73 (a) Consider p = 1. We know that T maps ℓp = ℓ∞ into (ℓ1 )∗ with
kµy k ≤ kykℓ∞ for each y ∈ ℓ∞ , so it remains to show that T is isometric and
surjective.
Choose y ∈ ℓ∞ . Then kδn k1 = 1, and we have
|hδn , µy i| = |yn |,
so kµy k ≥ |yn |. Since this is true for every n, we conclude that kµy k ≥ kykℓ∞ .
Hence T is an isometry.
To see that T is surjective, fix any µ ∈ (ℓ1 )∗ . Let yn = hδn , µi and set
y = (yn ). Since
|yn | = |hδn , µi| ≤ kδn kℓ1 kµk = kµk,
we have y ∈ ℓ∞ . Further, we have by definition that hδn , µy i = yn = hδn , µi.

By linearity µy and µ agree on the dense set c00 , and hence by continuity we
have µ = µy . Therefore T is surjective.
′
(a) Now consider p = ∞. We know that T maps ℓp = ℓ1 into (ℓ∞ )∗ with
kµy k ≤ kykℓ1 for each y ∈ ℓ1 , so it remains to show that T is isometric.
If y = 0 then kµy k = 0 = kykℓ1 , so assume that y 6= 0. Let αk ∈ F be
the scalar of unit modulus such that αk yk = |yk |. Then x = (αk ) ∈ ℓ∞ and
kxkℓ∞ = 1. Further,
X X
|hx, µy i| = αk yk = |yk | = kykℓ1 .
k k
This shows that kµy k ≥ kykℓ1 . Hence we have kµy k = kykℓ1 , which establishes
that T is an isometry.
1.74 By hypothesis, λ = hz, µi =6 0.
Given x ∈ X, let c = hx, µi, and set y = x − (c/λ) z. Then
c
hy, µi = hx, µi − hz, µi = c − c = 0,
λ
so y ∈ ker(µ).
To show this representation is unique, suppose that we also had x = y ′ +c′ z
where y ′ ∈ ker(µ) and c′ ∈ F. Then y − y ′ = (c′ − c)z, so since y − y ′ ∈ ker(µ)
we have
(c′ − c) λ = h(c′ − c) z, µi = hy − y ′ , µi = 0.
Therefore c = c′ , and hence y − y ′ = 0.
1.75 Given y ∈ ℓ1 , define µy : c0 → F by
∞
X
hx, µy i = xk yk , x ∈ c0 .
k=1
This converges since c0 ⊆ ℓ∞ . Clearly µy is linear, and we have

∞
X
|hx, µy i| ≤ kxk∞ |yk | = kxk∞ kyk1 .
k=1
Hence µy is bounded and kµy k ≤ kyk1 . Let ck be the scalars of unit modulus
such that ck yk = |yk |. Set xn = (c1 , . . . , cn , 0, 0, . . . ). Then xn ∈ c0 , kxn k∞ =
1, and
Xn X n
hxn , µy i = ck y k = |yk | → kyk1 as n → ∞.
k=1 k=1
Hence kµy k = kyk1 .

Therefore, if we define T : ℓ1 → c0 ∗ by T (y) = µy , then we have that T is
an antilinear isometry, and hence is injective. So, it remains only to show that
T is surjective. Suppose that µ ∈ c0 ∗ is given. Let yk = hek , µi, where ek is
the kth standard basis vector. Set y = (yk )k∈N . Let ck be the scalars of unit
modulus such that ck yk = |yk |. Set xn = (c1 , . . . , cn , 0, 0, . . . ). Then xn ∈ c0 ,
kxn k∞ = 1, and
n
X n
X n
X
|yk | = ck y k = ck hek , µi
k=1 k=1 k=1
X
n
= ck e k , µ
k=1
= hxn , µi
≤ kxn k∞ kµk = kµk.
Hence kyk1 ≤ kµk < ∞, so y ∈ ℓ1 . Therefore,P it just remains to show that

µ = µy . Given x ∈ c0 we have that x = xk ek , with convergence of this
series in ℓ∞ -norm. Since µ is continuous we therefore have
∞
X ∞
X
hx, µi = xk hek , µi = xk yk = hx, µy i.
k=1 k=1
Hence T (y) = µy = µ, so T is surjective. Thus ℓ1 ∼

= c0 ∗ .
P PN
1.76 If x = xn converges in X, then the partial sums sN = n=1 xn
converge to x. If µ ∈ X ∗ , then we have by continuity and linearity that
∞
X N
X
hxN , µi = lim hxN , µi = lim hsN , µi = hx, µi.
N →∞ N →∞
n=1 n=1
1.77 (a) We have that µh is linear, and |µh (f )| = |hf, hi| ≤ kf k khk, so µh is
bounded and kµh k ≤ khk. If h = 0, then µh = 0 and so kµh k = 0 = khk. On
the other hand, if h 6= 0 then g = h/khk is a unit vector, and
hh, hi
|µh (g)| = |hg, hi| = = khk,
khk
so we have kµh k ≥ khk. Thus kµh k = khk.

(b) Choose any µ ∈ H ∗ . If µ = 0 then h = 0 is the required vector, so
assume that µ 6= 0 (i.e., µ is not the zero operator). Then µ does not map
every vector to zero, so the kernel of µ is a closed subspace of H that is not
all of H.
Choose any g ∈/ ker(µ), and write
g = p + e, p ∈ ker(µ), e ∈ ker(µ)⊥ .
Note that µ(p) = 0 by definition, and therefore we must have µ(e) 6= 0 (since
µ(g) 6= 0). Set u = e/µ(e), and note that u ∈ ker(µ)⊥ and µ(u) = 1.
Given f ∈ H, since µ is linear and µ(u) = 1 we have

µ f − µ(f ) u = µ(f ) − µ(f ) µ(u) = 0.
Therefore f − µ(f ) u ∈ ker(µ). However, u ⊥ ker(µ), so
0 = hf − µ(f ) u, ui = hf, ui − hµ(f ) u, ui = hf, ui − µ(f ) kuk2.
Hence,
1 D u E
µ(f ) = hf, ui = f, = hf, hi = µh (f )
kuk2 kuk2
where
u
h = .
kuk2
Thus L = Lh , and from part (a), we have that kLk = kLh k = khk.
It remains only to show that h is unique. Suppose that we also had µ = µh′ .
Then for every f ∈ H we have
hf, h − h′ i = hf, hi − hf, h′ i = µh (f ) − µh′ (f ) = µ(f ) − µ(f ) = 0.
Consequently, h − h′ = 0.
(c) Parts (a) and (b) show that T is surjective and T is norm-preserving.
Finally, the part of the proof of Theorem 1.75 given in the text shows that T
is antilinear.
1.78 ⇐. This follows from Hölder’s Inequality.
⇒. Case 1 ≤ p < ∞. The hypotheses imply that T x = hx, yi is a bounded
linear map of (c00 , k·kℓp ) into F. By Exercise 1.72, T has a unique extension to
′
a bounded linear functional on ℓp , which we also call T. Hence T ∈ (ℓp )∗ = ℓp ,
′
so there exists some z ∈ ℓp such that T x = hx, zi for all x ∈ ℓp . Therefore,
′
by considering the standard basis vectors, we see that y = z ∈ ℓp .
Case p = ∞. Let αn be scalars of unit modulus such that αn yn = |yn |.
Fix N and set α = (α1 , . . . , αN , 0, 0, . . . ). Then
N
X N
X
|yn | = αn yn = hα, yi ≤ C kαk∞ = C < ∞.
n=1 n=1
Letting N → ∞, we see that y ∈ ℓ1 .

1.79 ⇒. This follows from Exercise 1.62.
⇐. Suppose that ker(µ) is closed. If µ = 0 then µ is certainly continuous, so
assume that µ 6= 0. In this case ker(µ) is a proper closed subspace of X, so by
Exercise 1.44 there exists some unit vector z ∈ X such that dist(z, ker(µ)) >
1/2. By Exercise 1.74, if x ∈ X then x can be written uniquely as x = y + cz
where y ∈ ker(µ) and c ∈ F. Then kz + (y/c)k ≥ dist(z, ker(µ)), so
1 1
kxk = kz + (y/c)k ≥ dist(z, ker(µ)) > .
c 2
Therefore, if we set λ = |hz, µi| then
|hx, µi| = |hy, µi + c hz, µi| = |c| |hz, µi| = λ |c| ≤ 2λ kxk.
Therefore µ is bounded.
1.80 Define T : X ∗ × Y ∗ → (X × Y )∗ by
T (µ, ν)(x, y) = hx, µi + hy, νi.
Since
|T (µ, ν)(x, y)| ≤ kxk kµk + kyk kνk

≤ k(x, y)k∞ kµk + kνk
= k(x, y)k∞ k(µ, ν)k1 .
Hence T (µ, ν) is a bounded linear functional on X ×Y, so T does map X ∗ ×Y ∗

into (X × Y )∗ . Further, this shows that T is continuous, and kT k ≤ 1.
Fix ε > 0. Then by definition of operator norm, there exist unit vectors
x ∈ X, y ∈ Y such that
|hx, µi| ≥ kµk − ε and |hy, νi| ≥ kνk − ε.
Let α, β be scalars of unit modulus such that
α hx, µi = |hx, µi| and β hy, νi = |hy, νi|.
Then (αx, βy) is a unit vector in X × Y, and
|T (µ, ν)(αx, βy)| = |hαx, µi + hβy, νi

= |hαx, µi| + |hβy, νi|
≥ kµk + kνk − 2ε
= k(µ, ν)k1 − 2ε.
Hence kT (µ, ν)k ≥ k(µ, ν)k1 − 2ε, and ε > 0 is arbitrary. Combined with our
calculation above, this implies kT (µ, ν)k = k(µ, ν)k1 . Hence T is an isometry.
It remains to show that T is surjective. Suppose that λ ∈ (X × Y )∗ . Define
µ : X → F and ν : Y → F by

hx, µi = (x, 0), λ and hy, νi = (0, y), λ .
Then
|hx, µi| ≤ k(x, 0)k∞ kλk = kxk kλk,
so µ is bounded and therefore belongs to X ∗ . Similarly, ν ∈ Y ∗ . Given (x, y) ∈
X × Y we have

T (µ, ν)(x, y) = hx, µi + hy, νi = (x, 0), λ + (0, y), λ = (x, y), λ .
Hence λ = T (µ, ν), so T is surjective.

2.1 Suppose that M is a subspace of a Hilbert space H and λ ∈ M. Then
H = M ⊕ M ⊥ . Given z ∈ H, write z = x + y where x ∈ M and y ∈ M ⊥ .
Define Λ : H → F by hz, Λi = hx, λi, i.e., Λ = λ ◦ P where P is the orthogonal
projection onto M . Then Λ is well defined and linear, and by construction
we have hx, Λi = hx, λi for x ∈ M. Therefore Λ|M = λ, so kΛk ≥ kλk. The
opposite inequality follows from the calculation
kΛk = kλ ◦ P k ≤ kλk kP k = kλk,
where we have used the fact that the norm of λ as an element of H coincides
with the operator norm of the functional on H that it induces.
2.2 Since x0 ∈
/ M, we can write x0 uniquely as
x0 = y0 + z0 , y0 ∈ M, z0 ∈ M ⊥ .
Note that
ky0 k = kx0 − z0 k = dist(x0 , M ) = d.
Set S = span{y0 }. Then S is one-dimensional, and hence is a closed sub-
space of H. Therefore we have H = S ⊕S ⊥ . Hence every x ∈ H can be written
uniquely as
x = cx y 0 + z x , cx ∈ F, zx ∈ S ⊥ .
Define µ : H → F by
hx, µi = cx , x ∈ H.
Then µ is linear and hx0 , µi = 1. Given x ∈ H, we have by the Pythagorean
Theorem that
kxk2 = kcx y0 + zx k2 = |cx |2 ky0 k2 + kzx k2 ≥ |hx, µi|2 d2 + 0.
Rearranging,
1
|hx, µi| ≤ kxk, x ∈ H,
d
so kµk ≤ 1/d. Also, y0 /d is a unit vector and
D y E 1
0
, µ = ,
d d
so kµk ≥ 1/d.
Thus µ ∈ H ∗ and kµk = 1/d. By the Riesz Representation Theorem, µ is
identified with an element of H that has the same norm. This µ has all of the
required properties.
2.3 (a) Suppose that X ∗ is separable. Then X ∗ has a countable dense sub-
set, say {µn }n∈N . Setting λn = µn /kµn k, it follows that S = {λn }n∈N is a
countable dense subset of the closed unit sphere

D∗ = µ ∈ X ∗ : kµk = 1 .
Since kλn k = 1, there must exist some xn ∈ X with

1
|hxn , λn i| ≥ .
2
Define
M = span {xn }n∈N .
This is a closed subspace of X. Since {xn }n∈N is a countable complete subset
of M, we have that M is separable by Theorem 1.27.
Suppose that M 6= X. Then there exists some x0 ∈ X\M, and we claim
that by rescaling x0 we may assume that d = dist(x0 , M ) = 1. To see this,
let mn ∈ M be such that kx0 − mn k → d. Let c = 1/d. Then kcx0 − cmn k =
c kx0 − mn k → cd = 1. Hence dist(cx0 , M ) ≥ 1. On the other hand, if m ∈ M
then dm ∈ M as well
kcx0 − mk = kcx0 − cdmk = c kx0 − dmk ≤ c d = 1.
Therefore dist(cx0 , M ) ≤ 1. Thus cx0 is exactly a distance 1 from M.

By Hahn–Banach, there exists some λ ∈ X ∗ such that
λ|M = 0, hx0 , λi = 1, kλk = 1.
In particular, λ ∈ D∗ . However, for any n we have xn ∈ M, so
kλ − λn k = sup |hx, λ − λn i|
kxk=1
≥ sup |hxn , λi − hxn , λn i|

n∈N
≥ sup |0 − hxn , λn i|
n∈N
1
≥ .
2
But this contradicts the fact that {λn }n∈N is dense in D∗ .
Therefore we must have M = X, and therefore X is separable since M is
separable.
(b) The space ℓ1 is separable, but its dual space is ℓ∞ , which is not sepa-
rable.
2.4 Suppose that µ ∈ C[0, 1]∗ is such that hx2k , µi = 0 for every k ≥ 0.
Extend µ symmetrically to [−1, 1]. That is, given f ∈ C[−1, 1], define g,
h ∈ C[0, 1] by g(x) = f (x) and h(x) = f (−x), and set
hf, νiC[−1,1] = hg, µi + hh, µi.

Since x2k is an even function, we then have
hx2k , νiC[−1,1] = hx2k , µi + hx2k , µi = 0.
On the other hand, x2k+1 is odd, so
hx2k+1 , νiC[−1,1] = hx2k+1 , µi + h−x2k+1 , µi = 0.
Thus hxk , νiC[−1,1] = 0 for every k ≥ 0. Since {xk }k≥0 is complete in C[−1, 1],
we conclude that ν = 0.
Now, given f ∈ C[0, 1], extend f symmetrically to [−1, 1], i.e., set f (−x) =
f (x) for x ∈ [−1, 0). Then
0 = hν, f iC[−1,1] = hµ, f i + hµ, f i = 2hµ, f i,
and therefore µ = 0. Hence {x2k }k≥0 is complete in C[0, 1].

2.5 The fact that A⊥ is a subspace follows from the linearity of functionals
µ ∈ X ∗ , and the fact that A⊥ is closed follows from the continuity of µ ∈ X ∗ .
In the terminology of this exercise, Corollary 2.5 states that {xn } is com-
plete in X if and only if {xn }⊥ = {0} in X ∗ .
2.6 Since the intersection of closed sets is closed,
T
F = {ker(µ) : µ ∈ X ∗ and S ⊆ ker(µ)}
is a closed set, and it contains S by definition. Since S̄ is the smallest closed

set containing S, we therefore have that S̄ ⊆ F.
Alternatively, given x ∈ S̄ there exist xn ∈ S such that xn → x. If µ ∈ X ∗
and S ⊆ ker(µ), then hxn , µi = 0 for every n, so since µ is continuous this
implies hx, µi = 0. Hence x ∈ ker(µ), and therefore x ∈ F.
Suppose that x ∈ / S̄. Then by Hahn–Banach, there exists some µ ∈ X ∗
such that µ = 0 on S̄ and hx, µi = 6 0. Hence S ⊆ ker(µ) but x ∈ / ker(µ), so
x∈ / F. This shows that F ⊆ S̄.
2.7 Fix 1 < p < ∞, and let p′ be the dual index. Let
T : ℓp → (ℓp )∗∗
y 7→ ye
be the canonical embedding of ℓp into (ℓp )∗∗ . That is, ye : (ℓp )∗ → F is given
by
hµ, ye i = hy, µi, µ ∈ (ℓp )∗ .
′
Given y ∈ ℓp , we know that
X
hx, µy i = xk yk x ∈ ℓp ,
k
defines a bounded linear functional on ℓp . The mapping

′
Tp : ℓp → (ℓp )∗
y 7→ µy
is an antilinear isometric isomorphism. Likewise, we have an antilinear iso-

′
metric isomorphism Tp′ : ℓp → (ℓp )∗ .
We already know that T is an isometry, so our goal is to show that T is
surjective. Choose F ∈ (ℓp )∗∗ , so F : (ℓp )∗ → F is a bounded linear functional.
′
Then F ◦ Tp : ℓp → F is a linear functional, and it is bounded since both F
′
and Tp are bounded. Hence F ◦ Tp ∈ (ℓp )∗ , and therefore F ◦ Tp = Tp′ y = µy
p
for some y ∈ ℓ . We will show that F = ye.
′
To see this, choose ν ∈ (ℓp )∗ . Then ν = µz = Tp z for some z ∈ ℓp .
Therefore,
hν, F i = hTp z, F i = hz, F ◦ Tp i
= hz, Tp′ yi
X
= zk yk dx
k
= hy, Tp zi
= hy, νi
= hν, ye i.
Therefore F = ye, so T is surjective.

2.8 Suppose that X is separable and reflexive. Then X ∗∗ is separable, so by
Exercise 2.3 we must have that X ∗ is separable.
Since (ℓ1 )∗ = ℓ∞ , if we had (ℓ∞ )∗ = ℓ1 then ℓ1 would be reflexive. But ℓ1 is
separable, so this would imply that ℓ∞ is separable, which it is not. Therefore,
although we have ℓ1 ⊆ (ℓ∞ )∗ , we cannot have equality.
2.9 Suppose that B : Y ∗ → X ∗ satisfies
∀ f ∈ X, ∀ µ ∈ Y ∗, hAf, µi = hf, Bµi.
Then with µ ∈ Y ∗ fixed, we have
hf, A∗ µ − Bµi = 0, f ∈ X.
Hence A∗ µ − Bµ is the zero operator, or, in other words, A∗ µ = Bµ. This is

true for every µ ∈ Y ∗ , so B = A∗ .
2.10 Fix µ ∈ X ∗ = B(X, F). Then its adjoint µ∗ is a bounded mapping of
F∗ = F into X ∗ . Given c ∈ F, the functional µ∗ c is defined by the rule
hx, µ∗ ci = hµ(x), ci = c µ(x),

because the inner product on F is simply multiplication of scalars.

2.11 Suppose that M is invariant under L, and note that L∗ : X ∗ → X ∗ .
Choose any µ ∈ M ⊥ . Then for any x ∈ M we have Lx ∈ M, so
hx, L∗ µi = hLx, µi = 0.
Hence L∗ µ ∈ M ⊥ , so M ⊥ is invariant under L∗ .

2.12 Let ǫ : M → X be the embedding map of a closed subspace M into X.
Given x ∈ M and µ ∈ X ∗ , we have that µ|M ∈ M ∗ since
|hx, µ|M i| = |hx, µi| ≤ kxk kµk.
Also,
hx, ǫ∗ µi = hǫx, µi = hx, µi = hx, µ|M i,
so ǫ∗ µ = µ|M .
2.13 Let hx, yi = x · y = xT y be the usual sesquilinear inner product on Cn .
Then given x ∈ Cn and y ∈ Cm , we have
T
hAx, yi = (Ax)T y = xT AT y = xT (A y) = xT (AT y) = hx, AT yi.
Hence, in the Hilbert space adjoint sense, A∗ = AT .

For the Banach space adjoint, we repeat the same proof but use the bilinear
form hx, yi = xT y. The result is the A∗ = AT .
2.14 Given x, y ∈ ℓ2 , we have
∞
X
hLx, yi = xk+1 yk = hx, Ryi.
k=1
Hence L = R∗ .
2.15 Given x, y ∈ H we have
X
∞
hMλ x, yi = λn hx, en i en , y
n=1
∞
X
= λn hx, en i hen , yi
n=1
∞
X
= λn hx, en i hen , yi
n=1
∞
X
= x, λ̄n hy, en i en = hx, Mλ̄ yi,
n=1
so Mλ∗ = Mλ̄ .
Thus, Mλ is self-adjoint if and only if λ = λ̄, i.e., λ has all real components.
Finally, Mλ is positive if and only if λn ≥ 0 for all n, and is positive definite
if and only if λn > 0 for all n.
2.16 (a), (b) For any x ∈ H and y ∈ L we have

∗ ∗
(A ) x, y = hx, A∗ yi = hAx, yi
and
hBAx, yi = hAx, B ∗ yi = hx, A∗ B ∗ yi,
so (A∗ )∗ = A and (BA)∗ = A∗ B ∗ .
(c) Assume that x ∈ ker(A) and let z ∈ range(A∗ ), i.e., z = A∗ y for some
y ∈ K. Then since Ax = 0, we have hx, zi = hx, A∗ yi = hAx, yi = 0. Thus
x ∈ range(A∗ )⊥ , so ker(A) ⊆ range(A∗ )⊥ .
Now assume that x ∈ range(A∗ )⊥ . Then for any z ∈ H we have hAx, zi =
hx, A∗ zi = 0. But this implies Ax = 0, so x ∈ ker(A). Thus range(A∗ )⊥ ⊆
ker(A).
(d) Using part (c), we have
⊥
ker(A)⊥ = range(A∗ )⊥ = range(A∗ ),
the final equality following from Lemma 1.44.

(e) We have
A is injective ⇐⇒ ker(A) = {0}

⇐⇒ ker(A)⊥ = H
⇐⇒ range(A∗ ) = H.
(f) The fact that kAk = kA∗ k follows from the corresponding result for
Banach adjoints, but we can also give a direct proof. Given any y ∈ K,
kA∗ yk = sup |hx, A∗ yi| = sup |hAx, yi|

kxk=1 kxk=1
≤ sup kAk kxk kyk = kAk kyk.

kxk=1
Hence A∗ is bounded, with kA∗ k ≤ kAk. Applying this reasoning to A∗ ,

we see that (A∗ )∗ = A is bounded, and kAk = k(A∗ )∗ k ≤ kA∗ k. Therefore
kAk = kA∗ k.
Now, since the operator norm is submultiplicative,
kA∗ Ak ≤ kAk kA∗ k = kAk2 .

Also,
kAxk2 = hAx, Axi = hA∗ Ax, xi ≤ kA∗ Axk kxk.
Taking the supremum over all unit vectors, we obtain
kAk2 = sup kAxk2 ≤ sup kA∗ Axk kxk = kA∗ Ak.

kxk=1 kxk=1
Consequently kAk2 = kA∗ Ak. The final equality, kAk2 = kAA∗ k, follows by
interchanging the roles of A and A∗ .
2.17 ⇒. If P is the orthogonal projection onto M, then it is clear that P 2 = P
and range(P ) = M. If x, y ∈ H, let e = x − P x ∈ M ⊥ and f = y − P y ∈ M ⊥ .
Then
hP x, yi − hx, P yi = hP x, P y + f i − hP x + e, P yi
= hP x, P yi + hP x, f i − hP x, P yi − he, P yi
= hP x, P yi + 0 − hP x, P yi − 0 = 0.
Hence hP x, yi = hx, P yi, so P ∗ = P.

⇐. Suppose that P 2 = P, P ∗ = P, and range(P ) = M. Choose any x ∈ H,
and let p = P x. If y ∈ M, then y = P z for some z, so
hx − p, yi = hx, P zi − hP x, P zi
= hx, P zi − hx, P ∗ P zi
= hx, P zi − hx, P 2 zi = 0.
Hence x − p ⊥ M, so p is the orthogonal projection of f onto M.

2.18 (a) If A, B are self-adjoint, then
(ABA)∗ = A∗ B ∗ A∗ = ABA,
so ABA is self-adjoint.
(b) ⇒. Suppose A, B are self-adjoint and AB is self-adjoint. Then AB =
(AB)∗ = B ∗ A∗ = BA.
⇐. If A, B are self-adjoint and AB = BA, then (AB)∗ = B ∗ A∗ = BA =
AB, so AB is self-adjoint.
(c) Let P, Q be orthogonal projections onto one-dimensional nonorthogo-
nal lines in H. That is, P f = hf, gi g and Qf = hf, hi h for some nonorthogonal
unit vectors g, h. Then range(P Q) ⊆ span{g} while range(QP ) ⊆ span{h}.
But P Q and QP are not the zero operators, so these ranges are these 1-
dimensional lines. To see this directly, we compute that
P Qf = hQf, gi g = hf, hi hh, gi g,

so P Qg = |hg, hi|2 g 6= 0. Therefore P Q is not the zero operator, so

range(P Q) = span{g}. Similarly,
QP f = hP f, hi h = hf, gi hg, hi h,
so QP h = |hh, gi|2 h 6= 0. Therefore QP is not the zero operator, so

range(QP ) = span{h}. Therefore QP is not the zero operator, so range(QP ) =
span{h}. Hence P Q 6= QP.
2.19 (a) Let λ be an eigenvalue of the self-adjoint operator L, and let f be
a corresponding eigenvector. Then we have
λ kf k2 = hλf, f i = hLf, f i = hf, Lf i = hf, λf i = λ̄ kf k2 .
Since f 6= 0, we must have λ = λ̄, and hence λ is real.

Now suppose that λ 6= µ are eigenvalues of L, and that f is a λ-eigenvector
while g is a µ-eigenvector. Then, since the eigenvalues are real,
λ hf, gi = hLf, gi = hf, Lgi = µ hf, gi.
Since λ 6= µ, we must have hf, gi = 0.

(b) Suppose that A is positive and λ is an eigenvalue for A with corre-
sponding eigenvector f. Then 0 ≤ hAf, f i = λ kf k2 , so we must have λ ≥ 0.
(c) Suppose that A is positive definite and λ for A is an eigenvalue with
corresponding eigenvector f. Then 0 < hAf, f i = λ kf k2 . Since f 6= 0, we
must have λ > 0.
2.20 We have
hA∗ Af, f i = hAf, Af i = kAf k2 ≥ 0
and
hAA∗ g, gi = hA∗ g, A∗ gi = kA∗ gk2 ≥ 0,
so A∗ A, AA∗ ≥ 0.
2.21 (a) We certainly have that Af = 0 implies A∗ Af = 0. On the other
hand, if A∗ Af = 0, then
kAf k2 = hAf, Af i = hA∗ Af, f i = 0,
so Af = 0. Thus ker(A) = ker(A∗ A).

(b) Combining part (a) with Exercise 2.16, we have
range(A∗ A) = ker(A∗ A)⊥ = ker(A)⊥ = range(A∗ ).
2.22 Let H, K be Hilbert spaces, and fix U ∈ B(H, K).

⇒. Suppose that U is unitary, i.e., U is an isometric isomorphism. Then U

is a bijection by definition, and also hU x, U yi = hx, yi for all x, y ∈ H. Given
x ∈ H and k ∈ K, let y = U −1 k. Then
hx, U ∗ ki = hU x, ki = hU x, U yi = hx, yi = hx, U −1 ki.
Since the adjoint is unique, it follows that U −1 = U ∗ .

⇐. Suppose that U is a bijection and U −1 = U ∗ . Then given x ∈ H we
have
kU xk2 = hU x, U xi = hU ∗ U x, xi = hU −1 U x, xi = hx, xi = kxk2 .
Hence U is also an isometry, and therefore is a unitary map.

2.23 (a) Since A is self-adjoint and 0 ≤ A ≤ I, we can use Theorem 2.15 to
compute that
kAk = sup hAx, xi ≤ sup hIx, xi = 1.

kxk=1 kxk=1
(b) Suppose that A ≥ 0. If we fix 0 < t < kAk−1 , then for each x ∈ H we
have
htAx, xi ≤ ktAxk kxk = t kAk kxk2 < kxk2 = hIx, xi.
Hence tA ≤ I. Further, since t is a positive real scalar, the operator A has a
square root if and only if the operator tA has a square root.
(c) We show by induction that Tn+1 − Tn is a polynomial in B with all co-
efficients nonnegative. Since T1 − T0 = T1 = 21 B, this is true for the base step.
Assume that Tn − Tn−1 is a polynomial in B with all nonnegative coefficients.
Then, using the fact that B, Tn , and Tn−1 commute, we have
1 1
Tn+1 − Tn = (B + Tn2 ) − (B + Tn−1 )
2 2
1
= (Tn2 − Tn−1
2
)
2
1
= (Tn − Tn−1 ) (Tn + Tn+1 ).
2
Now, by the inductive hypothesis, Tn − Tn−1 is a polynomial in B with all
coefficients nonnegative. The same is true for Tn and Tn−1 individually, and
therefore Tn + Tn−1 is also a polynomial in B with all coefficients nonnegative.
Therefore the product (Tn − Tn−1 ) (Tn − Tn−1 ) is a polynomial in B with all
coefficients nonnegative.
(d) We have kT0 k = 0 and
1 1 1 1
kT1 k = kBk = kI − Ak ≤ (kIk + kAk) ≤ (1 + 1) = 1.
2 2 2 2
Suppose that kTn k ≤ 1 for some n. Then

1 1 1
kTn+1 k = kB + Tn2 k ≤ kBk + kTn k2 ≤ (1 + 1) = 1.
2 2 2
Hence we conclude that kTn k ≤ 1 for every n. Consequently hTn x, xi ≤
kTn k kxk2 ≤ kxk2 . Since Tn ≤ Tn+1 we have hTn x, xi ≤ hTn+1 x, xi, so
hTn x, xi is a bounded, increasing sequence of scalars.
(e) We have
T (x + y) = lim Tn (x + y) = lim Tn (x) + lim Tn (y) = T (x) + T (y),

n→∞ n→∞ n→∞
and similarly T (cx) = cT (x), so T is linear. By continuity of the norm we

have
kT xk = lim kTn k ≤ 1,
n→∞
so T is bounded.
By continuity of the inner product,
hT x, xi = lim hTn x, xi ≥ 0,
n→∞
so T is a positive operator.
We will show that T commutes with B. Note that
kBT x − BTn xk ≤ kBk kT x − Tn xk → 0,
and similarly Tn Bx → T Bx for each x. Since Tn commutes with B, we con-

clude that BT x = T Bx for each x. Since Tn is a polynomial in B, we conclude
that T commutes with Tn as well. If L is an operator that commutes with B,
then L commutes with each Tn since Tn is a polynomial in B. Taking limits
as above, it follows that LT = T L.
We claim next that Tn2 x → T 2 x. Since Tn and T commute, we have
kT 2 x − Tn2 xk = k(T + Tn ) (T − Tn )xk

≤ kT + Tn k kT x − Tn xk
≤ 2 kT x − Tn xk → 0.
Therefore
1
T x = lim Tn+1 x = lim (B + Tn2 )x
n→∞ n→∞ 2
1 1
= Bx + lim Tn2 x
2 2 n→∞
1 1
= Bx + T 2 x.
2 2
Setting S = I − T and recalling that B = I − A, we therefore have
1 1 A S2
I −S = I − A + (I − S)2 = I − A + I − 2S + S 2 = I − − S + .
2 2 2 2
Rearranging, we find that A = S 2 .
Since kT k ≤ 1, we therefore have that
hSx, xi = h(I − T )x, xi = hx, xi − hT x, xi ≥ kxk2 − kT k kxk2 ≥ 0.
Hence S ≥ 0.
2.24 (a) Suppose that A ≥ 0 and hAx, xi = 0. By Theorem 2.18, there exists
a positive operator S such that S 2 = A. Hence
kSxk2 = hSx, Sxi = hS 2 x, xi = hAx, xi = 0.
Therefore Sx = 0, and hence Ax = SSx = 0.

(b) Suppose that B is also a positive operator such that B 2 = A. Note
that AB = B 2 B = BB 2 = BA, so B commutes with A. Since S commutes
with every operator that commutes with A, we conclude that S commutes
with B.
Fix x and let y = (S − B)x. Then since S and B commute,

hSy, yi + hBy, yi = h(S + B)y, yi = (S + B)(S − B)x, y

= (S 2 − B 2 )x, y

= (A − A)x, y = 0.
Since B, S ≥ 0, this implies hSy, yi = 0 = hBy, yi. Part (a) therefore implies
that Sy = 0 = By. Hence

kSx − Bxk2 = (S − B)x, (S − B)x = (S − B)(S − B)x, x

= S(S − B)x, x − hB(S − B)x, xi
= hSy, xi − hBy, xi = 0.
Therefore Sx = Bx.
2.25 (a) ⇒ (b). Suppose that E contained a nonempty open subset. Then
there is some ball Br (x) contained in E. Hence X \ E cannot be dense since
every element of X \ E lies at least a distance r from x. Thus E is not nowhere
dense.
(b) ⇒ (a). Suppose that E contains no nonempty open subsets. Choose
any x ∈ X and r ≥ 0. Then Br (x) is not a subset of E, so there must exist
some y ∈ X \ E that belongs to Br (x), i.e., d(x, y) < r. Thus X \ E is dense
in X, so E is nowhere dense.
2.26 For each N ∈ N, define

C[−N, N ] = f ∈ C(R) : supp(f ) ⊆ [−N, N ] .
This is a closed subspace of C0 (R) with respect to the uniform norm. Given
ε > 0, let g be the hat function supported on [N, N + 1] with height ε/2.
Given any f ∈ C[−N, N ] we have h = f + g ∈ / C[−N, N ] yet
kf − hk∞ = kgk∞ = ε/2 < ε.
Therefore h ∈ Bε (f ), so Bε (f ) is not contained in C[−N, N ]. Thus C[−N, N ]

is a closed subset of C0 (R) that has no interior, which says that C[−N, N ] is
a nowhere dense subset of C0 (R). Finally,
S
Cc (R) = C[−N, N ],
N ∈N
so Cc (R) is a meager subset of C0 (R).

2.27 For each n ≥ 0, set
En = {t ∈ R : f (n) (t) = 0}.
Since f (n) is continuous, En is closed. However, R = ∪En by hypothesis, so

the Baire Category Theorem implies that some set En contains an open set,
and hence must contain some open interval (a, b). But then f (n) (t) = 0 for all
t ∈ (a, b), so f equals a polynomial on (a, b). ⊓
⊔
2.28 Define

Fn = f ∈ C[0, 1] : ∃ x0 s.t. |f (x) − f (x0 )| ≤ n (x − x0 ), x0 ≤ x < 1 .
Suppose that fk ∈ Fn and fk → f in C[0, 1]. Then for each k ∈ N there exists
some xk such that
|fk (x) − fk (xk )| ≤ n (x − xk ), all xk ≤ x < 1.
The sequence {xn } has a convergent subsequence, say xnk → x0 .

Fix ε > 0. Then since f is continuous, there exists some δ > 0 such that
|x − x0 | < δ =⇒ |f (x) − f (x0 )| < ε.
Since xnk → x, there exists some N1 such that
k ≥ N1 =⇒ |xnk − x0 | < δ.
Since fk → f uniformly, there exists some N2 such that
k ≥ N2 =⇒ kf − fnk k∞ < ε.
Suppose x0 < x < 1. Then there exists some k (in fact, infinitely many) so
that x0 ≤ xnk < x < 1. Hence we can choose k so that we have in addition
that n = nk ≥ N1 , N2 . Then we have |xn − x0 | < δ, so
|f (x) − f (x0 )| ≤ |f (x) − fn (x)| + |fn (x) − fn (xn )| + |fn (xn ) − f (xn )|
+ |f (xk ) − f (x0 )|
≤ kf − fn k∞ + n (x − xn ) + kf − fn k∞ + ε
≤ n (x − xn ) + 3ε.
Since ε is arbitrary, we conclude that f ∈ Fn .

Therefore Fn is closed. However, we can show that Fn has no interior.
Suppose that f ∈ Fn and ε > 0. Since f is uniformly continuous, we can
find some continuous g whose graph consists of finitely many straight lines
such that kf − gk∞ < ε/2. Since g is Lipschitz, there is a finite K such that
|g(x) − g(y)| ≤ K |x − y| for all x, y ∈ [0, 1]. Create a function h that equals g
everywhere except on a suitably small interval, where the graph of h is a
triangle of height ε/2 with the slope of at least one side greater than n. Then
kg − hk∞ < ε/2, so kf − hk∞ < ε. However, h ∈ / Fn . Consequently, ∪Fn
cannot be all of C[0, 1].
Suppose that f ∈ D, i.e., f has a right-hand derivative at some point x0 .
Then there exists some δ > 0 such that
|f (x) − f (x0 )|
x0 < x < δ =⇒ < 1.
x − x0
If δ ≤ x < 1, then we have
|f (x) − f (x0 )| 2 kf k∞
≤ = C.
x − x0 δ
Let n ∈ N be larger than C. Then we have |f (x) − f (x0 )| ≤ n (x − x0 ) for
all x0 ≤ x < 1, so f ∈ Fn . Thus D is contained in ∪Fn , and therefore D is
a proper subset of C[0, 1]. Consequently, there must exist functions in C[0, 1]
that have no right derivative at any point.
2.29 By hypothesis, Ax = limn→∞ An x exists for each x ∈ X. This implies
that A is linear, and also
∀ x ∈ X, sup kAn xk < ∞.

n
By the Uniform Boundedness Principle, we have M = sup kAn k < ∞. Con-

sequently, for each x ∈ X,
kAxk ≤ kAx − An xk + kAn xk

≤ kAx − An xk + kAn k kxk
≤ kAx − An xk + M kxk
→ 0 + M kxk as n → ∞.
Therefore A is bounded and kAk ≤ M.

2.30 The fact that PN x → x for every x follows from Theorem 1.50. However,
PN (eN +1 ) = 0, so keN +1 − PN eN +1 k = 1. Since keN +1 k = 1, this implies that
kI − PN k ≥ 1, and in fact kI − PN k = 1 for every N.
2.31 (a) We are given that Ax = limn→∞ An x exists for each x in the dense
subspace S, and we are also given that R = supn kAn k < ∞. In this case, if
x ∈ S then
kAxk = lim kAn xk ≤ lim sup kAn k kxk ≤ R kxk.

n→∞ n→∞
Therefore A is bounded on S, and consequently by Exercise 1.72 has a unique

extension to a bounded linear mapping on X, which we also call A. Further,
kAk ≤ R.
Now let x be an arbitrary elements of X. Then given ε > 0, we can find
some y ∈ S such that kx − yk < ε. Also, since An y → Ay, there exists some
N > 0 such that kAy − An yk < ε for all n ≥ N. Consequently, for n ≥ N we
have
kAx − An xk ≤ kAx − Ayk + kAy − An yk + kAn y − An xk

≤ kAk kx − yk + ε + kAn k ky − xk
≤ Rε + ε + Rε.
Hence An x → Ax as n → ∞.
(b) Let {ek } be an orthonormal basis for a separable Hilbert space H, and
let S = span{ek }, which is a dense subspace of H. Define
n
X
An x = k hx, ek i ek .
k=1
Then An ∈ B(H), and

n
X n
X
kAn xk2 = |k hx, ek i|2 ≤ n2 |hx, ek i|2 = n2 kxk2 .
k=1 k=1
Also An en = nen , so kAn k = n. Hence supn kAn k = ∞.

PN
Given x ∈ S we have x = k=1 hx, ek i ek for some finite N. Therefore
An x = AN x for all n ≥ N, so Ax = limn→∞ An x = AN x exists. However, A
is not bounded on S, and cannot be extended to a bounded mapping on H.
P
2.32 Case p = 1. Assume that xn yn converges for all x ∈ ℓ1 . Define TN ,
1
T : ℓ → F by
∞
X N
X
Tx = xn yn and TN x = xn yn .
n=1 n=1
Clearly TN is linear, and for x ∈ ℓ1 we have

X
N
|TN x| ≤ |xn | sup |yn | ≤ CN kxkℓ1 ,
n=1 n=1,...,N
where CN = supn=1,...,N |yn | is a finite constant independent of x (though not

independent of N ). Therefore for each N we have TN ∈ B(ℓ1 , F) = (ℓ1 )∗ .
By hypothesis, for each x ∈ ℓ1 we have TN x → T x as N → ∞. The
Banach–Steinhaus Theorem therefore implies that T ∈ B(ℓ1 , F) = (ℓ1 )∗ . Since
(ℓ1 )∗ = ℓP
∞
, we conclude that there must exist some z ∈ ℓ∞ such that T x =
hx, zi = xn zn for all x ∈ ℓ1 . Letting {δn } denote the standard basis vectors
on ℓ1 , we have yn = T δn = zn for every n, so y = z ∈ ℓ∞ .
P ∞
Case p = ∞. Assume that xn yn converges
P for all xP∈ ℓ . Let |αn | = 1
be such that αn yn = |yn | for each n. Then αn yn = |yn | converges, so
y ∈ ℓ∞ .
2.33 (a) If S ⊆ X ∗ is bounded, then, by definition,
R = sup kx∗ k < ∞.

x∗ ∈S
Hence, if x ∈ X then

sup |hx, x∗ i| : x∗ ∈ S ≤ sup kxk kx∗ k| : x∗ ∈ S ≤ R kxk < ∞.

Conversely, suppose that sup |hx, x∗ i| : x∗ ∈ S < ∞ for each x ∈ X.
Then the family
S = x∗ x∗ ∈S ⊆ X ∗ = B(X, F)
satisfies the hypotheses of the Uniform Boundedness Principle, so we conclude
that R = supx∗ ∈X ∗ kx∗ k < ∞. Hence S is a bounded set.
(b) Let π : X → X ∗∗ be the natural map, and note that kxk = kπ(x)k for
every x.
Suppose that S ⊆ X is bounded. Then S ∗∗ = π(x) : x ∈ S is a bounded
subset of X ∗∗ . Since X ∗ is a Banach space, we have by part (a) that for each
x∗ ∈ X ∗ ,

sup |hx∗ , π(x)i| : x ∈ S = sup |hx∗ , x∗∗ i| : x∗∗ ∈ S ∗∗ < ∞.

Conversely, suppose that sup |hx, x∗ i| : x ∈ S < ∞ for each x∗ ∈ X ∗ .
Then for each x∗ ∈ X ∗ we have

sup |hx∗ , x∗∗ i| : x∗∗ ∈ S ∗∗ = sup |hx, x∗ i| : x ∈ S < ∞.
Since X ∗ is a Banach space, part (a) therefore implies that S ∗∗ is a bounded

subset of X ∗∗ . Therefore S is a bounded subset of X.
2.34 We are given 1 ≤ p, q ≤ ∞. For each i ∈ N, set ai = (aij )j∈N . Then by
hypothesis (a), X
(Ax)i = hx, ai i = xj aij
j
′
converges for every x ∈ ℓp . Applying Theorem 2.24, we conclude that ai ∈ ℓp
for each i ∈ N.
(i) Consider the case q < ∞. Define

AN x = hx, a1 i, . . . , hx, aN i, 0, 0, . . .
X∞ X∞
= a1j xj , . . . , aN j xj , 0, 0, . . . .
j=1 j=1
Since q < ∞, we have AN x → Ax as N → ∞. Each AN is linear, and we have

X
N 1/q
kAN xkq = |hx, an i|q
n=1
X
N 1/q
q
≤ kxkℓp kan kℓp′
n=1
X
N 1/q
= kan kqℓp′ kxkℓp
n=1
= CN kxkℓp ,
so AN ∈ B(ℓp , ℓq ). The Banach–Steinhaus Theorem therefore implies that

A ∈ B(ℓp , ℓq ).
(ii) Now consider the case q = ∞. With AN as above, we have
kAN xk∞ = sup |hx, an i| ≤ sup kxkℓp kan kℓp′ = CN kxkℓp .

1≤n≤N 1≤n≤N
Therefore AN ∈ B(ℓp , ℓ∞ ). By hypothesis, Ax ∈ ℓ∞ for each x ∈ ℓp , so for

each such x we have
sup kAN xkℓ∞ = kAxkℓ∞ < ∞.

N
Therefore M = sup kAN k < ∞ by the Uniform Boundedness Principle. Con-

sequently,
kAxkℓ∞ = sup kAN xkℓ∞ ≤ M kxkp ,
N
so A ∈ B(ℓp , ℓ∞ ).
2.35 ⇒. If A is surjective, then range(A) = Y, which is not meager.
⇐. Suppose that range(A) is not meager in Y. Then
S
∞
range(A) = A(BkX (0)).
k=1
By definition of nonmeager, some set A(BkX (0)) must contain an open ball
in Y. Lemma 2.26 therefore implies that A(BkX (0)) contains an open ball. But
then range(A) contains an open ball. Since range(A) is a subspace of Y, it
follows that range(A) = Y. Hence A is surjective.
Remark: The same reasoning shows that if range(A) is not closed, then it
is a meager subset of range(A).
2.36 Suppose that T : X → Y is a topological isomorphism and {xn } is com-
plete in X. Suppose µ ∈ Y ∗ and hT xn , µi = 0 for every n. Then hxn , T ∗ µi = 0
for every n. Since {xn } is complete, it follows that T ∗ µ = 0. But then for
every y ∈ Y we have
hy, µi = hT T −1y, µi = hT −1 y, T ∗µi = 0,
so µ = 0. Hence {T xn } is complete in Y. Since T −1 is also a topological

isomorphism, the converse argument is symmetric.
2.37 Assume T : X → Y is a topological isomorphism. We have kT xk ≤
kT k kxk since T is bounded. Since T −1 is also bounded, we have kT −1 yk ≤
kT −1k kyk for all y ∈ Y. Therefore, for x ∈ X we have
kxk = kT −1 T xk ≤ kT −1 k kT xk,
so the desired inequality follows by rearranging.

2.38 ⇒. Suppose that kLxk ≥ ckxk for all x ∈ X. Suppose that yn ∈ range(L)
and yn → y ∈ Y. Then yn = Lxn for each n, so by hypothesis,
c kxm − xn k ≤ kL(xm − xn )k = kLxm − Lxn k = kym − yn k.
Since {yn }n∈N is Cauchy in Y, this implies that {xn }n∈N is Cauchy in X.
Since X is complete, there must be some x ∈ X such that xn → x. Since L is
continuous, this implies that yn = Lxn → Lx. But we also have yn → y, so we
conclude that y = Lx ∈ range(L). Hence range(L) is closed. The hypotheses
also imply that L is injective, so L is a linear bijection of X onto the Banach
space range(L), and therefore is a topological isomorphism by the Inverse
Mapping Theorem.
⇐. Assume that range(L) is closed and L : X → range(L) is a topological
isomorphism. Then by Exercise 2.37 we have kT −1 k−1 kxk ≤ kT xk ≤ kT k kxk
for all x ∈ X.
2.39 (a) Suppose that λ ∈ ℓ∞ , i.e., λ is a bounded sequence. Then for any f
we have
∞
X ∞
X
|λn hf, en i|2 ≤ kλk2∞ hf, en i|2 = kλk2∞ kf k2 < ∞,
n=1 n=1
so the series defining Mλ f converges (because {en } is an orthonormal se-

quence). Moreover, the preceding calculation also shows that
∞
X
kLf k2 = |λn hf, en i|2 ≤ kλk2∞ kf k2 ,
n=1
so we see that kLk ≤ kλk∞ . On the other hand, by orthonormality we have

Len = λn en (i.e., each en is an eigenvector for L with eigenvalue λn ). Since
ken k = 1 and kLen k = |λn | ken k = |λn | we conclude that
kLk = sup kLf k ≥ sup kLen k = sup |λn | = kλk∞ .

kf k=1 n∈N n∈N
(b) Suppose that λ ∈ / ℓ∞ , i.e., λ is not a bounded sequence. Note that

T δn = λn δn . Therefore kδn kℓ2 = 1 but kT δn k = |λn |, so kT k ≥ |λn |. Hence
kT k = ∞ since λ is unbounded.
(c) If T is a topological isomorphism then T is continuous, and therefore

λ ∈ ℓ∞ by part (a). Now, T −1 is also continuous and T −1 x = λ−1 k xk , so we

must have λ−1 k ∈ ℓ∞ as well. This implies that 0 < inf |λk | ≤ sup |λk | < ∞.
2.40 We are given T ∈ B(X) with kT k < 1. Since
∞
X ∞
X
kT n k ≤ kT kn < ∞,
n=0 n=0
P
the series S = ∞ T n converges absolutely in the Banach space B(X).
PNn=0 n
Set SN = n=0 T . Then SN → S in operator norm, so
N +1 N
X X
ST − n
T = S− T T
n
≤ kS − SN k kT k → 0 as N → ∞.
n=1 n=0
P∞
Hence ST = n=1 T n , with convergence
P∞ of the series in operator norm. A
similar calculation shows that T S = n=1 T n . Therefore
∞
X ∞
X
(I − T ) S = S − T S = Tn − Tn = I
n=0 n=1
and
∞
X ∞
X
S (I − T ) = S − ST = Tn − T n = I.
n=0 n=1
Hence I − T has a continuous left and right inverse, so is topological isomor-

phism of X onto itself.
2.41 We are given a topological isomorphism T : X → Y from a Banach
space X onto a normed space Y. Suppose that {yn } is Cauchy in Y. Then
yn = T xn for some xn ∈ X, and since T −1 is continuous we have that
kxm − xn k = kT −1 ym − T −1 yn k ≤ kT −1k kym − yn k,
so {xn } is Cauchy in X. Therefore xn converges to some x ∈ X, and since T

is continuous we have that yn = T xn → T x. Therefore Y is complete.
2.42 (a) Suppose that T : X → Y is a topological isomorphism. Then
T −1 : Y → X is also a topological isomorphism. We know that T ∗ is bounded
and kT ∗ k = kT k.
Given ν ∈ Y ∗ and g ∈ Y, we have
|hg, νi| = hT T −1g, νi = |hT −1 g, T ∗νi

≤ kT −1 gk kT ∗νk
≤ kT −1 k kgk kT ∗νk.
Hence, kνk ≤ kT −1k kT ∗νk. Thus we have

1
kνk ≤ kT ∗νk ≤ kT k kνk, ν ∈ X ∗.
kT −1 k
Therefore T ∗ is injective and has closed range.
Now we show that T ∗ is surjective. Fix µ ∈ X ∗ , and define ν : Y → C by
hg, νi = hT −1 g, µi, g ∈ Y.
That is, ν = µ ◦ T −1 . Then ν is linear since both µ and T −1 are. Since
|hg, νi = |hT −1 g, µi ≤ kT −1 gk kµk ≤ kT −1k kgk kµk,
we see that ν is bounded. Hence ν ∈ X ∗ , and furthermore
hg, T ∗ νi = hT g, νi = hT −1 T g, µi = hg, µi.
Thus T ∗ ν = µ, so T ∗ is surjective.
Hence, we have shown that T ∗ : Y ∗ → X ∗ is a bounded bijection. The
Inverse Mapping Theorem therefore implies that T is a topological iso-
morphism. Alternatively, to prove this directly, note that if we apply the
above reasoning to the topological isomorphism T −1 : Y → X then we have
that (T −1 )∗ : X ∗ → Y ∗ is a bounded bijection. Therefore, if we have that
(T −1 )∗ = (T ∗ )−1 , then we can conclude that T ∗ is a topological isomorphism.
Therefore, fix µ ∈ X ∗ and g ∈ Y. Then g = T f for some f ∈ X, so we
have

g, (T −1 )∗ µ = T −1 g, µ = hf, µi

= f, T ∗ (T ∗ )−1 µ

= T f, (T ∗ )−1 µ = g, (T ∗ )−1 µ .
Therefore (T −1 )∗ = (T ∗ )−1 , so the result follows.

(b) Now suppose that T : X → Y is an isometric isomorphism. By part (a),
we know that T ∗ is a topological isomorphism, so we just have to show that
it is an isometry.
Fix any µ ∈ Y ∗ . Suppose that f ∈ X. Then
|hf, T ∗ µi| = |hT f, µi| ≤ kT f k kµk = kf k kµk.
Hence kT ∗ µk ≤ kµk.
Also, since T −1 : Y → X is an isometric isomorphism,
|hf, µi| = hT T −1f, µi = |hT −1 f, T ∗ µi ≤ kT −1 f k kT ∗µk = kf k kT ∗µk.
Hence kµk ≤ kT ∗ µk.

2.43 (a), (b) Given y ∈ K, write y = p + e where p = P y ∈ range(A) and
e ∈ range(A)⊥ . Then since B −1 p ∈ ker(A)⊥ we have
AA† y = AB −1 P y = AB −1 p = BB −1 p = p = P y.
Thus AA† = P.
(c) By Theorem 2.32, we have that range(A∗ ) is closed, and there-
fore range(A∗ ) = ker(A)⊥ . Let Q be the orthogonal projection of H onto
range(A∗ ). Given x ∈ H, write x = p + e where p = Qx ∈ range(A∗ ) and
e ∈ ker(A). Then since Ap ∈ range(A) we have P Ap = Ap, so
A† Ax = A† Ap = B −1 P Ap = B −1 Ap = B −1 Bp = p = Qx.
Thus A† A = Q.
2.44 (a) If x ∈ ker(A† ) then B −1 P x = A† x = 0. Since P x ∈ range(A) and
B −1 : range(A) → ker(A)⊥ is a topological isomorphism, this implies that
P x = 0. Since P is the orthogonal projection onto range(A), we conclude that
x ∈ range(A)⊥ .
Conversely, if x ∈ range(A)⊥ then P x = 0 and therefore A† x = B −1 P x =
0, so x ∈ ker(A† ).
(b) Note that P : K → range(A) is surjective and B −1 → range(A) →
ker(A)⊥ is a topological isomorphism, so
range(A† ) = range(B −1 P ) = range(B −1 ) = ker(A)⊥ .
(c) This follows from Theorem 2.33.

(d) Uniqueness. Suppose that C ∈ B(K, H) also satisfies statements (a)–

(c). Then for every y ∈ range(A) we have
AA† y = y = ACy.
Also, ker(A† ) = range(A⊥ ) = ker(C), so if y ∈ range(A)⊥ then
AA† y = 0 = ACy.
As K = range(A) ⊕ range(A)⊥ , we conclude that AA† y = ACy for all y ∈ K.

Thus
A† y − Cy ∈ ker(A), y ∈ K.
However, A† y ∈ range(A) = ker(A)⊥ and similarly Cy ∈ range(A) = ker(A)⊥ ,
so
A† y − Cy ∈ ker(A)⊥ , y ∈ K.
Therefore A† y − Cy = 0 for every y ∈ K.
2.45 (a) If A is positive definite then it is bounded and self-adjoint by def-
inition. If x ∈ H and Ax = 0, then hAx, xi = 0. By definition, this implies
x = 0, so A is injective. Hence A∗ = A has dense range.
(b) If A is a topological isomorphism, then it is surjective by definition.
On the other hand, if A is surjective, then by the preceding remarks it is a
bounded mapping of H onto itself, and hence is a topological isomorphism by
the Inverse Mapping Theorem.
Now consider H = L2 [0, 1] and Af (t) = tf (t). This is bounded since t is
bounded on [0, 1]. It is self-adjoint and positive since
Z
hAf, f i = t |f (t)|2 dt ≥ 0.
If hAf, f i = 0 then we must have t |f (t)|2 = 0 a.e., so f = 0 a.e. Hence A

is positive definite. However, A is not surjective, e.g., the constant function 1
does not belong to the range of A because 1/t ∈ / L2 [0, 1]. Another example is
2
Ax = (xn /n) acting on ℓ .
(c) Suppose that A ∈ B(H) is surjective and positive definite. We are given
2
(x, y) = hAx, yi and |||x||| = (x, x) = hAx, xi. Then (x, x) ≥ 0 for every x and
(x, x) = 0 if and only x = 0 by definition of positive definiteness. We have
(cx, y) = hAcx, yi = c hAx, yi = c (x, y), and
(x + y, z) = hA(x + y), zi = hAx, zi + hAy, zi = (x, z) + (y, z),
so (·, ·) is an inner product.

Since A is bounded with respect to the original norm,
2
|||x||| = (x, x) = hAx, xi ≤ kAxk kxk ≤ kAk kxk2 .
We need to show the complementary inequality.

Since A−1 is bounded, by applying the Cauchy–Bunyakovski–Schwarz
identity to both inner products we have
kxk4 = |hx, xi|2 = |hAA−1 x, xi|2

= |(A−1 x, x)|
2 2
≤ |||A−1 x||| |||x|||
2 2
= hx, A−1 xi |||x||| ≤ kA−1 k kxk2 |||x||| .
2
Consequently kxk2 ≤ kA−1 k |||x||| , and so the two norms are equivalent.
An alternative approach to this inequality is to consider the positive square
root of A. Since A1/2 A1/2 = A, we must have that A1/2 is a topological
isomorphism. We have
2
|||x||| = hAx, xi = hA1/2 x, A1/2 xi = kA1/2 xk2 .
Therefore, the operator norm of A−1 with respect to ||| · ||| is
|||A−1 ||| = sup |||A−1 x|||

|||x|||=1
= sup kA−1/2 xk
kA1/2 xk=1
= sup kA−1 xk = kA−1 k < ∞.

kyk=1
Hence
kxk2 = hAA−1 x, xi = (A−1 x, x) ≤ |||A−1 x||| |||x||| ≤ kA−1 k kxk2 .
2.46 Suppose that {fn }n∈N is a Cauchy sequence with respect to ||| · |||. Since
kfm − fn kX ≤ |||fm − fn |||, it follows that {fn }n∈N is Cauchy with respect to
the original norm in X. Since X is complete, it follows that there exists some
f ∈ X such that fn → f.
Similarly, kAfm − Afn kY ≤ |||fm − fn |||, so {Afn }n∈N is Cauchy in Y.
Since Y is complete, there exists some g ∈ Y such that Afn → g.
The hypotheses therefore imply that g = Af. Hence
|||f − fn ||| = kf − fn kX + kAf − Afn k → 0 as n → ∞.
Therefore X is complete with respect to ||| · |||.

2.47
P We are given 1 ≤ p ≤ ∞ and a sequence of scalars y = (yn ) such that
xn yn converges for all x ∈ ℓp .
Given x ∈ ℓp , let αn be the scalar of unit modulus such that αn xn yn =
|xn yn |. Then αx = (αn xn ) ∈ ℓp , so
X X X
xn yn = αn xn yn = |xn yn |
n n n
converges. Hence xy = (xn yn ) ∈ ℓ1 , so T x = (xn yn ) is a linear map of ℓp into

ℓ1 .
To show that T is continuous,
suppose
that xN → x in ℓp and T xn → z
in ℓ . Write xN = xN (k) , x = x(k) , and z = z(k) . Since ℓq convergence
1
implies componentwise convergence, for each k ∈ N we have limN →∞ xN (k) =

x(k). Therefore
z(k) = lim (T xN )(k) = lim xN (k) y(k) = x(k) y(k).

N →∞ N →∞
Hence z = (x(k) y(k)) = T x, so the Closed Graph Theorem implies that T is

continuous. P
Now define µ(x) = xn yn . Then
X
|µ(x)| ≤ |xn yn | = kT xk1 ≤ kT k kxk1.
n
′ ′
If 1 ≤ p < ∞ then µ ∈ (ℓp )∗ = ℓp , so there exists some z ∈ ℓp such that
µ(x) = hx, zi for x ∈ ℓp . By considering the standard basis vectors, it follows
′
that y = z ∈ ℓp . P
If p = ∞, then by hypothesis αn yn converges whenever |αn | = 1. By
choosing αn yn = |yn |, it follows that y ∈ ℓ1 .
2.48 (a) Just as in the solution to Exercise 2.34, it follows from Theorem 2.24
′
that ai ∈ ℓp for each i.
(b) By hypothesis, A maps ℓp into ℓq . Suppose that xN → x in ℓp and
Axn → z in ℓq . Sincee ℓq convergence
implies componentwise convergence, we
have that AxN = hxN , ai i i∈N converges componentwise to z as N → ∞. On
′
the other hand, since ai ∈ ℓp , we have for each fixed i that hxN , ai i → hx, ai i
as N → ∞. Hence zi = hx, ai i for each i, which says that z = Ax. Therefore
A is continuous by the Closed Graph Theorem.
2.49 Fix fn , f ∈ Cb1 (R) and assume that fn → f uniformly and fn′ → g
uniformly. Given x ∈ (0, 1), we have
Z x Z x Z x

g(t) dt − ′ ′
fn (x) dx ≤ kg − fn kL dx ≤ kg − fn′ kL∞ → 0.
∞
0 0 0
On the other hand, by the Fundamental Theorem of Calculus we have

Z x
fn′ (x) dx = fn (x) − fn (0) → f (x) − f (0).
0
Hence Z x
f (x) = g(t) dt + f (0),
0
and this implies that f ′ = g. Therefore D has a closed graph.

w*
2.50 Let X be a Banach space and suppose that x∗n −→ x∗ . If we also had
w*
x∗n −→ y ∗ , then for each f ∈ X we would have
hf, x∗ − y ∗ i = hf, x∗ i − hf, y ∗ i = lim hf, x∗n i = lim hf, x∗n i = 0.

n→∞ n→∞
Therefore x∗ = y ∗ since they define the same function.

Now, for each f ∈ X we have that hf, x∗n i → hf, x∗ i. Hence for each
f ∈ X we have sup |hf, x∗n i| < ∞. Since {x∗n }n∈N is a family of bounded
linear functionals on the Banach space X, the Uniform Boundedness Theorem
therefore implies that sup kx∗n k < ∞.
2.51 (a) Fix 1 < p < ∞ and xn , y ∈ ℓp . Let {δn }n∈N denote the standard
basis vectors for ℓp .
w ′
⇒. Suppose that xn → y. Since δk ∈ ℓp , we therefore have that
y(k) = hy, δk i = lim hxn , δk i = lim xn (k),

n→∞ n→∞
i.e., xn converges componentwise to y. Also, since we know that weakly con-

vergent sequences are bounded, we have sup kxn kp < ∞.
⇐. Suppose that xn converges componentwise to 0 and that we have K =
′
sup kxn kp < ∞. Choose any z ∈ ℓp and ε > 0. Since p′ < ∞, there exists an
PN
N > 0 such that kz − zN kp′ < ε, where zN = k=1 z(k)δk . Then

lim sup |hxn , zi| ≤ lim sup |hxn , z − zN i| + |hxn , zN i|
n→∞ n→∞
≤ lim sup kxn kp kz − zN kp′ plus lim sup |hxn , zN i|

n→∞ n→∞
N
X
≤ Kε + lim sup |xn (k) z(k)|
n→∞
k=1
N
X
≤ Kε + lim sup |xn (k) z(k)|
n→∞
k=1
= Kε + 0.
w
Since ε is arbitrary, we conclude that hxn , zi → 0. Hence xn → 0. The general
case follows by replacing xn with xn − y.
w
Case p = 1. If p = 1 then the “⇒” argument remains valid, i.e., xn → y in
1
ℓ implies that xn converges componentwisePto y and sup kxn k < ∞.
n
However, the converse fails. Set xn = n1 k=1 δk . Then kxn k1 = 1 for all n
and xn converges componentwise to 0. However, xn does not converge weakly
to 0, for if we take z = (1, 1, 1, . . . ) ∈ ℓ∞ then we have hxn , zi = 1 →
/ h0, zi.
(b) Since ℓp is reflexive for 1 < p < ∞, we only have to consider the cases
p = 1 and p = ∞.
w*
Case p = 1. ⇒. Suppose that xn −→ y in ℓ1 = c0 ∗ . Since δk ∈ c0 , we
therefore have that
y(k) = hy, δk i = lim hxn , δk i = lim xn (k),

n→∞ n→∞
i.e., xn converges componentwise to y. Also, since we know that weak* con-

vergent sequences are bounded, we have sup kxn kp < ∞.
⇐. Suppose that xn ∈ ℓ1 converges componentwise to 0 and that K =
sup kxn k1 < ∞. Choose any z ∈ c0 and ε > 0. Then there exists an N > 0
such that |z(k)| < ε for all k > N. Hence
N
X ∞
X
|hz, xn i| ≤ |z(k)| |xn (k)| + |z(k)| |xn (k)|
k=1 k=N +1
N
X ∞
X
≤ |z(k)| |xn (k)| + ε |xn (k)|
k=1 k=N +1
N
X
≤ |z(k)| |xn (k)| + ε kxn k1
k=1
N
X
≤ |z(k)| |xn (k)| + Kε.
k=1
Therefore
X
N
lim |hz, xn i| ≤ lim |z(k)| |xn (k)| + Kε = Kε.
n→∞ n→∞
k=1
w*
Since ε is arbitrary, we conclude that hz, xn i → 0. Hence xn −→ 0. The general
w* ∗
Case p = ∞. ⇒. Suppose that xn −→ y in ℓ∞ = ℓ1 . Since ek ∈ ℓ1 , we
therefore have that
y(k) = hy, ek i = lim hxn , ek i = lim xn (k),

n→∞ n→∞
i.e., xn converges componentwise to y. Also, since we know that weak* con-

vergent sequences are bounded, we have sup kxn k∞ < ∞.
⇐. Suppose that xn ∈ ℓ∞ converges componentwise to 0 and that K =
sup kxn k∞P< ∞. Choose any z ∈ ℓ1 and ε > 0. Then there exists an N > 0
such that ∞ k=N +1 |z(k)| < ε. Hence
N
X ∞
X
|hz, xn i| ≤ |z(k)| |xn (k)| + |z(k)| |xn (k)|
k=1 k=N +1
N
X ∞
X
≤ |z(k)| |xn (k)| + K |z(k)|
k=1 k=N +1
N
X
≤ |z(k)| |xn (k)| + Kε.
k=1
Therefore
X
N
lim |hz, xn i| ≤ lim |z(k)| |xn (k)| + Kε = Kε.
n→∞ n→∞
k=1
w*
Since ε is arbitrary, we conclude that hz, xn i → 0. Hence xn −→ 0. The general
2.52 Let {xn } be an orthonormal sequence in a Hilbert space H, and choose
any y ∈ H. Then by Bessel’s Inequality, we have
∞
X
|hxn , yi|2 ≤ kyk2 < ∞,
n=1
w
so hxn , yi → 0. Hence xn → 0.
P P
3.1 Theorem 1.49 tells us that cn xn converges if and only if |cn |2 < ∞.
However, the latter sum is a series of nonnegative scalars, so if is converges
then it converges absolutely
P and hence unconditionally.
P Thus, if σ is P any bi-
jection of N, then |2 < ∞ and only if
|cnP |cσ(n) |2 < ∞. Hence cn xn
converges if and only if cσ(n) xσ(n) converges for every bijection σ.
P
3.2 We are given a conditionally convergent series cn of real scalars.
(a) Let (pn ) be the sequence of nonnegative terms of (cn ) in order,Pand
let (qP
n ) be the sequence of negativeP terms of P(cn ) in order. If bothP pn
and qn converge, then If both pn and (−qn ) converge, so (pn −
qn ) converges. But this is a series containing only nonnegative terms, so it
converges absolutely. Therefore P every rearrangement of this P sum converges,
One of these rearrangements is |cn |, so we conclude that cn converges
absolutely and hence unconditionally, which is a contradiction.
P
(b) Fix x ∈ R. Since cn converges, we must have cn → 0. Let (pn ) be
the positive terms of (cn ) in decreasing order, and let (qn ) be the negative
terms of (cn ) in increasing order. Since (pn ) diverges, there is a first integer
M1 such that
S1 = p1 + · · · + pM1 ≥ x.
Then since (qn ) diverges, there is a first integer N1 such that
T1 = p1 + · · · + pM1 + q1 + · · · + qN1 < x.
Continuing in this way (and interleaving any cn that are zero), we obtain a
bijection σ of N, and the corresponding partial sums converge to x because
the terms pn , qn decrease to zero.
(c),
P (d) The proof of Lemma 3.3 constructs a permutation σ of N such
that cσ(n) diverges to ∞. Modifying that proof slightly, we can find an M1
such that
p1 + · · · + pM1 > 1,
then an N1 such that
p1 + · · · + pM1 + q1 + · · · + qN1 < −1,
then an M2 such that
p1 + · · · + pM1 + q1 + · · · + qN1 + pM1 +1 + · · · + pM2 > 1,

P
and so forth. This gives a rearrangement of cn that does not converge and
does not diverge to ±∞.
P
3.3 ⇒. Suppose that kfn k < ∞, and set
N
X N
X
sN = fn and tN = kfn k.
n=1 n=1
Then for N > M we have

X
N N
X

ksN − sM k = fn
≤ kfn k = |tN − tM |.
n=M+1 n=M+1
Since {tN }N ∈N is a Cauchy sequence of scalars, this implies that {sN }N ∈N

is a CauchyP series of vectors in X, and hence converges. By definition, this
∞
means that n=1 fn converges in X.
⇐. Suppose that every absolutely convergent series in X is convergent.
Let {fn }n∈N be a Cauchy sequence in X. Then there exists an N1 such that
1
m, n ≥ N1 =⇒ kfm − fn k < .
2
Set n1 = N1 . Then there exists an N2 such that
1
m, n ≥ N2 =⇒ kfm − fn k < .
22
Without loss of generality, we may assume that N2 > N1 . Let n2 = N2 . Then
we have n1 , n2 ≥ N1 , so
1
kfn2 − fn1 k < .
2
Continuing in this way, we can construct n1 < n2 < · · · such that
∀ k ∈ N, kfnk+1 − fnk k < 2−k .

P
Then k (fnk+1 − fnk ) is absolutely convergent, hence converges, say to f.
Therefore, the partial sums of this series, which are
M
X
gM = (fnk+1 − fnk ) = fnM +1 − fn1 ,
k=1
converge to f, so we have
fnM +1 − fn1 → f as M → ∞.
Hence
fnM → g = f + fn1 as M → ∞.
Thus {fn }n∈N has a subsequence {fnk }k∈N that converges to g ∈ X.
But {fn }n∈N is also Cauchy sequence, so given ε > 0, there exists a K > 0
such that kg − fnk k < ε for all nk > K. Also, there exists an N such that
kfm − fn k < ε for all m, n > N. Suppose that n > N. Then since the nk
are strictly increasing, there exists some nk greater than both K and N. For
this nk we have
kg − fn k ≤ kg − fnk k + kfnk − fn k < ε + ε = 2ε.
Hence fn → g, so X is complete.
P P
3.4 Assume that m n kxmn k < ∞. Choosing any ordering σ of N × N,
P P
we have that k kxσ(k) k < ∞. Thus x = k xσ(k) converges absolutely in X,
hence converges regardless of the ordering. For each n, the series
X
ym = xmn
n
converge absolutely in X. Moreover,

X XX
kym k ≤ kxmn k < ∞,
m m n
so the series X
y = ym
m
also converges absolutely. Now,

N X
N N X
∞ X
X X N X
∞ N X
X N

y −
xmn ≤ y −
xmn + xmn − xmn

m=1 n=1 m=1 n=1 m=1 n=1 m=1 n=1
N
X N
X ∞
X

≤ y −
ym + kxmn k
m=1 m=1 n=N +1
N ∞ X
∞
X X
≤
y − y
m + kxmn k
m=1 n=N +1 m=1
→ 0 as N → ∞.
Let σ be a permutation of N×N obtained by enumerating the “square annuli”

{1, . . . , N + 1} × {1, . . . , N + 1} \ {1, . . . , N } × {1, . . . , N }
in turn. Then
P the preceding calculation says that a subsequence of the partial
sums of k xσ(k) converge to y. However, we know that this series converges
and equals x, so we conclude that y = x and therefore
XX X
xmn = y = x = xσ(k) .
m n k
Further, the final series on the line above is independent of the choice of σ.
3.5 (a) Since the operator norm is submultiplicative, we have

∞ k
X ∞
X ∞
X
A f kAk k kf k kAkk
≤ = kf k = kf k ekAk < ∞.
k! k! k!
k=0 k=0 k=0
Therefore the series defining eA (f ) converges absolutely. Since X is a Banach

space, we conclude that the series defining eA (f ) converges in X.
(b) Since A is linear, it is clear that eA is linear as well. Also, using the
calculation from part (a), we have
∞ ∞ k
X Ak f X A f
keA (f )k =

≤ ≤ ekAk kf k.
k! k!
k=0 k=0
Hence keA k ≤ ekAk < ∞, so eA is bounded.

Alternatively, we can work directly with the operator norm. We have that
∞
X ∞
X
kAk k kAkk
≤ = ekAk < ∞,
k! k!
k=0 k=0
P∞ Ak
so the series eA = k=0 k! converges absolutely in operator norm. Since
B(X) is a Banach space under the operator norm, we conclude that the series
converges in B(X), hence eA ∈ B(X).
(c) Let sN , tN be the N th partial sums for the series defining eA and eB ,
respectively:
XN XN
Ak Bk
sN = and tN = .
k! k!
k=0 k=0
A B
Then since sN → e , tN → e , and {tN }N ∈N is a bounded sequence (since
it converges), we have that
keA eB − sN tN k ≤ keA eB − eA tN k + keA tN − sN tN k

≤ keA k keB − tN k + keA − sN k ktN k → 0.
That is, sN tN → eA eB . On the other hand, we have sN tN = tN sN since A

and B commute, and tN sN → eB eA , so we conclude that eA eB = eB eA .
The series defining eA and eB converge absolutely. Moreover,
k j
∞ X
X
k A B k−j
=
j k!
k=0 j=0
∞ X
X k
k! kAj B k−j k
=
j! (k − j)! k!
k=0 j=0
X∞ X∞
kAkj kBkk−j
≤
j=0
j! (k − j)!
k=j
∞
X ∞
kAkj X kBkk−j
=
j=0
j! (k − j)!
k=j
∞
X ∞
kAkj X kBkk
=
j=0
j! k!
k=0
∞
X kAkj
= ekBk
j=0
j!
= ekAk ekBk < ∞.

Exercise 3.4, combined with the fact that A and B commute, implies that we
can manipulate the series involved as follows:
X∞ X∞ X k j
A+B (A + B)k k A B k−j
e = =
k! j=0
j k!
k=0 k=0
∞ X
X k
k! Aj B k−j
=
j! (k − j)! k!
k=0 j=0
X∞ X∞
Aj B k−j
=
j=0
j! (k − j)!
k=j
∞
X ∞
Aj X B k−j
=
j=0
j! (k − j)!
k=j
∞
X ∞
Aj X B k
=
j=0
j! k!
k=0
∞
X j
A B
= e
j=0
j!
= eA eB .
(d) Suppose that A ∈ B(H) is self-adjoint. Then

∞
X ∞
X
((iA)k )∗ (−iA)k
(eiA )∗ = = = e−iA .
k! k!
k=0 k=0
Since iA and −iA commute, we have by part (c) that

eiA (eiA )∗ = eiA e−iA = eiA−iA = e0 = I
and similarly (eiA )∗ eiA = I. In particular, eiA is bijection. Hence eiA is uni-
tary by Exercise 2.22.
3.6 We are given an orthonormal set {xi }i∈I in a Hilbert space H.
(a) Fix x ∈ H. Suppose that hx, xi i =
6 0 for uncountably many i. Let
Jk = {i ∈ I : |hx, xi i| > 1/k}.
Since ∪Jk = I, some Jk must be uncountable. Let {yn }n∈N be a countable

set of distinct elements of {xi }i∈Jk . Then {yn } is an orthonormal sequence
and |hx, yn i| > 1/k for each n. However, by Bessel’s Inequality we must have
P
|hx, yn i|2 ≤ kxk2 , so we have a contradiction.
(b) By part (a), given x ∈ H there are at most countably many i ∈ I such
that hx, xi i =
6 0. Applying Bessel’s
P Inequality to this countable orthonormal
subset of {xi }i∈I , we obtain i∈I |hx, xi i|2 ≤ kxk2 .
(c) Fix x ∈ H. By part (a), there exists a countable sequence J = {ij }j∈N
/PJ. Since {xij } is a countable orthonormal se-
such that hx, xi i = 0 if i ∈
quence, we know that p = j hx, xij i xij is the orthogonal projection of x
onto span{xij }, and this series converges unconditionally. Fix any ε > 0.
Then there exists some finite F0 ⊆ N such that if F0 ⊆ F where F is finite,
then
X
p − hx, xij i xij
< ε.
j∈F
Let K0 = {ij : j ∈ F0 }. Let F = {j ∈ J : ij ∈ K0 }. Suppose that K is any

finite subset of K that contains F0 . Then since hx, xi i = 0 for any i ∈ K \ J,
we have
X X
p −
hx, xi i xi = p − hx, xij i xij
< ε.
i∈K j∈F
P
Hence the series p = i hx, xi i xi converges with respect to the net of finite
subsets of I, and, as previously stated, is the orthogonal projection of x onto
span{xij }, so hx − p, xij i = 0 for all j. Further, if ℓ ∈ I \ J, then hx − p, xℓ i =
hx, xℓ i − hp, xℓ i = 0. Therefore x − p ∈ {xi }⊥ . Since p ∈ span{xi }, it follows
that p is the orthogonal projection of x onto span{xi }.
3.7
P Since {xi }i∈I is orthogonal, we know from Exercise 3.7 that p =
i∈I hx, xi i xi converges with respect to the net of finite subsets of I and
is the orthogonal projection of x onto span{xi }.
(a) ⇒ (b). If {xi }i∈I is complete, then x = p for every x ∈ H.
P
(b) ⇒ (c). If x = i∈I hx, xi i xi , then since at most countably many terms
are nonzero, we have that
X X
2
kxk = hx, xi = hx, xi i xi , x = |hx, xi i|2 .
i∈I i∈I
(c) ⇒ (a). If hx, xi i = 0 for every i then statement (c) implies that x = 0,
and therefore {xi }i∈I is complete.
′
3.8 (a) We are given that T (µ) = hxn , µi ∈ ℓp for every µ ∈ X ∗ . Consider
the case 1 < p ≤ ∞, so 1 ≤ p′ < ∞. For each N > 0, define

TN (µ) = hx1 , µi, . . . , hxN , µi, 0, 0, . . . .
′
Then TN : X ∗ → ℓp is linear, and
N
X N
X
′ ′ ′ ′ ′
kTN (µ)kpℓp′ = |hxn , µi|p ≤ kxn kp kµkp = CN kµkp ,
n=1 n=1
where
N
X ′
CN = kxn kp < ∞.
n=1
Hence each TN is bounded. Further, for each µ ∈ X ∗ we have by hypothesis

that TN (µ) → T (µ) as N → ∞, so supN kTN (µ)kℓp′ < ∞ for each µ. The
Uniform Boundedness Principle therefore implies that
B = sup kTN k < ∞.

N
Hence,
∞
X N
X
′ ′ ′
kT (µ)kpℓp′ = |hxn , µi|p = lim |hxn , µi|p
N →∞
n=1 n=1
′
= lim kTN (µ)kpℓp′
N →∞
′ ′ ′ ′
≤ sup kTN kp kµkp = B p kµkp .
N
′ ′
Therefore T (µ) ∈ ℓp , and T : X ∗ → ℓp is bounded, with kT k ≤ B.
For the case p = 1, we have p′ = ∞ and
kTN (µ)kℓ∞ = sup |hxn , µi| ≤ sup kxn k kµk = CN kµk,

1≤n≤N 1≤n≤N
where
CN = sup kxn k.
1≤n≤N
Hence each TN is bounded. The Uniform Boundedness Principle therefore

implies that
B = sup kTN k < ∞.
N
Hence,
kT (µ)kℓ∞ = sup |hxn , µi| = sup kTN (µ)k ≤ sup kTN k kµk = B kµk.
n N N
Therefore T (µ) ∈ ℓ∞ , and T : X ∗ → ℓ∞ is bounded, again with kT k ≤ B.

(b) Fix 1 < p < ∞. Choose c = (cn ) ∈ ℓp and sequence of scalars (λn )
with |λn | ≤ 1 for every n. Let T and B be as in part (a). Then for 0 ≤ M < N
we have
N N
X X

λn cn xn = sup hλn cn xn , µi

n=M+1 kµk=1 n=M+1
N
X
≤ sup |λn cn | |hxn , µi|
kµk=1 n=M+1
X
N 1/p X 1/p′
p p′
≤ sup |cn | |hxn , µi|
kµk=1 n=M+1 n
X
N 1/p
p
≤ sup B kµk |cn |
kµk=1 n=M+1
X
N 1/p
p
= B |cn | ,
n=M+1
and a similar estimate holds for p = 1. As long as p < ∞, we conclude that

P
λn cn xn is a Cauchy series and hence converges. Since this is true for all
such choices of λn , we conclude that the series converges unconditionally.
Taking M = 0 and letting N → ∞, we find that
∞ N
X X
kU ck =
c x
n n = lim
c x
n n
N →∞
n=1 n=1
X
N 1/p
p
≤ lim B |cn |
N →∞
n=1
X
∞ 1/p
p
= B |cn | = B kckℓp .
n=1
Therefore U is bounded, with kU k ≤ B.

(c) Now consider the case p = ∞. Given (cn ) ∈ ℓ∞ and µ ∈ X ∗ , we have
Cµ = supn |hxn , µi| < ∞. Therefore, given 0 ≤ M < N and any bounded
sequence of scalars (λn ),
X
N N
X
hλn cn xn , µi ≤ |λn cn | |hxn , µi|

n=M+1 n=M+1
X
N

≤ sup |hxn , µi| |cn |
n
n=M+1
N
X
= Cµ |cn |.
n=M+1
P
Therefore
P the series hcn xn , µi converges unconditionally, which says that
series n cn xn converges weakly unconditionally in X.
Now let X = c0 and let {en } be the standard basis vectors. If µ = (µn ) ∈
(c0 )∗ = ℓ1 then T (µ) = hen , µi =Pµ ∈ ℓ1 , so part (a) implies that T is
bounded. However, it is not true that cn en converges for all (cn ) ∈ ℓp = ℓ∞ .
(d) Assume that 1 ≤ p < ∞. By part (b) we know that U : ℓp → X, so
′
U : X ∗ → (ℓp )∗ = ℓp . Given c = (cn ) ∈ ℓp and µ ∈ X ∗ , we have
∗
X ′
Uc = cn xn ∈ X and T µ = hxn , µi ∈ ℓp .
n
Therefore
X
hc, U ∗ µi = hU c, µi = cn xn , µ
n
X
= cn hxn , µi
n
D E
= (cn ), hxn , µi = hc, T µi.
Therefore T = U ∗ .
(e) Now assume that 1 < p < ∞ and X is reflexive. Then 1 < p′ < ∞, so
p′ ∗ ′
(ℓ ) = ℓp . Therefore, since T : X ∗ → ℓp , we have that T ∗ : ℓp → X ∗∗ = X.
Given c ∈ ℓp and x ∈ X, we have
X D X E
hx, T ∗ ci = hT x, ci = hx, xn i cn = x, cn xn = hx, U ci,
n n
P
where we have used the fact that the series U c = cn xn converges in norm.
Hence U = T ∗ .
∞
3.9 (a) Let {δn } be the sequence
P of standard basis vectors in ℓ . Then for
every finite set F ⊆ N we have k n∈F δn kℓ∞ = 1.PThus R = 1 (and similarly
RE = RΛ = 1). However, by the same reasoning, δn is not a Cauchy series,
hence does not converge in ℓ∞ .
(b) The same argument as in part (a) is valid in the space c, which is
separable.
3.10 Suppose thatP0 < A ≤ kxn k ≤ B < ∞ for some sequence {xn } in a
Hilbert space H. If cn xn converges unconditionally, then we have by Orlicz’s
Theorem that X X
A |cn |2 ≤ kcn xn k2 < ∞,
n
2
so (cn )ℓ .
2
On
P the other hand, if we fix any nonzero vector x ∈ H, then (1/n) ∈ ℓ
but x/n does not converge.
P
3.11 We are given complex-valued functions in Lp (E) such that fn con-
verges
P unconditionally. Write f n = g n + ih n where g n , h n are real-valued. Let
f = fn and similarly write f = g + ih. Given any permutation σ of N, we
have
XN XN

g −
gσ(n)
≤ f − fσ(n)
p,
n=1 Lp n=1 L
P P
so g = gn converges unconditionally, and similarly h = hn converges
unconditionally. Using the fact that the result has been proved for real-valued
functions, we therefore have that
Z X ∞ p/2
|fn (x)|2 dx
E n=1
Z X
∞ p/2
= |gn (x)|2 + |hn (x)|2 dx
E n=1
Z X
∞ ∞
X p/2
2 2
= 2 max |g n (x)| , |h n (x)| dx
E n=1 n=1
Z X
∞ p/2 X
∞ p/2
p/2 2 2
= 2 max |gn (x)| , |hn (x)| dx
E n=1 n=1
Z X
∞ p/2 Z X
∞ p/2
p/2 2 p/2 2
≤ 2 |gn (x)| dx + 2 |hn (x)| dx
E n=1 E n=1
< ∞.
Alternatively, we could simplify the proof by appealing to the Triangle In-
equality in ℓp/2 , noting that 21 ≤ p2 ≤ 1.
P
3.12 Suppose that f = cn fn converges unconditionally in Lp (E). Applying
Orlicz’s Theorem, we have
X X
A2 |cn |2 ≤ kcn fn k2Lp < ∞.
n n
P
Hence |cn |2 < ∞.
P PN
Since f = cn fn converges in P Lp (E), the partial sums SN =P n=1 cn fn
converge in Lp -norm to f. Since |cn |2 < ∞, we know that g = cn fn con-
verges in L (E). Hence SN converges in L2 -norm to g. Since Lp convergence
2
implies the existence of a pointwise almost everywhere subsequence, there is

some subsequence of SN that converges a.e. to f, and a subsequence of that
subsequence that converges a.e. to g. Hence f = g a.e., so f ∈ L2 (E).
3.13 Since 2 < p, we have k · kℓp ≤ k · kℓ2 by Exercise 1.12. Using this and
Tonelli’s Theorem, we therefore have
X XZ
kfn kpLp = |fn (x)|p dx
n n E
Z X p/p
= |fn (x)|p dx
E n
Z X p/2
≤ |fn (x)|2 dx
E n
< ∞,
where the conclusion of finiteness follows from Lemma 3.26.

3.14 Suppose that X is a finite-dimensional normed space. We already know
that all absolutely convergent series in any Banach space are unconditionally
convergent, so we need only prove the converse. Fix a basis {y1 , . . . , yN } for X,
and let {a1 , . . . , aN } be its biorthogonal system. Since all norms on X are
equivalent, we fix the norm as
N
X
kxk = |hx, ak i|.
k=1
P
Suppose that xn converges unconditionally, and let F be any finite
subset of N. With k fixed, we have

X X
hx , a i ≤ |hxn , ak i|
n k
n∈F n∈F
N
XX
≤ |hxn , ak i|
n∈F k=1
X

=
xn
.
n∈F
P
Therefore, with k fixed, the series n hxn , ak i is an unconditionally convergent
series of scalars. Hence it must converge absolutely, so we have
∞
X ∞ X
X N
kxn k = |hxn , ak i|
n=1 n=1 k=1
N X
X ∞
= |hxn , ak i| < ∞.
k=1 n=1
P
Therefore xn converges absolutely.
4.1 Let
S = A ⊆ X : A is finitely linearly independent .
Note that S is nonempty since any singleton {x} with x 6= 0 is linearly inde-
pendent. The inclusion relation ⊆ is a partial order on S.
Suppose that C is a chain in S, say C = {Ai }i∈I where I is some index set.
By definition, each set Ai is finitely independent, and we claim that A = ∪Ai
is also finitely independent. To see this, suppose that x1 , . . . , xn ∈ A. Then
for each k = 1, . . . , n we have xk ∈ Aik for some ik ∈ I. Since C is a chain, it
is linearly ordered by inclusion. Therefore there is a largest set Aik , i.e., there
is a j such that Aik ⊆ Aij for k = 1, . . . , n. Hence x1 , . . . , xn belong to Aij
and therefore are independent. Therefore A ∈ S, and since we have Ai ⊆ A
for each i ∈ I, the set A is an upper bound for the chain C.
Therefore, Zorn’s Lemma implies that S contains a maximal element B.
By definition, B is finitely independent, so if its finite span is X then it is a
Hamel basis for X. Suppose that there exists some x ∈ X \ span(B). Then
B ′ = B∪{x} is finitely independent and hence belongs to S. However, B ( B ′ ,
which implies that B is not a maximal element in S. This is a contradiction,
so we must have X = span(B).
4.2 (a) First proof. Let X be an infinite-dimensional Banach space. By defini-
tion, any Hamel basis Hamel basis for X is infinite. Suppose that {fn }n∈N was
a countable Hamel basis for X. Since {fn }n∈N is finitely linearly independent,
we must have fn 6= 0 for every n, so kfn k 6= 0 for every n.
For each N ∈ N, define
FN = span{f1 , . . . , fN }.
Since FN is finite dimensional, it is a closed subspace of X. Given any f ∈ FN

and r > 0, let
r fN +1
g = f+ .
2 kfN +1 k
Then g ∈/ FN , but kf − gk = r/2 < r. Hence FN contains no open subsets,
and therefore is nowhere dense. But then X = ∪FN is a countable union of
nowhere dense sets, contradicting the Baire Category Theorem.
Second proof, assuming continuous functionals. Let X be an infinite-
dimensional Banach space. By definition, any Hamel basis Hamel basis for X
is infinite. Suppose that {fn }n∈N was a countable Hamel basis for X. By def-
inition, for each f ∈ X there exist unique scalars cn (f ), with at most finitely
many nonzero, such that
∞
X
f = cn (f ) fn ,
n=1
Assume that each cn is a bounded linear functional on X (this is proved in

Section 4.3).
Since X is a vector space, we can define vectors gn in X by
1
gn = fn .
2n kfn k
Then ∞ ∞
X X 1
kgn k = < ∞,
n=1 n=1
2n
P∞
so the series n=1 gn converges absolutely in X. Since X is complete, this
sum must therefore converge in X. Set
∞
X ∞
X 1
g = gn = f .
n kf k n
n=1 n=1
2 n
If we fix m ∈ N then, since am is continuous, we have

∞
X 1 1
am (x) = a (e ) = m
n ke k m n
.
n=1
2 n 2 ke mk
But then we have am (x) 6= 0 for infinitely many m, contradicting the fact
that only finitely many am (x) can be nonzero.
(b) Suppose that S is an infinite-dimensional subspace of X that has a
countable Hamel basis {fn }n∈N . Then just as in part (a) we can write S =
∪Fn where Fn = span{f1 , . . . , fN } is closed because it is finite dimensional.
Hence S is a meager subset of X.
(c) We can write S
Cc (R) = C[−N, N ],
N ∈N
where C[−N, N ] is the set of continuous functions on R that are supported

in [−N, N ]. Each set C[−N, N ] is a closed proper subspace of C0 (R), and
consequently cannot contain any open balls in C0 (R) Hence we have written
Cc (R) as a countable union of nowhere dense sets, so it is a meager subset of
C0 (R).
Suppose that Cc (R) had a countable Hamel basis {fn }n∈N . Fix M ∈ N,
and set
FN = C[−M, M ] ∩ span{f1 , . . . , fN }.
Then FN is a closed subspace of C[−M, M ] since it is finite dimensional, and
since the finite span of {fn }n∈N is all of Cc (R), we have C[−M, M ] = ∪FN .
However, C[−M, M ] is a Banach space, so this contradicts the Baire Category
Theorem. Therefore Cc (R) cannot have a countable Hamel basis.
4.3 We are given a Hamel basis {xi }i∈I for an infinite-dimensional Banach
space X, with associated coefficient functionals {ai }i∈I .
(a) Let {δn } be the standard basis for ℓ1 (any Banach space with a basis
can be substituted in this example). Then {δ1 , . . . , δN } is a Hamel basis for
E = span{δ1 , . . . , δN }. Let M = span{δn }n>N , and let {yi }i∈I be a Hamel
basis for M. Since every vector in X can be written uniquely as x = e + m
where e ∈ E and m ∈ M, it follows that {δ1 , . . . , δN } ∪ {yi }i∈I is a Hamel
basis for X. In fact, every vector x ∈ X can be written uniquely as
X
x = x1 δ1 + · · · + xN δN + bi (x) yi .
i∈I
The first N coefficient functionals x 7→ xn are continuous.

P
(b) We have xj = i∈I ai (xj ) xi for a unique choice of scalars ai (xj ), so
we must have ai (xj ) = δij .
(c) Let J = {i ∈ I : ai is continuous}. Given x ∈ X, we have by definition
that only finitely many ai (x) are nonzero. Hence
∀ x ∈ X, sup kaj (x)k < ∞.

j∈J
Since aj is continuous for each j ∈ J, the Uniform Boundedness Principle

implies that supj∈J kaj k < ∞.
(d) Suppose that the set J was infinite. By selecting a countable subset,
we can without loss of generality simply assume that J = N. Then the series
∞
X xj
x =
j=1
2j kxj k
converges absolutely in X. If k ∈ J then ak is continuous, so by using part (b)

we have that
X∞
ak (xj ) 1
ak (x) = j kx k
= k 6= 0.
j=1
2 j 2 kxkk
But then ak (x) 6= 0 for infinitely many k, contradicting the definition of a

Hamel basis.
(e) Consider the infinite-dimensional normed linear space (c00 , k·kp ) where
1 ≤ p ≤ ∞. The standard basis {δn } is a HamelP basis for this space. Each
vector x ∈ c00 can be written uniquely as x = xn δn with only finitely many
xn 6= 0, and each coefficient mapping x 7→ xn is continuous since |xn | ≤ kxkp .
4.4 (a) We are given that k(x, µ(x))kY = kxkX ≥ 0.
If k(x, µ(x))kY = 0 then kxkX = 0, so x = 0 and therefore µ(x) = 0 since
µ is linear. Hence (x, µ(x)) = (0, 0). By linearity,
kt (x, µ(x))kY = k(tx, µ(tx))kY = ktxk = |t| kxk = |t| k(x, µ(x)kY .
The Triangle Inequality follows since

k(x, µ(x)) + (y, µ(y))kY = kx + y, µ(x + y)kY

= kx + ykX ≤ k(x, µ(x))kY + k(y, µ(y))kY .
Therefore Y is a normed
space.

Suppose that (xn , µ(xn )) is a Cauchy sequence in Y. Then {xn } is a
Cauchy sequence in X, so there exists some x ∈ X such that xn → x. But
then
k(x, µ(x)) − (xn , µ(xn ))kY = kx − xn kX → 0,
so Y is complete.
(b) Since µ is unbounded, we can find xn → x in X such that µ(xn ) does
not converge. Hence, although (xn , µ(xn )) converges to (x, µ(x)) in Y, it does
not converge in the norm of X1 . Therefore the embedding I(x) = x of Y into
X1 is not continuous.
4.5 Let {xi }i∈I be a basis for R over the field Q. Choose any countable
subsequence J = {j1 , j2 , . . . } of I. Define f (xjn ) = n for n ∈ N and f (xi ) = 0
for i ∈ I\J0 . Given a nonzero x ∈ R, by definition of Hamel basis we can
PN
write x = k=1 ck xik for some unique P i1 , . . . , iN ∈ I and unique nonzero
N
rational scalars c1 , . . . , cN . Define f (x) = k=1 ck f (xik ). This f is a Q-linear
function on R, i.e., f (x + ry) = f (x) + rf (y) for all x, y ∈ R and r ∈ Q.
If f was R-linear, then we would have f (x) = ax for some a ∈ R. Hence
m = f (xjm ) = axjm and n = f (xjn ) = axjn , so a 6= 0 and
a a
xj − xj = 0.
m m n n
Therefore
1 1
xj − xj = 0,
m m n n
so {xi }i∈I is not Q-independent, which is a contradiction. Hence f cannot be
R-linear.
P
4.6 (a) Given xP∈ X, there exist unique scalars (cn ) such that x = cn xn .
Therefore x = (cn /λn ) (λn xn ),Pso we need only show P that this represen-
tation is unique. If we have x = dn (λn xn ), then x = (dn λn ) xn , so by
uniqueness dn λn = cn for every n, and therefore dn = cn /λn . Hence {λn xn }
is a basis.
(b) Note that the partial sums associated with {λn xn } are
N
X N
X
SN x = (cn /λn ) (λn xn ) = cn xn ,
n=1 n=1
which are exactly equal to the partial sums associated with {xn }. Similarly,
if we consider a different ordering of the index set, the partial sums do not
change. Hence if {xn } is unconditional, then so is {λn xn }.
(c) By the same reasoning, if {xn } is an absolutely convergent basis, then

X X
k(cn /λn ) (λn xn )k = kcn xn k < ∞,
n n
so {λn xn } is absolutely convergent.
P (a) We already know that an

4.7 Porthonormal basis {en } isP
a basis, and that
cn en converges if and only if |cn |2 < ∞ if and only if cn en converges
unconditionally. Hence {en } is an unconditional basis, and it is normalized
since ken k = 1 for every n.
(b) Let {en } be an orthonormal basis for a separable Hilbert space H.
Define f1 = e1 + e2 and fn = en for n > 1. Then {fn } is not an orthogonal
sequence. Given x ∈ H we have
X ∞
X
x = hx, en i en = hx, e1 i f1 − e2 + hx, fn i fn
n n=2
∞
X
= hx, e1 i f1 − e2 + hx, fn i fn
n=2
∞
X
= hx, e1 i f1 + hx, e2 − e1 i f2 + hx, fn i fn .
n=3
This representation is unique, so {fn } is a Schauder basis for H. In fact, it

is an unconditional basis. On the other hand, {fn } does contain orthonormal
subsequences.
(c) Let {en } be an orthonormal basis for a separable Hilbert space H.
Define
e1
f1 = e1 and fn = en − n−1 for n ≥ 2.
2
Given x ∈ H, note that the series
X∞
hx, en i
e1
n=1
2n−1

converges absolutely since the sequence hx, en i is bounded. Therefore, the
following series converge and can be manipulated as follows:
∞
X ∞
X ∞
X
e1 e1
x = hx, en i en − hx, en i + hx, en i
n=1 n=2
2n−1 n=2
2n−1
∞
X e1 X∞
e1
= hx, en i en − n−1 + hx, e1 i e1 + hx, en i n−1
n=2
2 n=2
2
∞
X X∞
hx, en i
= hx, en i fn + f1 .
n=2 n=1
2n−1
To show that {fn } is a Schauder basis, we must showPthat this is the

unique representation of x. Suppose that we also had x = cn fn . Then for
m ≥ 2 we have
∞
X
hx, em i = cn hfn , em i
n=1
∞
X he1 , em i
= c1 he1 , em i + cn hen , em i − n−1 = cm ,
n=2
2
and for m = 1 we have

∞
X
hx, e1 i = cn hfn , e1 i
n=1
∞
X he1 , e1 i
= c1 he1 , e1 i + cn hen , e1 i − n−1
n=2
2
X∞
cn
= c1 − n−1
.
n=2
2
Consequently,
X∞ X∞
hx, en i hx, en i
c1 = hx, e1 i + n−1
= .
n=2
2 n=1
2n−1
Therefore the coefficients are unique.

Finally, hf1 , f1 i = he1 , e1 i = 1 6= 0 and for n > 1 we have
D e1 E 1
hf1 , fn i = e1 , en − = − 6= 0.
2n−1 2n−1
If m, n > 1, then
D e1 e1 E 1
hfm , fn i = em − , en − = δmn − 6= 0,
2m−1 2n−1 2m+n−2
since m + n − 2 ≥ 2. Hence {fn } contains no orthogonal subsets.
4.8 (a) If x = (xk ) ∈ ℓp then
N p ∞
X X
kx − SN xkpℓp
= x − xk δk |xk |p → 0
= as N → ∞.
k=1 ℓp k=N +1
Uniqueness is clear, so {δn } is a basis for ℓp . Because absolute and uncondi-

tional convergence of series of scalars are equivalent, if x ∈ ℓp and σ is any
permutation of N, then (xσ(k) ) ∈ ℓp , so
N
X p ∞
X

x − xσ(k) δσ(k) |xk |p → 0
p = as N → ∞,
k=1 ℓ k=N +1
so the basis is unconditional.

(b) Define δ0 = (1, 1, 1, . . . ) ∈ c. We will show that {δn }n≥0 is a normalized
unconditional basis for c.
Fix x = (xn )n∈N ∈ c and set x0 = lim xn . Set c0 = x0 and cn = xn − x0
P∞n ∈ N. Since cn → 0 as n ∈ ∞, we have by Exercise 1.20 that the series
for
n=1 cn δn converges in c0 . Moreover, since every permutation (xσ(n) ) belongs
to c0 and since x0 is independent of σ, this series converges unconditionally.
Hence the following series converges unconditionally in c and equals x;
∞
X ∞
X
x = cn δn = x0 δ0 + (xn − x0 ) δn . (A)
n=0 n=1
Clearly this is the unique representation of x in terms of {δn }n≥0 , so this

sequence is an unconditional basis for c. Since kδn kℓ∞ = 1 for every n ≥ 0,
we conclude that it is a normalized basis.
(c) Given x ∈ c, let x0 be as in part (b). Since the series in equation (A)
converges, if µ ∈ c∗ then
∞
X
hx, µi = x0 hδ0 , µi + (xn − x0 ) hδn , µi. (B)
n=1
Moreover, since the series in equation (A) converges unconditionally, it follows

that the series in equation (B) converges unconditionally as well, and since it
is a series of scalars it must therefore converge absolutely.
Alternatively, we can see this directly by considering the vector zN =
(c1 , . . . , cN , 0, 0, . . . ) where cn ∈ F is the scalar of unit modulus such that
P
cn hδn , µi = |hδn , µi|. We have kzN kℓ∞ = 1, and since zN = N n=1 cn δn ,
N
X N
X
hzN , µi = cn hδn , µi = |hδn , µi|.
n=1 n=1
Hence for every N,

N
X
|hδn , µi| = hzN , µi ≤ kzN kℓ∞ kµk = kµk.
n=1
Since this is independent of N, we have

∞
X
|hδn , µi| ≤ kµk < ∞. (C)
n=1
Therefore
∞
X ∞
X
|xn − x0 | |hδn , µi| ≤ 2 kxk ℓ∞ |hδn , µi| ≤ 2 kxkℓ∞ kµk < ∞,
n=1 n=1
so the series in equation (B) converges absolutely.

In any case, we can rewrite equation (B) as
X∞ X∞ ∞
X
hx, µi = x0 hδ0 , µi − hδn , µi + xn hδn , µi = xn ynµ , (D)
n=1 n=1 n=0
where ∞
X
y0µ = hδ0 , µi − hδn , µi and ynµ = hδn , µi.
n=1
Given µ ∈ c∗ , define yµ = (ynµ )n≥0 . Considering equation (C), we have

yµ ∈ ℓ1 . Therefore T (µ) = yµ maps c∗ into ℓ1 , and it is linear. Further, since
|x0 | ≤ kxkℓ∞ , we have by equation (D) that
∞
X
|hx, µi| ≤ |xn ynµ | ≤ kxkℓ∞ kyµ kℓ1 .
n=0
Hence kµk ≤ kyµ kℓ1 .

To show equality, for each n ≥ 0 let cn be a scalar of unit modulus such
that cn ynµ = |ynµ |. Define
N
X
xN = (c1 , . . . , cN , c0 , c0 , c0 , . . . ) = c0 δ0 + (cn − c0 ) δn .
n=1
Then xN ∈ c and kxN kℓ∞ = 1. Further,

N
X
hxN , µi = c0 hδ0 , µi + (cn − c0 ) hδn , µi
n=1
N
X N
X
= c0 hδ0 , µi − hδn , µi + cn hδn , µi
n=1 n=1
N
X N
X
= c0 hδ0 , µi − hδn , µi + |ynµ |
n=1 n=1
∞
X
→ c0 y0µ + |ynµ |
n=1
∞
X
= |y0µ | + |ynµ | = kyµ kℓ1 .
n=1
Therefore kµk = kyµ kℓ1 , so T is isometric.

It remains only to show that T is surjective. Given y ∈ ℓ1 , define µ : c → F
by
∞
X X∞
hx, µi = xn yn = x0 y0 + xn yn , x ∈ c.
n=0 n=1
Since
|hx, µi| ≤ kxk∞ kykℓ1 ,
we have µ ∈ c∗ . If m ∈ N then
∞
X
hδm , µi = 0 · y0 + δmn yn = ym ,
n=1
and for m = 0 we have

∞
X ∞
X
hδ0 , µi = y0 + yn = y0 + hδm , µi,
n=1 n=1
so ∞
X
y0 = hδ0 , µi − hδm , µi.
n=1
Therefore T µ = y, so T is surjective.
4.9 We are given yn = (1, . . . , 1, 0, 0, . . . ). Suppose that x ∈ c0 , and set
cn = xn − xn+1 , n ∈ N.
Note that cn → 0 since x ∈ c0 , and therefore

N
X N
X
cn = (xn − xn+1 ) = x1 − xN +1 → x1 .
n=1 n=1
P
Hence cn converges, and for 1 ≤ k ≤ N we have
N
X
xk − cn = xk − (xk − xN +1 ) = xN +1 → 0.
n=k
PN
Set SN x = n=1 cn yn . Then
N
X X
N N
X
SN x = cn y n = cn , cn , . . . , cN , 0, 0, . . . ,
n=1 n=1 n=2
and therefore
N
X N
X
x − SN x = x1 − cn , x2 − cn , . . . , xN − cN , xN +1 , xN +1 , . . .
n=1 n=2
= (xN +1 , xN +1 , . . . ),
so
kx − SN xkℓ∞ = |xN +1 | → 0.
Consequently,
∞
X
x = cn y n ,
n=1
with convergence of the series in norm.

P
Suppose that we also had x = dn yn . Since norm convergence implies
componentwise convergence, for each fixed k we have
∞
X ∞
X
cn yn = xk = dn yn .
n=k n=k
Hence
∞
X ∞
X ∞
X ∞
X
c1 y 1 = cn y n − cn yn = x1 − x2 = dn yn − dn yn = d1 y1 .
n=1 n=2 n=1 n=2
Consequently c1 = d1 , and by induction we obtain ck = dk for every k.

Thus {yn } is a Schauder basis for c0 , so it remains only to show that this
basis is conditional. Consider
1 1 1
x = (−1)k+1 /k = (1, − , , − , . . . ).
2 3 4
This vector belongs to c0 , and its representation in the basis {yn } is
1 1 1
x = 1+ y1 − + y2 + · · · .
2 2 3
However, the series
1 1 1
1+ y1 + + y2 + · · ·
2 2 3
does not converge componentwise (consider the first component), and there-
fore cannot converge in the norm of c0 . Hence the basis representation of this x
does not converge unconditionally, so {yn } is not an unconditional basis for c0 .
4.10 We are given zn = (0, . . . , 0, 1, 1, . . . ), where the 0 is repeated n − 1
times.
Fix x ∈ c, and let x0 = limn→∞ xn . Define
c1 = x1 and cn = xn − xn−1 for n > 1.

PN
Set SN x = n=1 cn zn . For the first component (m = 1) we have
X
N
(SN x)1 = cn z n = c1 = x1 .
n=1 1
For 2 ≤ m ≤ N, the mth component is

X
N m
X
(SN x)m = cn z n = x1 + (xn − xn−1 ) = xm .
n=1 m n=2
For m > N,
X
N N
X
(SN x)m = cn z n = x1 + (xn − xn−1 ) = xN .
n=1 m n=2
Hence
SN x = (x1 , . . . , xN , xN , xN , . . . ),
so
x − SN x = (0, . . . , 0, xN +1 − xN , xN +2 − xN , . . . ),
Now, xN → x0 , so given ε > 0 there exists an N0 such that
n ≥ N0 =⇒ |xn − x0 | < ε.
Hence for N > N0 we have
kx − SN xkℓ∞ = sup |xn − xN | ≤ 2ε.

n>N
P
Therefore kx − SN xkℓ∞ → 0, so x = cn zn .
P
Suppose that we also had x = dn zn . Since the first component of zn is
zero for each n > 1, we have d1 = x1 = c1 . By induction, dn = cn for every n.
Thus {zn } is a Schauder basis for c, so it remains only to show that this
basis is conditional. Consider the vector
X n
(−1)k+1 1 1 1
x = = 1, 1 − , 1 − + , . . .
k n∈N 2 2 3
k=1
1 1
= z1 − z2 + z3 − · · · .
2 3
The vector x belongs to c since
n
X (−1)k+1
lim = ln 2.
n→∞ k
k=1
However,
1 1 1 1
z1 + z2 + · · · + zn = 1, 1 + , . . . , 1 + · · · + , 0, 0, . . . ,
2 n 2 n
which does not converge in c. Hence the basis representation of this x does
not converge unconditionally, so {zn } is not an unconditional basis for c0 .
4.11 (a) Fix 1 ≤ p < ∞, and let {δn } be the standard basis for ℓp . Given
x ∈ ℓp and N ∈ N, we have
N
X N
X
kSN xkpp = |xn |p ≤ |xn |p = kxkpp .
n=1 n=1
Hence kSN k ≤ 1 for each N, and therefore this basis is monotone.

For p = ∞ and x ∈ c0 we have that
kSN xk∞ = sup{|x1 |, . . . , |xN |} ≤ kxk∞ .
Hence the standard basis for c0 is monotone.

Now let δ0 = (1, 1, . . . ), so {δn }n≥0 is the basis for c given in Exercise 4.8.
As shown
P in that exercise, if x ∈ c and we let x0 = limn→∞ xn , then x =
x0 δ0 + (xn − x0 ) δn . Therefore, given N ∈ N, we have
N
X
SN x = x0 δ0 + (xn − x0 ) δn
n=1
= (x0 , x0 , . . . ) + (x1 − x0 , . . . , xN − x0 , 0, 0, . . . )
= (x1 , . . . , xN , x0 , x0 , . . . ).
Since |x0 | ≤ kxk∞ , we therefore have that
kSN xk∞ = sup{|x1 |, . . . , |xN |, |x0 |} ≤ kxk∞ .
Hence kSN k ≤ 1, so this basis is monotone.

4.12 Let {xnk } be aP subsequence of {xn }. Suppose that x ∈ M = span{xnk }.
Then x ∈ X, so x = cn xn for some unique scalars (cn ). Letting {an } denote
the coefficient functionals for {xn }, we have cm = hx, am i. Now, if m 6= nk
for any k, then hxnk , am i = 0 for every k, and therefore am = 0 on M.
Hence we must have cm = hx, am i = 0 for every P m not in the sequence {nk }.
OmittingPzero terms, the P partial sums of cn xn are therefore the partial
sums of cnk xnk , so x = cnk xnk . Since {ank } is biorthogonal to {xnk }, it
follows that these representations are unique, so {xnk } is a basis for its closed
span.
4.13 We are given a Schauder basis {xn } for a Hilbert space H with dual
system {yn } such that kxn k = kyn k = 1 for every n. By the Polar Identity
and the fact that hxn , yn i = 1, we have
kxn − yn k2 = kxk2 − 2 Re(hxn , yn i) + kyk2 = 1 − 2 + 1 = 0.
Hence xn = yn for every n, so hxm , xn i = hxm , yn i = δmn . Thus {xn } is or-

thonormal, and since we already know that it is complete, it is an orthonormal
basis for H.
4.14 (a) Every element of a Schauder basis must be nonzero, so xn 6= 0 for
every n. Define T : ℓ1 → X by
∞
X yn
Ty = xn .
n=1
kxn k
This series converges absolutely in X because

X∞ X∞
|yn |
kxn k = |yn | = kyk1 < ∞.
n=1
kxn k n=1
Hence T is well defined, and furthermore the calculation above shows that
kT (y)k ≤ kyk1 , so T is bounded.
Suppose that T (y) = 0. Then, by the uniqueness of the coefficients in a
basis expansion, we must have yn /kxn k = 0 for every n, which implies yn = 0
for every n and hence y = 0. Thus T is injective.
Let {an } denote the coefficient functionals
P associated with the basis {xn }.
Suppose that x ∈ X, so we have x = an (x) xn . Set yn = an (x) kxn k. Then,
∞
X ∞
X
|yn | = |an (x)| kxn k < ∞,
n=1 n=1
so y ∈ ℓ1 . Further, T (y) = x, so T is surjective.

Thus, T is a bounded bijection of ℓ1 onto X. Since both ℓ1 and X are
Banach spaces, the Inverse Mapping Theorem implies that T has a bounded
inverse, and hence is a topological isomorphism.
(b) Conversely, suppose that X is a Banach space and T : X → ℓ1 is
1
a topological isomorphism. Let {δn } be the standard basis for ℓP . Then by
−1
Lemma 4.18, {T P δ n } is a basis for X. Given
P x ∈ X, write x = cn T −1 δn .
Then T x = cn δn , so we must have |cn | < ∞. Since kT −1 δn k ≤
−1 −1
kT k kδn k = kT k, we therefore have that
X X X
kcn T −1 δn k = |cn | kT −1 δn k ≤ kT −1 k |cn | < ∞.
n n n
P
Hence the series x = cn T −1 δn converges absolutely, so {T −1an } is an ab-
solutely convergent basis.
4.15 (a) It is easy to see that ||| · ||| is a seminorm. To see that it is a norm,
suppose that kxk = 0. Then for every N we have
X
N

kSN xk = ak (x)xk
= 0.
k=1
In particular, for N = 1 we have
|a1 (x)| kx1 k = ka1 (x) x1 k = 0.
We cannot have x1 = 0, so kx1 k = 6 0 and therefore a1 (x) = 0. By induction

we obtain ak (x) = 0 for all k, so x = 0.
Finally, if x ∈ X, then, by definition, SN x → x with respect to k · k, so
kSN xk → kxk, and therefore
kxk = lim kSN xk ≤ sup kSN xk = |||x|||.

N →∞ N
(b) Suppose that {yn }n∈N is Cauchy with respect to ||| · |||. For each N we
have
kSN ym − SN yn k = kSN (ym − yn )k ≤ |||ym − yn |||.
Since {yn }n∈N is Cauchy with respect to ||| · |||, we conclude that {SN yn }n∈N
is Cauchy with respect to k · k. However, X is complete with respect to k · k,
so this implies that {SN yn }n∈N converges with respect to k · k, i.e., there
exists some zN ∈ X such that limn→∞ kzN − SN yn k = 0. Since each SN yn
belongs to range(SN ) = span{x1 , . . . , xN }, which is closed since it is finite
dimensional, we must have zN ∈ span{x1 , . . . , xN } as well.
(c) Fix ε > 0. Then since {yn }n∈N is Cauchy with respect to ||| · |||, there
exists an n0 such that
m, n > n0 =⇒ |||ym − yn ||| < ε.
Fix any N ∈ N. Then for each n > n0 we have
kzN − SN yn k = lim kSN ym − SN yn k ≤ |||ym − yn ||| < ε.

m→∞
Therefore, for n > n0 ,

sup kzN − SN yn k ≤ ε.
N

This shows that limn→∞ supN kzN − SN yn k = 0.
With ε and n0 as above, fix any particular n > n0 . Now, we know that
SN yn → yn as N → ∞, so {SN yn }N ∈N is Cauchy with respect to k · k.
Therefore, there exists an N0 > 0 such that
M, N > N0 =⇒ kSM yn − SN yn k ≤ ε.
Therefore, for M, N > N0 we have
kzM − zN k ≤ kzM − SM yn k + kSM yn − SN yn k + kSN yn − zN k

≤ ε + ε + ε = 3ε.
Therefore {zN }N ∈N is Cauchy with respect to k · k. Hence there exists some

y ∈ X such that ky − zN k → 0.
(d) Fix N. We have seen that SN +1 yn → zN +1 as n → ∞, with respect
to k · k. Now, SN is linear, and therefore is continuous when restricted to the
finite-dimensional domain span{x1 , . . . , xN +1 }. Each of the vectors SN +1 yn
and zN +1 belong to this domain, so with respect to k · k we have that

SN (zN +1 ) = SN lim SN +1 yn
n→∞
= lim SN (SN +1 yn ) = lim SN yn = zN .

n→∞ n→∞
Hence,
N
X N
X
ak (zN ) xk = zN = SN zN +1 = ak (zN +1 ) xk .
k=1 k=1
By uniqueness, ak (zN ) = ak (zN +1 ) for k = 1, . . . , N. This is true for every N,

so for each k ∈ N we have that
ck = ak (zN ) = ak (zN +1 ) = ak (zN +2 ) = · · · .

PN
Hence zN = k=1 ck xk
with ck independent of N.
PN
(e) By parts (c) and (d), k=1 ck xk = zN → y with respect to k · k. By
definition of infinite series and the fact that {xk }k∈N is a Schauder basis, we
therefore have that
∞
X ∞
X
ck xk = y = ak (y) xk ,
k=1 k=1
with convergence with respect to k · k. By uniqueness, we must therefore have

ck = ak (y) for each k. Hence
N
X N
X
SN y = ak (y) xk = ck xk = zN .
k=1 k=1
Consequently, by part (c) we have
|||y − yn ||| = sup kSN y − SN yn k = sup kzN − SN yn k → 0 as n → ∞.

N N
This shows that X is complete with respect to ||| · |||.

(f) We have seen that k · k ≤ ||| · |||, and that X is complete with respect
to both of these norms. Theorem 2.30 therefore implies that these two norms
are equivalent. Hence there exists a constant C1 > 0 such that
∀ x ∈ X, kxk ≤ |||x||| ≤ C1 kxk.

In particular, for each N we have
kSN xk ≤ |||x||| ≤ C1 kxk,
so SN is bounded and kSN k ≤ C1 . Hence C = supN kSN k < ∞.

Note that the fact that C is finite also follows the Uniform Boundedness
Principle and the fact that supN kSN xk ≤ |||x||| < ∞ for each x.
4.16 (a) Suppose that X has a basis {xn }. Let {SN } denote the associated
partial sum operators, and let C = sup kSN k be the basis constant. Note that
each SN has finite rank and is continuous. Let K ⊆ X be compact.
First proof. The function fN (x) = kx − SN xk is a continuous real-valued
function on X. On the compact set K, it must therefore achieve a maximum,
say at xN . Since K is compact, the sequence {xN } must have a convergent
subsequence, say xNk → y ∈ K. Then
sup kx − SNk xk = sup fNk (x)

x∈K x∈K
= fNk (xNk
= kxNk − SNk xNk k
≤ kxNk − yk + ky − SNk yk + kSNk y − SNk xNk k
≤ kxNk − yk + ky − SNk yk + C ky − xNk k

→ 0 as k → ∞.
Therefore SNk → I uniformly on K. Since SNk has finite rank, we conclude

that X has the approximation property.
Second proof. We improve on the first proof by showing that SN → I
uniformly on K, without needing to select a subsequence of the partial sums.
Since K is compact, it is totally bounded. Hence, given ε > 0 we can find
finite many points y1 , . . . , yM ∈ X such that
S
M
K ⊆ Bε (yk ).
k=1
Since for every x we know that SN x → x as N → ∞, we can find a single N0

such that
∀ N ≥ N0 , ∀ k = 1, . . . , M, kyk − SN yk k ≤ ε.
Given x ∈ K, there exists some k such that kx−yk k < ε. Hence for all N ≥ N0
we have
kx − SN xk ≤ kx − yk k + kyk − SN yk k + kSN yk − SN xk ≤ ε + ε + Cε.

Therefore we have shown that
∀ N ≥ N0 , sup kx − SN xk ≤ (C + 2) ε.
x∈K
Thus SN → I uniformly on K.
(b) Suppose that Y has the approximation property and X is an arbi-
trary Banach space. Let T : X → Y be a compact operator. Then there exist
finite-rank operators TN ∈ B(Y ) that converge uniformly to the identity I on
compact subsets of Y. Let B = {x ∈ X : kxk ≤ 1} be the closed unit sphere
in X. Since T is compact, D = T (B) is a compact subset of Y. Therefore
kT − TN T k = sup kT x − TN T xk
kxk≤1
= sup ky − TN yk
kxk≤1,y=T x
≤ sup ky − TN yk
y∈D
→ 0 as N → ∞.
Therefore TN T are continuous, finite-rank operators that converge to T in

operator norm.
4.17 (a) Let {xn } be a basis for a Banach space X, with associated coefficient
w
functionals {an }. Suppose that xn → y. Since each am is continuous, we then
have
0 = lim δmn = lim hxn , am i = hy, am i.
n→∞ n→∞
P
But since {xn } is a basis for X, this implies that y = hy, an i xn = 0.
(b) Suppose instead that {xn } is a basic sequence and xn → y. Then y is a
weak limit point of M = span{xn }. By Theorem 2.39, every closed subspace
of X is weakly closed, so M must contain y. Therefore {xn } is a basis for M,
w
y ∈ M, and xn → y. Part (a) therefore implies that y = 0.
w
(c) Let {xn } be an orthonormal basis for a Hilbert space H. Then xn → 0
by Exercise 2.52. Alternatively, let {δn } be the standard basis for ℓp , where
′ w
1 < p < ∞. If y ∈ ℓp , then hδn , yi = yn → 0, so δn → 0.
∗
(d) Let {δn } be the standard basis for ℓ1 . Let y = (1, 1, 1, . . . ) ∈ ℓ∞ = ℓ1 .
Then hδn , yi = 1 →
/ 0, so δn does not converge weakly to 0.
4.18 (a) ⇒ (b). Suppose that {xn } is equivalent to {yn }. Then there is a
topological isomorphism T : X → Y P such that T xn = yn for every n. Since
T
P is continuous, the convergence of cn xn in X implies the convergence
P of
cn T xn in Y. Similarly, T −1Pis continuous, so the convergence of cn yn in
Y implies the convergence of cn T −1 yn in X. Therefore statement (b) holds.
4.19 Exercise 2.41 implies that Y is a Banach space and Lemma 4.18 implies
that {yn } is a basis for Y. Since we are given that T xn = yn for every n, the
two bases {xn } and {yn } are equivalent by definition.
4.20 (c) Suppose {xn } is an absolutely convergent basis for a Banach space X
that is equivalent
P to a basis {yn } for a Banach space Y. Given y ∈ Y, we can
write y = cn yn . Let T : X → Y be a topological isomorphismP such that
T xn = yn for every n. By Theorem 4.20, it follows that x = P cn xn , where
x = T −1 y. Since {xn } is absolutely convergent, we must have |cn | < ∞.
Therefore
X X X X
kcn yn k = |cn | kT xn k ≤ kT k |cn | kxn k = kT k kcn xn k < ∞.
Hence {yn } is absolutely convergent as well.

4.21 ⇒. If {xn } is equivalent to {yn }, then there exists a topological isomor-
phism T : X → Y such that T xn = yn for every n ∈ N. Hence
X X X
N N N

cn y n =
cn T xn ≤ kT k = cn T xn
,
n=1 n=1 n=1
and the opposite inequality follows similarly.

⇐. Suppose that the right-hand statement holds. Then we have for any
M < N that
X X X
N N N

C1 cn y n cn y n
≤ cn xn ≤ C2 .
n=M+1 n=M+1 n=M+1
P P
Hence cn xn is Cauchy if and only if cn yn is Cauchy, which implies by
Theorem 4.20 that {xn } and {yn } are equivalent.
4.22 (a) Let {δn } be the standard basis for c0 . If we set δ0 = (1, 1, . . . ), then
{δn }n≥0 is a basis for c. In fact, both of these bases are unconditional.
Given x ∈ c, let x0 = limn→∞ xn . Define T : c → c0 by
T x = (x0 , x1 − x0 , x2 − x0 , . . . ).
Since x 7→ x0 is linear, it follows that T is linear. Also,
kT xk∞ = sup{|x0 |, |xn − x0 |} ≤ sup{|x0 |, |xn | + |x0 |} ≤ 2kxk∞ ,

n n
so T is bounded.
Suppose that T x = 0. Then by definition of T we have x0 = 0, and hence
0 = T x = (0, x1 , x2 , . . . ). Therefore x = 0, so T is injective.
Let y ∈ c0 be given. Then y0 = 0, so if we set x = (0, y1 , y2 , . . . ) then we
have x ∈ c and T x = y. Hence T is injective.
Thus T is a continuous bijection of the Banach space c onto the Banach

space c0 . By the Inverse Mapping Theorem, T is a topological isomorphism.
Finally,
T δ0 = (1, 1 − 1, 1 − 1, . . . ) = (1, 0, 0, . . . ) = δ1 ,
and
T δn = (0, . . . , 0, 1 − 0, 0, 0, . . . ) = δn+1 .
Hence T maps the basis {δn }n≥0 onto the basis {δn }n∈N , so these two bases
are equivalent.
(b) Suppose that x ∈ c0 and kxk∞ = 1. Then |xn | ≤ 1 for every n and
limn→∞ xn = 0. Hence there is some k such that |xk | = 1 and some n such
that |xn | < 1. Then we can find scalars yn , zn not equal to xn and with |yn |,
|zn | < 1 such that xn = yn +z
2
n
. Let ym = xm and zm = xm for all m 6= n. For
m = k in particular we have |yk | = |zk | = |xk | = 1, so kyk∞ = kzk∞ = 1.
Now consider the unit ball in c, and consider the vector x = (1, 1, . . . ).
Suppose that y, z ∈ c satisfy kyk∞ = kzk∞ = 1 and y+z 2 = x. Then yn + zn =
2xn = 2 for every n. But |yn |, |zn | ≤ 1 for every n, so this is only possible if
yn = zn = 1. To see this explicitly, write yn = an + ibn and zn = cn + idn .
Then we have an + cn = 2 and bn + dn = 0, but −1 ≤ an , cn ≤ 1, so this
implies an = cn = 1 and hence bn = dn = 0.
(c) Suppose that T : c0 → c was an isometric isomorphism. Fix x ∈ c0 with
kxk∞ = 1, and let y 6= z ∈ c0 be as constructed in part (b). Then kT xk∞ =
kT yk∞ = kT zk∞ = 1 since T is an isometry. Further, T x = (T y + T z)/2
since T is linear. But then part (b) implies that T y = T x = T z. Since T is
invertible, it follows that y = x = z, which is a contradiction. Therefore no
such isometric isomorphism can exist.
4.23 We have seen that f ∈ C[0, 1] can be written
n
∞ 2X
X −1
f = aχ + bℓ + cn,k sn,k ,
n=0 k=0
with uniform convergence of this series. Every element of the Schauder system
except χ vanishes at t = 0, so f (0) = aχ(0) = a. Hence a is unique. Now
consider n
X∞ 2X −1
f − aχ = bℓ + cn,k sn,k .
n=0 k=0
Each element of the right-hand side vanishes at t = 1 except for the function ℓ.
Therefore f (1) − aχ(1) = bℓ(1), so b = f (1) − a = f (1) − f (0). Hence b is
unique. Now consider
n
∞ 2X
X −1
f − aχ − bℓ = cn,k sn,k .
n=0 k=0
Each element of the right-hand side vanishes at t = 1/2 except for s0,0 . There-
fore
f (1/2) − aχ(1/2) − bℓ(1/2) = c0,0 s0,0 (1/2),
so
c0,0 = f (1/2) − a − b = f (1/2) − f (0) − f (1).
Hence c0,0 is unique. Continuing in this way we obtain the uniqueness of the
representation of f.
4.24 (a) Suppose that fn ∈ C(T) and fn → g uniformly. Then g is contin-
uous. Since we have fn (t + 1) = fn (t) for every t, it follows from pointwise
convergence that g is 1-periodic. Hence g ∈ C(T), so C(T) is a closed sub-
space of C(R) and therefore is a Banach space with respect to the uniform
norm.
(b) Let S = {f ∈ C[0, 1] : f (0) = f (1)}. Define T : C(T) → S by T f =
f |[0,1] . Clearly T is linear and isometric. Given g ∈ S we have g(1) = g(0).
Therefore if we set f (t+k) = g(t) for t ∈ [0, 1) and k ∈ Z, then g is a 1-periodic
function and T f = g. Hence T is surjective, so is an isometric isomorphism.
(c) The proof that Lp (T) and Lp [0, 1] are isometrically isomorphic is sim-
ilar.
(d) Consider fα (t) = tα where α ∈ R. If αp + 1 > 0 then
Z 1 Z 1 1
p αp tαp+1 1
|fα (t)| dt = t dt = = < ∞.
0 0 αp + 1 0 αp + 1
On the other hand, if αp + 1 = 0 then
Z 1 Z 1
|fα (t)|p dt = t−1 dt = ∞,
0 0
and similarly if αp + 1 < 0 then fα is not p-integrable. Hence fα ∈ Lp (T)

exactly for α > −1/p.
If p < q then −1/p > −1/q so there exists an α with −1/p > α > −1/q.
For this α we will have fα ∈ Lq (T) but fα ∈
/ Lp (T).
4.25 Assume that {xn } is a strong basis for X. Then {xn } is a strong Schauder
basis by Theorem 4.13, so the associated coefficient functionals {an } are all
strongly continuous linear functionals on X. We will show that {xn } is a weak
basis and that {an } is the sequence of coefficient functionals associated with
this weak basis. Since we already know that these functionals are strongly
continuous, they are necessarily weakly continuous, and hence it will follow
automatically from this that {xn } is a weak
P Schauder basis.
Therefore, fix any x ∈ X. Then x = hx, an i xn converges strongly. Since
strong convergence implies weak convergence, this series must also converge
weakly to x. Or, to see this explicitly, simply note that if x∗ is an arbitrary
element of X ∗ then, by the continuity of x∗ ,
DX
N E D N
X E
lim hx, an i xn , x∗ = lim hx, an i xn , x∗
N →∞ N →∞
n=1 n=1
DX
∞ E
= hx, an i xn , x∗ = hx, x∗ i.
n=1
P
It therefore remains only to show that the Prepresentation x = hx, an i xn
is unique. Suppose that we also had x = cn xn , with weak convergence of
m. Then since am ∈ X ∗ , we have by the weak
this series. Fix any particularP
convergence of the series x = cn xn that
DX
N E
hx, am i = lim cn xn , am
N →∞
n=1
N
X
= lim cn hxn , am i
N →∞
n=1
N
X
= lim cn δnm = cm .
N →∞
n=1
Hence the representation is unique, and therefore {xn } is a weak basis for X.
4.26 (a) Recall that weakly convergent sequences are bounded
P (Theo-
rem 2.38). Therefore, if (cn ) ∈ Y then k(cn )kY < ∞ since cn xn =
PN
limN →∞ n=1 cn xn converges weakly. The remainder of the proof is now
identical to the proof of Theorem 4.12(a).
(b) Suppose
P that {xn } is a weak basis for X. Define the map T : Y → X
by T (cn ) = cn xn , where this series converges weakly. This mapping is well
defined by the definition of Y. It is clearly linear, and it is bijective because
{xn } is a weak basis. Finally, if (cn ) ∈ Y then
∞ N
X X

kT (cn )k =
cn xn ≤ sup cn xn
N
= k(cn )kY ,
n=1 n=1
again since weakly convergent series are bounded. Therefore T is bounded,

hence is a topological isomorphism of Y onto X.
P
4.27 Suppose that y ∈ ℓ1 and y = cn δn , where the series P converges weak*.
Then since δm ∈ c0 and ℓ1 = c0 ∗ , we have ym = hδm , yi = cn hδm , δn i = cm .
Hence the representation of y is unique.

4.28 (a) Note that y1 = (−1)n+1 .
For m, n > 1 we have
hx1 , y1 i = hδ1 , y1 i = 1,

hxn , y1 i = (−1)n δ1 + δn , y1 = 0,

hxm , yn i = (−1)m δ1 + δm , δn = δmn ,
so {yn } is biorthogonal to {xn }.

Given z = (zn ) ∈ ℓ1 , we have
N
X
SN z = hz, y1 i x1 + hz, yn i xn
n=2
X
∞ ∞
X
= (−1)n+1 zn δ1 + zn (−1)n δ1 + δn
n=1 n=2
X
∞ N
X N
X
n+1 n
= (−1) zn + (−1) zn δ1 + z n δn
n=1 n=2 n=2
∞
X N
X
= z1 + (−1)n+1 zn δ1 + + z n δn .
n=N +1 n=2
Therefore
∞ ∞
X X
kz − SN zkℓ1 =
z n δn − (−1)n+1 zn δ1

n=N +1 n=N +1 ℓ1
X
∞ ∞
X

≤ z n δn + |zn |

n=N +1 ℓ1 n=N +1
→ 0 as N → ∞.
Hence {xn } is a strong basis for ℓ1 .

P
(b) Suppose x ∈ ℓ1 and we had x = Pcn xn for some scalars cn , with
weak*Pconvergence of this series. Since x = hx, yn i xn converges weak*, we
have dn xn = 0 converging weak*, where dn = cn − hx, yn i. In particular,
since δm ∈ c0 , we have for m > 1 that
D N
X E
0 = lim δm , dn xn
N →∞
n=1
N
X
= lim hδm , d1 δ1 i + dn hδm , xn i
N →∞
n=2
N
X
= lim dn hδm , (−1)n+1 δ1 i + hδm , δn i = dm .
N →∞
n=2
P
Therefore dm = 0 for m > 1. Since dn xn = 0, this implies d1 x1 = 0 as well.
However, x1 = δ1 6= 0, so d1 = 0. Hence cn = hx, yn i for every n, so the weak*
representation of x is unique.
(c) The functional y1 is not weak* continuous on ℓ1 because it does not
w*
belong to c0 . For example, although δn −→ 0, we have hδn , y1 i = (−1)n+1 →
/ 0.
4.29 (a) A check of cases shows that hxm , yn i = δmn . Fix z = (zn ) ∈ ℓ1 .
Then
N
X
SN z = hz, y1 i δ1 + hz, yn i xn
n=2
X
∞ N X
X ∞
= z n δ1 + zk (δn − δn−1 )
n=1 n=2 k=n
X
∞ N X
X ∞ N X
X ∞
= z n δ1 + z k δn − zk δn−1
n=1 n=2 k=n n=2 k=n
X
∞ N X
X ∞ N
X −1 X
∞
= z n δ1 + z k δn − z n δn
n=1 n=2 k=n n=1 k=n+1
X
∞ N
X −1 X
∞ ∞
X
= z n δ1 + zk − z n δn
n=1 n=2 k=n k=n+1
X
∞ X
∞
− z k δ1 − z k δN
k=1+1 k=N +1
N
X −1 X
∞
= z 1 δ1 + z n δn − z k δN
n=2 k=N +1
N
X −1 X
∞
= z n δn − z k δN .
n=1 k=N +1
Therefore
∞ X
∞
X
kz − SN zkℓ1
= z n δn + z k δN

n=N k=N +1 ℓ1
∞ ∞
X X
≤
z δ
n n + |zk |
n=N ℓ1 k=N +1
→ 0 as N → ∞.
Hence {xn } is a strong basis for ℓ1 .

P
(b) We claim that xn = 0 with weak* convergence of the series. Given
y ∈ c0 , we have
N
X N
X
hy, xn i = hy, δ1 i + hy, δn − δn−1 i
n=1 n=2
N
X
= y1 + (yn − yn−1 )
n=2
= yN → 0 as N → ∞.
Hence weak* representations of vectors with respect to {xn } are not unique,
so {xn } is not a weak* basis for ℓ1 .
5.1 ⇒. Suppose there is a unique sequence {an } ⊆ X ∗ that is biorthogonal
to {xn }. We know that {xn } is minimal by part (a), so it remains to show
that {xn } is complete. Suppose that x∗ ∈ X ∗ is a continuous linear functional
such that hxn , x∗ i = 0 for every n. Then
hxm , x∗ + an i = hxm , x∗ i + hxm , an i = 0 + δmn = δmn .
Thus {x∗ +an } is also biorthogonal to {xn }. By our uniqueness assumption, we

must have x∗ = 0. Corollary 2.5, which is a consequence of the Hahn–Banach
Theorem, therefore implies that {xn } is complete.
⇐. Suppose that {xn } is both minimal and complete. By part (a) we know
that there exists at least one sequence {an } ⊆ X ∗ that is biorthogonal to {xn },
so we need to show that this sequence is unique. Suppose that {bn } ⊆ X ∗ is
also biorthogonal to {xn }. Then hxn , am − bm i = δmn − δmn = 0 for every m
and n. However, {xn } is complete, so Corollary 2.5 implies that am − bm = 0
for every m. Thus {an } is unique.
5.2 Fix 1 ≤ p < ∞ and ε > 0.
(a) Since C[0, 1] is dense in Lp [0, 1], given f ∈ Lp [0, 1], there exists some
g ∈ C[0, 1] such that kf − gkLp < ε. Since {xk }k≥0 is complete in C[0, 1],
there exists some polynomial q such that kg − qk∞ < ε. Then
Z 1 Z 1
kg − qkpLp = |g(x) − q(x)|p dx ≤ kg − qkp∞ 1 dx = εp .
0 0
Hence kg − qkLp ≤ ε, so kf − qkLp ≤ 2ε. Therefore the space of polynomials

is dense in Lp [0, 1], so {xk }k≥0 is complete in Lp [a, b].
(b) By Exercise 2.4 or the Müntz–Szász Theorem, we know that {x2k }k≥0
is complete in C[0, 1]. Hence there exist polynomials qn whose terms contain
only even degree monomials such that qn converges uniformly to the func-
tion x. As above, we then have that
kx − qn kLp ≤ kx − qn k∞ → 0 as n → ∞,
so x ∈ span{x2k }k≥0 where the closure is in Lp -norm. Therefore {xk }k≥0 is

not minimal in Lp [0, 1].
(c) Since the closure of (xk )k≥0 in the uniform norm is contained in C[0, 1],
this sequence is not complete in L∞ [0, 1].
5.3 (a) Suppose that there exist scalars such that
∞
X
c0 e 1 + cn (en + en+1 ) = 0,
n=1
where the series converges in norm. Applying the Pythagorean Theorem, the
square of the norms of the partial sums of this series are
N
X 2

tN = c0 e1 + (cn en + cn en+1 )
n=1
= kc0 e1 + c1 e1 + c1 e2 + c2 e2 + c2 e3 + · · · + cN eN + cN eN +1 k2
= k(c0 + c1 )e1 + (c1 + c2 )e2 + · · · + (cN −1 + cN )eN + cN eN +1 k2
= |c0 + c1 |2 + · · · + |cN −1 + cN |2 + |cN |2 .
We must have tN → 0 as N → ∞. Since for N ≥ k we have tN ≥ |ck + ck+1 |,

no term |ck +ck+1 | can be strictly positive. Hence ck+1 = −ck for every k ∈ N.
But then tN = |cN |2 , so cN → 0. However, |c0 | = |c1 | = · · · = |cN |, so this
implies that c0 = c1 = · · · = 0. Hence {e1 }∪{en +en+1 }n∈N is ω-independent.
However, this sequence cannot be minimal, because we claim that the
sequence {en + en+1 }n∈N is complete, and hence e1 belongs to its closed span.
To see this, suppose that hx, en + en+1 i = 0 for every n. Then
|hx, en+1 i| = |hx, en i|, all n.

P
However, by Bessel’s Inequality we have |hx, en i|2 ≤ kxk < ∞, so the only
way this is possible is if hx, en i = 0 for every n. Since {en } is an orthonormal
basis, this implies that x = 0. This proves the claim of completeness.
(b) Set f1 = e1 , and for n > 2 define fn = e1 + (en /n). Let F = {fn }n∈N .
For n > 1 we have
en = n (fn − e1 ) = n (fn − f1 ).
Therefore span(E) = span(F), so F is complete.

Suppose that there existed a sequence F = {fen }n∈N that was biorthogonal
to F. Then we would have
he1 , fe1 i = hf1 , fe1 i = 1,
and for n > 1,

D en E 1
0 = hfn , fen i = e1 + , fe1 = 1 + hen , fe1 i,
n n
from which it follows that hen , fe1 i = −n. But then
∞
X ∞
X
kfe1 k2 = |hfe1 , en i|2 = n2 = ∞,
n=1 n=1
which is a contradiction. Therefore F is not minimal.

P
Now suppose that there exist scalars cn such that cn fn = 0. Then
∞
X en X∞
c1 e 1 + cn e 1 + = cn fn = 0.
n=2
n n=1
Therefore, by definition, the partial sums

N
X XN
cn
sN = cn e 1 + en
n=1 n=2
n
converge to zero in norm. By the Pythagorean Theorem,

N N 2 N N
X X cn X 2 X |cn |2
ksN k 2
= cn e 1 +
en = cn + .
n=1 n=2
n n=1 n=2
n2
Therefore
X∞ XN
|cn |2 |cn |2
0 ≤ = lim ≤ lim ksN k2 = 0.
n=2
n2 N →∞
n=2
n2 N →∞
This is only possible if cn = 0 for all n ≥ 2. But then we are left with c1 e1 = 0,
so we must also have c1 = 0 since e1 6= 0. Thus F is ω-independent.
5.4 Define
yn = nen − (n + 1)en+1 .
Then for m, n ∈ N we have
m D
X E
ek
hxm , yn i = , nen − (n + 1)en+1
k
k=1
m
X m
X
nδk,n (n + 1)δk,n+1
= −
k k
k=1 k=1


0 − 0, n > m,
= 1 − 0, n = m,


1 − 1, n < m,
= δmn .
Thus {xn } and {yn } are biorthogonal, so each is minimal.

Note that x1 = e1 , and for n ≥ 1 we have
n+1
X n
X
ek ek en+1
xn+1 − xn = − = = .
k k n+1
k=1 k=1
Hence en ∈ span{xn } for each n ∈ N, which implies that {xn } is complete.

By the Plancherel Equality,

n
X 1 π2
1 ≤ kxn k2 = 2
≤ .
k 6
k=1
Thus {xn } is bounded above and below in norm. The same is not true of {yn }
since the Plancherel Equality tells us that
kyn k2 = n2 + (n + 1)2 .
Therefore {xn } cannot be a Schauder basis since if it was then its dual se-
quence would have to be bounded above and below P as well.
Also, {yn } is incomplete, because if we set x = en /n, then for each m
we have
X∞

1
x, yn = hem , nen − (n + 1)en+1 i = 1 − 1 = 0.
m=1
m
Thus x is orthogonal to {yn }. Again we see that {xn } cannot be a Schauder

basis, because if it was then its dual sequence would also be a Schauder basis
for the Hilbert space H.
5.5 If {xn } is ω-independent
P and for every x ∈ X there exist some scalars
(cn ) such that x = cn xn , thenPit remains only
P to show that these scalars
are unique. But if we have x = dn xn , then (cn − dn ) xn converges and
equals 0, so cn = dn for every n.
5.6 (b) Suppose that {Pn } is a family of partial sum projections and there
exist nonzero vectors x1 ∈ range(P1 ) and xn ∈ range(Pn ) ∩ ker(Pn−1 ) for
n > 1.
Fix N and 1 ≤ n < N. Then xn ∈ range(Pn ), so xn = Pn y for some y ∈ X.
Hence
Pn xn = Pn2 y = Pn y = xn .
Therefore
PN xn = PN Pn xn = Pn xn = xn ,
so xn ∈ range(PN ). Since we know that xN ∈ range(PN ), we conclude that
x1 , . . . , xN ∈ range(PN ).
Also, since xN ∈ ker(PN −1 ) we have
Pn xN = Pn PN −1 xN = Pn 0 = 0.
Therefore
xN ∈ ker(P1 ) ∩ · · · ∩ ker(PN −1 ).
Combining these statements, we see that
(
xm , m ≤ n,
Pn xm =
0, m > n.
P
We claim that {xn } is ω-independent. Suppose that cn xn = 0. Then
since P1 is continuous,
X X
0 = P1 cn xn = cn P1 xn = c1 x1 .
n n
Since x1 6= 0, this implies c1 = 0. Suppose that c1 = · · · = cN −1 = 0. Then

X
∞ ∞
X
0 = PN cn xn = cn PN xn = cN xN ,
n=N n=N
so cN = 0 as well. By induction, cn = 0 for every n.

Since the linearly independent vectors x1 , . . . , xN belong to range(PN ),
which is an N -dimensional space, we conclude that {x1 , . . . , xN } is a basis for
range(PN ). Fix x ∈ X. Then P1 x = c1 x1 for some unique scalar c1 .
Now, P2 x = ax1 + bx2 , for some unique a, b, but since
c1 x = P1 x = P1 P2 x = aP1 x1 + bP1 x2 = ax1 ,
we conclude that a = c1 . Let c2 = b. Continuing

PN in this way, we see there
exist unique scalars (cn )Psuch that PN x = n=1 cn xn . However, PN x → x
by hypothesis, so x = cn xn , and these scalars are unique since {xn } is
ω-independent. Therefore {xn } is a basis for X.
5.7 (a) ⇒ (b). This follows from the definition of basis and the fact that
every basis is a Schauder basis (Theorem 4.13).
(b) ⇒ (a). Assume
P that statement (b) holds. We need only show that the
representation
P x = hx, an i xn is
Punique. However,
P each am is continuous,
so if x = cn xn , then hx, am i = cn hxn , am i = cn δmn = cm .
PN
(b) ⇒ (c). If statement (b) holds then the partial sums n=1 an (x) xn
converge to x, so span{xn } is dense and hence {xn } is complete.
(c) ⇒ (d). This follows from the fact that convergent sequences are
bounded.
(d) ⇒ (e). Each SN is a bounded linear operator mapping X into itself,
so this implication follows from the Uniform Boundedness Principle.
5.8 (a) Since

xm ⊗ yn , am′ ⊗ bn′ HS = xm , am′ yn , bn′ = δmm′ δnn′ ,
we have that {xm ⊗ yn }m,n∈N is biorthogonal to {am ⊗ bn }m,n∈N .

Choose any nonzero T ∈ H ⊗ K = B2 (H, K) and ε > 0. Since the finite

rank operators are dense in B2 (H, K), there exists some operator L of the
form
XN
L = (pk ⊗ qk )
k=1
such that kT − LkHS < ε. We may assume that pk , qk 6= 0 for each k.

Since {xm } is a basis, there exists some M1 > 0 and scalars cm,k such that
M
X ε
pk − cm,k xm
< N kqk k
, k = 1, . . . , N.
m=1
Let X
M1
rk
= cm,k xm

m=1
as long as this quantity is nonzero, otherwise set rk = 1. Since {yn } is a basis,

there exists some M2 > 0 and scalars dn,ℓ such that
M
X
ε
qk − dn,k yn
< N rk
, k = 1, . . . , N.
n=1
Applying Exercise B.6, we have

N M1 X
N X M2
X X
p ⊗ q − c d (x ⊗ y )
k k m,k n,k m n
k=1 k=1 m=1 n=1 HS
N
X X
M1 X
M2

≤ dn,k yn
pk ⊗ qk − cm,k xm ⊗
k=1 m=1 n=1 HS
N
X M1
X
N
X M2
X
M1
X
≤ pk − c x kq k + q − d y c x
m,k m k k n,k n m,k m
k=1 m=1 k=1 n=1 k=1
N
X N
X
ε ε
≤ kqk k + rk
N kqk k N rk
k=1 k=1
= 2ε.
Therefore span{xm ⊗ yn }m,n∈N is dense in B2 (H, K). Thus {xm ⊗ yn }m,n∈N

is both minimal and complete, so is exact.
(b), (c) We are given that {zk }k∈N denotes {xm ⊗ yn }m,n∈N according to
the given ordering of N × N. By part (a), its biorthogonal system is {ck }k∈N
where this is {xm ⊗ yn }m,n∈N following the same ordering of N × N.
Z
By the uniqueness statement in Exercise B.9(b), to show that SN 2 =
X Y
SN ⊗ SN it suffices to check that equality holds when these operators are
applied to simple tensors x ⊗ y. We compute that
2
N
X
Z

SN 2 (x ⊗ y) = x ⊗ y, ck zk
k=1
N X
X N

= x ⊗ y, am ⊗ bn (xm ⊗ yn )
m=1 n=1
N X
X N
= hx, am i hy, bn i (xm ⊗ yn )
m=1 n=1
X
N X
N
= hx, am i xm ⊗ hy, bn i yn
m=1 n=1
X Y
= SN x ⊗ SN y
X Y

= SN ⊗ SN (x ⊗ y).
Therefore equality of operators holds, and Exercise B.9(b) also tells us that
Z X Y
S 2 ≤ SN kSN k ≤ CX CY .
N
Now we will show that

Z X Y X X Y
SN 2 +ℓ = SN ⊗ SN + (SN +1 − SN ) ⊗ Sℓ
for ℓ = 1, . . . , N + 1. This follows from the fact that

2
N
X ℓ
X
Z

SN 2 +ℓ (x ⊗ y) = x ⊗ y, ck zk + x ⊗ y, cN 2 +j zN 2 +j
k=1 j=1
ℓ
X
X Y

= SN ⊗ SN (x ⊗ y) + x ⊗ y, aN +1 ⊗ bj (xN +1 ⊗ yj )
j=1
ℓ
X
X Y

= SN ⊗ SN (x ⊗ y) + hx, aN +1 i hy, bj i (xN +1 ⊗ yj )
j=1
ℓ
X
X Y

= SN ⊗ SN (x ⊗ y) + hx, aN +1 i xN +1 ⊗ hy, bj i yj
j=1
X
ℓ
X Y

= SN ⊗ SN (x ⊗ y) + hx, aN +1 i xN +1 ⊗ hy, bj i yj
j=1
X Y
X X
Y
= SN ⊗ SN (x ⊗ y) + SN +1 x − SN x ⊗ Sℓ y

X Y X X Y
= SN ⊗ SN (x ⊗ y) + SN +1 − S N ⊗ S ℓ (x ⊗ y).
Therefore
Z
S 2 ≤ S X ⊗ S Y + (S X − S X ) ⊗ S Y
N +ℓ N N N +1 N ℓ
X Y X
≤ SN S + S
N
X Y
N +1 − SN Sℓ
≤ CX CY + 2CX CY = 3 CX CY .
Finally, we must show that

Z X Y X Y Y
SN 2 +N +1+ℓ = SN +1 ⊗ SN +1 − Sℓ ⊗ (SN +1 − SN )
for ℓ = 1, . . . , N. We compute that

Z X Y X X Y

SN 2 +N +1+ℓ (x ⊗ y) = SN ⊗ SN + (SN +1 − SN ) ⊗ SN +1 (x ⊗ y)
ℓ
X

+ x ⊗ y, aj ⊗ bN +1 (xj ⊗ yN +1 )
j=1
X Y X X Y

= SN ⊗ SN + (SN +1 − SN ) ⊗ SN +1 (x ⊗ y)
ℓ
X
+ hx, aj i hy, bN +1 i (xj ⊗ yN +1 )
j=1
X Y X X Y

= SN ⊗ SN + (SN +1 − SN ) ⊗ SN +1 (x ⊗ y)
ℓ
X
+ hx, aj i xj ⊗ hy, bN +1 i yN +1
j=1
X Y X X Y

= SN ⊗ SN + (SN +1 − SN ) ⊗ SN +1 (x ⊗ y)
X ℓ

+ hx, aj i xj ⊗ hy, bN +1 i yN +1
j=1
X Y X X Y

= SN ⊗ SN + (SN +1 − SN ) ⊗ SN +1 (x ⊗ y)
+ (SℓX x) ⊗ (SN
Y Y
+1 y − SN y)
X Y X X Y

= SN ⊗ SN + (SN +1 − SN ) ⊗ SN +1 (x ⊗ y)

+ SℓX ⊗ (SNY Y
+1 y − SN y) (x ⊗ y).
Therefore
Z X
S 2 ≤ 3 CX CY + (SN X Y
+1 − SN ) ⊗ Sℓ
N +N +1+ℓ
X
≤ 3 CX CY + SN X Y
+1 − SN Sℓ
≤ 3 CX CY + 2CX CY = 5 CX CY .
5.9 Without loss of generality, we consider g(t) = |t|α and e
g (t) = |t|−α and
the corresponding systems of weighted exponentials in L2 (T) = L2 [− 21 , 12 ].
(a) Since 0 < α < 1/2, we have 0 < 2α < 1 and 1 − 2α > 0. Therefore
Z 1/2 Z 1/2 1/2
2 2t2α+1 2 2−2α−1
|t|α dt = 2 t2α dt = = < ∞,
−1/2 0 2α + 1 0 2α + 1
and
Z 1/2 Z 1/2 1/2
2 2t1−2α 2 22α−1
|t|−α dt = 2 t−2α dt = = < ∞.
−1/2 0 1 − 2α 0 1 − 2α
α −α
Hence |t| and |t| both belong to L2 (T).
Further,

α 2πimt

|t| e , |t|−α e2πint = e2πimt , e2πint = δmn ,
so {|t|α e2πint }n∈Z and {|t|−α e2πint }n∈Z are biorthogonal.

Suppose that f ∈ L2 (T) and f, |t|α e2πint = 0 for every n. Then we have
f (t) |t|α , e2πint = 0 for every n. The function f (t) |t|α belongs to L1 (T),
and Fourier coefficients of functions in L1 (T) are unique, so f (t) |t|α = 0 for
every n. Therefore f (t) |t|α = 0 a.e., which implies that f = 0 a.e. hence
{|t|α e2πint }n∈Z is complete. A similar argument shows that {|t|−α e2πint }n∈Z
is complete.
Note: In fact, since |t|α is bounded, the proof for the completeness of
α 2πint
{|t| e }n∈Z only needs to appeal to the fact that the exponentials are
complete in L2 (T). However, |t|−α is unbounded, so this case requires more
care.
(b) It suffices to consider intervals I that contain the origin, since that is
the only zero or singularity of g or eg . Since functions in L2 (T) are 1-periodic,
it also suffices to consider intervals of length at most 1. So, fix I = [−a, b]
where 0 < a, b < 1.
Because 0 < 1 − 2α < 1 + 2α < 2, we have that
Z b Z a Z b
1 1
|t|−2α dt = t−2α dt + t−2α dt
a + b −a a+b 0 0
1−2α a
1−2α b
1 t
= + t
a + b 1 − 2α 0 1 − 2α 0
1 a1−2α + b1−2α
= .
a+b 1 − 2α
Combining with a similar calculation for the reciprocal exponent, we find that
Z Z
1 2 1 1
|g(t)| dt dt
|I| I |I| I |g(t)|2
Z b Z b
1 2α 1 −2α
= |t| dt |t| dt
a + b −a a + b −a
1 1
= 2
(a1+2α + b1+2α ) (a1−2α + b1−2α )
(a + b) (1 + 2α) (1 − 2α)
1 a2 + a1+2α b1−2α + a1−2α b1+2α + b2

= 2
= (A).
1 − 4α (a + b)2
Now, since 0 < a < 1 and 1 < 1 + 2α < 2, we have a1+2α ≤ a1 = a, and also
a1−2α ≤ 1. Hence we can continue equation (A) as follows:
1 a2 + a + b + b 2 1
(A) ≤ 2 2 2
≤ .
1 − 4α a + 2ab + b 1 − 4α2
Since this quantity is independent of a and b, we conclude that |g(t)|2 = |t|2α
is an A2 weight.
5.10 Let BC = {xn } and BR = {x1 , ix1 , x2 , ix2 , . . . }. Let us write R-span
and C-span to denote spans with respect to the real or complex field, etc.
⇒. Suppose that BC is a basis for X. Then BC is C-complete, so it follows
that BR is R-complete. Also, we have that xn 6= 0 for every n, so to prove
that BR is a basis we can use Theorem 5.17.
Given real scalars an , bn , set cn = an + ibn . Let C be the basis constant
for BC . Then for M ≤ N we have
M M M
X X X
an xn +
bn ixn = cn xn

n=1 n=1 n=1
N
X
≤ C
c x
n n
n=1
X
N

= C (an + ibn ) xn

n=1
X
N N
X
= C
a x
n n + b ix
n n .
n=1 n=1
Theorem 5.17 therefore implies that BR is a basis for XR .

⇐. Suppose that BR is a basis for XR , and let {an , bn } be the associated
coefficient functionals. Then given x ∈ X, we have that
x = hx, a1 i x1 + i hx, b1 i x1 + hx, a2 i x2 + i hx, b2 i x2 + · · · ,
meaning that the partial sums of this series converge to x. Considering the
“even” partial sums, we see that
x = (hx, a1 i + i hx, b1 i) x1 + (hx, a2 i + i hx, b2 i) x2 + · · · .
In order to show that this is the unique representation of x, it suffices to show

that BR is minimal.
Given n ∈ N, we have xn = 1xn + 0ixn . By uniqueness, we conclude that
hxm , an i = δmn and hxm , bn i = 0 for all m, n ∈ N. Therefore {an + ibn } is a
sequence of continuous, complex-valued linear functionals on X that satisfy
hxm , an + ibn i = hxm , an i + ihxm , bn i = δmn .
Hence {xn } ⊆ X has a biorthogonal sequence in X ∗ and therefore is minimal.

5.11 ⇒. If {xn } ⊆ X is a monotone basis for X then its basis constant is
C = 1, and hence the result follows from Theorem 5.12.
PN
⇐. Suppose {xn } is complete, xn 6= 0 for every n, and n=1 cn xn ≤
PN +1
for all N ∈ N and c1 , . . . , cN ∈ F. Then by induction we
n=1 cn xn PN
P M
obtain n=1 cn xn ≤ n=1 cn xn for all M ≤ N. Theorem 5.12 then
implies that {xn } is a basis, and its basis constant satisfies C ≤ 1. Since we
always have C ≥ 1, we conclude that C = 1, so the basis is monotone.
5.12 We will show that the Schauder system is a monotone basis for C[0, 1].
Enumerate the Schauder system as {gn } = {χ, ℓ, s0,0 , s1,0 , s1,1 , . . . }. Fix any
P −1
scalars c1 , . . . , cN and let f = N
n=1 cn gn . We must show that
N −1 N
X X
kf k∞ =
c g
n n ≤
c g
n n = kf + cN gN k∞ (A).
n=1 ∞ n=1 ∞
The cases N = 1, N = 2 are similar to the general case, so fix N ≥ 3. Note

that f is piecewise linear, and gN is supported on an interval I on which f is
linear. Hence f (t) = f (t) + cN gN (t) when t ∈/ I.
On the interval I, the function |f (t)| is maximum at one of the endpoints
of I. However, gN (t) = 0 at each endpoint of I, so |f (t) + cN gN (t)| cannot
achieve a smaller minimum on I. Hence equation (A) follows.
5.13 ⇒. If {xn } is a monotone basis for X, then the sequence of partial sum
operators {SN } has the required properties.
⇐. Assume that {PN } is a sequence of operators satisfying (i)-(v). Let
x1 ∈ range(P1 ) be a unit vector.
Consider the restriction of P1 to range(P2 ). Since range(P1 ) ⊆ range(P2 )
and P12 = P1 , we have a linear operator P1 : range(P2 ) → range(P1 ) that maps
a 2-dimensional space onto a 1-dimensional space. Hence there must be a unit
vector x2 ∈ range(P2 ) such that P1 x2 = 0. Continuing in this way, for each

N > 0 we can find a unit vector xN +1 ∈ range(PN +1 ) such that PN xN +1 = 0.
If 1 ≤ n ≤ N then we have by the nestededness hypothesis (iv) that
xn ∈ range(PN ). Therefore xn = PN yn for some vector yn , and so
PN xn = PN2 yn = PN yn = xn , n = 1, . . . , N.
At this point we could try to use Exercise 5.6 to show that {xn } is a basis,
but we will appeal to Exercise 5.11 instead. By construction, each vector xn is
nonzero. Also by construction, x1 , . . . , xN are linear independent vectors in the
N -dimensional space
S range(PN ), so we have span{x1 , . . . , xN } = range(PN )
for each N. Since range(PN ) is dense, we conclude that span{xn } is dense
in X, so {xn } is complete.
PN +1
Given scalars c1 , . . . , cN +1 we have n=1 cn xn ∈ range(PN +1 ). Also,
PN xn = xn for 1 ≤ n ≤ N and PN xN +1 = 0, so
N N +1 N +1 N +1
X X X X
cn xn cn xn cn xn
= cn PN xn ≤ kPN k = .
n=1 n=1 n=1 n=1
Therefore the hypotheses of Exercise 5.11 are fulfilled, so {xn } is a monotone

basis for X.
5.14 (a) Set f (t) = (1 + t)p and g(t) = 2p−1 (1 + tp ). For t = 1 we have
f (1) = 2p ≤ 2p−1 (1 + 2p ) = g(1). For t ≥ 1 we have
f ′ (t) = p (1 + t)p−1 ≤ p (2t)p−1 = 2p−1 ptp−1 = g ′ (t).

Hence f (t) ≤ g(t) for all t ≥ 1.
(b) If a = 0 then
|0 + b|p ≤ 2p−1 (0p + |b|p ),
and a symmetric argument applies if b = 0.
Consider the case 0 < a ≤ b. By part (a),
(a + b)p 2p−1 (ap + bp )
p p−1 p p
p
= 1 + (b/a) ≤ 2 1 + b /a = ,
a ap
so the desired inequality follows. A symmetric argument applies if 0 < b ≤ a.
Now let a, b be arbitrary complex numbers. Applying the preceding case,
|a + b|p ≤ (|a| + |b|)p ≤ 2p−1 (|a|p + |b|p ).
(c) We have

2 |c|p = 21−p |2c|p = 21−p |(c + cN ) + (c − cN )|p ≤ |c + cN |p + |c − cN |p .
′
5.15 (a) Since hn is bounded, it belongs to Lp [0, 1] = Lp [0, 1]∗ for each
1 ≤ p < ∞. Since hhm , hn i = δmn , the Haar system is its own biorthogonal
system.
(b) By definition, {hn } is complete in its closed span in L∞ [0, 1], and
hn 6= 0 for every n. We will appeal to Theorem 5.17 to show that {hn } is a
basis for its closed span.
Given N ∈ N and scalars c1 , . . . , cN , just as in the discussion in Section 5.5,
if we set
N
X −1 XN
gN −1 = cn h n and gN = cn h n
n=1 n=1
then gN −1 and gN agree except possibly on the dyadic interval I where hN is
nonzero. Let I1 , I2 denote the left and right halves of I. Then gN −1 takes a
constant value c on I, while gN = c + dN on I1 and gN = c − dN on I2 for an
appropriate constant dN . Hence
sup |gN −1 (t)| = |c| and sup |gN (t)| = max{|c + dN |, |c − dN |}.
t∈I t∈I
Since
|2c| = |c + dN + c − dN | ≤ |c + dN | + |c − dN |,
we must have at least one of |c + dN | ≥ |c| or |c − dN | ≥ |c|. Hence |gN −1 (t)| ≤
/ I. Consequently, kgN −1 kL∞ ≤
|gN (t)| for each t ∈ I, and they are equal for t ∈
kgN kL∞ , so Theorem 5.17 implies that {hn } is a basis for its closed span.
(c) The characteristic function of every dyadic interval [ 2kn , k+1
2n ) belongs to
span{hn }. If f ∈ C[0, 1] then f is uniformly continuous, so we can approximate
it as closely as we like in the uniform norm by a step function of the form
n
2X −1
g = c k χ[ k
, k+1 .
2n 2n )
k=0
Hence C[0, 1] ⊆ span{hn }. Since {hn } is a basic sequence, each f ∈ C[0, 1]

has a Haar basis representation. Now, hhm , hn i = δmn , and hn ∈ L1 [0, 1] ⊆
L∞ [0, 1]∗ , so {hn } is its own biorthogonal system as a basis for span{hn }
P
in L∞ [0, 1]. Therefore the basis representation of f ∈ C[0, 1] is hf, hn i hn ,
where the series converges uniformly.

5.16 (a) Since supp(ψn,k ) = 2−n k, 2−n (k + 1) , we compute that
Z

|hf, ψn,k i| = f (x) 2 ψ(2 x − k) dx
n/2 n
Z 2−n (k+ 21 ) Z 2−n (k+1)

n/2

= 2 f (x) dx − 2 n/2
f (x) dx
2−n k 2−n (k+ 21 )
Z 2−n (k+ 21 )

= 2n/2 f (x) − f x + 2−n · 1
2 dx
2−n k
Z 2−n (k+ 21 )
n/2
≤ 2 f (x) − f x + 2−n−1 dx
2−n k
Z −n
2 (k+ 21 ) α
≤ 2n/2 K 2−n−1 dx
2−n k
= 2n/2 2−n−1 K 2−nα−α
= 2n/2 2−n 2−1 K 2−nα 2−α
= 2−1−α K 2−nα−n/2 ,
(b) If f is Hölder continuous at x, then the same conclusion holds for

those n and k such that

supp(ψn,k ) = 2−n k, 2−n (k + 1) ⊆ (x − δ, x + δ).
In essence, the conclusion holds for all n large enough, and those k such that
ψn,k is “localized at x.”
5.17 (a) Assume that {xn } is a bounded basis for X. Then, by defini-
tion, 0 < inf kxn k ≤ sup kxn k < ∞. Further, by Theorem 4.13 we have
1 ≤ kan k kxn k ≤ 2C, where C is the basis constant for {xn }. Therefore
0 < inf kan k ≤ sup kan k < ∞. Combined with Theorem 5.21, this implies
that {an } is a bounded basis for its closed span in X ∗ .
(b) Consider the Babenko example {|t|1/4 e2πint }n∈Z , which is a Schauder
basis for L2 [− 21 , 12 ]. Note that
Z 1/2 Z 1/2 1/2
1/4 2πint 2 2t3/2 4 1 21/2
|t| e dt = 2 t1/2 dt = 2 = = .
3 0 3 2 3/2 3
−1/2 0
The dual system is {|t|−1/4 e2πint }n∈Z , and we have

Z 1/2 Z 1/2 1/2
−1/4 2πint 2 1
|t| e dt = 2 t−1/2 dt = 2 · 2t1/2 = 4 1/2 = 23/2 .
−1/2 0 0 2
Since these two quantities are not reciprocals, if we normalize {|t|1/4 e2πint }n∈Z
then the dual system will not be normalized.
5.18 Let {xn } be a minimal system in a Banach space X with a biorthogonal
system {an } that is a basis for X ∗ . Then {π(xn )} ⊆ X ∗∗ is biorthogonal to
{an } ⊆ X ∗ , so Theorem 5.21 implies that {π(xn )} is a basis for its closed
span in X ∗∗ .
Suppose that µ ∈ X ∗ and hxn , µi = 0 for every n. Since µ ∈ X ∗ and {an }
is a basis for X ∗ , we have that
X
X
µ = µ, π(xn ) an = hxn , µi an = 0.
n n
Therefore {xn } is complete in X.

Now, π : X → π(X) ⊆ X ∗∗ is an isometry. Since {xn } is complete in X,

it follows that {π(xn )} is complete in π(X). Since {π(xn )} is a basis for its
closed linear span, by applying the isometry π −1 : π(X) → X, we see that
{xn } is a basis for X.
P
5.19 (a) Given scalars c1 , . . . , cN , set x = N
n=1 cn xn . Then
N N
X X

cn (xn − yn ) = hx, an i (xn − yn )

n=1 n=1
N
X
≤ |hx, an i| kxn − yn k
n=1
N
X
≤ kxk kan k kxn − yn k
n=1
X
N
≤ λ kxk =
c x
n n .
n=1
The result therefore follows from Theorem 5.24.

(b) Let {en } be the standard basis for ℓ2 . Set y1 = 0 and yn = en for
n ≥ 2. Then
N X
N 1/2 N
X X
cn (en − yn ) = kc1 e1 k = |c1 | ≤ 2
|cn |
= cn e n
,
n=1 ℓ2 n=1 n=1
so λ = 1 in Theorem 5.24. Also, since an = en for all n, we have

X
kan k ken − yn k = ke1 k ke1 − y1 k = 1,
so λ = 1 in Corollary 5.25 as well. However, {yn } is not a basis for ℓ2 , and

furthermore it is not a basis for span{yn } since y1 = 0.
P
5.20 (a) Given a basis {xn }, let εn be P small enough that εn kan k < 1.
Then whenever kxy − yn k < εn , we have kxn − yn k kank < 1, so {yn } is a
basis equivalent to {xn } by Corollary 5.25.
(b) Let {xn } be a basis for X, and let εn be as part (a). If E ⊆ X is dense,
then there exist yn ∈ E such that kxn − yn k < εn . Therefore {yn } is a basis
for X by part (a).
(c) By the Weierstrass Approximation Theorem, the set of polynomials is
dense in C[0, 1]. Hence the result follows from part (b).
5.21 (a) We have x ∼ x since x − x = 0 ∈ M.
If x ∼ y then x − y ∈ M, so y − x ∈ M and hence y ∼ x.
If x ∼ y and y ∼ z then x − y ∈ M and y − z ∈ M, so x − z = (x − y) +

(y − z) ∈ M and therefore x ∼ z.
Thus ∼ is an equivalence relation on V. By definition, the equivalence class
of x ∈ V is
[x] = {y ∈ V : x ∼ y} = {y ∈ V : x − y ∈ M }
= {x − m : m ∈ M }
= {x + m : m ∈ M } = x + M.
(c) Suppose that x + M = y + M. Then x − y ∈ M = ker(T ), so T x = T y.

Therefore ψ(x + M ) = T x is well defined.
Since T is linear, we have
ψ(ax + by + M ) = T (ax + by) = aT x + bT y = aψ(x + M ) + bψ(y + M ),
so ψ is linear.
If ψ(x + M ) = 0 then T x = 0 so x ∈ M and hence x + M = M, the zero
element of X/M. Hence ψ is injective.
By definition, range(ψ) = X/M, so ψ is surjective.
(d) Let M be a subspace of a vector space V, and let N ⊆ V be a sub-
space such that M ∩ N = {0} and M + N = V. Then each x ∈ V can be
written uniquely as x = m(x) + n(x) where m(x) ∈ M and n(x) ∈ N. The
mapping n : V → N is linear and surjective, and n(x) = 0 if and only if
x = m(x) ∈ M, so ker(n) = M. By the Isomorphism Theorem, there exists
a linear bijection ψ : V /M → N. Since linear bijections are vector space iso-
morphisms, they preserve dimension. Hence dim(N ) = dim(V /M ), and this
number is independent of N.
6.1 (a) ⇒ (b). Assume that {xn } is an unconditional basis for X,P and let σ be
any permutation of N. Choose any xP ∈ X. Then the series x = hx, an i xn
converges unconditionally, so x = hx, aσ(n) i xσ(n) converges by Corol-
lary 3.11. We must show that this is the unique
P representation of x in terms
of {xσ(n) }. Suppose that we also had x = cn xσ(n) for some scalars (cn ).
Then, since {xn } and {an } are biorthogonal and aσ(m) is continuous, we have
X X X
hx, aσ(m) i = cn hxσ(n) , aσ(m) i = cn δσ(n),σ(m) = cn δnm = cm .
n n n
which shows that the representation is unique.

(b) ⇒ (a). Assume that {xσ(n) } is a basis for every permutation σ of N.
Let {an } be the sequence of coefficient functionals associated withP the basis
{xn }. We must show that for each x ∈ X the representation x = hx, an i xn
converges unconditionally. Fix any permutation σPof N. Since {xσ(n) } is a
basis, there exist unique scalars cn such that x = cn xσ(n) . However, each
am is continuous and {an } is biorthogonal to {xn }, so
X X X
hx, aσ(m) i = cn hxσ(n) , aσ(m) i = cn δσ(n),σ(m) = cn δnm = cm .
n n n
P P
Therefore P
x= cn xn = hx, aσ(n) i xσ(n) converges for every permutation
σ, so x = hx, an i xn converges unconditionally.
6.2 (a) If σ is any permutation of N then we know that {xσ(n) } is a basis for
X. However, bases are preserved by topological isomorphisms (Lemma 4.18),
so {T xσ(n) } is a basis for Y. As this is true for every σ, the basis {T xσ(n) } is
unconditional.
(b) In light of part (a), we need only show that {T xn } is bounded if
{xn } is bounded. This follows from the facts kT xn k ≤ kT k kxn k and kxn k =
kT −1T xn k ≤ kT −1 k kT xn k.
P
6.3 (a), (c), (d), (e). These follow from the fact that x = hx, an i xn and
this series converges unconditionally.
(b) Follows from (a) by the Uniform Boundedness Principle.
(f) Follows from (a) and (b).
6.4 (a) Given any permutation σ of N, we have by Lemma 6.1 that {xσ(n) }
is a basis for X with coefficient functionals {aσ(n) }. By Theorem 5.21, {aσ(n) }
is a basis for its closed linear span Mσ . However, spans and closed spans are
independent of ordering, so M = Mσ is independent of σ. Therefore {aσ(n) }
is an unconditional basis for M.
(b) If X is reflexive, then we have in part (a) that {aσ(n) } is a basis for X ∗
for every permutation σ, hence is an unconditional basis for X ∗ .
6.5 (a) Fix 1 ≤ p < 2 and set en (t) = e2πint . Suppose that {en }n∈Z was
an unconditional basis for Lp (T). Since the exponentials are biothogonal to
P
themselves, it follows that f = n∈Z fb(n) en converges unconditionally for
each f ∈ Lp (T). However, since ken kLp = 1 for every n, we have by Theo-
P b
rem 3.28, which is a corollary to Orlicz’s Theorem, that |f (n)|2 < ∞ and
2 2 p
f ∈ L (T). Since L (T) is a proper subset of L (T) when p < 2, this is a
contradiction.
(b) Fix 2 < p < ∞. If {en }n∈Z is an unconditional basis for Lp (T), then
since Lp (T) is reflexive we would have that its biorthogonal system, which is
′
{en }n∈Z , would be an unconditional basis for Lp (T) by Exercise 6.4. This
contradicts part (a).
6.6 It is not hard to see that ≤ is a partial order on X and that requirements
(a) and (b) in Definition 3.35 are satisfied.
P P
(c) Given
P x = anP xn and y = bn xn in X, set cn = max{an , bn }. Since
the series
P an xn and
P bn yn converge unconditionally, Theorem P 3.10 implies
that |an |xn and |bn |xn converge unconditionally. Hence (|an |+ |bn |)xn
converges unconditionally.
P Since |cn | ≤ |an |+|bn |, it follows from Theorem 3.10
that z = cn xn converges unconditionally. We will show that z is a least
upper bound for x, y.
Since an , bn ≤ cn , by definition of the ordering we have x, y ≤ z. P Suppose
that x, y ≤ w. Since {xn } is an unconditional basis, we can write w = dn xn .
By definition of the order, an , bn ≤ dn for every n. Hence cn = max{an , bn } ≤
dn , so z ≤ w. Therefore z is a least upper bound. The construction of a greatest
lower bound is similar.
P P
(d)
PFix x = an xn and y = bn xn in X. As above, we know that
p = |an |xn converges unconditionally. Further, since scalars are real we
have |aPn | = max{an , −an }, so p = |x| in the sense of Definition 3.35. Similarly,
|y| = |bn |xn .
Suppose that |x| ≤ |y|, which means that |an | ≤ |bn | for every n. Then by
criterion (c) of Theorem 6.7, for each N we have
N N
X X

an xn ≤ KΛ bn xn
.
n=1 n=1
However, F = R, so we know that KΛ = KE = 1. Letting N → ∞, we

conclude that
∞ ∞
X X
kxk =
a x
n n ≤
b
n n = kyk.
x
n=1 n=1
6.7 No. Let h1 = χ[0,1] and h2 = χ[0,1/2) − χ[1/2,1) . Using the ordering from
Exercise 6.6 we have
χ[0,1] = h1 + 0h2 ≤ h1 + h2 = 2χ[0,1/2) .
However, the functions χ[0,1] and 2χ[0,1/2) are not comparable using the point-
wise a.e. comparison.
6.8 (a) We have a1 = 1, a2 = 3/2, and
aN −1 + aN −2
aN = 1 + .
2
We claim that aN ≥ N/2 for every N. This is true for N = 1 and N = 2, and
if it is true for all integers between 1 and N for some N ≥ 2 then we have
aN + aN −1 1 N N − 1
aN +1 = 1 + ≥ 1+ +
2 2 2 2
N 1 N 3 N +1
= 1+ − = + ≥ .
2 4 2 4 2
Hence the claim follows by induction, and therefore aN → ∞.
(b) If we set bN = hN (µN ), then we have

 bN −1 + bN −2

 + 1, N odd,
2
bN =

 bN −1 + bN −2 − 1, N even.

2
We claim that 1 ≤ bN ≤ 2 for N odd and −1 ≤ bN ≤ 0 for N even.
We have b1 = 1 and b2 = −1/2. Suppose the claim holds for all integers
between 1 and N for some N ≥ 2. If N is odd then N + 1 and N − 1 are even,
so
bN + bN −1 2+0
bN +1 = −1 ≤ −1 = 1−1 = 0
2 2
and
bN + bN −1 1−1
bN +1 = −1 ≥ − 1 = −1.
2 2
If N is even then N + 1 and N − 1 are odd, so
bN + bN −1 2+0
bN +1 = +1 ≤ +1 ≤ 1+1 = 2
2 2
and
bN + bN −1 1−1
bN +1 = −1 ≥ + 1 = 1.
2 2
In particular, |bN | ≤ 2 for every N.
Since hN is piecewise linear, its global max and min must occur at one of
its knots, which are all located at t = µn for some 1 ≤ n ≤ N. Hence
khN k∞ = max{|bn | : n = 1, . . . , N } ≤ 2.
7.1 Let {en } and {fn } be orthonormal bases for a Hilbert space H.
(a) {e1 , e2 , e2 , e3 , e3 , e3 , . . . } is norm-bounded above but is not Bessel.
(b) {e1 + e2 , e2 , e3 , . . . } is a Bessel sequence that is a nonorthogonal basis
for H.
(c) {2−1/2 en , 2−1/2 fn } is a Bessel sequence since
X 1X 1X
|hx, 2−1/2 en i|2 + |hx, 2−1/2 fn i|2 = |hx, en i|2 + |hx, fn i|2
n
2 n 2 n
= kxk2 ,
and a similar calculation shows that the frame operator is S = I.

(c), (d) The sequence
{2−n em }m,n∈N
is Bessel since
XX X X X
|hx, 2−n em i|2 = 2−n |hx, em i|2 = 2−n kxk2 = kxk2 .
m n n m n
On the other hand, if finitely many elements are removed from {2−n em }m,n∈N ,
then it still contains a multiple of every vector en , and hence is complete.
Moreover, a similar calculation shows that the frame operator is S = I.
(e) {nen } is anP
unconditional basis
P for H, but it is not a Bessel sequence.
For example, x = en /n ∈ H, but |hx, nen i|2 = ∞.
(f) Consider the Babenko example, {|t|α e2πint }n∈Z where 0 < α < 1/2,
which is a conditional basis for H = L2 (T) = L2 [− 21 , 21 ]. Each element of this
basis has the same norm, so except for a scaling factor it is a normalized basis.
Set en (t) = e2πint and fn (t) = |t|α en (t). Given f ∈ H, we have g(t) =
α
|t| f (t) ∈ H, and
Z Z
hf, fn i = f (t) |t|α en (t) dt = g(t) en (t) dt = hg, en i.
Since |t|α ≤ 1 on [− 12 , 21 ], we therefore have

X X
|hf, fn i|2 = |hg, en i|2 = kgk2L2 ≤ kf k2L2 .
n∈Z n∈Z
Hence {fn } is a Bessel sequence in H.

(g) Using the notation of part (e), the dual of the Babenko example,
{|t|−α e2πint }n∈Z , is a conditional basis for H. Each element of this basis
has the same norm, so except for a scaling factor it is a normalized basis.
For convenience, take α = 1/4. The function gr = χ[0,r] belongs to H, and

if we set hn (t) = |t|−1/4 en (t) then
X X

|hgr , hn i|2 = |t|−1/4 χ[0,r] (t), en 2
n∈Z n∈Z
Z r
2
= |t|−1/4 χ[0,r] L2 = t−1/2 dt = 2r1/2 .
0
However, Z r
kgr k2L2 = 1 dt = r.
0
Therefore
1 X
|hfr , hn i|2 = 2r−1/2 → ∞ as r → 0+ .
kgr k2L2
n∈Z
Therefore {fn } is not a Bessel sequence.

(h) Combining parts (f) and (g), we see that {|t|α e2πint }n∈Z is a Bessel
basis, but its biorthogonal sequence is not Bessel.
7.2 (a), (b), (c) Follow from Exercise 3.8.
(d) Suppose that {xn } is complete. If Cx = 0 then we have hx, xn i = 0 for
every n. By completeness, this implies x = 0, so C is injective. Since R = C ∗ ,
it follows from Theorem 2.13 that R has dense range.
7.3 (a) ⇒ (b). This follows from Theorem 7.2.
(b) ⇒ (a). Fix x ∈ H and N > 0. Since E is dense, there exist vectors
yk ∈ E such that yk → x. Hence
N
X N
X
|hx, xn i|2 = lim |hyk , xn i|2
k→∞
n=1 n=1
N
X
= lim |hyk , xn i|2
k→∞
n=1
≤ B lim kyk k2
k→∞
= B kxk2 .
PN
Since this is true for every N, we conclude that n=1 |hx, xn i|2 ≤ B kxk2 , so
{xn } is a Bessel sequence.
(a) ⇒ (c). Suppose that {xn } is a Bessel sequence with Bessel bound B.
Then by Theorem 7.2, the synthesis operator R for {xn } is bounded, with
P 2
kRk ≤ B 1/2 . We therefore have cn xn ≤ B kck2ℓ2 for all c ∈ ℓ2 , which
implies statement (c).
(c) ⇒ (d). Suppose that statement (c) holds. Fix c = (cn ) ∈ ℓ2 . Given
M < N, set d = (0, . . . , 0, cM+1 , . . . , cN , 0, 0, . . . ). Then
X 2
N N
X
c x = kRdk22 ≤ kRk2 kdk22 = kRk2 |cn |2 .
n n ℓ ℓ
n=M+1 n=M+1
P
Therefore cn xn is Cauchy, and therefore converges.
P
(d) ⇒ (e). Suppose statement (d) holds. Then T c = cn xn converges for
PN
each c ∈ ℓ2 , so T : H → ℓ2 and T is clearly linear. Define TN c = n=1 cn xn .
Then
N
X X
N 1/2 X
N 1/2
2 2
kTN ck ≤ |cn | kxn k ≤ |cn | kxn k ≤ CN kckℓ2 ,
n=1 n=1 n=1
so each TN is bounded. By hypothesis, TN c → T c as N → ∞, so the Banach–

Steinhaus Theorem implies that T is bounded. Since T em = xm , statement (e)
follows.
(e) ⇒ (a). Suppose there exists some R ∈ B(ℓ2 , H) such that Rem = xm
for every m. Since R is bounded, it follows that
X X X
Rc = R cn e n = cn Ren = cn xn .
n n n
For x ∈ H and c ∈ ℓ2 , the adjoint of R satisfies

X X D E
hR∗ x, ci = hx, Rci = hx, cn xn i = hx, xn i cn = hx, xn i , cn .
n n

Hence R∗ x = hx, xn i . Since R∗ is a bounded map of H into ℓ2 , we conclude
that {xn } is a Bessel sequence.
7.4 Let G be the Gram matrix for a Bessel sequence {xn }. Given a sequence
c = (cn ) ∈ ℓ2 , we have

Gc = CRc = Rc, xm m∈N
X
= cn xn , xm
n m∈N
X
= cn hxn , xm i
n m∈N

= hxn , xm i m,n∈N
(cm )m∈N .
7.5 Suppose {xn } is Bessel with Bessel bound B, and let {en } be the standard
basis for ℓ2 . The synthesis operator for {xn } is bounded, so kxn k2 = kRen k2 ≤
B ken k2ℓ2 = B. Hence all Bessel sequences are bounded. Alternatively, we note
that for any m we have
X
kxm k4 = |hxm , xm i|2 ≤ |hxm , xn i|2 ≤ B kxm k2 .
n
Therefore we must have kxm k2 ≤ B. Further, if kxmPk2 = B, then equality

holds in the estimate above, and hence we must have n6=m |hxm , xn i|2 = 0.
7.6 Let B be a Bessel bound for {xn }. Given y ∈ K, we have
X X
|hy, Lxn i|2 = |hL∗ y, xn i|2 ≤ B kL∗ yk2 ≤ B kLk2 kyk2 .
n n
Hence {Lxn } is a Bessel sequence in K.

7.7 We are given that H ⊆ K.
If {xn } is a Bessel sequence in K, then there exists some B > 0 such that
X
∀ x ∈ K, |hx, xn i|2 ≤ B kxk2 .
n
The above condition holds for all x ∈ H, so {xn } is a Bessel sequence in H.

Now suppose that {xn } is a Bessel sequence in H. Then there exists some
B > 0 such that
X
∀ x ∈ H, |hx, xn i|2 ≤ B kxk2 .
n
Suppose that y ∈ K, and let P denote the orthogonal projection of K onto H.

Then
X X X
|hy, xn i|2 = |hy, P xn i|2 = |hP y, xn i|2 ≤ B kP yk2 ≤ B kyk2 ,
n n n
so {xn } is a Bessel sequence in K.

7.8 (a) No. Let {en } be any infinite orthonormal sequence in H and consider
PN
{nen } and E = span{en }. If x = m=1 cm em ∈ E then
N N 2
X X X 2
X
|hx, nen i|2 = c e , ne = c he , me i
m m n m m m
n n m=1 m=1
X 2
N

= mcm < ∞.
m=1
P
However,
P {nen } is not a Bessel sequence. For example, if x = en /n, then
|hx, nen i|2 = ∞.
(b) No. Let {en } be an orthonormal basis for H and set xn = nen . The
sequence {xn } is not a Bessel sequence. However, if we set yn = en /n then
{yn } is complete, and for each m ∈ N we have
X X
|hym , xn i|2 = |hem /m, nen i|2 = |hem /m, memi|2 = 1.
n n
P P
7.9 (a) We are given that K = m n |hxn , xn i|2 < ∞. Let G = [hxn , xm i]
be the Gram matrix for {xn }. Given c ∈ c00 , applying Cauchy–Bunyakovski–
Schwarz we have
X X 2

2
kGckℓ2 = hxn , xm i cn

m n
X X X
≤ |hxn , xm i|2 |cn |2
m n n
= K kck2ℓ2 .
Theorem 7.6 therefore implies that {xn } is a Bessel sequence.

P
(b) We are given that K = supm n |hxn , xn i| < ∞. Note that the
roles of mPand n in the definition of this constant are symmetric, i.e.,
K = supn m |hxn , xn i| < ∞ as well.
Let G = [hxn , xm i] be the Gram matrix for {xn }. Given c ∈ c00 , applying
Cauchy–Bunyakovski–Schwarz we have
2
X X
2
kGckℓ2 = hxn , xm i cn

m n
X X 2
1/2 1/2
≤ |hxn , xm i| |hxn , xm i| |cn |
m n
X X X
≤ |hxn , xm i| |hxn , xm i| |cn |2
m n n
X X
≤ K |hxn , xm i| |cn |2
m n
X X
= K |hxn , xm i| |cn |2
n m
X
≤ K2 |cn |2
n
= K kck2ℓ2 .
2
Theorem 7.6 therefore implies that {xn } is a Bessel sequence.

7.10 (a) Given x ∈ H we have

X
2
kxk = hx, yn i xn , x
n
X
= hx, yn i hxn , xi
n
X X
2 2
≤ |hx, yn i| |hx, xn i|
n n
X
≤ B kxk2 |hx, yn i|2 .
n
(b) Using the biorthogonality, we have

N N X 2 X 2
X X N N
2
|cn | =
cm ym , xn ≤ B cm y m
.
n=1 n=1 m=1 m=1
7.11 By Theorem 7.4 there exist orthonormal bases {em }, {fn } for H, K and
bounded maps U ∈ B(H), V ∈ B(K) such that U em = xm and V fn = yn for
all m, n. By Theorem B.10, {em ⊗fn }m,n∈N is an orthonormal basis for H ⊗K.
By Exercise B.9 there exists a unique bounded operator U ⊗ V ∈ B(H ⊗ K)
that satisfies (U ⊗ V )(x ⊗ y) = U x ⊗ V y for all x ∈ H, y ∈ K. By Exercise 7.6
we know that bounded operators map Bessel sequences to Bessel sequences,
so

{xm ⊗ fn }m,n∈N = {U em ⊗ V fn }m,n∈N = (U ⊗ V )(em ⊗ fn ) m,n∈N
is Bessel.
7.12 (a) The (m, n)-entry of L is
(
1/m, m ≥ n,
Lmn =
1/n, m ≤ n.
On the other hand, the (m, n)-entry of CC ∗ is

 
1/n
 .. 
 .  
  m 1 1
= 1
 1/n  m n m, m≥n
1/m · · · 1/m 0 0 · · ·   =

 0  n 1 1
= 1
n, m ≤ n.
 0  m n
 
..
.
Hence CC ∗ = L.
(b) We have
I − (I − C)(I − C)∗ = I − I + C + C ∗ − CC ∗ = C + C ∗ − CC ∗ .
By inspection, this is the diagonal matrix with entries 1, 1/2, . . . , and therefore
it is a positive matrix.
(c) Let M = (I −C)(I −C)∗ . Then I −M ≥ 0, so we have h(I −M )x, xi ≥ 0
for every x. Therefore
hM x, xi ≤ kIxkx = kxk2 .
Since M = M ∗ , we therefore have by Theorem 2.15 that
kM k = sup hM x, xi ≤ 1.
kxk=1
Therefore, by Theorem 2.13,
kI − Ck2 = k(I − C)(I − C)∗ k = kM k ≤ 1.
(d) We have
kCk = k(I − C) − Ik ≤ kI − Ck + kIk ≤ 1 + 1 = 2.
Hence
kLk = kCC ∗ k = kCk2 ≤ 22 = 4.
7.13 This follows from Exercise 3.1 and the definition of Riesz bases.
7.14 Let {x1 , . . . , xd } be a basis for a finite-dimensional Hilbert space V. Then
it has a dual basis {y1 , . . . , yd }. Both of these are complete Bessel sequences,
so it follows from Theorem 7.13 that they are Riesz bases.
Another approach is to note that {x1 , . . . , xd } is a bounded unconditional
basis, and hence is a Riesz basis.
Yet another is to observe that
d
X
hGc, ci = cn xn
.
n=1
Hence hGc, ci ≥ 0, and hGc, ci = 0 implies c = 0 since {x1 , . . . , xd } is a basis.

Hence the Gram matrix is positive definite, and therefore is invertible.
7.15 Let {xn } be an orthonormal basis for H. Then {xn /n} is an uncondi-
tional basis for H (see Exercise 4.6). However, it is not a Riesz basis since it
is not bounded below.
7.16 By Theorem 7.13, we have that {xn } is a Riesz basis for H. Therefore
there exists an orthonormal basis {en } for H and a topological isomorphism
T : H → H such that T en = xn for every n. Given x ∈ span{en }, write

PN
x = n=1 cn en . Then
N 2 N 2 N
X X X
2
kT xk =
cn T e n = cn xn = |cn |2 = kxk2 .

n=1 n=1 n=1
Hence T is an isometry on the dense subspace span{en }, and therefore is an

isometry on H. Isometries are inner-product preserving, so we have
hxm , xn i = hT em , T en i = hem , en i = δmn .
Hence {xn } is orthonormal, and since it is complete, it is an orthonormal basis

for H.
7.17 Since {xm } and {yn } are each Schauder bases, we know from Exercise 5.8
that {xm ⊗ yn }m,n∈N is a Schauder basis for H ⊗ K, and by Exercise 7.11 it is
also a Bessel sequence. Likewise, if we let {e
xm } and {e
yn } denote the biorthog-
onal sequences then they are Riesz bases and therefore {e xm ⊗ yen }m,n∈N is a
Bessel Schauder basis, and it is biorthogonal to {xm ⊗yn }m,n∈N . Theorem 7.13
therefore implies that each of these sequences are Riesz bases.
7.18 The easy way is to observe that this follows immediately from Theorem
5.24. That theorem tells us that {yn } is a basis that is equivalent to {xn },
and since {xn } is an orthonormal basis, {yn } is a Riesz basis by definition.
7.19 Fix N ∈ N and choose any c1 , . . . , cN ∈ F. Since N is finite, by
hypothesis the series
X∞ X N
cn ank Tk en
k=1 n=1
converges in H. Applying Minkowski’s Inequality, we therefore compute that

N N ∞
X X X

cn (xn − yn ) = cn ank Tk en

n=1 n=1 k=1
∞ N
X X
=
c a T e
n nk k n
k=1 n=1
∞
X X
N

≤ cn ank en
Tk
k=1 n=1
∞ N
X X
≤ kTk k
c a e
n nk n
k=1 n=1
∞
X X
N 1/2
2
= kTk k |cn ank |
k=1 n=1
∞
X X
N 1/2
≤ kTk k sup |ank | |cn |2
n
k=1 n=1
X
N 1/2
= λ |cn |2 .
n=1
Since λ < 1, the result follows by applying Exercise 7.18.

7.20 (a) We have
Z 1/2 2k Z 1/2 2k
2 k 2 1 2 1
kTk f kL2 = |x f (x)| dx ≤ |f (x)| dx = kf k2L2 .
−1/2 2 −1/2 2
Therefore kTk f k2L2 ≤ 2−k kf kL2 , so kTk k ≤ 2−k .

Now set fε = χ[ 21 −ε, 21 ] . This function satisfies
Z 1
2
kfε k2L2 = dx = ε,
1
2 −ε
while Z 1 2k
2 1
kTk fε k2L2 = x2k dx ≥ −ε · ε,
1
2 −ε
2
so 2k
2 kTk fε k2L2 1
kTk k ≥ ≥ −ε .
kfε k2L2 2
Since ε > 0 is arbitrary, this implies that kTk k2 ≥ 2−2k . Taking square roots
and combining with the converse estimate, we see that kTk k = 2−k .
(b) Expanding e2πinx in a Taylor series, we see that pointwise we have
en (x) − eλn (x) = e2πinx − e2πiλn x

= e2πinx 1 − e2πi(λn −λ)x
∞ k
X 2πi(λn − n)
2πinx
= e 1− xk
k!
k=0
X
∞ k
2πinx 2πi(λn − n)
= −e xk
k!
k=1
∞ k
X 2πi(λn − n)
= − xk e2πinx
k!
k=1
∞
X
= ank Tk en (x).
k=1
The final series converges absolutely in L2 [− 21 , 12 ] because

∞
X ∞
X (2π)k |λn − n|k
|ank | kTk en kL2 ≤ kTk k ken kL2
k!
k=1 k=1
∞
X (2π)k δ k
≤ 2−k
k!
k=1
∞
X (2π)k δ k
≤ 2−k
k!
k=1
∞
X πk δk
=
k!
k=1
= eπδ < ∞.
In fact, the same calculations show that the series converges absolutely in
L∞ [− 21 , 12 ]. Therefore the pointwise and L2 -limits are the same a.e., so
∞
X
en − eλn = ank Tk en ,
k=1
with convergence both absolutely in L2 -norm and pointwise.

(c) We have
(2π)k |λn − n|k (2πδ)k
|ank | = ≤ ,
k! k!
and this is independent of n. Therefore
∞
X ∞
X (2πδ)k
∞
X (πδ)k
λ = kTk k sup |ank | ≤ 2−k = = eπδ − 1.
n k! k!
k=1 k=1 k=1
Applying Exercise 7.19, if λ < 1 then {eλn }n∈Z is a Riesz basis for L2 [− 21 , 12 ].
This happens if eπδ − 1 < 1, or
eπδ < 2.
Taking logarithms, this requires πδ < ln 2, or

ln 2
δ < .
π
8.1 Given v = (v1 , v2 ) ∈ R2 , we have by the Parallelogram Law that
X3 √ 2 √ 2
3 1 3 1
2 2
|hv, xn i| = |v2 | + − v1 − v2 + v1 − v2
n=1
2 2 2 2
3 1
= |v2 |2 + 2 |v1 |2 + 2 |v2 |2
4 4
3 3 3
= |v1 |2 + |v2 |2 = kvk2 .
2 2 2
8.2 Let A, B be frame bounds for {xn } and let C be a Bessel bound for {yn }.
Then for each x ∈ H we have
X
A kxk2 ≤ |hx, xn i|2
n
X X
≤ |hx, xn i|2 + |hx, yn i|2 ≤ B kxk2 + C kxk2 .
n n
Hence {xn } ∪ {yn } is a frame with frame bounds A, B + C.

8.3 If {en } is an orthonormal basis, then {en }∪{en /n} is a frame, but {en /n}
is not a frame sequence.
8.4 The fact that the upper frame bound holds follows from Exercise 7.3. We
can use the same technique to establish that the lower frame bound holds as
well. Fix x ∈ H and N > 0. Since E is dense, there exist vectors yk ∈ E such
that yk → x. Hence
N
X N
X
|hx, xn i|2 = lim |hyk , xn i|2
k→∞
n=1 n=1
N
X
= lim |hyk , xn i|2
k→∞
n=1
≥ A lim kyk k2
k→∞
= A kxk2 .
P∞
Since this is true for every N, we conclude that n=1 |hx, xn i|2 ≥ A kxk2 , so
{xn } is a frame.
8.5 (a) We know that the frame operator is bounded and self-adjoint. We
will show that kS − AIk = 0. Since {xn }n∈N is a tight frame,
X
hSx, xi = |hx, xn i|2 = A kxk2 .
n
Therefore,

(S − AI)x, x = hSx, xi − A hx, xi = A kxk2 − A kxk2 = 0,
so

kS − AIk = sup (S − AI)x, x = 0.
kxk=1
Consequently, X
Ax = Sx = hx, xn i xn .
n
8.6 We are given an A-tight frame {xn }.

(a) This follows immediately from Exercise 7.5 since {xn } is a Bessel se-
quence with Bessel bound A. Alternatively, we can compute directly that
X
A kxm k2 = |hxm , xn i|2 ≥ |hxm , xm i|2 = kxm k4 ,
n
and therefore kxm k2 ≤ A.

(b) Suppose that kxm k2 < A for some m. Then for x ∈ H,
X X
A kxk2 = |hx, xn i|2 = |hx, xm i|2 + |hx, xn i|2 .
n n6=m
Since
|hx, xm i|2 ≤ kxk2 kxm k2 ,
we therefore have that
X
|hx, xn i|2 = A kxk2 − |hx, xm i|2
n6=m

≥ A kxk2 − kxk2 kxm k2 = A − kxm k2 kxk2 .
Therefore {xn }n6=m has a lower frame bound of A − kxm k2 . Since

X X
|hx, xn i|2 ≤ |hx, xn i|2 = A kxk2 ,
n6=m n
the upper frame bound is A. Moreover, considering the vector x = xm , we

have
X X
|hxm , xn i|2 = |hxm , xn i|2 − |hxm , xm i|2
n6=m n

= A kxm k2 − kxm k4 = A − kxm k2 kxm k2 .
Therefore A − kxm k2 is the optimal lower frame bound.

(c) If kxm k2 = A then

X
kxm k4 = A kxm k2 = |hxm , xn i|2
n
X
= |hxm , xm i|2 + |hxm , xn i|2
n6=m
X
= kxm k4 + |hxm , xn i|2 ,
n6=m
P
so we must have n6=m |hxm , xn i|2 = 0.
8.7 We are given an A-tight frame {xn }.
(a) ⇒ (b). By Exercise 8.6, if kxm k2 = A for some m then xn ⊥ xm for all
n 6= m. Hence, if kxm k2 = A for every m then {xn } is an orthogonal sequence.
(b) ⇒ (a). Assume that {xn } is orthogonal with every xn 6= 0. Choose any
particular xm . Then hxm , xn i = 0 for n 6= m, so
X
A kxm k2 = |hxm , xn i|2 = |hxm , xm i|2 = kxm k4 .
n
Since kxm k2 6= 0, we can divide to get kxm k2 = A.

(b) ⇒ (c). Assume that {xn }n∈N is orthogonal with every xn 6= 0. Fix any
m ∈ N. Then hxm , xn i = 0 for n 6= m, so
∞
X
A kxm k2 = |hxm , xn i|2 = |hxm , xm i|2 = kxm k4 .
n=1
Since kxm k2 6= 0, we can divide to obtain kxm k2 = A.

Since S = AI, given any x ∈ H we have
∞
1 1 X
x = Sx = hx, xn i xn .
A A n=1
1
We must show that the scalars cn = A hx, xn i are unique. Assume that we
P
also had x = dn xn , and fix any m ∈ N. Then by orthogonality,
DX
∞ E
hx, xm i = dn xn , xm
n=1
∞
X
= dn hxn , xm i = dm hxm , xm i = dm kxm k2 = A dm .
n=1
1
Therefore dm = A hx, xm i = cm . Thus we have uniqueness, so {xn }n∈N is a
Schauder basis for H.
(c) ⇒ (d). This follows immediately from the fact that every basis is ω-
independent.
(d) ⇒ (b). Assume that {xn }n∈N is ω-independent. This implies in par-
ticular that xn 6= 0 for every n. Fix m ∈ N. Then since S = AI,
∞ ∞
1 1 X X
1

xm = Sxm = hxm , xn i xn = A hxm , xn i xn .
A A n=1 n=1
But we also have

∞
X
xm = δmn xn ,
n=1
so by ω-independence we conclude that

1
δmn = hxm , xn i.
A
Therefore {xn }n∈N is orthogonal.
(b) ⇒ (e). Suppose that {xn } is an orthogonal sequence. It is complete
since we have assumed that {xn } is a tight frame. No proper subset of an
orthogonal sequence can be complete, so {xn } must be an exact frame.
(e) ⇒ (a). Suppose that kxn k2 < A for some n. Then {xn }n6=m is a frame
by Exercise 8.6, so {xn } is not an exact frame.
Final statement. ⇒. Suppose that {xn } is an exact Parseval frame. If
kxm k2 < 1 for some m then we have by part (b) of Exercise 8.6 that {xn }n6=m
is still a frame, which contradicts the definition of exact frame. Hence we have
kxm k = 1 for every m, which implies by the equivalence of statements (a)
and (b) above that {xn } is orthogonal, and hence orthonormal since it is
normalized. We know that {xn } is complete since it is a frame, so it must be
an orthonormal basis.
⇐. Every orthonormal basis is a Parseval frame by the Plancherel Equality.
Further, for each m we have that {xn }n6=m is incomplete and therefore cannot
be a frame, so {xn } is exact.
8.8 Let P be the orthogonal projection of a Hilbert space H onto a closed
subspace M, and let {xn } be an orthonormal basis for H. If x ∈ M, then
P x = x. Therefore, since P is self-adjoint,
X X X
|hx, P xn i|2 = |hP x, xn i|2 = |hx, xn i|2 = kxk2 .
n n n
Hence {xn } is a Parseval frame for M.

8.9 (a) We will show that if b > 1 then {e2πibnt }n∈Z is incomplete in L2 [0, 1].
Each exponential e2πibnt is 1b -periodic. Let ε be small enough that both [0, ε]
and [ 1b , ε] are contained within [0, 1]. Define
f = χ[0,0+ε] − χ[ 1b , 1b +ε] .
Then Z Z 1
ε b +ε
2πibnt −2πibnt
hf, e i = e dt − e−2πibnt dt = 0.
1
0 b
Thus f is orthogonal to every e2πibnt , and therefore {e2πibnt }n∈Z is incomplete.

Alternatively, since each function e2πibnt is 1b -periodic, any finite linear
combination will be 1b -periodic. As limits of 1b -periodic functions are still 1b -

periodic, it follows that every function in span {e2πibnt }n∈Z is 1b -periodic.
However, not every function in L1 [0, 1] is 1b -periodic, e.g., consider f (t) = t.
Therefore {e2πibnt }n∈Z is incomplete.
(b) We will show that if 0 < b < 1, then {e2πibnt }n∈Z is a tight frame for
L [0, 1]. Since {e2πint }n∈Z is an orthonormal basis for L2 [0, 1], by making a
2
change of variables it follows that {b1/2 e2πibnt }n∈Z is an orthonormal basis

for the space L2 [0, 1b ]. Therefore,
X 1
∀ f ∈ L2 [0, 1b ], |hf, e2πibnt i|2 = kf k22 . (A)
b
n∈Z
Note that these are inner products in L2 [0, 1b ], i.e.,

Z 1/b Z 1/b
kf k22 = 2
|f (t)| dt and hf, e 2πibnt
i = f (t) e−2πibnt dt.
0 0
If we take f ∈ L2 [0, 1], we can regard it as an element of L2 [0, 1b ] by setting

f (t) = 0 for 1 < t ≤ 1b . Then we can apply equation (A), but because we
have extended by zero, the norm and inner product are from L2 [0, 1]. In other
words, equation (A) holds for f ∈ L2 [0, 1], so {e2πibnt }n∈Z is a tight frame for
L2 [0, 1] with frame bounds A = B = 1/b.
An alternative approach is to note that P : L2 [0, 1b ] → L2 [0, 1] given by
P f = f χ[0,1] is an orthogonal projection, and apply Exercise 8.8 to conclude
that {b1/2 e2πibnt }n∈Z is a Parseval frame for L2 [0, 1]. Hence {e2πibnt }n∈Z is
a tight frame with frame bound 1b .
Now we consider the question of whether {e2πibnt }n∈Z is a basis for L2 [0, 1].
First, since
Z 1
1
ke2πibnt k22 = |e2πibnt |2 dt = 1 < ,
0 b
it follows from Exercise 8.7 that {e2πibnt }n∈Z is not a basis, and it also follows
from Exercise 8.6 that {e2πibnt }n6=m is a frame for any m.
Second, {e2πibnt }n∈Z is not an orthogonal sequence, because
Z 1
he2πimbt , e2πibnt i = e2πi(m−n)bt dt
0
1
e2πi(m−n)bt
=
2πi(m − n)b 0
e2πi(m−n)b 1
= − 6= 0.
2πi(m − n)b 2πi(m − n)b
More precisely, this quantity is not zero for every choice of m and n, e.g.,
consider m = 1 and n = 0. Exercise 8.7 again implies we do not have a basis.
Third, we show directly that the constant function f = 1 does not have
a unique representation in terms of the exponentials {e2πibnt }. Since f (t) =
e2πi0bt , one expansion is
X
f (t) = δmn e2πibnt .
n∈Z
Another expansion is provided by the tight frame property (see Exercise 8.5):
1 X
f (t) = hf, e2πibnt i e2πibnt .
A
n∈Z
We must check whether this is the same expansion we found before. So we

check the coefficients:

Z 1 b, n = 0,
1
hf, e2πibnt i = b e−2πibnt dt = −2πibn
A 0 − e +
1
, n 6= 0.
2πin 2πin
Since these are not the same values as δmn we have indeed found two different
ways to write f = 1 in terms of the frame elements e2πibnt , and therefore this
frame cannot form a basis.
8.10 We are given a frame {xn } with frame bounds A, B and a sequence
{yn } such that {xn − yn } is Bessel with Bessel bound K < A. Writing yn =
xn + (yn − xn ), we have
X
∞ 1/2 X
∞ 1/2
2 2
|hf, yn i| = |hf, xn i + hf, yn − xn i|
n=1 n=1
X
∞ 1/2 X
∞ 1/2
≤ |hf, xn i|2 + |hf, yn − xn i|2
n=1 n=1
1/2 1/2
≤ B kf k2 + K kf k2
= (B 1/2 + K 1/2 ) kf k.
Therefore {yn } has an upper frame bound of (B 1/2 + K 1/2 )2 .

Now write xn = yn + (xn − yn ) to get:
X
∞ 1/2 X
∞ 1/2 X
∞ 1/2
2 2 2
|hf, xn i| ≤ |hf, yn i| + |hf, xn − yn i| .
n=1 n=1 n=1
Then,
X
∞ 1/2 X
∞ 1/2 X
∞ 1/2
2 2 2
|hf, yn i| ≥ |hf, xn i| − |hf, xn − yn i|
n=1 n=1 n=1
1/2 1/2
≥ A kf k2 − K kf k2
= (A1/2 − K 1/2 ) kf k.
Since A > K, this gives us a lower frame bound for {yn } of (A1/2 − K 1/2 )2 .
P
(b) We are given a sequence {zn } such that K = khn k2 < ∞. Using
Cauchy–Bunyakovski–Schwarz, we compute
X X X
|hx, zn i|2 ≤ kxk2 kzn k2 = kxk2 kzn k2 = K kxk2 .
n n n
Hence {zn } is Bessel with Bessel bound K.

(c) By Exercise 8.9, {e2πibnt }n∈Z is a tight frame for L2 [0, 1] with frame
bound A = 1/b. Therefore, from parts (a) and (b), {e2πiλn t }n∈Z will be a
frame if {e2πibnt − e2πiλn t } satisfies
X 1
K = ke2πibnt − e2πiλn t k22 < .
b
n∈Z
We can estimate these norms using that fact that |eit − 1| ≤ |t| for every t:
Z 1
ke2πibnt − e2πiλn t k22 = |e2πibnt − e2πiλn t |2 dt
0
Z 1
= |e2πi(bn−λn )t − 1|2 dt
0
Z 1
≤ |2π(bn − λn )t|2 dt
0
4π 2
= |bn − λn |2 .
3
Therefore, {e2πiλn t }n∈Z will be a frame if
X 3
|bn − λn |2 < .
n
4π 2 b
Surprisingly (perhaps), this result is not even close to being optimal. This
result requires the that λn become closer and closer to nb as |n| increases. The
optimal result does not require this, only that the λn have a certain maximum
distance from nb. Here is the optimal result for b = 1, the case where you get
both a frame and a basis.
Kadec’s 41 -Theorem. If supn |n − λn | ≤ L < 41 , then {e2πiλn t }n∈Z is an
exact frame for L2 [0, 1].
This result is optimal in the sense that the theorem is false if L = 1/4. Nat-
urally, the proof is more difficult than the proof of our result, but it isn’t ter-
ribly complicated. If you’re interested, look in the nicely written book [You01]
by Young.
8.11 Fix vectors v1 , . . . , vN ∈ Fd . Let h·, ·i denote the dot product on Fd ,
and let AH denote the Hermitian of a matrix A.
(a) If x ∈ Fd then
N
X N
X
|hx, xn i|2 ≤ kxk2 kxn k2 = B kxk2 ,
n=1 n=1
so {v1 , . . . , vN } is Bessel. Let C be the matrix that has v1H , . . . , vN

H
as rows.
Then  H   
v1 hx, v1 i
   
Cx =  ...  x =  ..
. ,
H
vN hx, vN i
so C is the analysis operator for {v1 , . . . , vN }. The synthesis operator is there-
fore R = C H , which has v1 , . . . vN as columns.
(b) Let S = C ∗ C be the frame operator, and note that S is a d × d positive

matrix. Therefore S has an orthonormal basis of eigenvectors {w1 , . . . , wd } and
corresponding nonnegative eigenvalues λ1 ≤ · · · ≤ λd . Hence for every x ∈ Fd
we have
N
X X
d
2
|hx, xn i| = hSx, xi = S hx, wj i wj , x
n=1 j=1
X
d
= hx, wj i λj wj , x
j=1
d
X
= λj |hx, wj i|2
j=1
d
X
≤ λd |hx, wj i|2 = λd kxk2 ,
j=1
so {v1 , . . . , vN } is a Bessel sequence with Bessel bound λd . Similarly

N
X
|hx, xn i|2 ≥ λ1 kxk2 ,
n=1
although it is possible that λ1 could be zero. Moreover,

N
X
|hw1 , xn i|2 = hSw1 , w1 i = λ1 kw1 k2 ,
n=1
so {v1 , . . . , vN } is a frame if and only if λ1 > 0.

(i) ⇒ (ii). Assume {v1 , . . . , vN } span Fd . Then {v1 , . . . , vN } is complete, so
the analysis operator is injective. Specifically, if Cx = 0 then we have hx, vn i =
0 for each n, so x is orthogonal to every vector in span{v1 , . . . , vN } = Fd , and
therefore x = 0. Suppose that hSx, xi = 0. Then
kCxk2 = hCx, Cxi = hC ∗ Cx, xi = hSx, xi = 0,
so x = 0 since C is injective. Hence S is positive definite.

(ii) ⇒ (iii). Suppose that S is positive definite. Then λ1 > 0, so {v1 , . . . , vN }
is a frame.
(iii) ⇒ (i). Assume that {v1 , . . . , vN } is a frame for Fd . Since all frames
are complete, the span of v1 , . . . , vN must be all of Fd .
(c) Note that G = CC ∗ is a positive matrix, and
N 2
X

hGc, ci = cn vn
.
n=1
(i) ⇒ (ii). Assume that {v1 , .P . . , vN } is linearly independent. By the equa-

tion above, if hGc, ci = 0 then cn vn = 0 and therefore c = 0. Hence G is
positive definite.
P
(ii) ⇒ (i). Assume that G is positive definite, If cn vn = 0 then we have
hGc, ci = 0 by the equation above, and therefore c = 0 since G is positive
definite. Hence v1 , . . . , vN are independent.
(i) ⇒ (iii). Suppose that v1 , . . . , vN are independent.PThen every vector x in
H = span{v1 , . . . , vN } can be written uniquely as x = cn vn , so {v1 , . . . , vN }
is a basis for H. In fact, it is a bounded unconditional basis because it is a

finite set, and therefore is a Riesz basis for H.
(iii) ⇒ (i). Assume that {v1 , . . . , vN } is a Riesz sequence in Fd . Then it is
a bounded unconditional basis for its span, and hence is linearly independent.
8.12 (a) This follows just as in Exercise 8.11(b) by replacing Fd in that
exercise with the span of v1 , . . . , vN .
(b) This is a consequence of Theorem 7.13, since a set of independent
vectors is a bounded unconditional basis for its span.
8.13 Let {en } be an orthonormal basis for a Hilbert space H. Then {en /n} ∪
{en /n} is neither a basis nor a frame, but for each x ∈ H we have
X en X en
x = hx, nen i + hx, 0i ,
n
n n
n
so it is a quasibasis.
8.14 ⇒. This is proved in Theorem 8.13.
(b) ⇒ (a). Assume that statementP (b) holds. By continuity of the inner
product, it follows that hSx, xi = |hx, xn i|2 . Since we have AI ≤ S ≤ BI,
we therefore have Then hAIx, xi ≤ hSx, xi ≤ hBIx, xi for every x ∈ H, which
implies that {xn } is a frame for H.
8.15 We are given positive operators U, V ∈ B(H) with U ≤ V. Since U
and V are self-adjoint, we have by Theorem 2.15 that
kU k = sup hU x, xi ≤ sup hV x, xi = kV k.
kxk=1 kxk=1
(b) Since AI ≤ S ≤ BI, we have
0 ≤ BI − S ≤ B− AI = (B − A) I.
Therefore
1 B−A
0 ≤ I− S ≤ I.
B B
Also,
(B + A) I − 2S = BI + AI − 2S ≤ BI + AI − 2AI = BI − AI = (B − A) I,
so
2 B−A
I− S ≤ I.
A+B B+A
Since 0 < A ≤ B < ∞, using part (a), we therefore have that
1 B−A B−A

I − S ≤ kIk = < 1
B B B
and
2 B−A B−A
I − S ≤ kIk = < 1.
A+B B+A B+A
By Neumann series, we therefore have that S is a topological isomorphism.
8.16 (a) We are given a frame {xn } with an alternative dual {yn } that is a
Bessel sequence. 2
Given y ∈ H, we have P hy, xn i ∈ ℓ since {xn } is a frame. Since {yn }
is Bessel, the sequence hy, xn i yn therefore converges, and hence for every
x ∈ H we have
X X X
hx, yi = hx, yn i xn , y = hx, yn i hxn , yi = x, hy, xn i yn .
n n n
P
Therefore y = hy, xn i yn .
Let A, B be frame
P bounds for {yn }. Given y ∈ H, since {yn } is a dual of
{xn } we have y = hy, yn i xn . Therefore
X 2

2 2
kyk = sup |hy, xi| = sup hy, yn i xn x
kxk=1 kxk=1 n
X X 2
2
≤ sup |hy, yn i| |xn x|
kxk=1 n n
X
≤ sup |hy, yn i|2 B kxk2
kxk=1 n
X
= B |hy, yn i|2 .
n
Thus {yn } has a lower frame bound of B −1 and hence is a frame. By the
above work, {xn } is a dual for {yn }, so {xn } is an alternative dual frame for
{yn }.
(b) ⇒. Suppose that {yn } = {e xn } is the canonical dual for {xn }. Then it
is certainly an alternative dual, and given x ∈ H we have

e = hx, x
Cx en i = hx, S −1 xn i = hS −1 x, xn i = CS −1 x.
e
Since S is a topological isomorphism, it follows that range(C) = range(C).
⇐. Suppose that {yn } is a Bessel sequence that is an alternative dual to
e Since {yn } is Bessel, by definition
{xn } that satisfies range(C) = range(C).
of alternative dual we have
X
x = e = C ∗ Cx,
hx, yn i xn = RCx x ∈ H.
n
Also, by part (a) we know that {xn } is an alternative dual to {yn }, so

X
x = e
hx, xn i yn = RCx e∗ Cx,
= C x ∈ H.
n
Hence
e = I = C
C∗C e ∗ C.
Therefore
e ∗C
CC e = C
e
e ∗ z = z for z ∈ range(C)
and hence CC e = range(C). Therefore
e ∗ Cx = Cx,
CC x ∈ H.
e ∗ C = C. Taking adjoints,
That is, CC
e ∗ C)∗ = C ∗ C C
C ∗ = (CC e∗ ,
so
e∗ δn = C ∗ Cyn = Syn ,
xn = C ∗ δn = C ∗ C C
where S is the frame operator for {xn }. Hence yn = S −1 xn = x
en , so {yn } is
the canonical dual frame for {xn }.
8.17 (a) Let A, B be frame bounds for {xn }. Then given x ∈ M we have
X X X
|hx, P xn i|2 = |hP x, xn i|2 = |hx, xn i|2 ≤ B kxk2 ,
n n n
and similarly for the lower frame bound. Hence {P xn } is a frame for M with
frame bounds A, B.
(b) Suppose that P S = SP. Since S is a topological isomorphism, it follows
that
S −1 P = S −1 P SS −1 = S −1 SP S −1 = P S −1 ,
so S −1 commutes with P as well. Let T be the frame operator for {P xn } as
a frame for M. Using the self-adjointness of P, given x ∈ H we compute that
X X
TPx = hP x, P xn i P xn = P hP 2 x, xn i xn
n n
= P SP x = SP 2 x = SP x.
In particular, if x ∈ M then T x = Sx, so T = S|M . Hence S maps M

bijectively onto itself and therefore T −1 = (S|M )−1 as mappings on M. The
elements of the canonical dual frame of {P xn } are therefore
T −1 (P xn ) = (S|M )−1 P xn = S −1 P xn = P S −1 xn = P x
en ,
en }.
so the canonical dual frame is {P x
8.18 (a) Given x ∈ H, we have

e =
hx, x
Given c ∈ ℓ2 , we have
X X X
e =
Rc en =
cn x cn S −1 xn = S −1 cn xn = S −1 Rc.
n n n
(b) Let A, B be frame bounds for {xn } and let C, P R be its analysis and
synthesis operators. Since {e xn } is a frame, the series en converges for all
cn x
e be the synthesis operator for {e
c = (cn ) ∈ ℓ2 . Let R xn }. Since B −1 , A−1 are
frame bounds for {e xn }, we have kRk e 2 ≤ A−1 . The operator
X

Pc = en , xk
cn x = hRc,e xk i = C Rc e
n k∈N
is well defined on ℓ2 , and it is bounded since
e 2 kck22 ≤ B kck22 .
kP ck2ℓ2 ≤ kCk2 kRk ℓ ℓ
A
Now, since {e e = I. Hence, if

xn } is the canonical dual frame, we have RC
c ∈ range(C), say c = Cx for some x ∈ H, then
e = C RCx
P c = C Rc e = Cx = c.
Also, if c ∈ range(C)⊥ = ker(R), then for any d ∈ ℓ2 we have
e = hRc, Rdi
hP c, di = hc, P di = hc, C Rdi e = h0, Rdi
e = 0,
so P c = 0. Hence P is the orthogonal projection of ℓ2 onto M.

8.19 (b) ⇒ (a). We are given that yn = V δn where V C = I. Since V is
bounded, Exercise 7.6 implies that {yn } is a Bessel sequence in H.
Taking adjoints of the equation V C = I, we see that RV ∗ = C ∗ V ∗ = I.
Given x ∈ H we therefore have that
X X
hx, yn i xn = hx, V δn i xn
n n
X
= hV ∗ x, δn i xn
n

= R hV ∗ x, δn i = RV ∗ x = x.
Therefore {yn } is an alternative dual to {xn }. Since {yn } is also Bessel, The-
orem 8.18(a) implies that {yn } is a frame.
(a) ⇒ (b). Suppose that {yn } is an alternative dual frame for {xn }, and
let V : ℓ2 → H be the synthesis operator for {yn }. Then V is bounded and
V δn = yn by definition. Further, Theorem 8.18(a) implies that {xn } is a dual
frame for {yn }, so if x ∈ H then
X
V Cx = V hx, xn i = hx, xn i yn = x.
n
Thus V is a bounded left-inverse of C.

8.20 Let {xn } be a frame for H with canonical dual frame {e xn } and frame
bounds A, B. Let {en } be an orthonormal basis for K. Assuming that the
interchange of series can be justified, we compute that
X XX
em i =
hT xm , T x em i
hT xm , en i hen , T x
m m n
XX
= hT ∗ en , x
em i hxm , T ∗ en i
n m
X X
∗ ∗
= T en , em
hT en , xm i x
n m
X
∗ ∗
= hT en , T en i
n
X
= kT ∗ en k2 = kT ∗ k2HS = kT k2HS.
n
We justify the interchange by showing that Fubini’s Theorem for series is

applicable:
XX
em i|
|hT xm , en i hen , T x
m n
XX 1/2 X 1/2

2 2
≤ |hT xm , en i| em i|
|hen , T x
m n n
XX 1/2 X 1/2

∗ 2 ∗ 2
≤ |hxm , T en i| em i|
|hT en , x
m n n
X
≤ B 1/2 kT ∗ em k A−1/2 kT ∗em k
m
1/2
B
= kT k2HS < ∞.
A
8.21 (a) ⇒ (c). If {xn } is an exact frame, then, by definition, {xn }n6=m is not
a frame for any m. It therefore follows from Theorem 8.22 that hxm , x em i = 1
for every m.
(c) ⇒ (b). This follows from Theorem 8.22(b).

(b) ⇒ (a). Suppose that {xn } is orthogonal. Since it is a frame it is com-
plete in H. But then no proper subsequence of {xn } can be complete, so {xn }
is an exact frame.
8.22 (a) All frames are Bessel sequences, and therefore are norm-bounded
above by Exercise 7.5. We can also see this directly from the following calcu-
lation: X
kxm k4 = |hxm , xm i|2 ≤ |hxm , xn i|2 ≤ B kxm k2 .
n
(b) If {xn } is an exact, then {xn } and {e xn } are biorthogonal by Corol-

lary 8.23. Applying the lower frame inequality to the vector x em , we therefore
have that
X
xm k2 ≤
A ke xm , xn i|2 = |he
|he xm , xm i|2 ≤ kexm k2 kxm k2 .
n
Since {xn } is exact we must have xm 6= 0. Since S is a topological isomor-

phism, we therefore have xem = S −1 xm 6= 0 as well, so we can divide by ke
xm k2
to obtain the desired inequality.
8.23 Suppose that {xn } is exact and {yn } is a dual frame. Then for every
x ∈ H we have X
x = hx, yn i xn .
n
Letting {txn } denote the canonical dual frame, biorthogonality implies that
X
em i =
hx, x em i = hx, ym i.
hx, yn i hxn , x
n
em = ym . Hence the canonical

Since this is true for every x we must have x
dual frame is the only dual frame.
8.24 Note that for each y ∈ K,
X X
T ST ∗y = T hT ∗ y, xn i xn = hy, T xn i T xn .
n n
If y ∈ H then hT ST ∗y, yi = hS(T ∗ y), (T ∗ y)i, so since AI ≤ S ≤ BI, it

follows that
A kT ∗ yk2 ≤ hT ST ∗y, yi ≤ B kT ∗yk2 .
Further, since T is a topological isomorphism,
kyk kyk
= ≤ kT ∗ yk ≤ kT ∗k kyk = kT k kyk.
−1
kT k kT ∗ −1 k
Combining these equations we find that
A kyk2
≤ hT ST ∗y, yi ≤ B kT k2 kyk2 ,
kT −1 k2
which is equivalent to the statement A kT −1 k−2 I ≤ T ST ∗ ≤ B kT k2 I. Rewrit-

ing this explicitly implies that {T xn } is a frame and that both statements (a)
and (b) hold. Finally, statement (c) regarding exactness follows from the fact
that topological isomorphisms preserve complete and incomplete sequences
(see Exercise 2.36).
8.25 Suppose that {xn } is a Riesz basis for H. By Theorem 8.19 we know
that {P xn } is a frame for M.
Suppose that {P xn } is a Riesz basis for M. Suppose that P x = 0 for some
x ∈ H. Then
X X
0 = Px = P en i xn =
hx, x en i P xn ,
hx, x
n n
and since {P xn } is ω-independent, it follows that hx, xen i = 0 for every n.

But the dual frame {e xn } is complete, so this implies that x = 0. Therefore
M ⊥ = {0}, and hence M = H.
8.26 We are given a frame {xn } with analysis and synthesis operators C,
e R
R, and C, e are the analysis and synthesis operators for the canonical dual
frame {e
xn }.
(a) Given x ∈ H we have

e =
hx, x
e
Since S is a topological isomorphism, it follows that range(C) = range(C).
(b) The synthesis operator R maps ℓ2 onto H. Let T be the operator R
restricted to ker(R)⊥ = range(C) = range(C).e Then T : range(C) → H is a
topological isomorphism. Since the range of T is all of H, the pseudoinverse
of R is T −1 : H → range(C). We must show that T −1 = C. e Since {e
xn } is the
e = I and RC
canonical dual, we have that RC e = I.
Note that T Ce : H → H. If x ∈ H, then
X
e = RCx
T −1 Cx e = en i xn = x.
hx, x
n
On the other hand, CT e : range(C) → range(C). If c ∈ range(C) =

e e
range(C), then c = Cx for some x ∈ H, so
e c = CR
CT e Cx
e = Cx
e = c.
e = A−1 , as desired.
Hence C
(c) The fact that kCk2 is the optimal upper frame bound for {xn } follows
from the definition:
X
|hx, xn i|2 = kCxk2ℓ2 ≤ kCk2 kxk2 ,
n
and kCk2 is the smallest constant having this property.

e 2 is the optimal upper frame bound
This reasoning also implies that kCk
for the canonical dual frame {exn }. Since A, B are frame bounds for {xn } if
and only if B −1 , A−1 are frame bounds for {e e −2 is the
xn }, it follows that kCk
optimal lower frame bound for {xn }.
Finally, since S = RC = C ∗ C is self-adjoint,
kSk = sup |hSx, xi| = sup |hC ∗ Cx, xi| = sup kCxk2 = kCk2 ,
kxk=1 kxk=1 kxk=1
e 2 since S −1 is the frame operator for the dual

and therefore kS −1 k = kCk
frame {e
xn }.
8.27 (a) ⇒ (b). If {xn } is a frame, then we know by Theorem 8.29 that C
maps H bijectively onto a closed subspace of ℓ2 . On the other hand, because
of uniqueness we know that Cxn = en for every n, so range(C) = ℓ2 .
(b) ⇒ (c). If C maps H bijectively onto ℓ2 , then {xn } is a Bessel sequence
by Theorem 7.4. Therefore C is a bounded bijection and hence is a topo-
logical isomorphism. Its adjoint R is therefore a topological isomorphism by
Theorem 2.31.
(c) ⇒ (d). Statement (c) implies that R is a topological isomorphism of ℓ2
onto H. Since Ren = xn , statement (d) follows.
(d) ⇒ (a). Statement (d) implies that the operator T is a topological
isomorphism, hence {xn } is a Riesz basis by definition.
(a) ⇒ (e). All Riesz bases are frames and are ω-independent.
(e) ⇒ (f). Follows trivially.
2
P (f) ⇒ (c). Suppose that {xn2} is an ℓ -independent frame. Then Rc =
cn xn converges for all (cn ) ∈ ℓ since {xn } is Bessel. Further, R is surjective
P {xn } is a frame. Suppose that Rc
since = 0 for some c ∈ ℓ2 . Then we have
2
cn xn = 0, so c = 0 since {xn } is ℓ -independent. Hence R is injective, so
statement (c) holds.
8.28 (a) ⇒. Theorem 7.2 shows that if {xn } is Bessel and complete then R
has dense range.
⇐. Suppose that R has dense range, and suppose that y ∈ H satisfies
hy, xn i = 0 for every n. Then hy, xi = 0 for every x ∈ span{xn }.
Suppose z ∈ H is given. Since R has dense range, there exists some y ∈
range(R) such that kz − yk < ε. Since Rδn = xn , we have xn ∈ range(R) for
every n. Hence span{xn } ⊆
P
(b) ⇒. Since {xn } is Bessel, we know that the series Rc = cn xn con-
verges unconditionally for every c = (cn ) ∈ ℓ2 . The statements “{xn } is
ℓ2 -independent” and “R is injective” are both equivalent to “Rc = 0 if and
only if c = 0.”
(c) Since {xP n } is a conditional basis, there exists some x0 ∈ H such that
the series x0 = hx, x en i xn does
not converge unconditionally. Since {xn } is
Bessel, it follows that hx0 , x / ℓ2 .
en i n∈Z ∈
Consider
P∞ the sequence {xn }n≥0 = {xn } ∪ {x0 }. Suppose that (cn ) ∈ ℓ2
and n=0 cn xn = 0. Then
∞
X
−c0 x0 = cn xn .
n=1
If c0 6= 0 then for every m > 0 we have

∞
X
em i =
−c0 hx0 , x em i = cm .
cn hxn , x
n=1
Hence
em i m∈N 6∈ ℓ2 ,
(cm )m∈N = −c0 hx0 , x
which is a contradiction. Therefore we must have c0 = 0. But then
∞
X
cn xn = 0,
n=1
which implies cn = 0 for all n > 0 since {xn } is a basis. Hence {xn }n≥0 is
ℓ2 -independent.
On the other hand, {xn }n≥0 is not ω-independent because we have
∞
X
−x0 + en i xn = −x0 + x0 = 0.
hx, x
n=1
8.29 (a) ⇐. Since PM is an orthogonal projection, Exercise 8.8 shows that

{PM δN } is a Parseval frame for M. Since T : M → H is a topological isomor-
phism and frames are preserved by topological isomorphisms, it follows that
{xn } is a frame for H.
(b) ⇒. Assume that {xn } is a Parseval frame, and let M and T be as in
the proof of part (a). In particular, T is the topological isomorphism obtained
by restricting the synthesis operator R to M = range(C). Given c ∈ range(C),
we have c = Cx for some x ∈ H. Since {xn } is Parseval, S = RC = I. Further,
kCxkℓ2 = kxk, again by Parsevality, so
kT ck = kRck = kRCxk = kxk = kCxkℓ2 = kckℓ2 .

Thus T is isometric on M.
(b) ⇐ . We have that {PM δn } is a Parseval frame for M, and since T
is an isometric isomorphism it follows that the image of {PM δn } under T is
Parseval as well.
8.30 (a) ⇒. Let M and T be as in Corollary 8.29. Define K = H × M ⊥ . By
Exercise 1.40, this is a Hilbert space with respect to the natural inner product.
Define U : K → ℓ2 by U (x, c) = T −1 x + c. Since T −1 x ∈ M and c ∈ M ⊥ , we
have
kU (x, c)k2ℓ2 = kT −1 x + ck2ℓ2 = kT −1 xk2ℓ2 + kck2ℓ2 .
Since T : M → H is a topological isomorphism, it follows that U is a topolog-
ical isomorphism as well. Let V = U −1 . Given a ∈ ℓ2 , write a = b + c where
b ∈ M and c ∈ M ⊥ . Then U (T b, c) = b + c = a, so V a = (T b, c).
Let δn = bn + cn where bn = PM δn ∈ M and cn = δn − bn ∈ M ⊥ . Let
en = V δn . Then {en } is a Riesz basis for K. Further,
PH en = PH V δn = PH (T bn , cn ) = T bn = T PM δn = xn .
Therefore the result follows if we identify H with H × {0} ⊆ K.

⇐. Assume that {xn } = {P en } where {en } is a Riesz basis for K and P
is the orthogonal projection of K onto H. Given x ∈ H we have
X X X X
|hx, xn i|2 = |hx, P en i|2 = |hP x, en i|2 = |hx, en i|2 .
n n n n
P
Now, {en } is a frame, so A kxk2 ≤ |hx, en i|2 ≤ B kxk2 for all x ∈ K.
Therefore {xn } is a frame for H with frame bounds A, B.
(b) ⇒. The mapping T in part (a) is now an isometric isomorphism, so U
and V are isometric isomorphisms as well. Therefore {en } = {V δn } is an
orthonormal basis, and hence its projection onto H is a Parseval frame.
⇐. In this case the frame bounds are A = B = 1, so {xn } is Parseval.
8.31 (a) For each m ∈ N let {em d
n }n=1,...,d be an orthonormal basis for F .
d
Given x ∈ F we have
∞ X
X d ∞
X
|hx, 2−m/2 em 2
n i| = 2−m kxk2 = kxk2 ,
m=1 n=1 m=1
so {2−m/2 em d
n }m∈N,n=1,...,d is a Parseval frame for F .
(b) If {xn }n∈I is a frame that is bounded below then its associated canon-
ical Parseval frame {S 1/2 xn }n∈I is also a frame that is bounded below, so it
suffices to assume that {xn }n∈I is Parseval.
By Corollary 8.29 there exists some Hilbert space K ⊇ Fd and some or-
thonormal basis {en }n∈I for K such that xn = P en where P is the orthogonal
projection of K onto Fd . But then Exercise 1.53 implies that
X X
d = dim(Fd ) = kP en k2 = kxn k2 = |I|,
n∈I n∈I
so |I| must be finite.

8.32 (a) ⇒. If T : H → K is a topological isomorphism such that T xn = yn
then we have

CY y = hy, yn i = hy, T xn i = hT ∗ y, xn i = CX T ∗ y.
Since T ∗ is a topological isomorphism, it follows that range(CX ) = range(CY ).

⇐. Assume that M = range(CX ) = range(CY ). By Corollary 8.33, there
exists a topological isomorphism T : M → H such that xn = T PM δn , and
similarly there exists a topological isomorphism U : M → K such that yn =
U PM δn , Therefore U T −1 : H → K is a topological isomorphism, and we have
U T −1 xn = U PM δn = yn , so {xn } is equivalent to {yn }.
(b) Since ker(RX ) = range(CX )⊥ and ker(RY ) = range(CY )⊥ , we have
that range(CX ) = range(CYP) if and only if ker(RX ) = ker(RY ). This
P is equiv-
alent to the statement that cn xn = 0 for (cn ) ∈ ℓ2 if and only if cn yn = 0
for (cn ) ∈ ℓ2 .
(c) Let {en } be an orthonormal basis for a Hilbert space H. Let {xn } =
{e1 , e1 , e2 , e3 , e4 , · · · } and {yn } = {e1 , e2 , e1 , e3 , e4 , · · · }. Then {yn } is ob-
tained from {xn } by the permutation σ = (1 2) of the index set N. However,
(1, 1, 0, 0, . . . ) belongs to range(CX ) but does not belong to range(CY ), so
these two frames are not equivalent.
All exact frames are Riesz bases, which are all equivalent, so this cannot
happen for exact frames.
(d) Let {en } be an orthonormal basis for a Hilbert space H. Let {xn } =
{e1 , e1 , e2 , e3 , e4 , · · · } and {yn } = {e1 , −e1 , e2 , e3 , e4 , · · · }. Then {yn } is ob-
tained from {xn } by a single change of sign. However, (1, 1, 0, 0, . . . ) belongs
to range(CX ) but does not belong to range(CY ), so these two frames are not
equivalent.
8.33 ⇒. Suppose that {xn } is a Parseval frame for H. By Corollary 8.33,
there exists a Hilbert space K ⊇ H and an orthonormal basis {eN } for K
such that xn = P en for every n. Then I − P is the orthogonal projection
onto H ⊥ , so if we set yn = (I − P )en then {yn } is a Parseval frame for H ⊥ .
Given (x, y) ∈ H × H ⊥ , we have
X
X
(x, y), (xn , yn ) 2 = |hx, xn i|2 + |hy, yn i|2
n n
2
= kxk2 + kyk2 = (x, y) H×H ⊥ .

Hence (xn , yn ) is a Parseval frame for H × H ⊥ . By the Pythagorean The-
orem,

(xn , yn ) 2 = kxn k2 + kyn k2 = kxn + yn k2 = ken k2 = 1.
H×H ⊥

Hence (xn , yn ) is an orthonormal basis.
⇐ . Suppose
that
there exists a Parseval frame {yn } for a Hilbert space K
such that (xn , yn ) is an orthonormal basis for H × K. Then
X X X
|hx, xn i|2 = |hx, xn i|2 + |h0, yn i|2
n n n
X 2
= |h(x, 0), (xn , yn i|2 = (x, 0) H×K = kxk2 ,
n
so {xn } is Parseval.
8.34 Since M and H are separable and infinite dimensional, there exists an
isometric isomorphism T : H → M. Since {PM δn } is a Parseval frame for M,
{xn } = {T −1 PM δn } is a Parseval frame for H. The analysis operator for {xn }
is

Cx = hx, xn i = x, T −1 PM δn

∗ −1
= (T ) x, PM δn

= hPM (T ∗ )−1 x, δn i
= PM (T ∗ )−1 x.
Since (T ∗ )−1 maps onto M, we conclude that range(C) = M.

8.35 Let CX be the analysis operator for the frame {xn } in H and CY the
analysis operator for the frame {yn } in K.
(a) The inner product and norm on H ×K are constructed in Exercise 1.40.
Let AX , BX be frame bounds for {xn } and let AY , BY be frame bounds for
{yn }. Let A = min{AX , AY } and B = max{BX , BY }. Let C be the analysis
operator for the sequence {(xn , yn )}. Then given (x, y) ∈ H × K we have

C(x, y) = (x, y), (xn , yn ) = hx, xn i + hy, yn i = CX x + CY y.
Therefore, if range(CX ) ⊥ range(CY ) then we have by the Pythagorean The-

orem that

C(x, y) 22 = CX x 22 + CY y 22
ℓ ℓ ℓ
≥ AX kxk2 + AY kyk2

≥ A kxk2 + kyk2
2
= A (x, y) H×K ,
and similarly B is an upper frame bound.


Consequently, if {xn } and {yn } are Parseval then A = B = 1, so (xn , yn )
is Parseval.
Conversely, suppose that (xn , yn ) is Parseval. Considering y = 0, we
have

CX x 22 = CX x 22 + CY 0 22 = C(x, 0) 22 = (x, 0) 2 = kxk2 .
ℓ ℓ ℓ ℓ H×K
Hence AX = 1 is a lower frame bound for {xn }, and similarly for the other
frame bounds. Therefore {xn } and {yn } are Parseval.
(b) Assume that range(CX ) ⊕ range(CY ) = ℓ2 . Then by the calculation in
part (a) we have range(C)
= range(CX ) + range(CY ) = ℓ2 , so it follows from
Theorem 8.32 that (xn , yn ) is a Riesz basis for H × K.
A Riesz basis is an orthonormal basis if and only
if it is a Parseval frame,
so by combining with part (a) we conclude that (xn , yn ) is orthonormal if
and only if both {xn } and {yn } are Parseval.
(c) Let M be a closed subspace of ℓ2 such that both M and M ⊥ are
infinite dimensional. Then {PM δn } is a Parseval frame for M, but since M is
proper subspace this frame cannot be a Riesz basis (Exercise 8.25). Likewise
{PM ⊥ δn } is a Parseval frame for M ⊥ that is not a Riesz basis. The analysis
operator for {PM δn } is

CM x = hx, PM δn i = hPM x, δn i = PM x,
i.e., CM = PM . Likewise the analysis operator for {PM ⊥ δ n } is CM ⊥ = PM ⊥ .

Hence range(CM )⊕range(CM ⊥ ) = ℓ2 , so (PM δ n , PM is an orthonormal
⊥ δn )
basis for M ×M ⊥ . Note that this basis is not PM δn ∪ PM ⊥ δn ) . The latter

system is a 2-tight frame for ℓ2 , but it is not orthonormal.
More trivially, consider {δ1 , 0, δ2 , 0, . . . } and {0, δ1 , 0, δ2 , 0, . . . } as Parseval
frames for their closed spans. Then

(δ1 , 0), (0, δ1 ), (δ2 , 0), (0, δ2 , ), . . .
is an orthonormal basis for ℓ2 × ℓ2 .

(d) Suppose {xn } is a frame for H, and let M = range(C). By Exer-
e = M ⊥ , where C
cise 8.34, there exists a frame {yn } for H such that range(C) e

is the analysis operator for {yn }. Part (b) therefore implies that (xn , yn ) is
a Riesz basis for H × H.
8.36 (a) Suppose that {xm } and {yn } are frames, and let SX and SY denote
their frame operators. It follows from Exercise B.6 that x ⊗ y is a sesquilinear
form (antilinear in the first variable) that is continuous is both variables. By
Exercise B.9 there exists a unique topological isomorphism SX ⊗ SY that
satisfies (SX ⊗ SY )(x ⊗ y) = SX x ⊗ SY y for all x ∈ H, y ∈ K. Also, by
Exercise 7.11, {xm ⊗yn }m,n∈N is Bessel in H ⊗K, and therefore has a bounded
frame operator that we will call T. Now,
(SX ⊗ SY )(x ⊗ y) = SX x ⊗ SY y
X X
= hx, xm i xm ⊗ hy, yn i yn
m n
XX
= hxm , xi hy, yn i (xm ⊗ yn )
m n
X

= x ⊗ y, xm ⊗ yn (xm ⊗ yn )
m,n
= T (x ⊗ y).
By the uniqueness statements in Exercise B.9, we conclude that T = SX ⊗ SY
is a topological isomorphism.
This implies that {xm ⊗ yn }m,n∈N is a frame. For example, if we let R
denote the analysis operator for {xm ⊗ yn }m,n∈N , then we have T = RR∗ and
therefore R must be surjective since T is a topological isomorphism. It follows
from Theorem 8.29 that {xm ⊗ yn }m,n∈N is a frame.
(b) This follows from the identification of L2 (E × F ) with L2 (E) ⊗ L2 (F )
given in Theorem B.13.
8.37 (a) ⇒ (b). Suppose Pthat {xn }n∈F/ is a Riesz basis forP H, where F is a
2
finite subset of N. Then n∈F / c n xn converges if and only if / |cn | < ∞.
n∈F
Since F is finite, it follows that the same equivalence holds for the summation
over all n.
w
(b) ⇒ (c). We verify the next step in the construction. Since φn → 0, there
exists an m3 > m2 such that
X
N2
hφ , δ i δ < 1.
m 3 n n 2 8
n=1 ℓ
Then, similarly to the calculation at the previous step,

N3 N3
X X
hφ , δ i δ ≥ hφ , δ i δ − 1 → 1 − 1,
m3 n n
2
m3 n n
8 8
n=N2 +1 ℓ n=1 ℓ2
so we can choose N3 large enough that this quantity exceeds 1/2. Similarly,
N3 N3
X X 1 1
hφm3 , δn i P δn hφm3 , δn i δn
2 ≤ 2 + 8 → 0 + 8,
n=N2 +1 ℓ n=1 ℓ
so we can choose N3 large enough that this quantity is at most 1/4.

To show that the series converges, set
Nk+1
X
ψk = k −1/2 hφmk , δn i P δn .
n=Nk +1
By hypothesis, we have
kψk kℓ2 ≤ k −1/2 2−k .
Also, for any M < N we have

X X
N N
hφmj , δn i P δn
≤ kP k hφmj , δn i δn

n=M ℓ2 n=M ℓ2
X
N 1/2
= |hφmj , δn i|2
n=M
≤ kφmj kℓ2 = 1.
Fix any N1 + 1 ≤ M < N < ∞. Let j be the smallest integer such that
Nj ≤ M < Nj+1 and let ℓ be the largest integer such that Nℓ < N ≤ Nℓ+1 .
Then
N Nj+1
X X −1/2
c P δ ≤ j hφ , δ i P δ
n n
2
mj n n
2
n=M ℓ n=M ℓ
ℓ Nk+1
X X
+
k −1/2 hφmk , δn i P δn

k=j+1 n=Nk +1 ℓ2
X
N

+ ℓ −1/2
hφmℓ , δn i P δn

n=Nℓ +1 ℓ2
ℓ
X
≤ j −1/2 + kψk kℓ2 + ℓ−1/2
k=j+1
ℓ
X
≤ j −1/2 + k −1/2 2−k + ℓ−1/2 .
k=j+1
Since j → ∞ as M → ∞, it follows that the series is Cauchy.

On the other hand,
∞ ∞ Nk+1
X X X
|cn |2 = k −1 |hφmk , δn i|2
n=N1 +1 k=1 n=Nk +1
∞ Nk+1 2
X X
= k −1
hφ , δ i
mk n n δ
k=1 n=Nk +1 ℓ2
∞
X 1
≥ k −1 = ∞.
4
k=1
(c) ⇒ (a). We verify that the sequence {P δn }n∈F ∪ {P δn }∞ n=N +1 is

a Riesz basis for M. We know that {P δn }∞ n=N +1 is a Riesz basis for its
span K within M, and we let F be a minimal subset of {1, . . . , N } such
that {P δn }n∈F ∪ {P δn }∞n=N +1 is complete in M. Let L = span{P δn }n∈F .
Then K + L = M.
For simplicity, let F = {1, . . . , m}. We cannot have P δ1 ∈ K since if it
was then F would not be minimal. Let K1 = span{K, P δ1 }. We cannot have
P δ2 ∈ K1 , and so forth. Suppose that there exist scalars cn such that
m
X ∞
X
cn P δ n + cn P δn = 0.
n=1 n=N +1
Then cm P δm ∈ Km−1 , which implies cm = 0. Continuing in this way, c1 =

· · · = cm = 0, and therefore cn = 0 for all n ≥ N + 1 since {P δn }∞ n=N +1 is a
Riesz sequence. Thus {P δn }n∈F ∪ {P δn }∞ n=N +1 is ω-independent.
If x ∈ M then we can write x P = y + z with yP∈ K and z ∈ L. Then
∞
there exist scalars cn such that x = n∈F cn P δn + n=N +1 cn P δn , and this
series converges unconditionally since F is finite and {P δn }∞ n=N +1 is a Riesz
sequence. The scalars must be unique since we have ω-independence. Thus
{P δn }n∈F ∪ {P δn }∞n=N +1 is an unconditional basis for M. It is a bounded
basis as well since F is finite, and therefore is a Riesz basis for M.
(b) ⇔ (d), assuming {xn } is bounded below. This follows immediately
from the definitions and Theorem 8.36.
8.38 Row-reduction will leave a matrix that has N pivots. The remaining
columns are all finite linear combinations of the pivot columns.
8.39 Fix ε > 0. Set m1 = 1. Since
lim hym1 , yk i = 0,
k→∞
1
we can choose vectors ym k
such that
1 ε
|hym1 , ym i| < , k > 1.
k
2k+1
Then we have X
1 ε
|hym1 , ym i| < .
k
2
k>1
1
Set m2 = ym 1
. Since
lim hym2 , yk i = 0,
k→∞
2 1
we can choose a subsequence {ym k
} of {ym k
} such that
2 ε
|hym2 , ym i| < , k > 2.
k
2k+2
Then we have
X ε
2
|hym2 , ym i| < .
k
4
k>2
Continuing in this way, we obtain a subsequence {ymk } such that

X ε
|hymk , ymj i| < .
2k
j>k
Therefore XX X ε
|hymk , ymj i| ≤ = ε.
2k
k j>k k
Hence XX XX
|hymk , ymj i| ≤ 2 |hymk , ymj i| ≤ 2ε.
k j6=k k j>k
8.40 (a) ⇒ (b). Assume that statement (a) holds. We are given that {xn }n6=nk
has a lower frame bound of L. Since S −1/2 is a topological isomorphsm and
kS 1/2 k2 ≤ B, Exercise 8.24 implies that {S −1/2 xn }n6=nk is a frame with
lower frame bound L/B. However, since {S −1/2 xn } is a Parseval frame, Ex-
ercise 8.6 implies that the optimal lower frame bound for {S −1/2 xn }n6=nk is
1 − kS −1/2 xnk k2 . Therefore we must have L/B ≤ 1 − kS −1/2 xnk k2 , so
2 L
enk i = S −1/2 xnk ≤ 1 − .
hxnk , x
B
(b) ⇒ (a). Assume that D = supk hxnk , x enk i < 1. Fix any particular n.
Then 1 − kS 1/2 xnk k2 ≥ 1 − D > 0. Since 1 − kS 1/2 xnk k2 is the optimal lower
bound for {S −1/2 xn }n6=nk , we conclude that 1 − D is a lower frame bound.
Since S 1/2 is a topological isomorphism with kS −1/2 k2 ≤ 1/A, it follows
by Exercise 8.24 that {xn }n6=nk is a frame for H with lower frame bound
L = A(1 − D).
Conclusion. Suppose that statements (a), (b) hold. Then by the argument
2
above, we have that S −1/2 xnk ≤ 1 − L/B for every k. Since {S −1/2 xn }
is a Parseval frame, Theorem 8.44 implies that there exists an infinite set
J ⊆ {nk } such that {S −1/2 xn }n∈J
/ is a frame with frame bounds L/B − ε, 1.
Since S 1/2 is a topological isomorphism, kS −1/2 k2 ≤ A, and kS 1/2 k2 ≤ B, it
follows that {xn }n∈J
/ is a frame with frame bounds LA/B − Aε, B.
8.41 Note that
RV ∗ = C ∗ V ∗ = (V C)∗ = (CU )∗ = U ∗ C ∗ = U ∗ R.
(a) If x ∈ ker(C), then CU x = V Cx = V 0 = 0, so U x ∈ ker(C).

(b) If c ∈ ker(R) = ker(C ∗ ), then RV ∗ c = U ∗ Rc = 0, so V ∗ c ∈ ker(R).
(c) Since ker(C) is invariant under U, Theorem 2.13 implies that
ker(C)⊥ = range(C ∗ ) = range(R)
is invariant under U ∗ .
(d) Similar to part (c).
8.42 (i) Suppose that U f = λf for some nontrivial f ∈ L2 (T) and some
λ ∈ C. Then e2πibt f (t) = λf (t) a.e. Since f is not the zero function, there
must be a set E with positive measure such that f (t) 6= 0 for t ∈ E. But then
e2πibt = λ for t ∈ E. Since the set of t such that e2πibt takes a common value
is countable, this is a contradiction. Hence U has no eigenvalues.
(ii) Suppose that V c = λc for some nonzero c ∈ ℓ2 (Z) and λ ∈ C. If λ = 0
then (cn−1 )n∈Z = V c = 0, which implies c = 0, so λ = 0 is not an eigenvalue.
But then we have (cn−1 )n∈Z = V c = (λcn )n∈Z , so cn = λn c0 for n ∈ Z. We
must have c0 6= 0 since otherwise cn = 0 for every n. But if c0 6= 0 then
we cannot have c ∈ ℓ2 (Z), which again is a contradiction. Hence V has no
eigenvalues. Note that it is important here that we are doing a shift-operator
on ℓ2 (Z), not ℓ2 (N) (compare Exercise 2.19).
(iii) The adjoint of U is multiplication by e−2πibt , so the same proof as
part (a) shows that U ∗ has no eigenvalues.
(iv) The adjoint of V is the left-shift operator on ℓ2 (Z), so the same proof
as part (b) shows that V ∗ has no eigenvalues.
(b) Exercise 8.9 tells us that {e2πibnt }n∈Z is a b−1 -tight frame. Renormal-
izing, {be2πibnt }n∈Z is a Parseval frame. However, each element has norm
kbe2πibnt kL2 (T) = b < 1.
Theorem 8.44 therefore implies that an infinite subset can be removed yet still
leave a frame.
9.2 (a) Suppose that f ∈ C0 (R), and choose ε > 0. Then there exists an
R > 0 such that |f (x)| < ε/2 for all |x| > R − 1. Define


 f (x), |x| ≤ R + 1,

g(x) = linear, R + 1 ≤ |x| ≤ R + 2,



0, |x| ≥ R + 2.
Then g ∈ Cc (R), and therefore is uniformly continuous. Hence, there exists

0 < δ < 1 such that
|x − y| < δ =⇒ |g(x) − g(y)| < ε.
Fix |a| < δ. If |x| < R then |x − a| ≤ |x| + |a| < R + 1, so
|f (x) − Ta f (x)| = |g(x) − Ta g(x)| < ε.
On the other hand, if |x| ≥ R, then |x − a| ≥ |x| − |a| > R − 1, so

ε ε
|f (x) − Ta f (x)| ≤ |f (x)| + |f (x − a)| < + = ε.
2 2
Thus kf − Ta f k∞ < ε whenever |a| < δ, so f is uniformly continuous.
The function f (x) = sin x2 is continuous and bounded, but is not uniformly
continuous.
(b) Fix 1 ≤ p < ∞, and choose f ∈ Lp (R). Given ε > 0, we can find
g ∈ Cc (R) such that kf − gkp < ε. Fix R > 0 such that supp(g) ⊆ [−R, R].
Since g is uniformly continuous, there exists a δ > 0 such that
ε
|a| < δ =⇒ kg − Ta gk∞ < .
(2R)1/p
Therefore, for such a we have

Z R Z R
εp
kg − Ta gkpp = |g(x) − Ta g(x)|p dx ≤ dx = εp .
−R −R 2R
Since translation is isometric on Lp (R), we therefore have for |a| < δ that
kf − Ta f kp ≤ kf − gkp + kg − Ta gkp + kTa g − Ta f kp
≤ ε + ε + ε = 3ε.
Hence Ta f → f in Lp (R) as a → 0.
9.3 If g ∈ L1 (R) then
Z Z
g(x − a) e−2πiξx dx = g(x) e−2πiξ(x+a) dx
∧
(Ta g) (ξ) =
Z
= e−2πiaξ g(x) e−2πiξx dx
= e−2πiaξ gb(ξ) = M−a gb(ξ)
and
Z
e2πibx g(x) e−2πiξx dx
∧
(Mb g) (ξ) =
Z
= g(x) e−2πi(ξ−b)x dx = b
g(ξ − b) = Tb b
g(ξ).
Also, making the change of variables y = rx,

Z
r1/2 g(rx) e−2πiξx dx
∧
(Dr g) (ξ) =
Z
1/2 dy
= r g(y) e−2πiξy/r
r
Z
= r−1/2 g(y) e−2πiyξ/r dy
= r−1/2 gb(ξ/r) = D1/r gb(ξ).
9.4 Consider the case m = 1. Assume that f (x), xf (x) ∈ L1 (R). We must
show that the limit
′ fb(ξ + η) − fb(ξ)
fb (ξ) = lim
η→0 η
Z
e−2πi(ξ+η)x − e−2πiξx
= lim f (x) dx (A)
η→0 η
exists. We do this by applying the Lebesgue Dominated Convergence Theo-
rem. Define ex (ξ) = e−2πiξx . Then the integrand in equation (A) converges
pointwise for almost every x, because
ex (ξ + η) − ex (ξ)
lim f (x) = f (x) e′x (ξ) = f (x) (−2πix) e−2πiξx .
η→0 η
Further, since we always have |eiθ − 1| ≤ |θ|, we have

−2πiηx
−2πi(ξ+η)x
− e−2πiξx − 1
f (x) e = |f (x)| |e −2πiξx e
|
η η
≤ 2π|xf (x)| ∈ L1 (R).

Hence the Lebesgue Dominated Convergence Theorem implies that the limit
in equation (A) exists, and that we have
Z
′ e−2πi(ξ+η)x − e−2πiξx
fb (ξ) = lim f (x) dx
η→0 η
Z
= f (x) (−2πix) e−2πiξx dx
∧
= (−2πix)f (x) (ξ).
Technically, the Dominated Convergence Theorem applies to sequences in-

dexed by the natural numbers, whereas above we need to let the continuous
variable η converge to zero. However, this can be justified by considering all
possible sequences ηk → 0.
In any case, we conclude that fb is differentiable. Further, since fb ′ is the
Fourier transform of the integrable function (−2πix)f (x), we have by the
Riemann–Lebesgue Lemma that fb ′ ∈ C0 (R). Since we also have fb ∈ C0 (R),
we have shown that fb ∈ C01 (R).
The result extends to m > 1 by induction.
9.5 (a) We have
Z Z
(fe) (ξ) =
∧
f (−x) e−2πiξx dx = f (x) e2πiξx dx
Z
= f (x) e−2πiξx dx = fb(ξ).
(b) The function f (y) g(x − y) is measurable on R2 , and we have

Z ∞Z ∞ Z ∞ Z ∞
|f (y) g(x − y)| dx dy = |f (y)| |g(x − y)| dx dy
−∞ −∞ −∞ −∞
Z ∞
= kgkL1 |f (y)| dy
−∞
= kf k1 kgk1 .
R
Fubini’s Theorem therefore implies that (f ∗ g)(x) = f (y) g(x − y) dy exists
for almost every x and is an integrable function of x.
(c) Fubini’s Theorem allows us to interchange integrals in the following
calculation:
Z
(f ∗ g)(x) e−2πiξx dx
∧
(f ∗ g) (ξ) =
ZZ
= f (y) g(x − y) e−2πiξx dy dx
Z Z
= f (y) e−2πiξy g(x − y) e−2πiξ(x−y) dx dy
Z Z
= f (y) e−2πiξy g(x) e−2πiξx dx dy
Z
= f (y) e−2πiξy gb(ξ) dy
= fb(ξ) gb(ξ).
(d) By definition, supp(f ) is the smallest closed set such that f = 0 a.e.
off supp(f ), and by hypothesis supp(f ) is compact. Therefore, we can fix a
representative of f such that f (x) = 0 for every x ∈/ supp(f ), and likewise we
fix similar representatives of g and f ∗ g. Let
K = supp(f ) + supp(g).
R
Suppose that (f ∗ g)(x) = f (y) g(x − y) dy was nonzero. Then there must be
a y such that y ∈ supp(f ) and x − y ∈ supp(g) simultaneously, for otherwise
f (y) g(x − y) would be zero for every y and therefore f ∗ g(x) would be zero.
Hence, x = y + (x − y) ∈ supp(f ) + supp(g). This shows that
{x ∈ R : (f ∗ g)(x) 6= 0} ⊆ supp(f ) + supp(g) = K.
Hence, for our chosen representative of f ∗ g, we have (f ∗ g)(x) = 0 for every

x∈/ K. But then, as an equivalence class of functions, f ∗ g = 0 a.e. off K.
Since K is closed, this implies that supp(f ∗ g) ⊆ K.
(e) We have that
Z
(f ∗ g)(x + h) − (f ∗ g)(x) g(x + h − y) − g(x − y)
= f (y) dy.
h h
The integrand converges pointwise a.e. to f (y) g ′ (x − y) as h → 0, and g ′ is

bounded by hypothesis. By the Mean Value Theorem, given x, y, and h there
exists a point c such that

g(x + h − y) − g(x − y)
= |g ′ (c)| ≤ kg ′ k∞ .
h
As a function of y,

f (y) g(x + h − y) − g(x − y) ≤ |f (y)| kg ′ k∞ ∈ L1 (R).
h
The Lebesgue Dominated Convergence Theorem therefore implies that

(f ∗ g)(x + h) − (f ∗ g)(x)
(f ∗ g)′ (x) = lim
h→0 h
Z
g(x + h − y) − g(x − y)
= lim f (y) dy
h→0 h
Z
= f (y) g ′ (x − y) dy = (f ∗ g ′ )(x).
Thus f ∗ g is differentiable.
9.6 (b) Because L1 (R) ∩ L2 (R) is dense in L2 (R), the relations from part (a)
continue to hold when we extend the Fourier transform to L2 (R). For example,
if we fix g ∈ L2 (R) then there exist functions fn ∈ L1 (R) ∩ L2 (R) such that
fn → g in L2 -norm. As the Fourier transform is unitary, we also have fbn → b g
in L2 -norm. By passing to subsequences if necessary, we can assume that these
convergences hold pointwise almost everywhere as well (see Theorem A.23).
Since (Ta fn ) = M−a fbn , it follows that (Ta g) = M−a g. This is an equality
∧ ∧
of functions in L2 (R), so is a pointwise almost everywhere equality.

9.7 (a) If f ∈ L1 (R) is even, then (making the change of variables x 7→ −x),
Z ∞
b
f (−ξ) = f (x) e−2πi(−ξ)x dx
−∞
Z −∞
= − f (−x) e−2πi(−ξ)(−x) dx
∞
Z ∞
−∞
If f ∈ L1 (R) is real, then

Z ∞
fb(−ξ) = f (x) e−2πi(−ξ)x dx
−∞
Z −∞
∞
If f is also even, then we know fb is even, so in this case we have fb(ξ) =

fb(−ξ) = fb(ξ), so fb is real.
(b) If f ∈ L2 (R) is real and even, then fn = f · χ[−n,n] is real and even
and fn → f in L2 -norm. Since fn ∈ L1 (R), its Fourier transform fbn is both
real and even by part (a). Since fbn → fb in L2 -norm and convergence in L2 -
norm implies the existence of a subsequence that converges pointwise a.e., we
conclude that fb is real and even (in the almost everywhere sense).
∨
9.8 (a) Suppose f, fb ∈ L1 (R), e.g., f ∈ Cc2 (R). Then we have f = fb from
the Inversion Formula, so
∨
fb (−ξ) = f (−ξ).
∧∧
f (ξ) =
(b) Since Cc2 (R) is dense in L2 (R), given g ∈ L2 (R) we can find
fn ∈ Cc2 (R) such that fn → g in L2 -norm. Then there is a subsequence that
converges pointwise almost everywhere, so by applying part (a) and taking
∧∧
pointwise limits we obtain g (ξ) = g(−ξ) for almost every ξ.
9.9 Assume that f, g are absolutely continuous on [a, b]. Then f ′ , g ′ ∈ L1 [a, b],
so f ′ (x) g ′ (y) ∈ L1 [a, b]2 . Letting E = {(x, y) ∈ [a, b]2 : x ≤ y}, we compute
that
ZZ Z b Z y
f ′ (x) g ′ (y) dx dy = f ′ (x) dx g ′ (y) dy
E a a
Z b
= f (y) − f (a) g ′ (y) dy
a
Z b Z b
= f (y) g ′ (y) dy − f (a) g ′ (y) dy
a a
Z b
= f (y) g ′ (y) dy − f (a) g(b) − g(a) .
a
On the other hand,

ZZ Z b Z b
f ′ (x) g ′ (y) dy dx = f ′ (x) g ′ (y) dy dx
E a x
Z b
= f ′ (x) g(b) − g(x) dy
a
Z b Z b
= g(b) f ′ (x) dx − f ′ (x) g(x) dx
a a
Z b

= g(b) f (b) − f (a) − f ′ (x) g(x) dx.
a
Fubini’s Theorem implies that these two integrals are equal, so by rearranging
we obtain
Z b Z b
′
f (x) g (x) dx = f (b)g(b) − f (a)g(a) − f ′ (x) g(x) dx.
a a
∨
10.1 Fix f ∈ L2[−Ω,Ω] (R). Then f, f ∈ L2 (R) and supp(f ) ⊆ [−Ω, Ω].
∨
Consequently f is bandlimited to [−Ω, Ω] because

∨ ∧
supp( f ) = supp(f ) ⊆ [−Ω, Ω].
∨
Hence f ∈ FL2[−Ω,Ω] (R), so the inverse Fourier transform maps L2[−Ω,Ω] (R)
into FL2[−Ω,Ω] (R).
To see that this mapping is surjective, fix g ∈ FL2[−Ω,Ω](R). Then g,
g ∈ L2 (R) and supp(b
b g ) ⊆ [−Ω, Ω]. Hence gb ∈ L2[−Ω,Ω](R), and it is mapped
to g by the inverse Fourier transform.
The argument for the Fourier transform is similar, based on the facts that
∨
fb(ξ) = f (−ξ) a.e., and the interval [−Ω, Ω] is symmetric about the origin.
10.2 Suppose f ∈ FL2[−Ω,Ω] (R), so in particular we have supp(fb) ⊆ [−Ω, Ω].
Since
(Dr f ) (ξ) = D1/r fb(ξ) = r−1/2 fb(ξ/r),
∧
∧ ∧
the function (Dr f ) (ξ) can only be nonzero if −Ω ≤ ξ/r ≤ Ω. Hence (Dr f )
is supported in [−rΩ, rΩ], and therefore Dr f ∈ FL2[−rΩ,rΩ](R). Thus Dr
maps FL2[−Ω,Ω] (R) into FL2[−rΩ,rΩ](R), and a similar argument shows that
Dr is surjective.
10.3 (a) Suppose that fn ∈ PW(R), f ∈ L2 (R), and fn → f in L2 -norm.
The Fourier transform is unitary, so fbn → fb in L2 -norm. Consequently, there
is subsequence that converges pointwise a.e. But fbn (ξ) = 0 for almost every
ξ∈ / [−1/2, 1/2], so it follows that fb(ξ) = 0 for almost every ξ ∈
/ [−1/2, 1/2].
∨
(b) First we use that fact that f = fb to give an explicit proof that
f ∈ PW(R) is continuous. Given x, y ∈ R we have
∨ ∨
|f (x) − f (y)| = | fb (x) − fb (y)|
Z Z

= fb(ξ) e2πiξx dξ − fb(ξ) e2πiξy dξ
Z 1/2

= fb(ξ) e2πiξx − e2πiξy dξ
−1/2
Z 1/2
≤ |fb(ξ)| |e2πiξy | |e2πiξ(x−y) − 1| dξ
−1/2
Z 1/2
≤ |fb(ξ)| 2π|ξ| |x − y| dξ
−1/2
Z 1/2 1/2 Z 1/2 1/2
≤ 2π |x − y| |fb(ξ)|2 dξ |ξ|2 dξ
−1/2 −1/2
1
= 2π |x − y| kfbkL2
2 · 31/2
= 3−1/2 π kf kL2 |x − y|.
Hence f is Lipschitz and therefore uniformly continuous.

Now we must justify the interchange of limit and integration in the calcu-
lation given in the proof of Theorem 10.4(c). Since f is continuous, the point
value f (x) is defined. Let x be fixed. The Fourier transform fb is an element of
L2 (R), so is only defined almost everywhere. However, for each ξ where fb(ξ)
is defined, the integrand converges pointwise since
2πiξx
e − e2πiξy d 2πiξx
lim fb(ξ) = fb(ξ) e = fb(ξ) 2πiξ e2πiξx .
y→x x−y dx
Since fb is both square integrable and compactly supported, it is integrable.

Also, for every x 6= y we have
2πiξx 2πiξ(x−y)
e − e2πiξy
2πiξy e − 1 2π|ξ| |x − y|
= |e | ≤ = 2π|ξ|.
x−y x−y |x − y|
Therefore the integrand is bounded by 2π|ξ fb(ξ)|, which is integrable since fb

is integrable and has compact support. Consequently the hypotheses of the
Lebesgue Dominated Convergence Theorem are satisfied, so the interchange
of limit and integration is justified by that theorem.
10.4 (a) Fix f ∈ PW(R). By Theorem 10.4 we know that kfbkL1 < ∞.
Further, f is infinitely differentiable, and the proof of Theorem 10.4 shows
that
f ′ = (2πiξ fb(ξ)) .
∨
Since f ′ ∈ PW(R), it follows by induction that

Z ∞
∨
f (n) (x) = (2πiξ)n fb(ξ) (x) = (2πiξ)n fb(ξ) e2πiξx dξ
−∞
Z 1/2
= (2πiξ)n fb(ξ) e2πiξx dξ.
−1/2
(b) Using part (a), we see that

Z 1/2
|f (n) (x)| ≤ |(2πiξ)n fb(ξ) e2πiξx | dξ
−1/2
Z 1/2
≤ (2π · 21 )n |fb(ξ)| dξ
−1/2
Z 1/2
= π n
|fb(ξ)| dξ = π n kfbkL1 .
−1/2
Hence kf (n) k∞ ≤ π n kfbkL1 .

(c) Since
X∞ X∞
|f (n) (a)| π n |x − a|n
|x − a|n ≤ kfbkL1 = kfbkL1 eπ|x−a| < ∞,
n=0
n! n=0
n!
the Taylor series for f converges absolutely at each x, and in fact it converges
uniformly on every compact set.
However, we still have to show that the Taylor series converges to f (x).
The remainder term Rn is defined by
∞
X f (k) (a)
f (x) = (x − a)k + Rn (x),
k!
k=0
and it has the form

f (n+1) (c)
Rn (x) = (x − a)n+1
(n + 1)!
for some c between a and x. Therefore

|f (n+1) (c)| π n+1 |x − a|n+1
|Rn (x)| ≤ |x − a|n+1 ≤ kfbkL1 ,
(n + 1)! (n + 1)!
which converges to zero as n → ∞. Hence the Taylor series converges to f (x).

(d) Suppose that f and fb are both compactly supported. By dilating f
we reciprocally dilate fb, so it suffices to assume that f ∈ PW(R). Since f is
compactly supported, there is an R such that f (x) = 0 for all |x| > R. Fix
a > R. Then f (n) (a) = 0 for every n. By part (c), if we consider the Taylor
series of f about the point a, we see that f (x) = 0 for every x.
10.5 Note that
(Ta dπ ) (ξ) = M−a dbπ(ξ) = e2πiax χ[− 21 , 21 ] (x),

∧
i.e., (Ta dπ ) = ea χ[− 12 , 12 ] . Since {eλn }n∈N is a frame for L2 (T), it follows that
∧
{eλn χ[− 1 , 1 ] }n∈N is a frame for L2 1 1 (R), and therefore {Tλn dπ }n∈N is a
2 2 [− 2 , 2 ]
frame for PW(R). The canonical dual frame is {e
sn }n∈N . Therefore, for each
f ∈ PW(R) we have
X X
f = hf, Tλn dπ i sen = f (λn ) sen ,
n n
where the series converges unconditionally in L2 -norm.

10.6 By Theorem 10.4, if b > 1 then {Tbn dπ }n∈Z is incomplete in PW(R).
Hence there exists a nonzero function f ∈ PW(R) that is orthogonal to Tbn dπ
for every n ∈ Z. Therefore f (bn) = hf, Tbn dπ i = 0 for every n ∈ Z, so f has
the same sample values as the zero function.
10.7 Suppose that Fn ∈ Hϕ and Fn → F in L2 (T). Then there exists a
subsequence that converges pointwise a.e., say Fnk (ξ) → F (ξ) a.e. as k → ∞.
Since each Fnk vanishes almost everywhere on Zϕ , the function F must be
zero almost everywhere on Zϕ as well, and therefore F ∈ Hϕ . Hence Hϕ is
closed.
10.8 (a) ⇐. Suppose that ϕ 6= 0 a.e. Suppose that f ∈ L2 (T) is such that
hf, ϕ en i = 0 for every n ∈ Z. Then
hf ϕ, en i = 0, n ∈ Z.
The function f ϕ belongs to L1 (T). The Uniqueness Theorem therefore implies

that f ϕ = 0 a.e. However, since ϕ is nonzero almost everywhere, this implies
that f = 0 a.e. Hence {ϕ en }n∈Z is complete.
(b) ⇐. Suppose that 1/ϕ ∈ L2 (T). Then ϕ e = 1/ϕ ∈ L2 (T), and {ϕ en }n∈Z
e en }n∈Z are biorthogonal, so {ϕ en }n∈Z is minimal.
and {ϕ
(c) ⇐. Suppose ϕ ∈ L∞ (T), and fix B so that |ϕ|2 ≤ B a.e. Then
X X
|hf, ϕ en i|2 = |hf ϕ, en i|2 = kf ϕk2L2 ≤ B kf k2L2 ,
n∈Z n∈Z
so {ϕ en }n∈Z is a Bessel sequence with Bessel bound B.

(d) ⇐. Suppose that A ≤ |ϕ(x)|2 ≤ B for a.e. t ∈ / Zϕ . We will show that
E(ϕ) is a frame for Hg with frame bounds A, B.
Choose any f ∈ L∞ (T) ∩ Hg . Then since f and ϕ are both zero on Zϕ ,
we have
X X
|hf, en ϕi|2 = |hf ϕ, en i|2
n∈Z n∈Z
= kf ϕk2L2
Z 1
= |f (x)|2 |ϕ(x)|2 dx
0
Z
= |f (x)|2 |ϕ(x)|2 dx
Zϕ
Z
≥ A |f (x)|2 dx
Zϕ
Z 1
= A |f (x)|2 dx
0
= A kf k2L2 .
Since L∞ (T) ∩ Hϕ is dense in Hϕ , this establishes that A is a lower frame

bound, and a similar argument shows that B is an upper frame bound.
(e) ⇐. If A ≤ |ϕ|2 ≤ B a.e., then parts (b) and (d) imply that E(ϕ) is an
exact frame, and hence is a Riesz basis.
(f) ⇒. Assume {ϕ en }n∈Z is an orthonormal basis. Then it is a Riesz basis
e so |ϕ| = 1 a.e.
and we must have ϕ = ϕ,
⇐. If |ϕ| = 1 a.e., then {ϕ en }n∈Z is its own biorthogonal system, hence is
orthonormal, and it is also complete since ϕ 6= 0 a.e.
10.9 This follows from the fact that 1/ϕ ∈ L2 (T) but ϕ is not bounded away
from zero.
P
10.10 We are given a series in L2 (T) of the form P n∈Z cn ϕ en , with
(cn )n∈Z ∈ ℓ2 (Z) and ϕ ∈ L∞ (T). The series m = cn en also converges
in L2 (T), so
N
X N
X
2 2
mϕ − cn ϕ en L2 ≤ kϕk2L∞ m − cn ϕ L2 → 0 as N → ∞.
n=−N n=−N
P
Hence mϕ = n∈Z cn ϕ en , and in fact this series converges unconditionally
to mϕ.
10.11 Since {e2πinx }n∈Z is an orthonormal basis for L2 (T), the mapping
F : ℓ2 (Z) → L2 (T) given by F(c) = b
c is a unitary mapping, where
X
b
c (ξ) = ck e−2πikx .
k∈Z
We compute that
X
ck−n e−2πikx
∧
(Tn c) (ξ) =
k∈Z
X
= ck e−2πi(k+n)x
k∈Z
X
= e−2πinx ck e−2πikx = e−2πinx b
c (ξ).
k∈Z
Therefore {Tn c}n∈Z = E(b c ), the system of weighted exponentials generated

c ∈ L2 (T). Theorem 10.10 tells us exactly when this system is
by the function b
complete, Bessel, etc., in L2 (T), and since F is unitary the system {Tn c}n∈Z
must have exactly the same properties in ℓ2 (Z). For example,
{Tn c}n∈Z is complete ⇐⇒ b

c (ξ) 6= 0 a.e.
10.12 Fix g ∈ L2 (R) and a > 0, and let h(x) = g(ax). If {Tak g}k∈Z is a frame
sequence, then since the dilation operator Da f (x) = a1/2 f (ax) is unitary, it
follows that If {Da (Tak g)}k∈Z is a frame sequence as well. Now,
Da Tak g(x) = a1/2 (Tak g)(ax) = a1/2 g(ax − ak) = a1/2 h(x − k).
Hence {h(x − k)}k∈Z = {a−1/2 Da Tak g}k∈Z is a frame sequence. If we let

V0 (h) = span{h(x − k)}k∈Z and V0 (g) = span{g(x − k)}k∈Z then V0 (h) =
Da (V0 (g)), so f ∈ V0 (g) if and only f (ax) ∈ V0 (h).
10.13 Given a function f ∈ L1 (R), set gn (x) = f (x + an). Then
X XZ a
kgn kL1 [0,a] = |f (x + an)| dx
n∈Z n∈Z 0
Z ∞
= |f (x)| dx = kf kL1(R) < ∞.
−∞
P
Thus, the series ϕ(x) = n∈Z f (x+an) converges absolutely in L1 [0, 1]. Using
Fubini’s Theorem to interchange the sum and integral, we have
Z 1 Z aX
ϕ(x) dx = f (x + an) dx
0 0 n∈Z
XZ a
= f (x + an) dx
n∈Z 0
XZ an+a
= f (x) dx
n∈Z an
Z ∞
= f (x) dx.
−∞
PN
10.14 Suppose that k=1 Tak g = 0 a.e. Taking the Fourier transform of both
sides, we have that
N
X
M−ak b
g = 0 a.e.
k=1
But then,
X
N
−2πiak ξ
e gb(ξ) = 0 a.e.
k=1
Since g is not the zero function, neither is gb, and therefore b

g must be nonzero
on a set of positive measure. But then the trigonometric polynomial m(ξ) =
PN −2πiak ξ
k=1 e is zero on a set of positive measure, which is a contradiction.
10.15 Set
N
X
mN (ξ) = ck e−2πikξ .
k=−N
P −2πikξ
Since b
c (ξ) = k∈Z ck e converges in L2 (T) we have mN → b c in L2 -norm
as N → ∞. Hence there exists some subsequence that converges pointwise a.e.,
say mNk (ξ) → b c (ξ) a.e. as k → ∞.
By hypothesis we also have that mN gb → F in L2 (R) as N → ∞, so
g → F in L2 -norm as well, and therefore some subsequence of this con-
mN k b
verges to F pointwise a.e. But mNk (ξ) gb(ξ) → b c (ξ) gb(ξ) a.e., so we must have
F (ξ) = bc (ξ) gb(ξ) a.e.
10.16 (b) ⇐. Suppose that A ≤ Φg ≤ B a.e. off Zg . Then by part (a) we know
that T (g) is a Bessel sequence with Bessel bound B. Note that T (g) ⊆ Hg , so
if we establish that the lower frame bound holds on Hg then T (g) is a frame
for Hg .
So, fix any F ∈ Hg . Then since {e−k }k∈Z is an orthonormal basis for
L2 (T), and F and Φg both vanish on Zg , we can compute that
X
X
2
F, Gk 2 = F Φ1/2
g , e−k

k∈Z k∈Z
2
= F Φ1/2
g
2
L (T)
Z 1
= |F (ξ)|2 Φg (ξ) dξ
0
Z 1
≥ A |F (ξ)|2 dξ
0
2
= A F L2 (T) .
This establishes the lower frame bound.

(c) ⇐. Suppose that 1/Φg ∈ L1 (T), which implies that Φg 6= 0 a.e. and
1/2 −1/2
Hg = L2 (T). Set Gk = U (Tk g) = e−k Φg . Then Fk = e−k Φg ∈ L2 (R),
and we have
hFj , Gk i = hFj Φ1/2
g , e−k i
Z 1
= Fj (ξ) Φ1/2
g (ξ) e
2πikξ
dξ
0
Z 1
= e−2πijξ ϕ−1/2
g (ξ) Φ1/2
g (ξ) e
2πikξ
dξ
0
= δjk .
Hence {Fk }k∈Z is biorthogonal to {Gk }k∈Z , and therefore these sequences are
minimal in L2 (T). Since U maps V0 (g) unitarily onto L2 (T), we conclude that
T (g) is minimal in V0 (g), and it is exact since T (g) is complete in V0 (g) by

definition.
(d) A sequence is a Riesz basis for V0 (g) if and only if it is minimal and is
a frame for V0 (g). Hence this part follows from statements (b) and (c).
(e) A sequence is an orthonormal basis if and only if it is Riesz basis with
frame bounds A = B = 1.
(f) The mapping U is a unitary transformation of V0 (g) onto L2 (T). Fur-
1/2
ther, U (Tk g) = e−k Φg , so U maps the sequence T (g) to the sequence
1/2
E(Φg ). By Theorem 4.18 we know that Schauder bases are preserved by
topological isomorphisms (and hence by unitary mappings). Therefore T (g)
1/2
is a Schauder basis for its closed span V0 (g) if and only if E(Φg ) is a Schauder
2
basis for L (T) with respect to the same ordering of the index set Z. Using the
1/2
given ordering Z = {0, −1, 1, −2, 2, . . . }, Theorem 5.15 tells us that E(Φg )
is a Schauder basis for L2 (T) if and only if Φg ∈ A2 (T).
10.17 We are given g ∈ L2 (R) such that T (g) is a frame sequence. Hence
A ≤ Φg ≤ B a.e. by Theorem 10.19.
⇒. If f ∈ V0 (g) then there exists a sequence c = (ck )k∈Z ∈ ℓ2 (Z) such
that X
f = ck Tk g,
k∈Z
where the series converges unconditionally in L2 (R), and therefore

X
fb(ξ) = ck e−2πikξ gb(ξ),
k∈Z
again with unconditional convergence of the series. Exercise 10.15 then implies
that fb = b
cbg.
⇐. Suppose f ∈ L2 (T) satisfies fb = m gb for some function m ∈ L2 (T).
c for some sequence c = ℓ2 (Z). Set
Then m = b
cN = (. . . , 0, 0, c−N , . . . , cN , 0, 0, . . . ).
Then
N
X
cc
N (ξ) = ck e−2πikξ → b
c (ξ) = m(ξ),
k=−N
where the convergence is in the norm of L2 (T). Hence

Z N 2
∞ X
fb(ξ) − ck e −2πikξ
gb(ξ) dξ

−∞ k=−N
Z ∞
= c (ξ) gb(ξ) − cc
|b b(ξ)|2 dξ
N (ξ) g
−∞
XZ 1
= c (ξ + j) b
|b g(ξ + j) − cc g(ξ + j)|2 dξ
N (ξ + j) b
j∈Z 0
Z 1 X
2
= c (ξ) − cc
|b N (ξ)| gb(ξ + j)|2 dξ
0 j∈Z
Z 1
2
= c (ξ) − cc
|b N (ξ)| Φg (ξ) dξ
0
Z 1
2
≤ B c (ξ) − cc
|b N (ξ)| dξ
0
→ 0 as N → ∞.
Therefore, in the sense of convergence of series in L2 (R) we have

X X
fb(ξ) = ck e−2πikξ b
g(ξ) = ck M−k b
g(ξ).
k∈Z k∈Z
In fact, all of the series above converge unconditionally, but the important
fact is that by applying the inverse Fourier transform we obtain
X
f = ck Tk g ∈ V0 (g).
k∈Z
10.18 (a) We are given that b g(ξ) 6= 0 for almost every ξ. Suppose that
f ∈ V0 (g). Then there exist functions fn ∈ V0 (g) such that fn → f in L2 -
norm. By passing to a subsequence, we may assume that fn → f pointwise a.e.
Now, each function fn can be written as
M
X
fn = ck T k g
k=−M
for some M and scalars ck . Applying the Fourier transform,

X
M
fbn (ξ) = ck e−2πikξ gb(ξ) = pn (ξ) gb(ξ),
k=−M
and pn is a trigonometric polynomial, hence 1-periodic. Since gb is nonzero

almost everywhere, the functions pn converge at almost every point:
fbn (ξ)
p(ξ) = lim pn (ξ) = lim .
n→∞ n→∞ gb(ξ)
Therefore fb(ξ) = p(ξ) gb(ξ) a.e. Finally, as each function pn is 1-periodic, the
same is true of p.
(b) Suppose that T (g) is complete in L2 (R). If gb vanishes on any set E

of positive and finite measure, then the function f whose Fourier transform
is fb = χE is orthogonal to M−k gb for every k, which is a contradiction. Hence
b
g must be nonzero almost everywhere.
The function f whose Fourier transform is fb(ξ) = e−ξ b
2
g(ξ) belongs to
L2 (R) = V0 (g), so by part (b) there exists some function p ∈ L2 (T) such that
fb = p b
g. But then p(ξ) = fb(ξ)/b
2
g (ξ) = e−ξ , which is not 1-periodic. This is a
contradiction, so we cannot have V0 (g) = L2 (R).
10.19 (a) ⇒. Suppose that span(T (f )) = span(T (g)), and call this space
V0 (g). Since translates of f generate V0 (g) and g ∈ V0 (g), Exercise 10.17
g = m fb where m ∈ L2 (T). On the other hand, by Theorem 10.19
implies that b
we must have Φg = 1 = Φf a.e. Hence for almost every ξ we have
X X
1 = Φg (ξ) = g (ξ + k)|2 = |m(ξ)|2
|b |fb(ξ + k)|2 = |m(ξ)|2 ,
k∈Z k∈Z
so |m(ξ)| = 1 a.e.
⇐. If gb = m fb where m is 1-periodic and |m| = 1 a.e., then, since T (f )
is a frame, Exercise 10.17 implies that g ∈ span(T (f )). Since we also have
fb = m gb, it follows symmetrically that f ∈ span(T (g)). Hence span(T (f )) =
span(T (g)).
(b) ⇐. If g = α Tk f where α is a scalar then
g(ξ) = α M−k fb(ξ) = α e−2πikξ fb(ξ),

b
g(ξ)| = |α| |fb(ξ)|. By part (a), if T (f ) and T (g) have the same spans then
so |b
we must have |α| = 1.
⇒. Since g belongs to span(T (f )) and the integer translates of f are
orthonormal, we have
X X
g = hg, Tk f i Tk f = αk Tk f.
k∈Z k∈Z
However, if f and g are each compactly supported, then only finitely many of
the inner products hg, Tk f i can be nonzero. This implies that
X
gb(ξ) = hg, Tk f i M−k fb(ξ) = α(ξ) fb(ξ)
k∈Z
where X X
α(ξ) = hg, Tk f i e−2πikξ = αk e−2πikξ ,
k∈Z k∈Z
and α is a trigonometric polynomial that satisfies |α(ξ)| = 1 a.e. Thus

XX
1 = |α(ξ)|2 = αj αk e−2πi(j−k)ξ
j∈Z k∈Z
XX
= αj αj−k e−2πikξ
j∈Z k∈Z
X X
= ek−j
αj α e−2πikξ
j∈Z k∈Z
X
= e)j e−2πikξ ,
(α ∗ α
j∈Z
ej = α−j . Only finitely many terms in this sum are nonzero, so by

where α
independence of the exponentials we must have
e)j = δ0j ,
(α ∗ α j ∈ Z.
Again, α and therefore α ∗ αe are finite sequences. Writing out α ∗ α

e, we see
that it can only equal a delta sequence if α contains only one nonzero term.
This implies that g = αk Tk f for some particular k, and by part (a) we must
have |αk | = 1
10.20 (a) Suppose g ∈ L2 (R) is compactly supported, say supp(g) ⊆
[−N, N ]. Applying the Triangle Inequality in L2 [0, 1],
Z 1 X 2 1/2 X

|g(x + k)| dx
= T−k g

0 k∈Z k∈Z L2 [0,1]
X
≤ kT−k gkL2 [0,1]
k∈Z
X Z 1 1/2
2
= |g(x + k)| dx
k∈Z 0
N Z
X 1 1/2
= |g(x + k)|2 dx
k=−N 0
< ∞.
Therefore g satisfies the hypotheses of Theorem 10.25, so T (g) is Bessel but

cannot be a redundant frame for V0 (g).
Alternatively, note that hg, Tk gi k∈Z is a finite sequence. Equation (10.20)
therefore implies that Φg is a trigonometric polynomial, hence is continuous
on T.
(b) Suppose gb is continuous and compactly supported, say supp(b g) ⊆
[−N, N ]. Then b
g satisfies the amalgam condition given in Lemma 10.24:
X N
X
g · χ[k,k+1] k2∞ =
kb g · χ[k,k+1] k2∞ < ∞.
kb
k∈Z k=−N
Therefore T (g) is Bessel but cannot be a redundant frame for V0 (g).

(c) Let g ∈ L2 (R) be the function whose Fourier transform is gb = χ[0,1/2) .
Then Φg (ξ) = χ[0,1/2) extended 1-periodically to R. This function is discon-
tinuous, but it is bounded above and below off of its zero set, so T (g) is a
frame but not a Riesz basis for V0 (g).
10.21 We have gb♯ = m b g where m ∈ L2 (R), so g ♯ ∈ V0 (g) by Exercise 10.17.

Therefore V0 (g ♯ ) = span(T (g ♯ )) is a closed subspace of V0 (g). On the other
hand,
gb = gb♯ Φ1/2
g ,
1/2
so since Φg ∈ L2 (T) we have that g ∈ V0 (g ♯ ). Hence V0 (g) = V0 (g ♯ ).
Let Z = ZΦg . For ξ ∈
/ Z we have
X
Φg♯ (ξ) = |gb♯ (ξ + j)|2
j∈Z
X |b
g (ξ + j)|2
=
Φg (ξ + j)
j∈Z
X |b
g (ξ + j)|2
= = 1,
Φg (ξ)
j∈Z
while for ξ ∈ Z we have Φg♯ (ξ) = 0. Theorem 10.19 therefore implies that
T (g ♯ ) is a Parseval frame for V0 (g ♯ ) = V0 (g).
If T (g) is a Riesz basis for V0 (g), then |Z| = 0, so Φg♯ = 1 a.e., and
therefore T (g ♯ ) is an orthonormal basis for V0 (g).
10.22 We compute that
Z 1
2
hw, wi = 2 (1 − x)2 dx =
0 3
and Z 1
1
hw, T−1 wi = hw, T1 wi = x (1 − x) = .
0 6
Since supp(w) = [−1, 1], we have hw, Tk wi for all |k| > 1. By equation (10.20),
we therefore have
X 2 1 2πiξ 2 + cos 2πξ
Φw = hw, Tk wi e−2πikξ = + e + e−2πiξ = .
3 6 3
k∈Z
Therefore
1
≤ Φw (ξ) ≤ 1, all ξ ∈ R,
3
so T (w) is a Riesz basis for its closed span.
We compute that
X
Φwe (ξ) = be + k)|2
|w(ξ
k∈Z
X |w(ξb + k)|2
=
Φw (ξ + k)2
k∈Z
P
b + k)|2
k∈Z |w(ξ
=
Φw (ξ)2
Φw (ξ)
=
Φw (ξ)2
1
=
Φw (ξ)
3
= .
2 + cos 2πξ
Similarly,
h i X
wc♯ , w
be (ξ) = c♯ (ξ + k) w(ξ
w be + k)
k∈Z
X b + k) w(ξ + k)
w(ξ
=
k∈Z
Φw (ξ + k)1/2 Φw (ξ + k)
P
k∈Z b + k)|2
|w(ξ
=
Φw (ξ + k)3/2
Φw (ξ)
=
Φw (ξ)3/2
1
=
Φw (ξ)1/2
1/2
3
= .
2 + cos 2πξ
10.23 Fix g ∈ L2 (R), and write Ta g(x) = g(x − a).

Suppose that {Tak g}k∈N is a Schauder basis for L2 (R), and let C be the
basis constant and SN the partial sum operators.
Fix ε > 0. Because translation is strongly continuous in L2 (R), there exists
a δ > 0 such that
kg − Ta gkL2 < ε whenever |a| < δ. (∗)

If (ak ) is not uniformly separated then there must exist some j 6= k such
that |aj − ak | < δ, and without loss of generality we may assume that j < k.
Let fjk = Taj g −Tak g , and note that the basis hypothesis implies that fjk 6= 0.
Further, we have kfjk kL2 < ε by equation (∗). Considering the basis expan-
sions, since j < k we have
Sj (fjk ) = Sj (Taj g) − Sj (Tak g) = Taj g.
Therefore
kSj (fjk )kL2 kTaj gkL2 kgkL2
C ≥ kSj k ≥ > = .
kfjk kL2 ε ε
Since ε is arbitrary, this contradicts the fact that C is finite.

11.1 Recall that Ta Mb = e−2πiab Mb Ta . Since
|hf, Mbn Tak gi| = |hf, Tak Mbn gi|,
we have that {Mbn Tak g}k,n∈Z is a frame if and only if {Tak Mbn g}k,n∈Z is a
frame. In this case, we have for f ∈ L2 (R) that
X X
hf, Mbn Tak gi Mbn Tak g = hf, e2πiakbn Tak Mbn gi e2πiakbn Tak Mbn g
k,n k,n
X
= e−2πiakbn e2πiakbn hf, Tak Mbn gi Tak Mbn g
k,n
X
= hf, Tak Mbn gi Tak Mbn g.
k,n
Therefore the frame operators for these two frames are identical.
11.2 (b) The general linear group of degree 3 is the set of all 3 × 3 invertible
real matrices. It is denoted by GL3 (R), and it is a group under matrix mul-
tiplication. Since H3 ⊆ GL3 (R), we just have to show that H3 is a subgroup.
We have
    
1 b t 1 d u 1 b + d t + u + bc
0 1 a 0 1 c  = 0 1 b + d  ∈ H3 ,
0 0 1 0 0 1 0 0 1
so H3 is closed under composition. Also,

    
1 b t 1 −b ab − t 100
0 1 a 0 1 −a  =  0 1 0  ,
0 0 1 0 0 1 001
so H3 is closed under inverses. Therefore H3 is a group.

(c) By construction, H2 is isomorphic to H3 .
(d) Because
(e2πit Ta Mb ) (e2πiu Tc Md ) = e2πi(t+u) Ta Mb Tc Md

= e2πi(t+u) Ta e2πibc Tc Mb Md
= e2πi(t+u+bc) Ta+c Mb+d ,
we have that H1 is isomorphic to H2 and H3 .

(e) We have
(a, b, 0) ∗ (a, b, 0) = (2a, 2b, ab),
so aZ × bZ × {0} is not a subgroup of H2 .

On the other hand, we will show that G = aZ × bZ × abZ is a subgroup.
Given k, n, i, ℓ, m, j ∈ Z, we have
(ak, bn, abi) ∗ (aℓ, bm, abj) = (ak + aℓ, bn + bm, abi + abj + bnaℓ)

= a(k + ℓ), b(n + m), ab(i + j + nℓ) ∈ G,
so G is closed under the group operation.

By our work above, the inverse operation in H2 is
(a, b, t)−1 = (−a, −b, ab − t).
Therefore, given k, n, i ∈ Z, we have

(ak, bn, abi)−1 = (−ak, −bn, akbn−abi) = a(−k), b(−n), ab(kn−i) ∈ G.
Hence G is closed under inverses, so G is a subgroup of H2 .

(f) Fix (c, d, u) ∈ H2 , and let F be an integrable function on H2 = R3 .
Then by translation-invariance we have
ZZZ ZZZ

F (c, d, u) ∗ (a, b, t) da db dt = F (a + c, b + d, t + u + bc) da db dt
ZZZ
= F (a, b, t + u + bc) da db dt
ZZZ
= F (a, b, t + u + bc) dt da db
ZZZ
= F (a, b, t) dt da db
ZZZ
= F (a, b, t) da db dt.
A similar calculation establishes that da db dt is also the right Haar measure.

11.3 (a), (b) We will show that the frame operator S commutes with Ta and
Mb . Using Exercise 11.1, given f ∈ L2 (R) we have
X
S(Ta f ) = hTa f, Tak Mbn gi Tak Mbn g
k,n
X
= hf, T−a Tak Mbn gi Tak Mbn g
k,n
X
= hf, Ta(k−1) Mbn gi Tak Mbn g
k,n
X
= hf, Tak Mbn gi Ta(k+1) Mbn g
k,n
X
= hf, Tak Mbn gi Ta Tak Mbn g
k,n
X
= Ta hf, Tak Mbn gi Tak Mbn g
k,n
= Ta Sf.
Also,
X
S(Mb f ) = hMb f, Mbn Tak gi Mbn Tak g
k,n
X
= hf, M−b Mbn Tak gi Mbn Tak g
k,n
X
= hf, Mb(n−1) Tak gi Mbn Tak g
k,n
X
= hf, Mbn Tak gi Mb(n+1) Tak g
k,n
X
= hf, Mbn Tak gi Mb Mbn Tak g
k,n
X
= Mb hf, Mbn Tak gi Mbn Tak g
k,n
= Mb Sf.
Therefore S commutes with Ta and Mb , and hence commutes with Tak = (Ta )k
and Mbn = (Mb )n . Consequently,
S(Mbn Tak f ) = Mbn Tak Sf, f ∈ L2 (R).
Since S −1 is a topological isomorphism of L2 (R) onto itself, we can replace f

by S −1 f on the line above to obtain
S(Mbn Tak S −1 f ) = Mbn Tak f, f ∈ L2 (R),
and therefore
Mbn Tak (S −1 f ) = S −1 (Mbn Tak f ), f ∈ L2 (R).
Consequently, the canonical dual frame is

−1
S (Mbn Tak g) k,n∈Z = Mbn Tak (S −1 g) k,n∈Z = G(S −1 g, a, b).
(c) Suppose that G(g, a, b) is a Gabor frame. By part (b), its canonical
dual is G(e
g , a, b). By Corollary 8.23, these frames are exact if and only if
hMbn Tak g, Mbn Tak ge i = 1 for every k, n. However, all of these inner products
are equal:
hMbn Tak g, Mbn Tak ge i = hTak g, M−bn Mbn Tak e
gi
= hTak g, Tak ge i = hg, T−ak Tak ge i = hg, e
g i.
Hence G(g, a, b) is a Riesz basis if and only if hg, e
g i = 1.
(d) We have seen that the frame operator commutes with Tak and Mbn ,
and that the same is true for S −1 . Since S −1 is a positive operator, it has a
square root S −1/2 . By Theorem 2.18, S −1/2 commutes with every operator
that commutes with S −1 . Therefore S −1/2 commutes with Tak and Mbn , and
hence
S −1/2 (Mbn Tak f ) = Mbn Tak (S −1/2 f ), f ∈ L2 (R).
Corollary 8.28 therefore implies that G(S −1/2 g, a, b) is a Parseval frame.
11.4 (a) The Fourier transform is a unitary operator on L2 (R), so the image
of a frame under the Fourier transform is still a frame. Applying equations
(9.2) and (9.3), we have
∧
F G(g, a, b) = (Mbn Tak g) k,n∈Z

= Tbn M−ak bg k,n∈Z

= Tbn Mak gb k,n∈Z .
Hence

G(g, a, b) is a frame ⇐⇒ Tbn Mak b
g k,n∈Z is a frame

⇐⇒ Mak Tbn b
g k,n∈Z is a frame
⇐⇒ G(b
g , b, a) is a frame.
Note that since the Fourier transform is unitary, the frame bounds of G(g, a, b)
are the same as those of G(bg , b, a).
11.5 (a) The fact that G(χ[0,1] , 1, 1) is orthonormal follows from the fact
that {e2πinx }n∈Z is an orthonormal sequence in L2 [a, a + 1] for any a ∈ R. It
is a frame by Theorem 11.4, or we can prove this directly as follows. Given
f ∈ L2 (R), let fk = f χ[k,k+1] . Note that
Z k+1
hfk , en iL2 [k,k+1] = fk (x) e−2πinx dx
k
Z ∞
= f (x) χ[k,k+1] (x) e−2πinx dx
−∞
Z ∞
= f (x) Mn Tk χ[0,1] (x) dx
−∞

= f, Mn Tk χ[0,1] L2 (R) .
Since {e2πinx }n∈Z is an orthonormal basis for L2 [k, k + 1], we can use the
Planchel Equality to compute that
Z ∞ X Z k+1
2 2
kf kL2 = |f (x)| dx = |fk (x)|2 dx
−∞ k∈Z k
X
= kfk k2L2 [k,k+1]
k∈Z
X
= kfbk k2ℓ2 (Z)
k∈Z
XX
= |fbk (n)|2
k∈Z n∈Z
XX
= |hfk , en iL2 [k,k+1] |2
k∈Z n∈Z
X X

= f, Mn Tk χ[0,1] 2 2
L (R)
k∈Z n∈Z
Hence G(χ[0,1] , 1, 1) = {Mn Tk χ[0,1] }k,n∈Z is an orthonormal Parseval frame,

and hence is an orthonormal basis.
(b) Let χ = χ[0,1] . If we fix b = 1, then we have supp(χ) = [0, 1] = [0, b−1 ].
If 0 < a ≤ 1 then the intervals [ak, a(k + 1)] have a finite number of overlaps.
In fact, they can overlap at most B = 1 + (1/a) times. Therefore
X
1 ≤ |g(x − ak)|2 ≤ B,
k∈Z
so Theorem 11.4 implies that G(g, a, 1) is a frame.

11.6 (a) The proof of Theorem 11.4 shows that
X
Z

hSf, f i = f, Mbn Tak g 2 = b−1 |f (x)|2 G0 (x) dx = hb−1 G0 f, f i.
k,n∈Z
Since b−1 G0 ≥ 0, multiplication by G0 is a self-adjoint (in fact, positive)

operator on L2 (R). We have

(S − b−1 G0 )f, f = 0, f ∈ L2 (R),
and S − b−1 G0 is self-adjoint, so Corollary 2.16 implies that S − b−1 G0 = 0.

(b) Since Sf = b−1 G0 f, we have S −1 f = bf /G0 (note that G0 is bounded
away from zero and infinity, so the same is true of 1/G0 .) Using the periodicity
of G0 we therefore have that
Mbn Tak g(x)

S −1 (Mbn Tak g)(x) = b
G0 (x)
e2πibnx g(x − ak)

= b
G0 (x − ak)
= b e2πibnx (g/G0 )(x − ak)
= b Mbn Tak (g/G0 )(x)
= Mbn Tak ge(x),
where e
g = bg/G0 . Hence the dual frame is
−1
S (Mbn Tak g) k,n∈Z = Mbn Tak ge k,n∈Z = G(g, a, b).
Alternatively, we can simply appeal to Exercise 11.3, and note that ge =

S −1 g = bg/G0 .
(c) We are given ab = 1, i.e., a = b−1 , and supp(g) ⊆ [0, b−1 ] = [0, a]. This
implies that G0 (x) = |g(x)|2 on [0, a], and G0 is the a-periodic extension of
this to R. Also, since supp(g) ⊆ [0, a], we have
hTak g, Tℓa gi = 0, k 6= ℓ ∈ Z.
If k = ℓ and m 6= n, then
Z ak+a

g(x − ak)
Mbm Tak g, Mbn Tak e
g = e2πibmx g(x − ak) e−2πibnx dx
ak G0 (x − ak)
Z ak+a
|g(x − ak|2
= e2πi(bm−bn)x dx
ak G0 (x − ak)
Z ak+a
= e2πi(bm−bn)x dx = 0.
ak
Hence G(g, a, b) has a biorthogonal system. Since we are given that G(g, a, b)
is a frame, it therefore is an exact frame and hence is a Riesz basis.
(c), (d) Suppose 0 < ab ≤ 1, and g satisfies the hypotheses of the
Painless Nonorthogonal Expansions. Since the canonical dual of G(g, a, b) is
G(e
g , a, b), Corollary 8.23 implies that these frames are exact if and only if
hMbn Tak g, Mbn Tak ge i = 1 for every k, n. Now, all of these inner products are
equal to hg, ge i, and since ge = bg/G0 we have
Z b−1 Z b−1
|g(x)|2
hg, e
gi = g(x) ge(x) dx = b P dx.
0 0 k |g(x − ak)|2
P
Since |g(x)|2 ≤ k |g(x − ak)|2 , the integrand on the line above lies between 0
and 1 almost everywhere. Hence
Z b−1
|g(x)|2
b 1− P dx ≥ 0,
0 k |g(x − ak)|2
and the integrand is nonnegative a.e. Therefore, this integral is zero if and
only if
|g(x)|2
1− P 2
= 0 a.e. on [0, b−1 ].
k |g(x − ak)|
That is,
G(g, a, b) is a Riesz basis ⇐⇒ hg, e

gi = 1
X
⇐⇒ |g(x)|2 = |g(x − ak)|2 a.e. on [0, b−1 ]
k∈Z
⇐⇒ g(x − ak) = 0 a.e. on [0, b−1 ] for all k 6= 0.
In case these hold, by examining the supports of g(x − ak) we see that G0 = 0
a.e. on [0, b−1 − a] and [a, b−1 ]. However, to be a frame we must at least
have that G0 is nonzero a.e. Therefore, if G(g, a, b) is a Riesz basis then these
intervals must have zero measure, which requires a = b−1 . Thus, we can only
have a Riesz basis when ab = 1.
11.7 Suppose g ∈ Cc (R) is not the zero function. Let b be small enough that
supp(g) is contained in some interval I of length 1/b. Since g is continuous
and not everywhere zero, there exists some interval J such that g(x) 6= 0
for all x ∈ J. Let a > 0 be smaller than the length of J. Then G0 (x) =
P 2
k |g(x − ak)| is continuous, a-periodic, and everywhere nonzero. Therefore
0 < inf G0 ≤ sup G0 < ∞, so G(g, a, b) is a frame by Theorem 11.4.
11.8 Suppose 0 < a < b−1 0 . The support condition on g combined with the
fact that g(x) > 0 on (0, b−1
0 ) implies that
X
0 < A ≤ |g(x − ak)|2 ≤ B < ∞
k∈Z
for all x ∈ R. Also, if 0 < b < b0 then supp(g) = [0, b−1

0 ] ⊆ [0, b
−1
]. Therefore
the hypotheses of Theorem 11.4(a) are satisfies, which implies that G(g, a, b)
is a frame.
2
11.9 We are given f (x) = e−1/x χ(0,∞) (x). For x > 0, we have
−2 −2
f ′ (x) = 2x−3 e−x = p1 (x−1 ) e−x ,
where p1 (t) = 2t3 is a polynomial of degree 3.

−2
Suppose that f (n) (x) = pn (x−1 ) e−x for x > 0, where pn is a polynomial
of degree 3n. Then for x > 0,
−2 −2 −2
f (n+1) (x) = 2x−3 pn (x−1 ) e−x − x−2 p′n (x−1 ) e−x = pn+1 (x−1 ) e−x ,
where
pn+1 (t) = 2t3 pn (t) − t2 p′n (t).
Since 2t3 pn (t) has degree 3n + 3 while t2 p′n (t) has degree 3n + 1, we conclude
that their difference has degree 3n + 3.
Finally, since pn is a polynomial,
−2 2
lim f (n) (x) = lim+ pn (x−1 ) e−x = lim pn (t) e−t = 0.
x→0+ x→0 t→∞
Therefore f (n) is differentiable at x = 0, and f (n) (0) = 0.

11.10 (a), (b) Given 0 < ab < 1, Exercise 11.9 implies that there exists a
function f ∈ Cc∞ (R) such that f > 0 on (0, b−1 ) and f = 0 outside of this
interval. The function
X
F0 (x) = |f (x − ak)|2
k∈Z
is infinitely differentiable because for any given x there are at most a fixed
finite number of terms in the sum. Because a < b−1 it is also strictly positive
everywhere, so
f
g = 1/2
F0
is infinitely differentiable and supported in [0, b−1 ]. Further,
X
G0 (x) = |g(x − ak)|2
k∈Z
X |f (x − ak)|2
= P 2
k∈Z j∈Z |f (x − aj)|
P
|f (x − ak)|2
= Pk∈Z 2
= 1.
j∈Z |f (x − aj)|
Theorem 11.4 therefore implies that G(g, a, b) is a b−1 -tight frame. By rescal-
ing, we can make it a Parseval frame.
11.11 (a) If G(g, a, b) is a Riesz frame, then it is a frame, and ab = 1 by
Corollary 11.7(e).
Conversely, if ab = 1 and G(g, a, b) is a frame then Corollary 11.7(e) implies
that it is a Riesz basis.
(b) If G(g, a, b) is an orthonormal basis, then it is a tight frame and kgkL2 =

1. Further, G(g, a, b) is a Riesz basis, so ab = 1 by part (a).
Conversely, suppose that G(g, a, b) is an A-tight frame, ab = 1, and we
have kgk2 = 1. Then G(g, a, b) is a Riesz basis by part (a). Therefore by
Corollary 8.24 we have
A ≤ kgk2L2 ≤ B.
But the frame is tight so A = B, and by hypothesis we have kgk22 = 1, so
G(g, a, b) is a Parseval frame. A Parseval frame that is a Riesz basis is an
orthonormal basis by Corollary 8.23.
11.12 The dilation Da f (x) = a1/2 f (ax) is a unitary mapping on L2 (R), and
Da (Tak g)(x) = a1/2 (Tak g)(ax) = a1/2 g(ax − ak)
= a1/2 g(a(x − k))
= (Da g)(x − k) = Tk (Da g)(x).
Applying Theorem 10.19, we therefore have:
{Tak g}k∈Z is a Riesz sequence with frame bounds A, B
⇐⇒ {Da Tak g}k∈Z is a Riesz sequence with frame bounds A, B
⇐⇒ {Tk (Da g)}k∈Z is a Riesz sequence with frame bounds A, B

X
|(Da g) (ξ + k)|2 ≤ B a.e.
∧
⇐⇒ A ≤
k∈Z
X
⇐⇒ A ≤ g(ξ + k)|2 ≤ B a.e.
|D1/a b
k∈Z
X 2
⇐⇒ A ≤ a−1/2 b
g ξ
+ k ≤ B a.e.
a a
k∈Z
X 2
⇐⇒ Aa ≤ bg ξ + ak ≤ Ba a.e.
k∈Z
∧
11.13 Since (Ta g) = M−a gb, this follows by applying the Fourier transform
to Theorem 11.8.
11.14 This follows by taking A = B in Theorem 11.9.
11.15 For simplicity, fix p, q, and write W = W (Lp , ℓq ). It is straightforward
to check that k · kW is a norm on W, so we concentrate on showing that this
space is complete. For simplicity, we consider only 1 ≤ q < ∞, as the case
q = ∞ is similar.
∞
By Exercise
P 3.3, it suffices to show
P that if {fn }n=1 is a sequence of functions
in W with kfn kW < ∞, then fn converges to an element of W. Now, for
any fixed j we have
X X X 1/q X
kfn · χ[j,j+1] kLp ≤ kfn · χ[k,k+1] kqLp = kfn kW < ∞.
n n k∈Z n
P
Therefore the series gj = n fn · χ[j,j+1] converges absolutely in Lp [j, j + 1].
Define g on R by setting g(x) = gj (x) for x ∈ [j, j + 1]. Then g · χ[j,j+1] =
gj ∈ Lp [j, j + 1], and by applying the Triangle Inequality in ℓq we obtain
X 1/q
kgkW = kg · χ[k,k+1] kqLp
k∈Z
X X q 1/q

= fn · [k,k+1]
χ
p
k∈Z n L
X X q 1/q
= kfn · χ[k,k+1] kLp
k∈Z n
X X 1/q
≤ kfn · χ[k,k+1] kqLp
n k∈Z
X
= kfn kW < ∞.
n
Hence g ∈ W, and we have

X q 1/q
X X
g − fn = fn · χ[k,k+1]
g−
n W k∈Z n Lp
X 1/q
= 0 = 0,
k∈Z
P
so g = fn with convergence in the norm of W. Therefore W is complete.
11.16 Let
E = [0, 12 ) ∪ [1 + 21 , 1 + 43 ) ∪ [2 + 34 , 2 + 87 ) ∪ + · · · .
The sets E + k for k ∈ Z are disjoint, so {Tk g}k∈Z is orthonormal. Also,
E mod 1 = {x mod 1 : x ∈ E} = [0, 1),
and the “pieces” have no overlap when they are translated back to the inter-
val [0, 1). This implies that {Mn g}n∈Z is orthonormal. Together, G(g, 1, 1) is
orthonormal.
Suppose that f is orthogonal to Mn Tk g for every k, n ∈ Z. Fix k = 0.
Take f · χE and translate each “piece” back into the interval [0, 1). Because
the exponentials e2πinx are 1-periodic, the resulting function h on [0, 1) is
orthogonal to e2πinx for each n ∈ Z, and therefore is zero a.e. Repeating for
n 6= 0 shows that f = 0 a.e. everywhere. Hence G(g, 1, 1, ) is complete.
11.17 (a) If f ∈ W (L∞ , ℓ1 ) then
Z ∞ X Z k+1 X
|f (x)| dx = |f (x)| dx ≤ sup |f (x)| = kf kW (L∞ ,ℓ1 ) ,
−∞ k∈Z k k∈Z x∈[k,k+1]
so f ∈ L1 (R). Also
sup |f (x)| = sup sup |f (x)| ≤ kf kW (L∞ ,ℓ1 ) ,
x∈R k∈Z x∈[k,k+1]
so f ∈ L∞ (R). It follows that f ∈ Lp (R) for every p.

Finally, Cc (R) ⊆ W (L∞ , ℓ1 ), so W (L∞ , ℓ1 ) is dense in Lp (R) for all fi-
nite p.
(b) Fix f ∈ W (L∞ , ℓ1 ) and a > 0. Let j ∈ Z be such that j ≤ a ≤ j + 1.
Given k ∈ Z and x ∈ [k, k + 1], we have
k − j − 1 ≤ x − j − 1 ≤ x − a ≤ x − j ≤ k + 1 − j.
Hence
kTa f · χ[k,k+1] kL∞ = sup |f (x − a)|
x∈[k,k+1]
≤ sup |f (y)|
y∈[k−j−1,k−j+1]
= kf · χ[k−j−1,k+j+1] kL∞
≤ kf · χ[k−j−1,k−j] kL∞ + kf · χ[k−j,k−j1 ] kL∞ .

Therefore
X
kTa f kW (L∞ ,ℓ1 ) = kTa f · χ[k,k+1] kL∞
k∈Z
X X
≤ kf · χ[k−j−1,k−j] kL∞ + kf · χ[k−j,k−j1 ] kL∞
k∈Z k∈Z
= 2 kTaf kW (L∞ ,ℓ1 ) .
(c) If f ∈ L∞ (R) and g ∈ W (L∞ , ℓ1 ) then

X
kf gkW (L∞ ,ℓ1 ) = kf g · χ[k,k+1] kL∞
k∈Z
X
≤ kf kL∞ kg · χ[k,k+1] kL∞
k∈Z
= kf kL∞ kgkW (L∞ ,ℓ1 ) .

(d) The argument for the lower inequality is entirely symmetrical to the
one given for the upper inequality.
11.18 There are several points in the computation that should be justified.
(a) Since g belongs to W (L∞ , ℓ1 ), it is integrable. As f is bounded, the
function f (x) e−2πibnx g(x − ak) is integrable on R, and therefore the equality
Z ∞
f (x) e−2πibnx g(x − ak) dx
−∞
Z b−1 X
j
= f x− j
b e−2πibn x− b
g x − ak − j
b dx
0 j∈Z
is justified.
(b) Because f and Fk are bounded and g is integrable, the function
f (x) g(x − ak) Fk (x)
is integrable. This justifies the equality
Z b−1 X

f x − jb g x − ak − jb Fk x − jb dx
0 j∈Z
Z ∞
= f (x) g(x − ak) Fk (x) dx
−∞
(c) Now we justifty the interchange in order occuring at the line

X Z b−1 X
f x − jb g x − ak − jb Fk x − jb dx
k∈Z 0 j∈Z
Let R > 0 be such that supp(f ) ⊆ [−R, R]. Then

XXZ ∞
|f (x)| |g(x − ak)| |f x − jb | |g x − ak − jb | dx
k∈Z j∈Z −∞
XXZ R
j
= |f (x)| |g(x − ak)| |f x − b | |g x − ak − jb | dx
k∈Z j∈Z −R
XXZ R
≤ |f (x)| kTak g · χ[−R,R] kL∞
k∈Z j∈Z −R

× |f x − jb | kTak+ j g · χ[−R,R] kL∞ dx
b
X X
2
≤ 2R kf kL∞ kTak g · χ[−R,R] kL∞ kT j g · χ[−R,R] kL∞
b
k∈Z j∈Z
≤ 2R kf k2L∞ C kgk2W (L∞ ,ℓ1 )

where C is some appropriate constant. Therefore Fubini’s Theorem justifies

the change in order that occurs at this step of the computation.
(d) The fact that the series defining Lf converges absolutely in L2 -norm
justifies the interchange
XZ ∞
Z ∞ X
f (x) f x − jb Gj (x) dx = f (x) f x − jb Gj (x) dx.
j∈Z −∞ −∞ j∈Z
11.19 Suppose
G(g, a, b) = Mbn Tak h k,n∈Z
is a frame with frame bounds A, B. By Theorem 11.18, if g − h ∈ W (L∞ , ℓ1 )

then
G(g − h, a, b) = Mbn Tak (g − h) k,n∈Z
is a Bessel sequence with Bessel bound
2
K = C1/a Cb kg − hk2W (L∞ ,ℓ1 ) .
b
If we choose δ small enough and require kg − hkW (L∞ ,ℓ1 ) < δ, then we will
have K < A. Exercise 8.10 therefore implies that

G(h, a, b) = Mbn Tak h k,n∈Z
is a frame.
11.20 Since G(χ[0,1] , 1, 1) and {Enk }k,n∈Z are each orthonormal basis, it
follows from Theorem 4.20 that they are equivalent bases. Hence there is a
topological isomorphism Z such that Z(Mn Tk χ[0,1] ) = Enk . The fact that
both systems are orthonormal bases implies that this mapping is unitary, and
that Z is unique.
11.21 Fix f ∈ L2 (R) or f ∈ W (Lp , ℓ1 ). Given x, ξ ∈ R and m, n ∈ Z, we
have
X
Zf (x + m, ξ + n) = f (x + m − j) e2πij(ξ+n)
j∈Z
X
= f (x + m − j) e2πijξ
j∈Z
X
= f (x − j) e2πi(j+m)ξ
j∈Z
X
= e2πimξ f (x + j) e2πijξ
j∈Z
2πimξ
= e Zf (x, ξ).
The series above converge unconditionally in the norm of Lp (Q). We can

justify the “pointwise” calculations above in several ways, e.g.,

Zf (x + m, ξ + n) − e2πimξ Zf (x, ξ) p
L (Q)

X X
=
f (x + m − j) e 2πij(ξ+n)
− e 2πimξ
f (x − j) e 2πijξ

j∈Z j∈Z Lp (Q)
X X

=
f (x + m − j) e 2πijξ
− f (x − j) e 2πi(j+m)ξ

j∈Z j∈Z Lp (Q)
= 0.
11.22 (a) Because f ∈ L1 (R), we have
XZ 1 XZ 1
2πijξ
|f (x − j) e | dξ = |f (x − j)| dξ = kf kL1 < ∞.
j∈Z 0 j∈Z 0
Fubini’s Theorem therefore tells us that

Z 1X XZ 1
f (x − j) e2πijξ dξ = f (x − j) e2πijξ dξ.
0 j∈Z j∈Z 0
Fix any a ∈ R. Since Zf ∈ L1 (Q) and Zf is quasiperiodic we have

Z a+1 Z 1
|Zf (x, ξ)| dx dξ < ∞.
a 0
Fubini’s Theorem therefore implies that

Z 1
F (x) = Zf (x, ξ) dξ
0
1
is defined a.e. and belongs to L [a, a + 1]. We compute that
Z 1

kf − F kL1 [a,a+1] = f (x) − Zf (x, ξ) dξ

0 L1 [a,a+1]
Z a+1 Z 1 X

= f (x) − f (x − j) e2πijξ dξ dx

a 0 j∈Z
Z a+1 X Z 1

= f (x) − f (x − j) e2πijξ dξ dx

a j∈Z 0
Z a+1 X

= f (x) − f (x − j) δ0,j dx

a j∈Z
= 0.
Consequently f (x) = F (x) for almost every x ∈ [a, a + 1].

(b) If Zf is continuous then quasiperiodicity implies that it is uniformly
continuous. Using part (a) we have that
kf − Ta f kL∞ = sup |f (x) − f (x − a)|

x∈R
Z 1
≤ sup |Zf (x, ξ) − Zf (x − a, ξ)| dξ
x∈R 0
≤ sup sup |Zf (x, ξ) − Zf (x − a, ξ)|

x∈R ξ∈[0,1]
≤ kZf − T(a,0) Zf kL∞
→ 0 as a → 0,
where T(a,b) denotes the two-dimensional translation operator. Therefore f is

uniformly continuous.
(c) It follows immediately from part (a) that if Zf = 0 then f = 0, so Z
is injective.
Since Z(Mn Tk χ[0,1] ) = Enk , the range of Z contains span{Enk }k,n∈Z .
The two-dimensional analogue of Theorem 13.21 tells us that {Enk }k,n∈Z is
complete in L1 (Q), so this implies that range(Z) is dense.
Suppose that Z : L1 (R) → L1 (Q) was surjective. Choose any f ∈ L2 (R).
Then Zf ∈ L2 (Q) ⊆ L1 (Q). Since Z is a surjective map of L1 (R) onto L1 (Q),
there must be some g ∈ L1 (R) such that Zg = Zf.
Now, since f belongs to L2 (R), the series defining Zf converges (uncon-
ditionally) in L2 (Q). Since Enk ∈ L2 (Q), we can use the continuity of the
inner product and Fubini’s Theorem to compute the Fourier coefficients of
Zf. Specifically, given k, n ∈ Z,
X

c
Zf(n, −k) = Zf, Enk = f (x − j)e 2πijξ
, e 2πinx −2πikξ
e
j∈Z L2 (Q)
X

= f (x − j)e2πijξ , e2πinx e−2πikξ L2 (Q)
j∈Z
XZ 1 Z 1
= f (x − j) e2πijξ e−2πinx e2πikξ dx dξ
j∈Z 0 0
XZ 1 Z 1
= f (x − j) e−2πinx dx e2πi(j+k)ξ dξ
j∈Z 0 0
XZ 1
= f (x − j) e−2πinx dx (δ−j,k )
j∈Z 0
Z 1
= f (x + k) e−2πinx dx
0

= f χ[k,k+1] , e2πinx L2 [k,k+1] .
∧
= f χ[k,k+1] (n).
Thus, the (n, −k)th Fourier coefficient of Zf ∈ L2 (Q) is the nth Fourier
coefficient of f χ[k,k+1] ∈ L2 [k, k + 1] ⊆ L1 [k, k + 1].
On the other hand, the series defining Zg converges absolutely in the
c
norm of L1 (Q), so an entirely similar calculation shows that Zg(n, −k) =
∧
χ
g [k,k+1] (n). Since Zf = Zg, we therefore have
∧ ∧
f χ[k,k+1] (n) = g χ[k,k+1] (n), k, n ∈ Z.
Now, f χ[k,k+1] and g χ[k,k+1] both belong to L1 [k, k + 1], and the Uniqueness
Theorem for Fourier coefficients (Theorem 13.26) tells us that functions in
L1 [k, k+1] are uniquely determined by their Fourier coefficients, so this implies
that f χ[k,k+1] = g χ[k,k+1] a.e. for each k ∈ Z. Hence f = g a.e.
But this tells us that every function f ∈ L2 (R) equals a function
g ∈ L1 (R), and hence L2 (R) ⊆ L1 (R). This is false, so we have obtained
a contradiction.
(d) Suppose Z −1 : range(Z) → L1 (R) was continuous. By definition of in-
verse, we have Z −1 Zf = f for f ∈ L1 (R) and ZZ −1 F = F for F ∈ range(Z).
However, range(Z) is dense, so Z −1 has a unique extension to a continuous
mapping Z −1 : L1 (Q) → L1 (R). By continuity, we obtain ZZ −1 F = F for all
F ∈ L1 (Q). But then Z is surjective, which contradicts part (c).
11.23 We are given f ∈ L2 (R) such that Zf is continuous.
(a) If f is even then
X 1
Zf (1/2, 1/2) = f ( 21 − j) e2πi 2 j
j∈Z
X 1
= f (− 21 − j) e2πi 2 j
j∈Z
X 1
= f (− 21 − (j − 1)) e2πi 2 (j−1)
j∈Z
X 1
= f ( 21 − j) e2πi 2 j e−πi
j∈Z
X 1
= − f ( 21 − j) e2πi 2 j = −Zf (1/2, 1/2).
j∈Z
Therefore Zf (1/2, 1/2) = 0.

(b) If f is odd then

X
Zf (0, 0) = f (−j)
j∈Z
X
= − f (j)
j∈Z
X
= − f (−j) = −Zf (0, 0).
j∈Z
Therefore Zf (0, 0) = 0.
Since Zf is 1-periodic in the second variable,
X 1
Zf (0, 1/2) = f (−j) e2πi 2 j
j∈Z
X 1
= − f (j) e2πi 2 j
j∈Z
X 1
= − f (−j) e2πi 2 (−j)
j∈Z
= Zf (0, −1/2) = Zf (0, 1/2).
Therefore Zf (0, 1/2) = 0.

Finally,
X
Zf (1/2, 0) = f ( 21 − j)
j∈Z
X
= − f (− 21 + j)
j∈Z
X
= − f (− 21 − (j − 1))
j∈Z
X
= − f ( 21 − j) = −Zf (1/2, 0).
j∈Z
Therefore Zf (1/2, 0) = 0.
(c) If f is real-valued then
X 1
Zf (x, 1/2) = f (x − j) e2πi 2 j
j∈Z
X 1
= f (x − j) e−2πi 2 j = Zf (x, −1/2) = Zf (x, 1/2).
j∈Z
Therefore Zf (x, 1/2) is real for each x ∈ [0, 1]. However, by quasiperiodicity,
Zf (1, 1/2) = Zf (0 + 1, 1/2) = eπi Zf (0, 1/2) = −Zf (0, 1/2).
Thus, Zf (0, 1/2 and Zf (1, 1/2) have opposite signs. Since Zf (x, 1/2) is con-
tinuous and real-valued, the Intermediate Value Theorem implies that there
must be some x ∈ [0, 1] such that Zf (x, 1/2) = 0.
11.24 The proof is very similar to the proof of Exercise 10.8 so we will only
expand on the following point.
(d) ⇒. Suppose that G(g, 1, 1) is a frame. Then it is a Bessel sequence,
so by part (b) there exists some B > 0 such that |Zg|2 ≤ B a.e., and B is
a Bessel bound. An almost identical argument based on the lower the frame
bound shows that there exists some A > 0 such that |Zg|2 ≥ A a.e., and A is
a lower frame bound.
Since Q has finite measure and we have A ≤ |Zg|2 ≤ B a.e., it follows that
1/Zg ∈ L2 (Q). Therefore G(g, 1, 1) is exact by part (b). Finally, exact frames
coincide with Riesz bases.
PN
11.25 (a) Suppose p(x) = k=−N ck e2πikx satisfies |p(x)| = 1 a.e. Set ck = 0
for |k| > N. Then for almost every x we have
N
X N
X
1 = |p(x)|2 = cj ck e2πi(j−k)x
j=−N k=−N
XX
= cj ck+j e2πi(j−(k+j))x
j∈Z k∈Z
XX
= cj ck+j e2πikx .
k∈Z j∈Z
Since |p|2 is continuous, it belongs to L2 (T). However, {e2πikx }k∈Z is an or-

thonormal basis for L2 (T), see Example 1.52. By uniqueness, we therefore
have X
cj ck+j = δ0k .
j∈Z
(We could also use the Fundamental Theorem of Algebra or properties of

analytic functions to arrive at the same conclusion.)
Suppose that cm 6= 0 and cn 6= 0, while ck = 0 for k < m and k > n.
Suppose m 6= n. If n 6= 0 then consider k = n − m. We can only have cj 6= 0
for m ≤ j ≤ n, but in this case n − m + j > n except for j = m, in which case
n − m + j = n. Therefore
X
0 = cj cn−m+j = cn cn = |cn |2 6= 0,
j∈Z
which is a contradiction. Similarly, if m 6= 0 then by considering k = m −

n we obtain a contradiction. Therefore we must have m = n, so p(x) =
cn e2πinx a.e. (in fact, equality must hold everywhere since both sides are
continuous). Finally, |cn | = |p(x)| = 1.
(b) ⇒. Suppose that g ∈ L2 (R) is compactly supported and G(g, 1, 1) is
an
P orthonormal basis for L2 (R). Then |Zg|2 = 1 a.e. by Theorem 11.31 and
|g(x − j)|2 = 1 a.e. by Theorem 11.6.
Fix any particular representative of g. If supp(g) ⊆ [−N, N + 1] then
N
X
Zg(x, ξ) = g(x − j) e2πijξ .
j=−N
With x ∈ R fixed, Zg(x, ·) is a trigonometric polynomial. For almost every x,

we |Zg(x, ·)| = 1 a.e. Part (a) therefore implies that for such an x, g(x − j)
can be nonzero for at most one j (with j depending on x), and |g(x − j)| = 1
for that j. Set E P
= {x ∈ R : g(x)P6= 0}. Then we have |g(x)| = 1 for x ∈ E,
so |g| = χE , and χE (x − j) = |g(x − j)|2 = 1 a.e. Finally, E is bounded
since g is compactly supported.
χ
P ⇐. Suppose that |g| = E where E is a bounded set that satisfies
χ
k E (x − k) = 1 a.e. Fix any representative of g. Then for any particu-
lar x we have
X
Zg(x, ξ) = g(x − j) e2πijξ = g(x − k) e2πikx ξ
j∈Z
for some kx ∈ Z. For almost every x we must have |g(x − kx )| = 1. Conse-

quently |Zg(x, ξ)| = 1 a.e., so G(g, 1, 1) is an orthonormal basis.
11.26 Suppose c ∈ ℓ2 (Z2 ) and Rc = 0. Since G(φ, 1, 1) is a Bessel sequence,
the series defining Rc converges unconditionally in L2 -norm, so
X
cnk Mn Tk φ = Rc = 0.
k,n∈Z
The Zak transform is unitary, so this implies that

X X
cnk Enk Zφ = cnk Z(Mn Tk φ) = Z(Rc) = 0.
k,n∈Z k,n∈Z
Since {Enk }k,n∈Z is an orthonormal basis for L2 (Q) and c ∈ ℓ2 (Z2 ), the series
X
F = cnk Enk
k,n∈Z
converges in L2 -norm. Further, Zφ is bounded since G(φ, 1, 1) is Bessel, so we

have X
F Zg = cnk Enk Zg = 0.
k,n∈Z
However, Zg(x, ξ) 6= 0 for almost every (x, ξ), so this implies that F = 0 a.e.
Finally, the fact that {Enk }k,n∈Z is an orthonormal basis therefore implies
that c = 0.
11.27 Suppose g ∈ L2 (R) and x |g(x)|2 → c as x → ∞. If c > 0, then
for x large enough we have |g(x)|2 ≥ c/x, which contradicts the fact that g is
square integrable.
11.28 Let fn = χ[n,n+1] . Then kfn kL2 = 1, and
Z n+1 n+1
x3 (n + 1)3 − n3
kP fn k2L2 = 2
x dx = = .
n 2 n 2
Hence fn ∈ DP , but kP fn kL2 → ∞ as n → ∞. Therefore P is unbounded.

Now consider gn (x) = n1/2 g(nx), where g is any differentiable function in
L (R) such that g ′ ∈ L2 (R) (such as the Gaussian function). Then kgn kL2 =
2
kgkL2 for every n, but gn′ (x) = n3/2 g ′ (nx), so kgn′ kL2 = nkg ′ kL2 . Hence
kM gn kL2 → ∞ as n → ∞, so M is unbounded.
11.29 The position and momentum operators map S(R) into itself simply
by definition of the Schwartz class. Given f, g ∈ S(R), we have
Z Z
hP f, gi = xf (x) g(x) dx = f (x) xg(x) dx = hf, P gi,
so P is self-adjoint on this domain. This calculation also holds on the larger

domain
DP = {f ∈ L2 (R) : xf (x) ∈ L2 (R)}.
Given f, g ∈ S(R), by using integration by parts we have
Z
1
hM f, gi = − f ′ (x) g(x) dx
2πi
∞ Z
f (x) g(x) 1
= + f (x) g ′ (x) dx = 0 + hf, M gi,
−2πi −∞ 2πi
so M is self-adjoint on this domain. This calculation is valid on the larger

domain
SM = L2 (R) ∩ C01 (R).
11.30 If f is differentiable everywhere, then
[P, M ]f (x) = P M f (x) − M P f (x)

1
= xM f (x) − (P f )′ (x)
2πi
1 1
= xf ′ (x) − xf ′ (x) + f (x)
2πi 2πi
1
= − f (x).
2πi
To relate this to equation (11.32), note that if f is smooth enough and has
sufficient decay then the following integrals are well defined, and by rearrang-
ing terms and using integration by parts we have
Z Z

1 1
M f, P f − P f, M f = f ′ (x) xf (x) dx − xf (x) f ′ (x) dx
2πi 2πi
Z Z
1 1
= xf ′ (x) f (x) dx − (P f )′ (x) f (x) dx
2πi 2πi
Z Z
= P M f (x) f (x) dx − M P f (x) f (x) dx

= P M f − M P f, f

= [P, M ]f, f
1
= − hf, f i
2πi
1
= − kf k2L2 .
2πi
This is equation (11.32), noting that the function g in that equation has unit
norm.
11.31 (a) For f, g ∈ S we have that

[A, B]f, f = ABf, f − BAf, f

= Bf, Af − Af, Bf

= 2i Im Bf, Af .
Hence

[A, B]f, f = 2 Im Bf, Af

≤ 2 Bf, Af
≤ 2 kBf k kAf k.
(b) Equality holds in the first inequality above if and only if hBf, Af i is
purely imaginary. Equality holds in the second inequality above if and only if
Bf is a scalar multiple of Af. So, we must have Bf = aAf for some a ∈ C,
and therefore hBf, Af i = akAf k2 . In order for this quantity to be purely
imaginary, we must have a = ci for some real c.
11.32 Given f ∈ S(R) we have
kxf (x)k2 kξ fb(ξ)k2 = kP f k2 kM f k2

1

≥ [P, M ]f, f
2
1
= kf k22 ,
4π
the last equality following from Exercise 11.30.
11.34 Set λk = (pk , qk ). Let z = λ1 , u = λ2 − λ1 , and v = λ3 − λ1 . Then since
λ1 , λ2 , λ3 are not collinear, the points 0, u, v are not collinear, which means
that u, v are independent. Therefore the matrix A that has u, v as columns
is invertible. If we let e1 = (1, 0) and e2 = (0, 1) be the standard basis vectors
in R2 (written as column vectors), then we have
A0 + z = 0 + λ1 = λ1 ,
Ae1 + z = u + λ1 = λ2 ,
Ae2 + z = v + λ1 = λ3 .
Hence λ1 , λ2 , and λ3 lie on A(Z2 ) + z.

11.35 (a) Suppose that G(g, Λ) is linearly independent. Write z = (a, b) and
PN
suppose that k=1 ck Mqk +b Tpk +a g = 0 a.e. Then we have
N
X
ck e2πi(qk +b)x g(x − pk − a) = 0 a.e.
k=1
Replacing x by x + a it follows that almost every x we have

N
X
0 = ck e2πi(qk +b)(x+a) g(x − pk )
k=1
N
X
= ck e2πi(qk +b)a e2πi(qk +b)x g(x − pk )
k=1
N
X
= e2πibx ck e2πi(qk +b)a e2πiqk x g(x − pk )
k=1
N
X
= e2πibx c′k Mqk Tpk g(x),
k=1
where c′k = ck e2πi(qk +b)a . Since e2πibx is everywhere nonzero and G(g, Λ) is
independent, this implies that c′k = 0 for each k, and hence ck = 0 for each k.
Therefore G(g, Λ + z) is linearly independent. The converse implication is
similar.
(b) Suppose that G(g, Λ) is independent. Define

ak pk 10 pk pk
= Sr = = ,
bk qk r1 qk rpk + qk
N PN
so G(h, Sr (Λ)) = Mbk Tak h k=1 . Suppose that k=1 ck Mbk Tak h = 0 a.e.
Then for almost every x we have
N
X
0 = ck Mbk Tak h(x)
k=1
N
X
= ck e2πibk x h(x − ak )
k=1
N
X 2
= ck e2πi(rpk +qk )x eπir(x−pk ) g(x − pk )
k=1
N
X 2 2
= ck e2πiqk x e2πirpk x eπirx e−2πirpk x eπirpk g(x − pk )
k=1
N
X
2 2
= eπirx ck eπirpk e2πiqk x g(x − pk )
k=1
N
X
2
= eπirx c′k Mqk Tpk g(x),
k=1
2 2
where c′k = ck eπirpk . Since eπirx is everywhere nonzero, we conclude that
PN ′
k=1 ck Mqk Tpk g = 0 a.e. Since G(g, Λ) is independent, this implies that
c′k = 0 for every k, and hence ck = 0 for every k. Therefore G(h, Sr (Λ)) is
independent. The converse implication is similar.
(c) Suppose that G(g, Λ) is independent. Note that

p 0 −1 pk −qk
R k = = ,
qk 1 0 qk pk
∨

∨ N PN ∨
so G( g , R(Λ)) = Mpk T−qk g k=1 . Suppose that k=1 ck M−pk Tqk g = 0 a.e.
Then for almost every x we have
N
X ∨
0 = ck Mpk T−qk g (ξ)
k=1
N
X ∨
= ck (Tpk Mqk g) (ξ)
k=1
N
X ∨
= c′k (Mqk Tpk g) (ξ),
k=1
where c′k
=e −2πipk qk
ck . Since the Fourier transform is unitary and G(g, Λ) is
independent, we conclude that c′k = 0 for every k, and therefore ck = 0 for
∨
each k. Hence G( g , R(Λ)) is independent.
(d) Suppose that G(g, Λ) is independent. Note that

pk a 0 pk apk
Da = = ,
qk 0 1/a qk qk /a
N PN
so G(h, Da (Λ)) = Mqk /a Tapk h k=1 . If k=1 ck Mqk /a Tapk h = 0 a.e., then for
almost every y = ax we have
N
X
0 = ck Mqk /a Tapk h(y)
k=1
N
X
= ck e2πi(qk /a)(ax) h(ax − apk )
k=1
N
X
= ck e2πiqk ax g(x − pk )
k=1
N
X
= ck Mqk Tpk g(x).
k=1
Since G(g, Λ) is independent, this implies that ck = 0 for each k, and therefore
G(h, Da (Λ)) is independent.
(e) Suppose that

x y
A = , xw − yz = 1.
zw
If y 6= 0, then

−y 0 1 0 0 −1 1 0
D−y Syw RSx/y =
0 −1/y yw 1 1 0 x/y 1
" #
x y
= xyw−y yw
y2 y
" #
x y
= xw−1
y w

x y
= = A.
zw
On the other hand, if y = 0 then xw = 1, so x 6= 0 and we have

1 0 x 0
Sz/x Dx =
z/x 1 0 1/x

x 0
=
z 1/x

x 0
= = A.
zw
11.36 Set Λ = A(Z2 ) + z. If G(g, Λ) is dependent, then G(g, A(Z2 )) is depen-

dent by Exercise 11.35(a). By part (e) of that exercise, A is some scalar mul-
tiple of a product of matrices of the form Sr , R, and Da . Applying parts (b),
(c), (d) of Exercise 11.35, we conclude that G(h, cZ2 ) is dependent, where h
is not the zero function. However, G(h, cZ2 ) = G(h, c, c), which we know is
independent by Theorem 11.36, so we have a contradiction.
PN
11.37 (a) Suppose that k=1 ck Mqk g = 0 a.e., where not all ck are zero.
PN
Define m(x) = k=1 ck e2πiqk x . Then, since the sum is finite we have
N
X N
X
m(x) g(x) = ck e2πiqk x g(x) = ck Mqk g(x) = 0 a.e.
k=1 k=1
However, a nontrivial trigonometric polynomial cannot vanish on any set of

positive measure, so this implies that g = 0 a.e.
PN
(b) Suppose that k=1 ck Tpk g = 0 a.e., where not all ck are zero. Applying
the Fourier transform, we then have
N
X X
N ∧
ck M−pk b
g(ξ) = ck Tpk g (ξ) = 0 a.e.
k=1 k=1
By part (a) this implies that gb = 0 a.e., and therefore g = 0 a.e. since the
Fourier transform is unitary.
(c) We are given N distinct but collinear points Λ = {(pk , qk )}N k=1 . By
Exercise 11.35(a), if z ∈ R2 then G(g, Λ) is independent if and only if G(g, Λ +
z) is independent. Therefore we can assume that Λ is contained in a line that
passes through the origin.
N
If this line is vertical, then pk = 0 for every k and G(g, Λ) = Mqk g k=1
is independent by part (a).
If this line is not vertical then it has some finite slope r. Define h(x) =
2
e−πirx g(x). By Exercise 11.35(b), G(g, Λ) is linearly independent if and only
if G(h, S−r (Λ)) is linearly independent. However,

pk 1 0 pk pk
S−r = = ,
qk −r 1 rpk 0
N
so G(h, S−r (Λ)) = Tpk h k=1 . This set is independent by part (b) of this
exercise, so we conclude that G(g, Λ) is independent as well.
11.38 Fix any nonzero g ∈ L2 (R) with g supported on a half-line of the
form [a, ∞). By taking a as large as possible we may assume that g is not
zero almost everywhere on [a, a + ε] for all ε > 0. Given finitely many distinct
points in R2 , we can write these points as
{(pk , qk,j )}j=1,...,Mk , k=1,...,N ,
i.e., for each distinct translate we group the corresponding modulates together.
We can assume that p1 < · · · < pN . Given scalars ck,j , suppose that
X Mk
N X N
X
0 = ck,j Mqk,j Tpk g(x) = mk (x) g(x − pk ) a.e., (∗)
k=1 j=1 k=1
where
Mk
X
mk (x) = ck,j e2πiqk,j x .
j=1
Since g is supported in a half-line, the supports of the functions g(x − pk )

overlap some places but not others. If we choose x ∈ [a + p1 , a + p1 + ε] with
ε small enough then only the terms with k = 1 can be nonzero. For such x,
the right-hand side of equation (∗) will contain only one nonzero term, i.e.,
it reduces to m1 (x) g(x − p1 ) = 0 a.e. on [a + p1 , a + p1 + ε]. However, the
trigonometric polynomial mk (x) cannot vanish on any set of positive measure
unless it is the trivial polynomial. Therefore we must have c1,j = 0 for all
j. We can then repeat this argument for k = 2 to obtain c2,j = 0, etc. A
symmetric argument applies if g is supported within (−∞, a].
11.39 (a) We have
2 2
Hn′ (x) = D eπx Dn e−2πx
2 2 2 2
= 2πxeπx Dn e−2πx + eπx Dn+1 e−2πx
= 2πxHn (x) + Hn+1 (x).
2 2
(b) We have that H0 (x) = e−πx = p0 (x) e−πx , where p0 (x) = 1 has
degree 0 and leading coefficient 1 = (−4π)0 .
2
Assume that for some n ≥ 0 we have Hn (x) = pn (x) e−πx for some
polynomial pn of degree n and leading coefficient (−4π)n . Then
Hn+1 (x) = Hn′ (x) − 2πxHn (x)

2 2 2
= p′n (x) e−πx − 2πx pn (x) e−πx − 2πx pn (x) e−πx
2 2
= p′n (x) e−πx − 4πx pn (x) e−πx
2
= pn+1 (x) e−πx ,
where
pn+1 (x) = p′n (x) − 4πxpn (x).
Since p′n has degree n−1 and 4πxpn (x) has degree n+1, we conclude that pn+1
has degree n + 1, and the leading coefficient is (−4π) · (−4π)n = (−4π)n+1 .
2
11.40 We are given h(x) = p(x) e−πx where p is a nontrivial polynomial.
Given a finite set of distinct points in R2 we can write these points as
Λ = {(pk , qk,j )}j=1,...,Mk , k=1,...,N ,
with p1 < · · · < pN . If N = 1 then Λ is a set of collinear points, and therefore

G(h, Λ) is independent by Exercise 11.37. Therefore we we may assume that
N > 1.
Given scalars ck,j , not all zero, suppose
Mk
N X
X
ck,j Mqk,j Tpk h(x) = 0 a.e.
k=1 j=1
Because of the special form of h, this simplifies to

N
X
2
e−πx mk (x) p(x − pk ) e2πpk x = 0 a.e., (A)
k=1
where
Mk
X 2
mk (x) = ck,j e−πpk e2πiqk,j x .
j=1
Without loss of generality we may assume that m1 and mN are nontrivial,

otherwise ignore those terms and reindex. Then dividing both sides of equa-
2
tion (A) by e−x e2pN x and rearranging, we have
N
X −1
mN (x) p(x − pN ) = − mk (x) p(x − pk ) e2(pk −pN )x a.e. (B)
k=1
However, pk − pN < 0 for k = 1, . . . , N − 1, so since each mk is bounded, the

right-hand side of equation (B) converges to zero as x → ∞. On the other
hand, as mN is a nontrivial trigonometric polynomial and p is a nontrivial
polynomial, the left-hand side of equation (B) does not converge to zero as
x → ∞, which is a contradiction.
11.41 Any finite set of vectors is a frame for its span, so let let A, B be frame
bounds for G(g, Λ) as a frame for its span in L2 (R). Fix 0 < δ < A1/2 /(2N 1/2 ).
Then by the strong continuity of the translation and modulation operators,

we can choose ε small enough that for all |r| ≤ ε we have
kTr g − gk2 ≤ δ and kMr Tpk g − Tpk gk2 ≤ δ, k = 1, . . . , N.
Now suppose that |pk − p′k | < ε and |qk − qk′ | < ε for k = 1, . . . , N. Then
for any scalars c1 , . . . , cN we have
XN XN

ck Mqk Tpk g ≤ ck (Mqk − Mqk′ )Tpk g
2 2
k=1 k=1
XN XN

+ ck Mqk′ (Tpk − Tp′k )g + ck Mqk′ Tp′k g
2 2
k=1 k=1
N
X
≤ |ck | kMqk −qk′ Tpk g − Tpk gk2
k=1
N
X XN

+ |ck | kTpk −p′k g − gk2 + ck Mqk′ Tp′k g
2
k=1 k=1
N
X XN

≤ 2δ |ck | + ck Mqk′ Tp′k g
2
k=1 k=1
X
N 1/2 XN

≤ 2δN 1/2 |ck |2 + ck Mqk′ Tp′k g .
2
k=1 k=1
However, we also have by Exercise 8.12 that

X
N 1/2 XN

A1/2 |ck |2 ≤ ck Mqk Tpk g .
2
k=1 k=1
Combining and rearranging these inequalities, we find that
X
N 1/2 XN
1/2 1/2 2
A − 2δN |ck | ≤ ck Mqk′ Tp′k g .
2
k=1 k=1
PN
Since A1/2 − 2δN 1/2 > 0, it follows that if k=1 ck Mqk′ Tp′k g = 0 a.e., then
c1 = · · · = cN = 0.
11.42 If m is differentiable at x = 0 and m(0) = 0, then we have
m(x) m(x) − m(0)

lim = lim = m′ (0).
x→0 x x→0 x−0
Set C1 = |m′ (0)|/2 and C2 = 2|m′ (0)|. Then there exists some δ > 0 such
that
|m(x)|
C1 ≤ ≤ C2 , |x| < δ.
|x|
Consequently
C1 |x| ≤ |m(x)| ≤ C2 |x|, |x| < δ,
and therefore
ln C1 + ln |x| ≤ ln |m(x)| ≤ ln C2 + ln |x|, |x| < δ.
Now,
Z δ Z δ
ln |x| dx = 2 ln x dx = 2δ (ln δ − 1).
δ 0
Rδ
Since ln C1 + ln |x| ≤ p(x) ≤ ln C2 + ln |x|, we conclude that −δ
p(x) dx exists
and is finite.
12.1 If W(ψ, a, b) is a wavelet frame then
XX
S(Da f ) = hDa f, Dan Tbk ψi Dan Tbk ψ
n∈Z k∈Z
XX
= hf, D1/a Dan Tbk ψi Dan Tbk ψ
n∈Z k∈Z
XX
= hf, Dan−1 Tbk ψi Dan Tbk ψ
n∈Z k∈Z
XX
= hf, Dan Tbk ψi Dan+1 Tbk ψ
n∈Z k∈Z
XX
= hf, Dan Tbk ψi Da Dan Tbk ψ
n∈Z k∈Z
XX
= Da hf, Dan Tbk ψi Dan Tbk ψ
n∈Z k∈Z
= Da Sf.
Hence the dual frame is

−1
S (Dan Tbk ψ k,n∈Z = Dan (S −1 Tbk ψ) k,n∈Z .
Thus, if we set ψek = S −1 (Tbk ψ) for k ∈ Z, then the canonical dual simply
consists of dilations of these functions ψek , i.e., the canonical dual is

Dan ψek k,n∈Z .
12.2 (a) The general linear group of degree 2 is the set of all 2 × 2 invertible
real matrices. It is denoted by GL2 (R), and it is a group under matrix mul-
tiplication. Since A2 ⊆ GL2 (R), we just have to show that A2 is a subgroup.
We have
a0 c0 ac 0
= ∈ A2 ,
b1 d1 bc + d 1
so A2 is closed under composition. Also,
−1
a0 1/a 0
= ∈ A2 ,
b1 −b/a 1
so A2 is closed under inverses. Therefore A2 is a group.

By construction, A3 is isomorphic to A2 .
Finally, because
Da Tb Dc Td = Da Dc Tbc Td = Dac Tbc+d ,

we have that A1 is isomorphic to A2 and A3 .

(b) Note that
(D1 Tb ) (Dan T0 ) = D1 Dan Tan b T0 = Dan Tan b .
However, since n ranges through Z, the value of an cannot be an integer for

every n. Therefore {Dan Tbk }k,n∈Z is not closed under compositions.
(c) Since G is a group, we have Dan = (Da )n ∈ G and Tbk = (Tb )k ∈ G
for every n, k ∈ Z. Consequently,
Dan T(am j+k)b = Dan−m Dam Tam bj Tbk = Dan−m Tbj Dam Tbk ∈ G.
(d) Fix (u, v) ∈ A3 , and note that u > 0. Given a function F that is
integrable with respect to da a db, by making the change of variable c = au
followed by d = b + (cv)/u, we compute that
Z ∞Z ∞ Z ∞Z ∞
da da
F (u, v) ∗ (a, b) db = F (au, av + b) db
−∞ 0 a −∞ 0 a
Z ∞Z ∞ dc/u
cv
= F c, +b db
−∞ 0 u c/u
Z ∞Z ∞ dc
cv
= F c, + b db
0 −∞ u c
Z ∞Z ∞
dc
= F (c, d) dd
0 −∞ c
Z ∞Z ∞
da
= F (a, b) db.
−∞ 0 a
Similarly, using the change of variable c = au followed by d = bu + v, we

have
Z ∞Z ∞ Z ∞Z ∞
da da
F (a, b) ∗ (u, v) 2 db = F (au, bu + v) 2 db
−∞ 0 a −∞ 0 a
Z ∞Z ∞
dc/u
= F (c, bu + v) db
−∞ 0 (c/u)2
Z ∞Z ∞
dc
= F (c, bu + v) u db 2
0 −∞ c
Z ∞Z ∞
dc
= F (c, d) dd 2
−∞ 0 c
Z ∞Z ∞
da
= F (a, b) 2 db.
−∞ 0 a
12.3 (a) By equation (12.3), the image of W(ψ, a, b) under the Fourier trans-
form is

c a, b) = a−n/2 e2πibka−n ξ ψ(a
W(ψ, b −n ξ)
k,n∈Z
n/2 2πibkan ξ
= a e b n ξ)
ψ(a .
k,n∈Z
Since the Fourier transform is unitary, our goal is to show that this collection
is a Parseval frame for L2 [0, ∞). Note that, with n fixed,
n/2 1/2 2πibkan ξ
a b e k∈Z
is an orthonormal basis for L2 (I) where I is any interval of length a−n b−1 .
Let f be any function in Cc (R). For simplicity let us take c = 1, but the
same argument applies to any positive c. Let d = 1 + b−1 , so ψb ∈ L2 (R)
is supported within [1, d], which is an interval of length b−1 . The dilated
b n ξ) belongs to L2 (In ) where In = [a−n , a−n c], an interval of
function ψ(a
b n ξ)
length a−n (d − 1) = a−n b−1 . Since f is bounded, the product f (ξ) · ψ(a
2
also belongs to L (In ). Applying the Plancherel Equality, we therefore have
Z ∞
b n ξ)|2 dξ =
|f (ξ) ψ(a b n ξ)
f (ξ) ψ(a 2
0 L (In )
X D E 2

= b n ξ), an/2 b1/2 e2πibkan ξ
f (ξ) ψ(a
L2 (In )
k∈Z
2
X Z
= an b b n ξ) e−2πibkan ξ
f (ξ) ψ(a

k∈Z In
2
X Z ∞
= an b b n ξ e−2πibkan ξ .
f (ξ) ψ(a

k∈Z 0
X D E 2

= b b n ξ) e2πibkan ξ
f (ξ), an/2 ψ(a .
L2 [0,∞)
k∈Z
Hence, using Tonelli’s Theorem to interchange the sum and integral,

X D E 2

b n ξ) e2πiban ξ
f (ξ), an/2 ψ(a
L2 [0,∞)
k,n∈Z
XZ ∞
= b−1 b n ξ)|2 dξ
|f (ξ) ψ(a
n∈Z 0
Z ∞ X
= b −1
|f (ξ)| 2 b n 2
ψ(a ξ)| dξ
0 n∈Z
Z ∞
= Ab−1 |f (ξ)|2 dξ
0
= b−1 kf k2L2 [0,∞) .
for all f ∈ Cc (R). Since Cc (R) is dense in L2 (R), it follows that G(g, a, b) is
a tight frame with frame bound Ab−1 .
(b) By applying the same techniques as in Example 11.5 and Exer-
cise 11.10, we can choose a, b, and ψ so that ψb is as smooth as we like,
even infinitely differentiable. We just need to make sure that the dilations of
the interval [c, c + b−1 ] have nontrivial overlaps, i.e., we need to choose a, b so
that ac < c + b−1 .
(c) We take a = 2 and b = 1. Define ψ1 ∈ L2 (R) by


 0, x < 1/2,





 linear, 1/2 ≤ x ≤ 3/4,

ψb1 (ξ)2 = 1, 3/4 ≤ x ≤ 1,





 linear, 1 ≤ x ≤ 3/2,


0, x > 3/2.
P
If we set c = 1/2 then we have c + b−1 = 3/2. Also, n∈Z |ψb1 (ξ)|2 = 1 on
2
(0, ∞), so W(ψ1 ) is a Parseval frame for H+ (R) by part (a).
Now let ψ2 ∈ L (R) be the function such that ψb2 (ξ) = ψ(−ξ).
2 b Then
2
W(ψ1 ) ∪ W(ψ2 ) is a Parseval frame for L (R). Explicitly, we just have to set
ψ2 (x) = ψ(−x), for then
Z Z
ψb2 (ξ) = ψ(−x) e−2πiξx dx = b
ψ(x) e2πiξx dx = ψ(−ξ).
12.4 (a) Let Ek = (E ∩ [k, k + 1]). We claim that, modulo sets of measure
zero, we have S
(Ek − k) = [0, 1],
k∈Z
and (Ej − j) ∩ (Ek − k) has measure zero when j 6= k.

The inclusion ∪(Ek − k) ⊆ [0, 1] follows from the definition of Ek . Con-
versely, if x ∈ [0, 1] then since E tiles by translation we have x ∈ E − k for
some k. Therefore x + k ∈ E and x + k ∈ [k, k + 1], so x + k ∈ Ek . Therefore
x ∈ Ek − k for that k, so [0, 1] ⊆ ∪(Ek − k).
Given f ∈ L2 [0, 1], for x ∈ Ek define U f (x) = f (x − k). Because the sets
Ek − k only have overlaps of measure zero, this defines U f on E except for a
set of measure zero. We have
Z 1
kf k2L2 [0,1] = |f (x)|2 dx
0
XZ
= |f (x)|2 dx
k∈Z Ek −k
XZ
= |U f (x)|2 dx
k∈Z Ek
Z
= |U f (x)|2 dx = kU f k2L2 (E) .
E
Hence U is a unitary mapping of L2 [0, 1] onto L2 (E). Now consider en (x) =

e2πinx . If x ∈ Ek then, since en is 1-periodic, we have
U en (x) = en (x − k) = en (x).
Since {en }n∈Z is an orthonormal basis for L2 [0, 1] and U is unitary, it follows
that {en }n∈Z is an orthonormal basis for L2 (E).
(b) By part (a), {ek }k∈Z is an orthonormal basis for L2 (E). Since the
dilation Da−n is unitary, {Da−n ek }k∈Z is an orthonormal basis for L2 (an E).
Since ∪(an E) = R with overlaps of measure zero, we conclude that

Da−n (ek χE ) k,n∈Z = Da−n Mk χE k,n∈Z
is an orthonormal basis for L2 (R). Letting ψ = (χE )

∧
and applying the
Fourier transform,
∧
Da−n Mk χE = Dan Tk ψ,
so
Dan Tk ψ k,n∈Z
is an orthonormal basis for L2 (R).

(c) Suppose that Dan Tk ψ k,n∈Z is an orthonormal basis for L2 (R), where
ψb = χE . Applying the inverse Fourier transform, it follows that

Da−n Mk χE k,n∈Z
(A)

Now, if F = R \ ∪(an E) has positive measure, then χF is orthogonal to
every element of equation (A), which is a contradiction. Therefore ∪(an E) =
R, up to a set of measure zero. Moreover, if am E∩an E has positive measure for
some m 6= n then the functions Da−m χE and Da−n χE will not be orthogonal,
which is a contradiction. Hence E must tile by dilation by a.
Fixing n = 0, we have that {e2πikx }k∈Z must be an orthonormal basis for
2
L (E). Arguing similarly to part (a), this implies that E tiles by translation.
(d) By inspection, E tiles by translation, and ∪(2n E) = R\{0}. We com-

pute that
∨
ψ(x) = χ[−1,− 12 ] + χ[ 21 ,1] (x)
∨
= χ[−1,1] − χ[− 12 , 12 ] (x)
Z 1 Z 1/2
2πiξx
= e dξ − e2πiξx dξ
−1 −1/2
e2πix − e2πix eπix − eπix

= −
2πix 2πix
sin(2πx) sin(πx)
= − .
πx πx
12.5 Clearly d(B, C) ≥ 0 and d(B, C) = d(C, B).
Suppose d(B, C) = 0 and choose x ∈ B. Given ε > 0 we have d(B, C) < ε,
so B ⊆ Cε . Thus x ∈ Cε , so dist(x, C) < ε. This is true for every ε > 0, so
dist(x, C) = 0. Since C is closed, this implies that x ∈ C. Thus B ⊆ C, and
similarly C ⊆ B.
Suppose B, C, D are compact subsets of Rd . Let a = d(B, C) and b =
d(C, D). Fix ε > 0 Then B ⊆ Ca+ε and C ⊆ Db+ε . Hence if x ∈ B then
dist(x, C) < a+ε, so there exists some y ∈ C such that |x−y| < a+ε. Similarly
there exists some z ∈ D such that |y − z| < b + ε. Therefore |x− z| < a+ b + 2ε,
so dist(x, D) < a + b + 2ε. Thus B ⊆ Da+b+2ε . Similarly D ⊆ Ba+b+2ε , and
since ε > 0 is arbitrary this implies that d(B, D) ≤ a + b = d(B, C) + d(C, D).
Thus d is a metric on H(Rd ).
12.8 Fix ψ ∈ L2 (R). Then
svee
ψε = ψb · χ[−ε,ε]C ∈ L2 (R)
is admissible, and ψε → ψ as ε → 0.
12.9 If ψb is continuous and ψ(0)
b b
6= 0, then |ψ(ξ)| > δ > 0 on some neighbor-
b
hood of the origin, which implies that ψ is not admissible.
12.10 We fill in the details of the claims made in Example 12.12.
Exercise 10.3 shows that V0 = PW(R) is a closed subspace of L2 (R).
Since Vn consists of functions whose Fourier transforms are supported
within [−2n−1 , 2n−1 ], the union of the Vn contains all functions in L2 (R)
with compactly supported Fourier transforms. Since the Fourier transform is
unitary and
L2c (R) = {f ∈ L2 (R) : supp(f ) is compact}
is dense in L2 (R), it follows that ∪n∈Z Vn is a dense subspace of L2 (R) as
well.
If f ∈ ∩n∈Z Vn then supp(fb) is contained in [−2n−1 , 2n−1 ] for every n,

which implies that fb(ξ) = 0 a.e. Hence fb is the zero function in L2 (R), and
therefore f = 0 as well. Thus ∩n∈Z Vn = {0}.
Combined with the other facts given in Example 12.12, this shows that
{Vn }n∈Z is an MRA.
By construction, Vn+1 = Vn ⊕ Wn . The argument of Example 12.11 then
shows that
{Tk ϕ}k∈Z ∪ {D2n Tk ψ}n≥0, k∈Z
Given n, we have
Vn = V−n ⊕ W−n ⊕ · · · ⊕ Wn−1 .
Hence
n−1
X X
Pn f = P−n f + bk,m D2m Tk ψ,
m=−n k∈Z
where
bk,m = hf, D2m Tk ψi.
Since Pn f → f and P−n f → 0 in L2 -norm, we conclude that
XX
f = bk,m D2m Tk ψ.
m∈Z k∈Z
Since {D2m Tk ψ}m,k∈Z is an orthonormal sequence, we conclude that it is an

orthonormal basis for L2 (R).
12.11 (a), (b), (c) follow immediately from the definition of an MRA.
(d) Fix f ∈ L2 (R) and ε > 0. Since ∪Vn is dense, there exist a functions
g ∈ ∪Vn such that kf − gkL2 < ε. By definition, g ∈ VN for some N ∈ Z.
Since the spaces in an MRA are nested, we have g ∈ Vn for all n ≥ N. Since
Pn f is the function in Vn that is closest to f, for all n ≥ N we have
kf − Pn f kL2 ≤ kf − gkL2 < ε.
Hence Pn → f in L2 -norm as n → ∞.
(e) The space of bounded, compactly supported functions is dense in
L2 (R), so suppose first that f is bounded and compactly supported, say
supp(f ) ⊆ [−R, R]. Then
X

kPn f k2L2 = f, D2n Tk ϕ 2
k∈Z
X

= D2−n f, Tk ϕ 2
k∈Z
Z 2
X ∞
= −n/2 −n
x) ϕ(x − k) dx
2 f (2
k∈Z −∞
Z 2n R 2
X
−n
= 2 f (2 −n
x) ϕ(x − k) dx
k∈Z −2n R
X Z 2n R Z 2n R
−n −n 2
≤ 2 |f (2 x)| dx |ϕ(x − k)|2 dx
k∈Z −2n R −2n R
X Z 2n R Z 2n R+k
≤ 2−n kf k2L∞ dx |ϕ(x)|2 dx
k∈Z −2n R −2n R+k
X Z 2n R+k
≤ 2R kf k2L∞ |ϕ(x)|2 dx
k∈Z −2n R+k
Z
= 2R kf k2L∞ |ϕ(x)|2 dx,
En
where S
En = −2n R + k, 2n R + k .
k∈Z
By the Lebesgue Dominated Convergence Theorem, since ϕ ∈ L2 (R) this final

quantity converges to zero as n → −∞.
Now let f be an arbitrary element of L2 (R), and fix ε > 0. Then there
exists a bounded, compactly supported function g such that kf − gkL2 < ε.
Hence
kPn f kL2 ≤ kPn gkL2 + kPn f − Pn gkL2

≤ kPn gkL2 + kf − gkL2
≤ kPn gkL2 + ε
→ 0+ε as n → −∞.
Since ε is arbitrary, we conclude that Pn f → 0 as n → −∞.

12.12 We are given that {Vn } is an MRA with scaling function ϕ. Suppose
that f ∈ L2 (R) is orthogonal to D2n Tk ϕ for every k, n ∈ Z. Since ∪Vn is
dense in L2 (R), there exist functions gn ∈ Vn such that gn → f as n → ∞. By
Lemma 12.10, {D2n Tk ϕ}k∈Z is an orthonormal basis for Vn , so f is orthogonal
to every element of Vn . In particular, hf, gn i = 0. But hf, gn i → hf, f i =
kf k2L2 , so we conclude that f = 0 a.e.
(b) This follows combining part (a) with the Haar MRA constructed in
Example 12.11.
12.13 If ϕ ∈ L2 (R) is refinable then we have
X
ϕ(x) = ck ϕ(2x − k),
k∈Z
and this series converges absolutely in L2 (R) by hypothesis. Hence

X
Tm ϕ(x) = ϕ(x − m) = ck ϕ(2(x − m) − k)
k∈Z
X
= ck ϕ(2x − k − m − m)
k∈Z
X
= ck (Tm ϕ)(2x − k − m)
k∈Z
X
= ck+m (Tm ϕ)(2x − k).
k∈Z
Therefore Tm ϕ satisfies a refinement equation with scalars (ck+m )k∈Z .

12.14 The refinement equation is
1 1
B1 (x) = B1 (x) + B1 (x − 1) + B1 (x − 2).
2 2
An indirect proof is given in Exercise 12.20.
12.15 We know that the Shannon scaling function ϕ has orthonormal integer
translates. Also ϕ ∈ V0 ⊆ V1 , so ϕ must satisfy a refinement equation. Since
{D2 Tk ϕ}k∈Z is an orthonormal basis for V1 , we have
X
ϕ = hϕ, D2 Tk ϕi D2 Tk ϕ.
k∈Z
b = χ[− 21 , 21 ] , we apply the unitary of the Fourier transform to

Recalling that ϕ
compute these inner products:

∧
21/2 ϕ, D2 Tk ϕ = 21/2 ϕ, b (D2 Tk ϕ)

= 21/2 ϕ, b D1/2 M−k ϕb
Z
= b e−2πik(ξ/2) ϕ(ξ/2)
ϕ(ξ) b
Z 1/2
= eπikξ dξ
−1/2
1/2
eπikξ
=
πik −1/2
2 sin(πk/2)
=
πk


0,
 k even,
= 2/(πk), k = . . . , −7, −3, 1, 5, . . . ,


−2/(πk), k = . . . , −5, −1, 3, 7, . . . .
Substituting these values, we see that

X 2 sin(πk/2) 1/2 X 2 sin(πk/2)
ϕ(x) = 2−1/2 2 ϕ(2x − k) = ϕ(2x − k),
πk πk
k∈Z k∈Z
where the series converges in L2 -norm.

We can compute m0 from these coefficients, or observe that
χ[− 1 , 1 ] (ξ) = m0 (ξ/2) ϕ(ξ/2)

b = m0 (ξ/2) χ[−1, 1](ξ).
2 2
We conclude that m0 (ξ/2) = χ[− 12 , 21 ] , at least on the interval [−1, 1]. Hence
m0 (ξ) = χ[− 41 , 14 ] , extended 1-periodically to a function on T.
12.16 It is the Cantor–Lebesgue function, reflected about the point x = 1 to
create a continuous function supported on [0, 2].
12.17 By hypothesis, there some sequence of partial sums such that
X
ϕ = lim 2−1/2 ck D2 Tk ϕ,
n→∞
k∈Fn
where this is a limit in L2 -norm. As the Fourier transform is unitary, we

therefore have
X
2−1/2 ck (D2 Tk ϕ) (ξ)
∧
b
ϕ(ξ) = lim
n→∞
k∈Fn
X
= lim 2−1/2 ck D1/2 M−k ϕ(ξ)
b
n→∞
k∈Fn
X
= lim 2−1/2 ck 2−1/2 e−2πik(ξ/2) ϕ(ξ/2).
b
n→∞
k∈Fn
1 X
= lim ck e−2πik(ξ/2) ϕ(ξ/2).
b
2 n→∞
k∈Fn
This is a limit in L2 (R), but by replacing the sets Fn with a subsequence

(that we also call Fn ), we may assume that the limit converges pointwise a.e.
Now, since {e−2πikξ }k∈Z is an orthonormal basis for L2 (T), the series
defining m0 converges unconditionally. Therefore any subsequence of partial
sums will converge in norm. Hence we have
1 X
m0 (ξ) = lim ck e−2πikξ ,
2 n→∞
k∈Fn
and by passing to a subsequence yet again we may assume that this conver-
gence is pointwise a.e. Hence for almost every ξ, since Fn is a finite set we can
compute that
X
1
b
ϕ(ξ) = lim ck e−2πik(ξ/2) ϕ(ξ/2)
b
2 n→∞
k∈Fn
X
1
= lim ck e−2πik(ξ/2) ϕ(ξ/2)
b
2 n→∞
k∈Fn
b
= m0 (ξ/2) ϕ(ξ/2).
12.18 (b) If (ck )k∈Z ∈ ℓ1 (Z), then P

the series defining m0 converges absolutely
in Lp (T) for every p, and the series ck D2 Tk ϕ converges absolutely in L2 (R)
since kD2 Tk ϕkL2 = kϕkL2 . We compute that
b
ϕ(ξ) b
= m0 (ξ/2) ϕ(ξ/2)
X
1
= ck e−2πik(ξ/2) ϕ(ξ/2)
b
2
k∈Z
X
= 2−1/2 ck 2−1/2 e−2πik(ξ/2) ϕ(ξ/2)
b
k∈Z
X
= 2−1/2 ck D1/2 M−k ϕ(ξ)
b
k∈Z
X
2−1/2 ck (D2 Tk ϕ) (ξ).
∧
=
k∈Z
Because all of the series converge absolutely, the justification of the factoring
that relates the second to third lines follows from an argument that is similar
to (but even more simple than) the ones used in Exercise 10.15 or 12.17. Since
the Fourier transform is a unitary mapping, it follows that
X
ϕ = 2−1/2 ck D2 Tk ϕ.
k∈Z
(a) Since (ck )k∈Z ∈ ℓ2 (Z), the series defining m0 converges uncondition-
P
ally in L2 (T). If {Tk ϕ}k∈Z is a Bessel sequence, then the series ck D 2 T k ϕ
converges unconditionally in L2 (R). Therefore we can compute just as in the
argument above, appealing this time to Exercise 10.15 for justification of the
factoring.
12.19 (a) We compute that
XX X X
|cj dk−j | = |cj | |dk−j |
k∈Z j∈Z j∈Z k∈Z
X X
= |cj | |dk |
j∈Z k∈Z
= kckℓ1 kdkℓ1 < ∞.
It follows that c ∗ d ∈ ℓ1 (Z).

(b) Exercise 9.5(b) shows that ϕ ∗ ψ ∈ L1 (R).
(c) We are given that ϕ and ψ are integrable functions that satisfy the
refinement equations
X X
ϕ(x) = ck ϕ(2x − k) and ψ(x) = dk ψ(2x − k),
k∈Z k∈Z
where the refinement coefficients are summable.

Assuming that the interchange of summations and integrals is justified, by
making the change of variables z = 2y − j and then reindexing the series we
compute that
Z ∞
(ϕ ∗ ψ)(x) = ϕ(y) ψ(x − y) dy
−∞
Z ∞ X X
= cj ϕ(2y − j) dk ψ(2x − 2y − k) dy
−∞ j∈Z k∈Z
XX Z ∞
= cj dk ϕ(z) ψ(2x − z − j − k) dy
j∈Z k∈Z −∞
Z ∞
1X X
= dk cj−k ϕ(z) ψ(2x − z − j) dz
2 −∞
k∈Z j∈Z
1X X
= dk cj−k (ϕ ∗ ψ)(2x − j)
2
k∈Z j∈Z

1X X
= dk cj−k (ϕ ∗ ψ)(2x − j)
2
j∈Z k∈Z
1X
= (c ∗ d)j (ϕ ∗ ψ)(2x − j).
2
j∈Z

Hence ϕ ∗ ψ is refinable with refinement coefficients 12 (c ∗ d)j j∈Z .
The use of Fubini’s Theorem is justified by performing a similar caculation
to show that
Z ∞ Z ∞ XX
|cj | |dk | |ϕ(2y − j)| |ψ(2x − 2y − k)| dy
−∞ −∞ j∈Z k∈Z
1
≤ kc ∗ dkℓ1 kϕ ∗ ψkL1 < ∞.
2
The symbol for this refinement equation is
∞
1 X 1
mc∗d (ξ) = (c ∗ d)k e−2πikξ
2 2
k=−∞
∞ ∞
1 X X
= cj dk−j e−2πi(j+k−j)ξ
4
k=−∞ j=−∞
∞ ∞
1 X X
= cj e−2πijξ dk−j e−2πi(k−j)ξ
4 j=−∞
k=−∞
∞
X ∞
X
1
= cj e−2πijξ dk e−2πikξ
4 j=−∞ k=−∞
∞
X ∞
1 −2πijξ 1 X −2πikξ
= cj e dk e
2 j=−∞
2
k=−∞
= mc (ξ) md (ξ).
12.20 Let χ = χ[0,1] .

(a) If x ∈ [0, 1], then we have
Z
χ χ
( ∗ )(x) = χ(y) χ(x − y) dy
Z
= χ[0,1] (y) χ[0,1] (x − y) dy
Z 1 Z x
= χ[0,1] (x − y) dy = dy = x.
0 0
Likewise if x ∈ [1, 2] then (χ ∗ χ)(x) = 2 − x. Also (χ ∗ χ)(x) = 0 for all other x.

Hence B1 is the hat function on the interval [0, 1].
(b) Since χ is refinable, it follows from Exercise 12.19 that Bn is refinable,
integrable, and compactly supported. Further, supp(B0 ) = [0, 1] and B1 =
B0 ∗B0 so supp(B1 ) ⊆ [0, 1]+[0, 1] = [0, 2]. By induction, supp(Bn ) ⊆ [0, n+1].
The refinement coefficients for B0 are c0 = c1 = 1 and all other ck = 0.
Setting c = (. . . , 0, 0, 1, 1, 0, 0, . . . ), we have by Exercise 12.19 that the nonzero
refinement coefficients for Bn have the pattern of a normalized Pascal’s tri-
angle where each row must sum to 2:
B0 1 1
B1 1/2 1 1/2
B2 1/4 3/4 3/4 1/4
B3 1/8 1/2 3/4 1/2 1/8

and so forth. Thus Bn satisfies a refinement equation of the form
n+1
X
Bn (x) = ck Bn (2x − k).
k=0
Exercise 12.21 therefore again shows that supp(Bn ) ⊆ [0, n + 1].

Let Mn denote the symbol for Bn . The symbol for B0 is
1
M0 (ξ) = (1 + e−2πiξ ).
2
Exercise 12.28 therefore implies that the symbol for Bn is
n+1 (1 + e−2πiξ )n+1
Mn (ξ) = M0 (ξ) = = M0 (ξ) Mn−1 (ξ).
2n+1
(j)
We have M0 (1/2) = 0. Suppose that Mn−1 (1/2) = 0 for j = 0, . . . , n − 1.
Applying the product rule,
j
X j (ℓ) (j−ℓ)
Mn(j) (ξ) = M0 (ξ) Mn−1 (ξ).
ℓ
ℓ=0
j
X
(j) j (ℓ) (j−ℓ)
= M0 (ξ) Mn−1 (ξ) + M0 (ξ) Mn−1 (ξ)
ℓ
ℓ=1
j
X
(j) j 1 + e−2πiξ (j−ℓ)
= M0 (ξ) Mn−1 (ξ) + (−2πiξ)ℓ Mn−1 (ξ).
ℓ 2
ℓ=1
Fix 0 ≤ j ≤ n. If 1 ≤ ℓ ≤ j then
0 ≤ j − ℓ ≤ n − 1,
(j−ℓ)
so Mn−1 (1/2) = 0. Therefore
j
X
(j) j (j−ℓ)
Mn(j) (1/2) = M0 (1/2) Mn−1(1/2) + (−πi)ℓ · 0 · Mn−1 (1/2)
ℓ
ℓ=1
= 0.
(j)
That is, Mn (1/2) = 0 for j = 0, . . . , n.
(c) By definition, B0 = χ. Exercise 9.5 shows that L1 (R) is closed under
convolution, so we have by induction that Bn = Bn−1 ∗ χ ∈ L1 (R) for every n,
and also Bn is compactly supported. Let
sin πξ
dπ (ξ) = ,
πξ
be the sinc function, which by equation (9.12) is the Fourier transform of

χ[−1/2,1/2] . Since χ = T1/2 χ[−1/2,1/2] , we have
sin πξ
b (ξ) = M−1/2 dπ (ξ) = e−πiξ
χ .
πξ
Therefore
n+1 n+1
cn (ξ) = sin πξ
B b
χ (ξ) = e−πi(n+1)ξ ,
πξ
and this is integrable for n > 0.

(d) The Fourier transform interchanges convolution with multiplication
(see Exercise 9.5). Therefore, we have that
∧
(Bn − T1 Bn ) (ξ) = B cn (ξ)
cn (ξ) − M1 B
cn (ξ) (1 − e−2πiξ )
= B
cn (ξ) e−πiξ (eπiξ − e−πiξ )
= B
cn (ξ) e−πiξ 2i sin(πξ)
= B
cn (ξ) e−πiξ 2πiξ sin(πξ)

= B
πξ
cn (ξ) χ
= 2πiξ B b (ξ)
∧
= 2πiξ (Bn ∗ χ) (ξ)
[
= 2πiξ Bn+1 (ξ)
[
= −B ′
n+1 (ξ).
∧
[
Therefore (T1 Bn − Bn ) = B ′
n+1 , so by the Uniqueness Theorem we have
′
T1 Bn − Bn = Bn+1 .
(e) We know that B1 = χ ∗ χ ∈ Cc (R). Also, the fact that Bcn has the form
c
given in part (a) implies that Bn , ξ n−1 c 1
Bn (ξ) ∈ L (R), so Theorem 9.15 and
the Inversion Formula imply that
cn )
Bn = (B
∨
∈ C0n−1 (R).
But also Bn is compactly supported, so Bn ∈ Ccn−1 (R).

Note that B1 = χ ∗ χ is piecewise linear. Letting D denote the differenti-
ation operator, we have
Bn(n−1) = Dn−2 Bn′

= Dn−2 (T1 Bn−1 − Bn−1 )
= T1 Dn−2 Bn−1 − Dn−2 Bn−1 .
(n−1)
It therefore follows by induction that Bn is piecewise linear.
12.21 (a) Let [a, b] be the smallest interval such that ϕ = 0 a.e. outside of
[a, b]. If
ha b + N i
x 6∈ , ,
2 2
then we have 2x < a and 2x − N > b. Evaluating the right-hand side of the
refinement equation, we see that ϕ(x) = 0. Hence we must have
ha b + N i
[a, b] ⊆ , ,
2 2
Therefore
a b+N
≤ a and b≤ .
2 2
Rearranging, we see that a ≥ 0 and b ≤ N. Hence supp(ϕ) ⊆ [0, N ].
(b) We are given that ϕ is continuous. Since supp(ϕ) ⊆ [0, N ], this implies
that ϕ(j) = 0 for all integer j ≤ 0 and j ≥ N. If ϕ(1) = · · · = ϕ(N − 1) = 0
then by applying the refinement equation we obtain ϕ(j/2ℓ ) = 0 for all integer
j and ℓ. Combined with continuity, this implies that ϕ is identically zero.
Therefore, we can assume that ϕ(1), . . . , ϕ(N − 1) are not all zero.
Setting ck = 0 for k < 0 and k > N, we have
P
[c2i−j ]i,j=1,...,N [ϕ(j)]j=1,...,N = j∈Z c2i−j ϕ(j) i=1,...,N
P
= j∈Z cj ϕ(2i − j) i=1,...,N
= [ϕ(j)]j=1,...,N .
T
Thus ϕ(1), . . . , ϕ(N − 1) is a 1-eigenvector of the matrix M.
(c) If ϕ is differentiable, then we have
N
X N
X
d
ϕ′ (x) = ck ϕ(2x − k) = 2ck ϕ′ (2x − k),
dx
k=0 k=0
so ϕ′ is refinable.
′ ′
(d) If ϕ is differentiable and ϕ is continuous, then applying part (b) to ϕ
we see that ϕ′ (1), . . . , ϕ′ (N −1) is an eigenvalue for 2M for the eigenvalue 1.
Hence 1/2 is an eigenvalue for M.
(e) Since M is an (N −1)×(N −1) matrix, it can have at most N −1 distinct
eigenvalues. In order for ϕ to have N − 2 continuous derivatives, each of 1,
1/2, . . . , 2−(N −1) must be eigenvalues of M. This is the maximum possible
number of continuous derivatives that ϕ can have.
12.22 Suppose ϕ ∈ L2 (R) is refinable and compactly supported and
{Tk ϕ}k∈Z is orthonormal. The refinement equation is
X
ϕ(x) = 2−1/2 ck 21/2 ϕ(2x − k).
k∈Z
Since {21/2 ϕ(2x − k)}k∈Z is orthonormal, we must have

2−1/2 ck = ϕ(x), 21/2 ϕ(2x − k) .
However, since ϕ is compactly supported, only finitely many of these inner

products can be nonzero.
12.23 The key facts underlying this proof are that ℓ1 (Z) is closed under con-
volution and the Fourier transform converts convolution in the time domain
into multiplication in the Fourier domain (compare Section 13.3).
Set M (ξ) = |m0 (ξ)|2 + |m0 (ξ + 21 )|2 . Since m0 ∈ L2 (T), we have M ∈
1
L (T). P
Define an = k∈Z ck ck−2n . This series converges because c = (ck ) ∈
ℓ2 (Z). Assume c ∈ ℓ1 (Z), the sequence a = (an ) belongs to ℓ1 (Z) because
X XX
|an | ≤ |ck ck−2n |
n∈Z n∈Z k∈Z
XX
≤ |ck ck−n |
n∈Z k∈Z
XX
= |ck ck−n |
k∈Z n∈Z
XX
= |ck cn |
k∈Z n∈Z
X X
= |cn | |ck | < ∞.
n∈Z k∈Z
Therefore, we can compute that

M (ξ) = |m0 (ξ)|2 + |m0 (ξ + 12 )|2

1XX 1XX
= cn e2πinξ ck e−2πikξ + cn e2πin(ξ+1/2) ck e−2πik(ξ+1/2)
4 4
k∈Z n∈Z k∈Z n∈Z
1XX 1XX
= cn ck e−2πi(n−k)ξ + cn ck e−2πi(n−k)(ξ+1/2)
4 4
1XX 1XX
= cn+k ck e−2πinξ + cn+k ck e−2πin(ξ+1/2)
4 4
1XX 1XX
= cn+k ck e−2πinξ + cn+k ck e−πin e−2πinξ
4 4
1XX 1XX
= cn+k ck e−2πinξ + cn+k ck (−1)n e−2πinξ
4 4
1XX
= (cn+k + (−1)n cn+k ) ck e−2πinξ
4
k∈Z n∈Z
1XX
= c2n+k ck e−2πinξ
2
k∈Z n∈Z

1X X
= c2n+k ck e−2πinξ
2
n∈Z k∈Z
1X
= a−n e−2πinξ
2
n∈Z
1X
= an e2πinξ .
2
n∈Z
c(n) = an
Since a ∈ ℓ1 (Z), it follows that the Fourier coefficients of M are M
for n ∈ Z. Consequently, M = 1 a.e. if and only if a/2 is the delta sequence,
i.e., an = 2δ0n for n ∈ Z.
12.24 Without loss of generality, we can assume that 0 < δ ≤ 1. Since
|eiθ − 1| ≤ min{|θ|, 2}, we compute that

1 X 1 X
|m0 (ξ) − 1| = ck e 2πikξ
− ck
2 2
k∈Z k∈Z
1 X 2πikξ
≤ |e − 1|δ |e2πikξ − 1|1−δ |ck |
2
k∈Z
X
≤ |πξ|δ |k|δ 21−δ |ck | = C|ξ|δ .
k∈Z
The remainder of the proof is similar to the proof of Theorem 12.25. We have
n
Y
|Pn (ξ)| = m0 (2−j ξ) − 1 + 1
j=1
n
Y
≤ C 2−jδ |ξ|δ + 1
j=1
n
Y −jδ
|ξ|δ
≤ eC 2
j=1
Pn
2−jδ |ξ|δ
= eC j=1
′
|ξ|δ
≤ eC .
With R > 0 fixed,
sup |Pn (ξ) − Pn−1 (ξ)| = sup |m0 (2−n ξ) − 1| |Pn−1 (ξ)|
ξ∈[−R,R] ξ∈[−R,R]
′
Rδ
≤ C 2−nδ Rδ eC .
P∞
Since n=1 2−nδ < ∞, this implies that {Pn }n∈Z is Cauchy in the uniform
norm on [−R, R].
12.25 (a) Since c = (. . . , 0, 0, c0 , . . . , cN , 0, 0, . . . ) is a finite sequence, equa-
tions (12.42) and (12.43) are equivalent. Suppose that N = 2m is even.
If ck and ck+2m are both nonzero then we must have 0 ≤ k ≤ 2m and
0 ≤ k + 2m ≤ 2m. This implies k = 0. Hence
X
0 = δ0,−2m = ck ck+2m = c0 c2m = c0 cN 6= 0,
k∈Z
which is a contradiction. Therefore N must be odd.

12.26 We set χ = χ[0,1) .
(a) Since χ = T−1/2 χ[− 21 , 21 ) , we have
sin πξ
b (ξ) = M1/2 (χ[− 1 , 1 ] )∧ (ξ) = eπiξ
χ .
2 2 πξ
(b) The box function satisfies the refinement equation χ(x) = χ(2x) +
χ(2x − 1). This is the refinement equation with c0 = c1 = 1 and all other
P
ck = 0. Therefore, we have k∈Z ck ck−2n = 2δ0n , which is equivalent to
|m0 (ξ)|2 + |m0 (ξ + 12 )|2 = 1.
(c) Fix ξ ∈ R. Using the double-angle formula sin 2x = 2 sin x cos x, we
have
n
Y
cos(2−j πξ) = cos(πξ) · · · cos(2−n πξ)
j=1
sin(πξ) sin(πξ/2) sin(2−n−1 πξ)

= ···
2 sin(πξ/2) 2 sin(πξ/4) 2 sin(2−n πξ)
sin(πξ)
=
2n sin(2−n πξ)
sin(πξ)
→ as n → ∞,
πξ
sin x
where at the last step we have used the fact that x → 1 as x → 0. This
yields Viète’s formula.
(d) Note that
1 eπiξ −πiξ
m0 (ξ) = 1 + e2πiξ ) = e + eπiξ
2 2
e πiξ
= 2 cos(πξ = eπiξ cos(πξ).
2
By parts (a), (b) and Theorem 12.29, we have
Y∞
sin πξ
eπiξ b (ξ) =
= χ m0 (2−j ξ)
πξ j=1
∞
Y −j
= eπi2 ξ
cos(2−j πξ)
j=1
P∞ ∞
Y
πi2−j ξ
= e j=1 cos(2−j πξ)
j=1
∞
Y
= eπiξ cos(2−j πξ),
j=1
and Viète’s formula follows from this.

12.27 The proof carries over with only a few changes of “=” to “≤” at
appropriate places.
12.28 (a) Except for a scaling factor, this proof of this exercise for j = 1 is
the same as Exercise 13.4, and it extends by induction to higher j. Since
(j) (−2πi)j X j
m0 (ξ) = k ck e−2πikx ,
2
k∈Z
we have
(j) (−2πi)j X j (−2πi)j X
m0 (1/2) = k ck e−πik = (−1)k k j ck .
2 2
k∈Z k∈Z
(b) See the proof of Exercise 12.20.

P P
12.29 We are given that (ck )k∈Z is summable and c2k = 1 = c2k+1 .
Since ϕ ∈ L1 (R), we have by Exercise 10.13 that its periodization
X
p(x) = ϕ(x + k)
k∈Z
1
belongs to L (T). In particular, p is 1-periodic. Also, by refinability we have
X
p(x) = ϕ(x + j)
j∈Z
XX
= ck ϕ(2x + 2j − k)
j∈Z k∈Z
X X
= ck ϕ(2x + 2j − k)
k∈Z j∈Z
X X X X
= c2k ϕ(2x + 2j − 2k) + c2k+1 ϕ(2x + 2j − 2k − 1)
k∈Z j∈Z k∈Z j∈Z
X X X X
= c2k ϕ(2x + 2j) + c2k+1 ϕ(2x + 2j − 1)
k∈Z j∈Z k∈Z j∈Z
X X
= ϕ(2x + 2j) + ϕ(2x + 2j − 1)
j∈Z j∈Z
X
= ϕ(2x + j)
j∈Z
= p(2x).
Hence p satisfies both p(x + 1) = p(x) and p(2x) = p(x) for a.e. x ∈ T. The
mapping τ x = 2x mod 1 from [0, 1) onto itself is ergodic. The Birkhoff Ergodic
Theorem [Wal82, Thm. 1.14] therefore implies that there is a constant C such
that
n−1
1X
p(x) = p(τ k x) → C a.e. as n → ∞.
n
k=0
Hence p(x) = C a.e.
To evaluate this constant, we integrate the periodization of ϕ over a period:
Z 1 Z 1X Z ∞
C = C dx = ϕ(x + k) dx = b
ϕ(x) dx = ϕ(0).
0 0 k∈Z −∞
Alternatively, the Birkhoff Ergodic Theorem tells us directly that C =

R1
0 p(x) dx.
12.30 Set X
q(x) = (j + a) ϕ(x + j).
j∈Z
Since ϕ is compactly supported, for any given x only finitely many terms in
this series can be nonzero. Similarly, all sums in this solution will have finitely
many nonzero terms, so there are no convergence issues to consider.
We have
X
q(x + 1) = (j − a) ϕ(x + 1 + j)
j∈Z
X
= (j − a − 1) ϕ(x + j)
j∈Z
X X
= (j − a) ϕ(x + j) − ϕ(x + j)
j∈Z j∈Z
= q(x) − 1.
Therefore, the function h(x) = q(x) − x satisfies
h(x + 1) = q(x + 1) − (x + 1) = q(x) − 1 − x + 1 = h(x),
and so is 1-periodic. Also,

Z 1 X Z 1 Z 1
|h(x)| dx ≤ (j + a) |ϕ(x + j)| dx + |x| dx < ∞,
0 j∈Z 0 0
because only finitely many terms in the sum are nonzero.

By refinability we have
X
q(x) = (j − a) ϕ(x + j)
j∈Z
XX
= (a + j) ck ϕ(2x + 2j − k)
j∈Z k∈Z
X X
= ck (j − a) ϕ(2x + 2j − k)
k∈Z j∈Z
X X
= c2k (j − a) ϕ(2x + 2j − 2k)
k∈Z j∈Z
X X
+ c2k+1 (j − a) ϕ(2x + 2j − 2k − 1)
k∈Z j∈Z
X X
= c2k (j − a + k) ϕ(2x + 2j)
k∈Z j∈Z
X X
+ c2k+1 (j − a + k) ϕ(2x + 2j − 1)
k∈Z j∈Z
X X
= c2k j ϕ(2x + 2j)
k∈Z j∈Z
X X
− a c2k ϕ(2x + 2j)
k∈Z j∈Z
X X
+ k c2k ϕ(2x + 2j)
k∈Z j∈Z
X X
+ c2k+1 j ϕ(2x + 2j − 1)
k∈Z j∈Z
X X
− a c2k+1 ϕ(2x + 2j − 1)
k∈Z j∈Z
X X
+ k c2k+1 ϕ(2x + 2j − 1)
k∈Z j∈Z
1X X
= 2j ϕ(2x + 2j) − a ϕ(2x + 2j)
2
j∈Z j∈Z
X
1 X
+ 2k c2k ϕ(2x + 2j)
2
k∈Z j∈Z
1X
+ (2j − 1 + 1) ϕ(2x + 2j − 1)
2
j∈Z
X
− a ϕ(2x + 2j − 1)
j∈Z
X
1 X
+ (2k + 1 − 1) c2k+1 ϕ(2x + 2j − 1)
2
k∈Z j∈Z
1X X
= 2j ϕ(2x + 2j) − a ϕ(2x + 2j)
2
j∈Z j∈Z
X
1 X
+ 2k c2k ϕ(2x + 2j)
2
k∈Z j∈Z
1X
+ (2j − 1) ϕ(2x + 2j − 1)
2
j∈Z
1X
+ ϕ(2x + 2j − 1)
2
j∈Z
X
− a ϕ(2x + 2j − 1)
j∈Z
X
1 X
+ (2k + 1) c2k+1 ϕ(2x + 2j − 1)
2
k∈Z j∈Z
X
1 X
− c2k+1 ϕ(2x + 2j − 1)
2
k∈Z j∈Z
1X X
= j ϕ(2x + j) − a ϕ(2x + j)
2
j∈Z j∈Z
aX aX
+ ϕ(2x + 2j) + ϕ(2x + 2j − 1)
2 2
j∈Z j∈Z
1X aX
= j ϕ(2x + j) − ϕ(2x + j)
2 2
j∈Z j∈Z
q(2x)
= .
2
Therefore
q(2x) 2x h(2x)
h(x) = q(x) − x = − = .
2 2 2
Hence h satisfies both h(x + 1) = h(x) and h(2x) = 2h(x) for a.e. x ∈ T.
The mapping τ x = 2x mod 1 from [0, 1) onto itself is ergodic. The Birkhoff
Ergodic Theorem [Wal82, Thm. 1.14] therefore implies that there is a constant
C such that
n−1 Z 1
1X
h(τ k x) → C = h(x) dx a.e. as n → ∞.
n 0
k=0
But
n−1 n−1 n−1
1X 1X 1X k 2n − 1
h(τ k x) = h(2k x) = 2 h(x) = h(x).
n n n n
k=0 k=0 k=0
Since this converges to the finite number C we must have h(x) = 0 a.e.
12.31 (a) The fact that the Hilbert transform is unitary follows from the
Plancherel Equality and the fact that | − i sign(ξ)| = 1 for every ξ.
If f ∈ L1 (R)∩L2 (R), then fb is a continuous function. However, if fb(0) 6= 0,
∧
then (Hf ) is not continuous, so we cannot have Hf ∈ L1 (R).
(b) By Lemma 12.15, we have
b
ϕ(ξ) b
= m0 (ξ/2) ϕ(ξ/2) a.e.
Let θ(ξ) = −i sign(ξ), and note that θ(2ξ) = θ(ξ) for every ξ. Then for almost
every ξ we have
∧ ∧
b
(Hϕ) (ξ) = θ(ξ) ϕ(ξ) b
= θ(ξ/2) m0 (ξ/2) ϕ(ξ/2) = m0 (ξ/2) (Hϕ) (ξ/2).
Since (ck )k∈Z ∈ ℓ1 (Z), Exercise 12.18 implies that Hϕ satisfies the refinement
equation.
12.32 Since the hypotheses of Theorem 12.25 are satisfied, we know that the
infinite product defining P converges uniformly on compact sets. Also,
P (ξ) = m0 (ξ/2) P (ξ/2).
By hypothesis, m0 is continuous, so km0 k∞ . Fix R > max{2, km0k∞ }. Since

P is continuous, we have
C = kP · χ[−1,1] k∞ < ∞.
Suppose 2n−1 ≤ |ξ| < 2n , where n > 1. Then 2−n ξ ∈ [−1, 1], so
n
Y
|P (ξ)| = |m0 (2−j ξ)| |P (2−n ξ)| ≤ C km0 kn∞ ≤ C Rn .
j=1
Set α = log2 R, and let M > α be an integer. Then since α > 0 and |ξ| ≥ 1,
we have
n−1
Rn = R 2log2 R = R 2(n−1)α ≤ R |ξ|α ≤ R |ξ|M .
Hence for all |ξ| ≥ 1 we have
|P (ξ)| ≤ CR |ξ|M .
12.33 Suppose that ϕ was an integrable, nonzero solution to the refinement

equation. Then ϕb is continuous and bounded.
P
We are given that ∆ = |m0 (0)| = 21 ck < 1. Fix ∆ < r < 1. Since m0
is continuous and m0 (0) = ∆, there exists some δ > 0 such that
|ξ| < δ =⇒ |m0 (ξ)| < r.
On the Fourier side, the refinement equation takes the form
b
ϕ(ξ) b
= m0 (ξ/2) ϕ(ξ/2).
Iterating, for |ξ| < δ we have for any n that

n
Y
b
|ϕ(ξ)| = |m0 (2−j ξ)| |ϕ(2
b −n ξ)| ≤ rn kϕk
b ∞ → 0 as n → ∞.
j=1
b = 0 on (−δ, δ). But ϕ(2ξ)

Hence ϕ b b
= m0 (ξ) ϕ(ξ), b=0
so this implies that ϕ
b = 0, so ϕ = 0 a.e. by the Uniqueness
on (−2δ, 2δ). Iterating, we obtain ϕ
Theorem.
12.34 (a) 2/(2x) = 1/x.
(b) By definition, we have for f ∈ Cc∞ (R) that
hf, 21/2 D2 δi = 21/2 hD1/2 f, δi = 21/2 (D1/2 f )(0)

= 21/2 2−1/2 f (0/2) = f (0) = hf, δi.
Hence 21/2 D2 δ = δ. In distributional shorthand notation, this is often written

as 2δ(2x) = δ(x), although it should be kept in mind that δ is only a functional
and is not a function.
(c) Note that
d −1/2
(D1/2 f )′ (x) = 2 f (x/2) = 2−3/2 f ′ (x/2),
dx
and by induction,
(D1/2 f )(j) (x) = 2−(2j+1)/2 f ′ (x/2j ),
If f ∈ Cc∞ (R), then we have
hf, 2j+1 2−1/2 D2 δ (j) i = 2(2j+1)/2 D1/2 f δ (j)

= 2(2j+1)/2 (−1)j (D1/2 f )(j) (0)
= 2(2j+1)/2 (−1)j 2−(2j+1)/2 f (j) (0)
= (−1)j f (j) (0)
= hf, δ (j) i.
Hence δ (j) = 2j+1 2−1/2 D2 δ (j) . In distributional shorthand notation, this says
that δ (j) (x) = 2j+1 δ (j) (2j x).
12.35 (a) Theorem 12.25 implies that P is a continuous function, hence is
locally integrable. By Exercise 12.32, P has polynomial growth at infinity, and
∨
therefore defines a tempered distribution. Its inverse Fourier transform µ = P
is therefore a tempered distribution as well. Given f ∈ S(R) we have
X X

f, 2−1/2 ck D2 Tk µ = 2−1/2 ck T−k D1/2 f, µ
k∈Z k∈Z
X

2−1/2 ck (T−k D1/2 f ) , µ
∧
= b
k∈Z
X

= 2−1/2 ck Mk D2 fb, µ
b
k∈Z
X Z ∞
= 2 −1/2
ck e2πikξ 21/2 fb(2ξ) P (ξ) dξ
k∈Z −∞
X Z ∞
dξ
= ck e2πikξ/2 fb(ξ) P (ξ/2)
−∞ 2
k∈Z
Z ∞
1X dξ
= fb(ξ) ck e2πikξ/2 P (ξ/2)
−∞ 2 2
k∈Z
Z ∞
dξ
= fb(ξ) m0 (ξ)/2) P (ξ/2)
−∞ 2
Z ∞
dξ
= fb(ξ) P (ξ)
−∞ 2

= fb, µ
b = hf, µi.
P
Therefore k∈Z 2−1/2 ck D2 Tk µ = µ.
(b) Rearranging the computations in part (a), if ν ∈ S ′ (R) satisfies the
refinement equation and νb is continuous, then we have νb(ξ) = νb(ξ/2) m0 (ξ/2).
Since the hypotheses of Theorem 12.25 are satisfied, iterating this equation
leads to
νb(ξ) = νb(0) P (ξ) = νb(0) µ
b(ξ).
Hence ν is a scalar multiple of µ.
(c) Since P is locally integrable with polynomial growth and θ is bounded,
the product θP is locally integrable with polynomial growth, and therefore
is a tempered distribution. If we set νb = θP, then νb(ξ) = νb(ξ/2) m0 (ξ/2).
Then a computation similiar to the one from part (a) shows that ν satisfies
the refinement equation in a distributional sense.
(d) Given any ξ, we have θ(ξ) = θ(2−n ξ) for every n, so
θ(ξ) = lim θ(2−n ξ) = θ(0).

n→∞
Hence θ is a constant function.

The function θ(ξ) = −i sign(ξ) is an example of a discontinuous “dila-
tionally periodic” function. More generally, if we fix the values of θ on [1, 2)
and define θ(2n ξ) = θ(ξ) for all n ∈ Z, we obtain a function that satisfies
θ(2ξ) = θ(ξ) for ξ > 0. Combining this with a similar construction for ξ < 0
gives us a dilationally periodic function on the entire line.
12.36 (a) Let ek (ξ) = e2πikξ . Then for f ∈ S(R) we have
Z ∞

∨
fb, δbk = hf, δk i = f (k) = fb (k) = fb(ξ) e2πikξ dξ = hfb, e−k i.
−∞
Hence δbk = e−k , and therefore

N N
1X b 1X
νb(ξ) = ck δk (ξ) = ck e−2πikξ = m0 (ξ).
2 2
k=0 k=0
Note that supp(ν) ⊆ {0, . . . , N } ⊆ [0, N ].

(b) As above,
N N
1X [ 1X −j
νbj (ξ) = ck δ2−j k (ξ) = ck e−2πi2 kξ = m0 (2−j ξ).
2 2
k=0 k=0
Since the distributional Fourier transform converts convolution to multiplica-

tion, we therefore have
n
Y
m0 (2−j ξ) = Pn (ξ).
∧
cn (ξ) = (ν1 ∗ · · · ∗ νn ) (ξ) = νb1 (ξ) · · · νc
µ n (ξ) =
j=1
Since supp(νj ) ⊆ [0, 2−j N ], we have supp(µn ) ⊆ [0, N ].

(c) By Exercise 12.32, the function P has polynomial growth at infinity.
The functions Pn are continuous and 2n -periodic, hence bounded, but we claim
that they obey the same polynomial growth as P, with the same constants.
To see this, note that, by Theorem 12.25, µcn = Pn → P = µ b uniformly on
compact sets. Therefore
C = sup kPn · χ[−1,1] k∞ < ∞,

n≥0
where P0 = 1, and we have |P (ξ)| ≤ C on [−1, 1] as well. Let R be large

enough that R > km0 k∞ and R > 2.
Now we argue similarly to Exercise 12.32. Note that
Pn (ξ) = m0 (ξ/2) m0 (ξ/4) · · · m0 (ξ/2n ) = m0 (ξ/2) Pn−1 (ξ/2).
Iterating, as long as m ≤ n we have
Pn (ξ) = m0 (ξ/2) · · · m0 (ξ/2m ) Pn−m (ξ/2m ).
Suppose that 0 < m ≤ n and 2m−1 ≤ |ξ| < 2m . Then 2−m ξ ∈ [−1, 1], so
m
Y
|Pn (ξ)| = |m0 (2−j ξ)| |Pn−m (2−m ξ)| ≤ C km0 km m
∞ ≤ CR .
j=1
Suppose on the other hand that m > n and 2m−1 ≤ |ξ| < 2m . Since Pn
is 2n -periodic, there is some integer ℓ such that |ξ − 2n ℓ| ∈ [0, 2n ]. Hence
2i ≤ |ξ − 2n ℓ| < 2i−1 for some i ≤ n, so
|Pn (ξ)| = |Pn (ξ − 2n ℓ)| ≤ C Ri ≤ C Rm .
Thus, for all m > 0 we have
2m−1 ≤ |ξ| < 2m =⇒ |Pn (ξ)| ≤ C Rm .
Just as in the proof of Exercise 12.32, if we set α = log2 R and let M > α be
an integer then we have
|Pn (ξ)| ≤ CR |ξ|M , |ξ| ≥ 1.
The constants C and M are independent of ξ.

Now fix any f ∈ S(R) and ε > 0. Since fb ∈ S(R) we have
D = kξ M+2 fb(ξ)k∞ < ∞,
so
D
|fb(ξ)| ≤ , |ξ| ≥ 1.
|ξ|M+2
Choose T > 1 large enough that
Z
1
dξ ≤ ε.
|ξ|>T |ξ|2
Then
|hf, µ − µn i|

= fb, µ
b−µ cn
Z T Z
≤ |fb(ξ)| |P (ξ) − Pn (ξ)| dξ + |fb(ξ)| |P (ξ) − Pn (ξ)| dξ
−T |ξ|>T
Z

≤ kfbkL1 P − Pn · χ[−T,T ] ∞ + |fb(ξ)| 2C |ξ|M dξ
|ξ|>T
Z
12
≤ kfbkL1 P − Pn · χ[−T,T ] ∞ + 2CD dξ
|ξ|>T |ξ|

≤ kfbkL1 P − Pn · χ[−T,T ] ∞ + 2CDε.
Hence
lim sup |hf, µ − µn i| ≤ 0 + 2CDε.
n→∞
w*
Since ε is arbitrary, we conclude that hf, µn i → hf, µi, and therefore µn −→ µ.
Suppose that f ∈ Cc∞ (R) satisfies supp(f ) ⊆ R\[0, N ]. Then since
supp(µn ) ⊆ [0, N ], we have hf, µn i = 0 by definition of support. Consequently
hf, µi = lim hf, µn i = 0.

n→∞
Hence supp(µ) ⊆ [0, N ].

(d) The space L1 (R) embeds injectively into S ′ (R). Therefore, if ϕ is an
integrable solution to the refinement equation then it is a distributional solu-
tion whose Fourier transform is a continuous function. However, the compactly
supported distribution µ constructed above is a solution to the refinement
equation and has a Fourier transform that is a continuous function. By the
uniqueness result in Exercise 12.35, ϕ must be a scalar multiple of µ. The
support of an integrable function coincides with its distributional support, so
supp(ϕ) ⊆ [0, N ].
(e) By Theorem 12.29, there exists a function ϕ ∈ L2 (R) that is a solution
to the refinement equation and which has a continuous Fourier transform ϕ. b
By part (c), there is a distribution, that we will also call ϕ, that is a solution
to the refinement equation, is compactly supported, and has a continuous
Fourier transform. By the uniqueness result proved in Exercise 12.35, these
two distributions must be equal (if we rescale them so that ϕ(0)b = 1). Hence
ϕ is compactly supported and belongs to L2 (R), and therefore is integrable.
12.37 (a) Set
dk = (−1)k−1 c1−k , k ∈ Z.
Then (dk )k∈Z ∈ ℓ2 (Z), so the series
X
ψ(x) = (−1)k−1 c1−k ϕ(2x − k)
k∈Z
converges unconditionally in L2 -norm.

(b) Just as in the proof of Lemma 12.15, part (a) implies that
b
ψ(ξ) b
= m1 (ξ/2) ϕ(ξ/2),
where
1X
m1 (ξ) = dk e−2πikξ .
2
k∈Z
Note that this series converges unconditionally in L2 (T). Further,

1X
m1 (ξ) = (−1)1−k c1−k e−2πijξ .
2
k∈Z
1X
= ck eπik e2πi(k−1)ξ
2
k∈Z
1X 1
= e−2πiξ ck e2πik(ξ+ 2 )
2
k∈Z
−2πiξ
= e m0 (ξ + 21 ).
(c) Just as in the proof of Theorem 12.27, by applying the refinement

equation and the periodicity of m0 , we have
X
Φψ (ξ) = b + k)|2
|ψ(ξ
k∈Z
X 2 2
= m1 ξ+k ϕ b ξ+k
2 2
k∈Z
X 2 2 X 2 2
= m1 ξ+2k ϕ b ξ+2k + m1 ξ+2k+1 ϕ b ξ+2k+1
2 2 2 2
k∈Z k∈Z
X
ξ 2
2 2 X 2
= m1 2
ϕb ξ
2 + k + m1 ξ+1
2
ϕb ξ+1
2 +k
k∈Z k∈Z
2 2
= m1 ξ
2
+ m1 ξ
2 + .1
2
2 2
= m0 ξ
2 + 1
2
+ m0 ξ
2
= 1 a.e.
Applying the definition of m1 , we see that
m1 (ξ) m0 (ξ) + m1 (ξ + 21 ) m0 (ξ + 12 )
1
= e−2πiξ m0 (ξ + 21 ) m0 (ξ) + e−2πi(ξ+ 2 ) m0 (ξ) m0 (ξ + 21 )
= e−2πiξ m0 (ξ + 21 ) m0 (ξ) − e−2πiξ m0 (ξ) m0 (ξ + 12 )

= 0.
Hence
X
bϕ
ψ, b (ξ) = b + k) ϕ(ξ
ψ(ξ b + k)
k∈Z
X
ξ+k ξ+k ξ+k ξ+k
= m1 2 b
ϕ 2 m0 2 b
ϕ 2
k∈Z
X
= m1 ξ
m0 ξ ϕb ξ
+k
2 2 2
k∈Z
X
+ m1 ξ+1
m0 ξ+1 ϕb ξ+1
+k
2 2 2
k∈Z
ξ
ξ
ξ 1
ξ 1

= m1 2 m0 2 + m1 2 + 2 m0 2 + 2
= 0,
which by Lemma 10.20 implies that hψ, Tk ϕi = 0 for every k ∈ Z.

12.38 (a) Since we are given that ϕ ∈ L2 (R) is refinable, we compute that
b
|ϕ(ξ)|2 b 2 = |m0 (ξ/2)|2 |ϕ(ξ/2)|
+ |ψ(ξ)| b 2
+ |m1 (ξ/2)|2 |ϕ(ξ/2)|
b 2
2
b
= |ϕ(ξ/2)| a.e.
(b) Applying part (a), we turn the series into a telescoping sum:
N
X N
X
b n ξ)|2 =
|ψ(2 b n−1 ξ)|2 − |ϕ(2
|ϕ(2 b n ξ)|2
n=1 n=1
2
b
= |ϕ(ξ)| b N ξ)|2 .
− |ϕ(2 (A)
If we know in addition that ϕ ∈ L1 (R) then we have by the the Riemann–
Lebesgue Lemma that ϕ b is continuous and ϕ(ω)
b → 0 as |ω| → ∞. Hence, in
this case we have
∞
X
b n ξ)|2 =
|ψ(2 lim b 2 − |ϕ(2
|ϕ(ξ)| b N ξ)|2 b
≤ |ϕ(ξ)|2
a.e.
N →∞
n=1
However, if we only know that ϕ ∈ L2 (R), then we can reach the same
conclusion by applying the Lebesgue Dominated Convergence Theorem. Note
that by equation (A), for every N we have
N
X
b n ξ)|2 = |ϕ(ξ)|
|ψ(2 b 2 − |ϕ(2
b N ξ)|2 ≤ |ϕ(ξ)|
b 2
.
n=1
P∞ b n ξ)|2 converges pointwise almost everywhere
Now, the series ρ(ξ) = n=1 |ψ(2
in the extended real sense since each term is nonnegative. For almost every ξ
we have that
N
X ∞
X
lim b n ξ)|2 =
|ψ(2 b n ξ)|2 ≤ |ϕ(ξ)|
|ψ(2 b 2 ∈ L1 (R).
N →∞
n=1 n=1
PN b n ξ)|2
The Dominated Convergence Theorem therefore implies that n=1 |ψ(2
P∞ b n ξ)|2 in L1 -norm. Hence
converges to n=1 |ψ(2
Z ∞ X∞
0 ≤ b
|ϕ(ξ)|2
− b n ξ)|2 dξ
|ψ(2
−∞ n=1
Z ∞ N
X
= b
lim |ϕ(ξ)|2
− b n 2
|ψ(2 ξ)| dξ
−∞ N →∞ n=1
Z ∞ N
X
= lim b
|ϕ(ξ)|2
− b n ξ)|2 dξ
|ψ(2
N →∞ −∞ n=1
Z ∞
= lim b N ξ)|2 dξ
|ϕ(2
N →∞ −∞
Z ∞
−N 2
= lim 2 b
|ϕ(ξ)| dξ
N →∞ −∞
= lim 2−N kϕk

b 2L2 = 0.
N →∞
Hence ∞
X
b n ξ)|2 = |ϕ(ξ)|
|ψ(2 b 2 a.e.
n=1
(c) Theorem 12.6 implies that if ψ is any function that generates a dyadic
P b n 2
wavelet orthonormal basis, then we must have n∈Z |ψ(2 ξ)| = 1 a.e.,
whether ψ is associated with an MRA or not. In this problem we are given the
additional information that ψ is associated with an MRA and asked to try to
prove that the same conclusion holds without appealing to Theorem 12.6. I’m
not sure how to do this if we assume only that the scaling function ϕ belongs
to L2 (R). However, if we assume ϕ ∈ L1 (R) ∩ L2 (R) then ϕ b is continuous
b
and we have ϕ(0) = 1. In this case we can use use part (b) and compute that
X ∞
X
b n ξ)|2 =
|ψ(2 lim b n ξ)|2
|ψ(2
N →∞
n∈Z n=−N
∞
X
= lim b n−N −1 ξ)|2
|ψ(2
N →∞
n=1
∞
X
= lim b n (ξ/2N +1 ))|2
|ψ(2
N →∞
n=1
N +1 2
= b
lim |ϕ(ξ/2 )|
N →∞
2
b
= |ϕ(0)| = 1.
12.39 Since the boundary points of intervals have measure zero, we can ignore
them throughout.
(a) The set E is congruent mod 1 to
5 1 5 2 1 2
7 , 1 ∪ 2 , 7 ∪ 7 , 2 ∪ 0, 7 = [0, 1].
Therefore E tiles the real line by integer translates with overlaps of measure
zero.
Also, dilates by 2 of E have overlaps of measure zero. Since
E/4 ∪ E/2 ∪ E ∪ 2E ∪ 4E
contains
1 4
4 8 2 16
2, 7 ∪ 1, 87 ∪ 72 , 21 ∪ 2, 16
7 ∪ 7, 1 ∪ 7, 2 = 7, 7
and dilations by 2 of this last interval cover the real line, we see that E tiles
the positive axis by dilations. Since E is symmetric, it also tiles the negative
axis.
Exercise 12.4 therefore implies that E is a wavelet set.
(b) Define ψb = χE , and suppose that ψ was associated with an MRA and
a scaling function ϕ. Then by Exercise 12.38, we have
∞
X
b
|ϕ(ξ)|2
= b n ξ)|2 = χF
|ψ(2
n=1
where
F = − 78 , −1 ∪ − 74 , − 12 ∪ − 72 , 72 ∪ 21 , 74 ∪ 1, 87 .
(c) Since
b
ϕ(ξ) b
= m0 (ξ/2) ϕ(ξ/2),
2 2
we conclude that m0 (ξ) = 1 on − 7 , 7 . Since m0 is 1-periodic, it is therefore 1

on the interval 75 , 79 . Hence if ξ ∈ 1, 87 then we have
b
0 = ϕ(2ξ) b
= m0 (ξ) ϕ(ξ) = 1,
which is a contradiction.

b
(d) Since |ϕ(ξ)| = 1 both for ξ ∈ 0, 71 and for ξ ∈ 1, 87 , we have
X
2
Φϕ (ξ) = b
|ϕ(ξ)| = 2, ξ ∈ 0, 17 .
k∈Z
Hence ϕ cannot have integer orthonormal translates, which is a contradiction.

12.40 The function ν(x) = x4 (35 − 84x + 70x2 − 20x3 ) satisfies:
ν(x) = x4 (35 − 84x + 70x2 − 20x3 ), ν(0) = 0, ν(1) = 1,

ν ′ (x) = −140 x3 (x − 1)3 , ν ′ (0) = ν ′ (1) = 0,
ν ′′ (x) = −420 x2 (x − 1)2 (2x − 1), ν ′′ (0) = ν ′′ (1) = 0,
ν ′′′ (x) = −840 x (5x3 − 10x2 + 6x − 1) ν ′′′ (0) = ν ′′′ (1) = 0,
ν ′′′′ (x) = −840 (20x3 − 30x2 + 12x − 1) ν ′′′′ (0) = 840, ν ′′′′ (1) = −840.
Thus ν ′′′ is continuous, so ν ∈ C 3 (R). Further, 0 < ν(x) < 1 for x ∈ (0, 1), so
ν is a C 3 sigmoid function.
12.33 (Details for the proof of Theorem 12.33).
(a) Given equation (12.58), which states that
mf (ξ) m0 (ξ) + mf (ξ + 12 ) m0 (ξ + 12 ) = 0 a.e.,

we will verify that the function

−mf (ξ + 1 )/m0 (ξ), m0 (ξ) 6= 0,
2
λ(ξ) =
m (ξ)/m (ξ + 1 ), otherwise.
f 0 2
which is defined just after equation (12.59), satisfies mf (ξ) = λ(ξ) m0 (ξ + 21 )

a.e.
Case 1. For those points ξ where m0 (ξ) 6= 0, by equation (12.58) and the
definition of λ we have
mf (ξ + 12 ) m0 (ξ + 12 )
λ(ξ) m0 (ξ + 12 ) = − = mf (ξ).
m0 (ξ)
Case 2. Now consider those points ξ where m0 (ξ) = 0. By the antialiasing
hypothesis, we have m0 (ξ + 21 ) 6= 0 for all these points. Equation (12.58)
implies that
0 = mf (ξ) m0 (ξ) + mf (ξ + 12 ) m0 (ξ + 12 ) = mf (ξ + 12 ) m0 (ξ + 12 ),
and since m0 (ξ + 12 ) 6= 0 this implies that
mf (ξ + 12 ) = 0.
Therefore
mf (ξ)
λ(ξ) m0 (ξ + 12 ) = m0 (ξ + 21 ) = mf (ξ)
1
m0 (ξ + 2)
in this case as well.
(b) Now we verify that λ(ξ + 12 ) = −λ(ξ) a.e.
Case 1. If m0 (ξ) 6= 0 and m0 (ξ + 12 ) 6= 0, then by applying the definition
of λ and equation (12.58) we have that
λ(ξ) = −mf (ξ + 21 )/m0 (ξ)
= mf (ξ)/m0 (ξ + 12 )

= mf (ξ + 21 ) + 1
2 /m0 (ξ + 12 ) = −λ(ξ + 21 ).
Case 2. If m0 (ξ) = 0 then m0 (ξ + 21 ) 6= 0, so by applying the definition of

λ we have

λ(ξ + 21 ) = −mf (ξ + 21 ) + 12 /m0 (ξ + 21 ) = −mf (ξ)/m0 (ξ + 12 ) = −λ(ξ).
Case 3. If m0 (ξ + 12 ) = 0 then m0 (ξ) 6= 0, so again by applying the
definition of λ we have

λ(ξ) = −mf (ξ + 21 )/m0 (ξ) = −mf (ξ + 12 )/m0 (ξ + 12 ) + 21 = −λ(ξ + 12 ).
13.1 (a) Fix f ∈ C(T) and ε > 0. Since f is uniformly continuous, there
exists 0 < δ < 1 such that
|x − y| < δ =⇒ |f (x) − f (y)| < ε.
Fix |a| < δ. Then
|f (x) − Ta f (x)| = |f (x) − f (x − a)| < ε.
Thus kf − Ta f k∞ ≤ ε whenever |a| < δ, so kTa f − f k∞ → 0.
(b) Fix 1 ≤ p < ∞, and choose f ∈ Lp (T). Given ε > 0, we can find
g ∈ C(T) such that kf − gkLp < ε. Since g is uniformly continuous, there
exists a δ > 0 such that
|a| < δ =⇒ kg − Ta gk∞ < ε.
Therefore, for such a we have
Z 1 Z 1
p p
kg − Ta gkLp = |g(x) − Ta g(x)| dx ≤ εp dx = εp .
0 0
Since translation is isometric on Lp (T), we therefore have for |a| < δ that
kf − Ta f kLp ≤ kf − gkLp + kg − Ta gkLp + kTa g − Ta f kLp
≤ ε + ε + ε = 3ε.
Hence Ta f → f in Lp (T) as a → 0.
13.2 (a) We have
Z 1
f (x − a) e−2πinx dx
∧
(Ta f ) (n) =
0
Z 1
= f (x) e−2πin(x+a) dx
0
Z 1
= e −2πina
f (x) e−2πinx dx = M−a fb(n).
0
(a’) We have
X
c(n − m) e−2πinx
∧
(Tm c) (x) =
n∈Z
X
= c(n) e−2πi(n+m)x
n∈Z
X
= e−2πimx c(n) e−2πinx = M−m b
c (x).
n∈Z
(b’) We have
∧
X
(Ma c) (x) = (Ma c)(n) e−2πinx
n∈Z
X
= e2πian c(n) e−2πinx
n∈Z
X
= c(n) e−2πin(x−a) = b
c (x − a).
n∈Z
13.3 Since f is differentiable, we can apply integration by parts to conclude

that
Z 1
b′
f (n) = f ′ (x) e−2πinx dx
0
1 Z 1
−2πinx
= f (x) e − f ′ (x) e−2πinx (−2πin) dx
0 0
= 0 + 2πinfb(n).
Therefore for n 6= 0 we have
|fb′ (n)| kfb′ kℓ∞ kf ′ kL1

|fb(n)| ≤ ≤ ≤ .
2π |n| 2π |n| 2π |n|
The result extends to higher derivatives by induction.

P
13.4 We are given that n∈Z |ncn | < ∞. Note that
c (ξ + h) − b
b c (ξ) X e−2πin(ξ+h) − e−2πinξ
= cn .
h h
n∈Z
Since e−n (ξ) = e−2πinξ is differentiable, the summand converges pointwise as

h → 0:
e−2πin(ξ+h) − e−2πinξ d
lim cn = cn e−2πinξ = −2πincn e−2πinξ .
h→∞ h dξ
Further, since |1 − eiθ | ≤ |θ|, we have

−2πinh
e−2πin(ξ+h) − e−2πinξ − 1
cn = |cn | |e−2πinh | e
h h
2π|nh|
≤ |cn | |e−2πinh |
|h|
= 2π|ncn | ∈ ℓ1 (Z).
Alternatively, we can obtain the same estimate by applying the Mean-Value

Theorem to e−2πinξ . The Dominated Convergence Theorem for series therefore
allows us to interchange the sum and integral in the following calculation:
b
c (ξ + h) − b
c (ξ) X e−2πin(ξ+h) − e−2πinξ
lim = lim cn
h→0 h h→0 h
n∈Z
X e−2πin(ξ+h) − e−2πinξ
= cn lim
h→0 h
n∈Z
X
= −2πi ncn e−2πinξ .
n∈Z
Since (ncn )n∈Z ∈ ℓ1 (Z), we conclude that cb ′ (ξ) exists and
b
cb ′ (ξ) = d(ξ)
where d = (−2πincn )n∈Z .

PM k
PN k
13.5 Let p(x) = k=0 ak x and q(x) = k=0 bk x be two P
polynomials.
M+N
Their pointwise product is a polynomial of the form p(x)q(x) = n=0 cn xn .
For 0 ≤ n ≤ M + N, the scalar cn has the form
n
X
cn = ak bn−k .
k=0
If we take ak = 0 for k ∈ / {0, . . . , M } and bk = 0 for k ∈

/ {0, . . . , N }, then
ak bn−k = 0 when k ∈
/ 0, . . . , n. Therefore we can express cn as
n
X X
cn = ak bn−k = ak bn−k = (a ∗ b)n .
k=0 k∈Z
Suppose that we define cn by this formula for all n. If cn 6= 0 then we must

have ak bn−k 6= 0. Therefore 0 ≤ k ≤ M and 0 ≤ n−k ≤ N, which implies that
0 ≤ n ≤ M + N. Hence c0 = 0 whenever n ∈ / {0, . . . , M + N }. Consequently,
this formula for cn is valid for all n ∈ Z.
13.6 f (x) = x−1/2 ∈ L1 (T), but f (x)2 = x−1 ∈
/ L1 (T).
13.9 (a) Fix f , g ∈ L1 (T). Then we have
Z 1 Z 1 Z 1 Z 1
2πinx
|f (y) g(x − y) e | dx dy = |f (y)| |g(x − y)| dx dy
0 0 0 0
Z 1 Z 1
= |f (y)| |g(x)| dx dy
0 0
Z 1
= kgkL1 |f (y)| dy
0
= kgkL1 kf kL1 < ∞.
(b) Given f , g ∈ ℓ1 (Z), the functions fb and gb are continuous, and the series
defining them converge absolutely. Hence we can compute pointwise that
X
(f ∗ g)(n) e−2πinx
∧
(f ∗ g) (x) =
n∈Z
X X
= f (m) g(n − m) e−2πinx
n∈Z m∈Z
X X
= f (m) e−2πimx g(n − m) e−2πi(n−m)x
m∈Z n∈Z
X X
= f (m) e−2πimx g(n) e−2πinx
m∈Z n∈Z
X
= f (m) e−2πimx gb(x)
m∈Z
= fb(x) gb(x).
13.11 Set f ∗ (x) = f (−x). Then, making the change of variables x 7→ −x,
Z 1/2
fc∗ (−n) = f (x) e−2πi(−n)x dx
−1/2
Z −1/2
= − f (−x) e−2πi(−n)(−x) dx
1/2
Z 1/2
= f (x) e−2πinx dx = fb(n).
−1/2
∨
Finally, f (n) = fb(−n) by definition.

13.12 Let δ = δ0 = δ0n n∈Z . If c ∈ ℓp (Z) then we have
X
(c ∗ δ)(n) = c(n − k) δ(k) = c(n),
k∈Z
so c ∗ δ = c.
13.13 (a) Assume that f ∈ L1 (T) and g ∈ C(T). Then g is uniformly

continuous, so
Z 1

|(f ∗ g)(x) − (f ∗ g)(x − a)| = f (y) g(x − y) − g(x − a − y) dy
0
Z 1
≤ |f (y)| kg − Ta gk∞ dy
0
= kf kL1 kg − Ta gk∞
→ 0 as a → 0.
(b) Assume that f ∈ L1 (T) and g ∈ C 1 (T). We have

Z
(f ∗ g)(x + h) − (f ∗ g)(x) g(x + h − y) − g(x − y)
= f (y) dy.
h h
The integrand converges pointwise a.e. to f (y) g ′ (x − y) as h → 0. Further, g ′

is bounded since it is continuous and periodic. Therefore, by the Mean Value
Theorem, given x, y, and h there exists a point c such that

g(x + h − y) − g(x − y)
= |g ′ (c)| ≤ kg ′ k∞ .
h
Therefore (as a function of y),

f (y) g(x + h − y) − g(x − y) ≤ |f (y)| kg ′ k∞ ∈ L1 (T).
h
The Lebesgue Dominated Convergence Theorem therefore applies, and we

find that
(f ∗ g)(x + h) − (f ∗ g)(x)
(f ∗ g)′ (x) = lim
h→0 h
Z
g(x + h − y) − g(x − y)
= lim f (y) dy
h→0 h
Z
= f (y) g ′ (x − y) dy = (f ∗ g ′ )(x).
Thus f ∗ g is differentiable, and furthermore (f ∗ g)′ = f ∗ g ′ ∈ C(T) by

part (a). Hence f ∗ g ∈ C 1 (T).
′
13.14 Fix 1 < p < ∞. Choose f ∈ Lp (T) and g ∈ Lp (T). Since 1 ≤ p′ < ∞,
′
translation is strongly continuous on Lp (T). Therefore, by Hölder’s Inequality
we have
|(f ∗ g)(x) − (f ∗ g)(x − a)|

Z 1

= f (y) g(x − y) − g(x − a − y) dy
0
Z 1 1/p Z 1 1/p′
p p′
≤ |f (y)| dy |g(x − y) − g(x − a − y)| dy
0 0
= kf kLp kg − Ta gkLp′
→ 0 as a → 0.
Therefore f ∗ g is continuous. The case p = 1 or p = ∞ are similar.

13.15 Exercise 11.9 shows that there exists an infinitely differentiable func-
tion gN such that 0 ≤ gN ≤ 1 everywhere, gN (x) 6= 0 for −1/N R < x < 1/N,
and gN (x) = 0 for all other x. For each N > 1 set cN = gN (x) dx and
let kN be the function gN /cN extended 1-periodically to the real line. Then
{kN }N ≥2 is an approximate identity, and each kN is infinitely differentiable.
13.16 Fix 1 < p < ∞ and let K = sup kkN kL1 . Using Hölder’s Inequality,
we have
kf − f ∗ kN kpLp
Z 1 Z 1 p

= f (x) − f (x − t) k (t) dt dx
N
0 0
Z 1 Z 1 p
′
≤ |f (x) − f (x − t)| |kN (t)|1/p |kN (t)|1/p dt dx
0 0
Z 1 Z 1 p/p Z 1 p/p′
p
≤ |f (x) − f (x − t)| |kN (t)| dt |kN (t)| dt dx
0 0 0
Z 1 Z 1
p/p′
= kkN kL1 |f (x) − f (x − t)|p |kN (t)| dt dx
0 0
Z 1 Z 1
p/p′ p
≤ K |f (x) − f (x − t)| dx |kN (t)| dt
0 0
Z 1
′
= K p/p kf − Tt f kpLp |kN (t)| dt.
0
From this point onwards, the proof is identical to the proof of Theorem 13.13,
using the fact that translation is strongly continuous in Lp (T) when p < ∞.
13.17 By Exercise 13.15 and 11.9, we can create an approximate identity
{kN } such that kN ∈ C ∞ (T) for every N. Given f ∈ Lp (T), it follows from
Exercise 13.13 that f ∗ kN ∈ C ∞ (T). By Theorem 13.13 we have f ∗ kN → f
in Lp -norm. Hence C ∞ (T) is dense in Lp (T) when 1 ≤ p < ∞, and the same
argument with p = ∞ establishes denseness in C(T).
13.18 If f is Hölder continuous on T, then, using equation (13.13), we have
Z Z α
1 1 1 1 1 1 α 1 1
|fb(n)| ≤ f (x) − f x − dx ≤ dx = .
2 0 2n 2 0 2n 2 2|n|
13.19 By definition and the fact that (δm ) (t) = e2πimt , we have
∨
N
X
∨
χN (x) = e2πimx = dN (x).
m=−N
∨
Also, χbN = χN since χN is even.
Let ω = e2πix , and let
N
X N
X
s = dN (x) = e2πimx = ωm.
m=−N m=−N
Then we have
sω = s − ω −N + ω N +1 ,
so
s(ω − 1) = ω N +1 − ω −N .
Multiplying both sides by ω −1/2 = e−πix , we obtain
s(ω 1/2 − ω −1/2 ) = ω N +1/2 − ω −N −1/2 .
Now,
ω 1/2 − ω −1/2 = eπix − e−πix = 2i sin πx,
and likewise
ω N +1/2 − ω N −1/2 = e2πi(N +1/2)x − e−2πi(N +1/2)x = 2i sin(2N + 1)πx,
so
sin π(2N + 1)x
s = .
sin πx
13.20 We have dN ∈ L1 (T) since it is continuous on T, and also
Z 1 Z 1 N
X
dN = e2πinx dx = 1.
0 0 n=−N
However, we will show that sup kdN kL1 = ∞. Using the fact that
Z 1 1
− cos πx 2
sin πx dx = = π,
0 π 0
together with the estimate | sin x| ≤ |x|, we have

Z 1/2
1 sin(2N + 1)πx
kdN kL1 = dx
2 sin πx
0
Z 1/2
| sin(2N + 1)πx|
≥ dx
0 |πx|
Z N + 21
| sin πx|
= dx
0 π|x|
N
X −1 Z k+1
| sin πx|
≥ dx
k π|x|
k=0
N −1 Z
1 X k+1 | sin πx|
= dx
π k k+1
k=0
N −1 Z
1 X 1 k+1
= sin πx dx
π k+1 k
k=0
N −1
2 X 1
=
π2 k+1
k=0
N
2 X1
= dx → ∞ as N → ∞.
π2 k
k=1
In fact, we can estimate using the Integral Test as follows:

N Z N
4 X1 4 1 4
kdN kL1 ≥ ≥ dx = 2 ln N.
π2 k π2 1 x π
k=1
To obtain an upper bound, the fact that

1 1
f (x) = −
sin πx πx
is odd and increasing on [−1/2, 1/2] means that |f (x)| reaches its maximum
value at x = 1/2. Hence we have
1 1 2 1

− = |f (x)| ≤ f (1/2) = 1 − , |x| ≤ .
sin πx πx π 2
Consequently,
1 1 2 1
≤ + 1− , |x| ≤ .
| sin πx| π|x| π 2
0.4
0.2
-0.4 -0.2 0.2 0.4
-0.2
-0.4
Fig. B.1. Graph of f (x).
Arguing similarly to before, we have

Z 1/2
1 sin(2N + 1)πx
kdN k1 = dx
2 sin πx
0
Z 1/2 Z
| sin(2N + 1)πx| 2 1/2
≤ dx + 1 − | sin(2N + 1)πx| dx
0 π|x| π 0
Z N + 21
| sin πx| 2 1
≤ dx + 1 −
0 π|x| π 2
Z 1 Z 1 1
N
sin πx 1 X k+1 | sin πx|
≤ dx + dx + −
0 πx π k k 2 π
k=1
2
N
X 1 1 1
≤ α + + −
π2 k 2 π
k=1
N
1 1 2 X1
= α+ − + 2 ,
2 π π k
k=1
where Z 1
sin πx
α = dx ≈ 0.58949 < 1.
0 πx
Hence
N N
2 4 X1 4 X1
kdN k1 ≤ 2α + 1 − + 2 ≈ 2 + 1.54236 . . . .
π π k π k
k=1 k=1
Now, for every n we have

N
X
≤ 1 + ln N,
k=1
and in fact Euler’s constant is

X
N
lim − ln N = γ ≈ 0.577 . . . .
N →∞
k=1
Hence
N
2 4 X1
kdN k1 ≤ 2α + 1 − + 2
π π k
k=1
2 4
≤ 2α + 1 − + 2 (1 + ln N )
π π
2 4 4
= 2α + 1 − + 2 + 2 ln N,
π π π
where γ ≈ 0.577 . . . is Euler’s constant. Numerically,
2 4 4 4
kdN k1 ≤ 2α + 1 − + 2 + 2 ln N ≈ 2 ln N + 1.94764 . . . .
π π π π
13.21 (a) We have
XN XN n
X
1 1
σN = sn = ak
N + 1 n=0 N + 1 n=0
k=−n
N
X N
X
1
= ak
N +1
k=−N n=|k|
N
|k|
N
X X
N − |k| + 1
= ak = 1− ak .
N +1 N +1
k=−N k=−N
(b) Suppose that a ∈ ℓ1 (Z). Set

n |k| o
WN (k) = max 1 − ,0 .
N +1
Then for each k we have
lim WN (k) ak = ak
N →∞
and
WN (k) |ak | ≤ |ak |.
Since a is summable, the Dominated Convergence Theorem for series implies

that
N
X |k|
lim σN = lim 1− ak
N →∞ N →∞ N +1
k=−N
∞
X
= lim WN (k) ak
N →∞
k=−∞
∞
X ∞
X
= lim WN (k) ak = ak .
N →∞
k=−∞ k=−∞
More generally, suppose that we know that the partial sums sN converge
to L. Choose any ε > 0. Then there exists an M such that
N > M =⇒ |L − sN | < ε.
Let C = sup |sN |. Then we have |L| ≤ C, so

(N + 1)L − (s0 + · · · + sN )

lim sup |L − σN | = lim sup
N →∞ N →∞ N +1
X N M
|L − sk | X |L − sk |
≤ lim sup +
N →∞ N +1 N +1
k=M+1 k=0
X
N
ε (M + 1) 2C
≤ lim sup + ≤ ε + 0 = ε.
N →∞ N +1 N +1
k=M+1
This is true for every ε > 0, so σN → L.

(c) We have s2N = 1 and s2N +1 = 0, so
s0 + · · · + s2N N
σ2N = = ,
2N + 1 2N + 1
s0 + · · · + s2N +1 N
σ2N +1 = = ,
2N + 2 2N + 2
1
and therefore σN → 2 as N → ∞.
13.22 We compute that
N
X |n| b
σN f (x) = 1− f (n) e2πinx
N +1
n=−N
N Z
X |n| 1
−2πint
= 1− f (t) e dt e2πinx
N +1 0
n=−N
Z N
1 X |n| 2πin(x−t)
= f (t) 1− e dt
0 N +1
n=−N
Z 1
= f (t) wN (x − t) dt
0
= (f ∗ wN )(x).
∨ PN
13.23 (a) Fix any N ∈ N. Note that χN (x) = n=−N e2πinx . Using Exer-
cise 13.21, we can therefore compute that
∨
N
X |n| 2πinx
WN (x) = 1− e
N +1
n=−N
∨ ∨
χ0 (x) + · · · + χN (x)
=
N +1
N
1 X sin(2N + 1)πx
=
N + 1 n=0 sin πx
N
1 X e(2n+1)πix − e−(2n+1)πix
=
N + 1 n=0 2i sin πx
XN X N
1 1
= eπix e2nπix − e−πix e−2nπix
N + 1 2i sin πx n=0 n=0
2πi(N +1)x −2πi(N +1)x

1 1 πix e −1 −πix e −1
= e − e
N + 1 2i sin πx e2πix − 1 e−2πix − 1
2πi(N +1)x
1 1 e −1 e−2πi(N +1)x − 1
= −
N + 1 2i sin πx eπix − e−πix e−πix − eπix
2πi(N +1)x
1 1 e − 2 + e−2πi(N +1)x
=
N + 1 2i sin πx eπix − e−πix
πi(N +1)x 2
1 1 (e − e−πi(N +1)x)
=
N + 1 2i sin πx 2i sin πx
1 (2i sin π(N + 1)x)2
=
N +1 (2i sin πx)2
2
1 sin π(N + 1)
= .
N +1 sin πx
(b) The first requirement to be an approximate identity follows from the

fact that
Z Z
|n|
1 N
X 1
wN (x) dx = 1− e2πinx dx
0 N +1 0
n=−N
N
X |n|
= 1− δ0n = 1.
N +1
n=−N
Since wN ≥ 0, the second requirement follows trivially.

Finally, choose 0 < δ < 1/2. Note that sin πx is increasing on [0, 1/2].
Therefore,
Z Z 1/2
2 sin2 (N + 1)πx
wN (x) dx = dx
δ≤|x|<1/2 N +1 δ sin2 πx
Z 1/2
2 1
≤ dx
N +1 δ sin2 πδ
1 1
≤ → 0 as N → ∞.
N + 1 sin2 πδ
13.24 (a) Suppose that f ∈ L1 (T) and fb ∈ ℓ2 (Z). Then we have fb ∈ ℓ1 (Z),
so the Inversion Formula applies. In particular, f is continuous and bounded
on T, so f ∈ L2 (T).
(b) If f ∈ L2 (T), then we know that the Plancherel Equality holds.
So, suppose that f ∈ L1 (T) \ L2 (T). Then we must have fb ∈ / ℓ2 (Z) by
b
part (a). Hence we have both kf kL2 = ∞ and kf kL2 = ∞, so again the
Plancherel Equality holds.
13.25 Integration by parts shows that



Z 1  i ,

n 6= 0,
xe −2πinx
dx = 2πn
0 

1
 , n = 0,
2
and 

 πin + 1
Z 1  , n 6= 0,
2π 2 n2
x2 e−2πinx dx =


0 1, n = 0.
3
Hence the Fourier coefficients of f (x) = π 2 (x2 −x+ 61 ) (extended periodically)
are 1 1 1
fb(0) = π 2 − + = 0
3 2 6
and, for n 6= 0,
πin + 1 i πin + 1 πin 1
fb(n) = π 2 − = π2 − = .
2π 2 n2 2πn 2
2π n 2 2
2π n 2 2n2
Hence fb ∈ ℓ1 (Z), and therefore the Fourier series for f converges uniformly
on T to the continuous function f . Combining positive and negative terms,
we therefore have that
X∞ X X
cos 2πnx 1 2πinx
= e = fb(n) e2πinx = f (x),
n=1
n2 2n 2
n∈Z,n6=0 n∈Z
where the series converges uniformly on [0, 1). Since f and each term in the
series is 1-periodic, we also have uniform convergence on any compact subset
of R, except that we must remember that f is periodic, and hence is given by
the formula f (x) = π 2 (x2 − x + 61 ) only for x ∈ [0, 1].
13.26 (a) If f ∈ C 2 (T), then applying Exercise 13.3 twice, we have
fc′′ (n) = (2πin)2 fb(n), n ∈ Z.
Therefore
X X |fc′′ (n)| X kfc′′ k∞ X kf ′′ kL1
kfbkL1 = |fb(n)| ≤ 2 2
≤ 2 2
≤ < ∞.
4π n 4π n 4π 2 n2
n∈Z n∈Z n∈Z n∈Z
Hence f ∈ A(T). Since C 2 (T) is dense in C(T), we conclude that A(T) is

dense as well.
(b) Suppose that f = g ∗ h where g, h ∈ L2 (T). Then f ∈ C(T) ⊆ L1 (T).
Also, gb, b
h ∈ ℓ2 (Z), so we have fb = gb b
h ∈ ℓ1 (Z). Therefore f ∈ A(T).
Conversely, suppose that f ∈ A(T). Then fb ∈ ℓ1 (Z). For each n ∈ Z, let
gn be any complex number such that gn2 = fb(n). Then
X X
|gn |2 = |fb(n)| < ∞.
n∈Z n∈Z
Consequently, the function

X
g(x) = gn e2πinx
n∈Z
belongs to L2 (T), and satisfies gb(n) = gn . By Exercise 13.14 we have g ∗ g ∈

C(T) ⊆ L1 (T), and its Fourier coefficients are
g(n) gb(n) = fb(n),

∧
(g ∗ g) (n) = b n ∈ Z.
By uniqueness, we must have f = g ∗ g ∈ L2 (T) ∗ L2 (T).

(c) Define
∞
X
FN = f ∈ A(T) : b
|f (n)| ≤ N .
n=−∞
Then we have
S
∞
A(T) = FN .
N =1
Fix N, and suppose that fk ∈ FN and fk → f in C(T), i.e., uniformly.

Then
kfb − fbk k∞ ≤ kf − fk kL1

Z 1
= |f (x) − fk (x)| dx
0
≤ kf − fk k∞ → 0 as k → ∞.
By Fatou’s Lemma for series, we therefore have

∞
X ∞
X
|fb(n)| = lim |fbk (n)|
k→∞
n=−∞ n=−∞
∞
X
≤ lim inf |fbk (n)|
k→∞
n=−∞
= lim inf N = N.
k→∞
Thus FN is closed.
Finally, since A(T) is a proper subspace of C(T), it cannot contain any
open subsets of C(T). Hence FN has no interior points, and therefore is
nowhere dense in C(T). Therefore A(T) is a countable union of nowhere
dense sets, and hence is meager by definition.
14.1 Using Lemma 13.5, we have

∧ ∧
(MN SN M−N f ) = TN (SN M−N f )

∧
= TN (M−N f ) · χ[−N,N ]

= TN (T−N fb) · χ[−N,N ]
= fb · χ[0,2N ]
o ∧
= (SN f ) . ⊓
⊔
14.2 (a) ⇒ (b) + (c). This follows because the basis constant for {e2πinx }n∈Z
a p
is the supremum of all kSN kLp →Lp and kSN kL →Lp .
(b) ⇐⇒ (d). If f ∈ Lp (T) then
o
kSN f kLp = kMN SN M−N f kLp
= kSN M−N f kLp
≤ kSN kLp→Lp kM−N f kLp
= kSN kLp→Lp kf kLp .
o p
Hence kSN kL →Lp ≤ kSN kLp →Lp . Conversely,
kSN f kLp = kMN SN M−N MN f kLp

o
= kSN MN f kLp
o
≤ kSN kLp →Lp kMN f kLp
o
= kSN kLp →Lp kf kLp ,
o p
so kSN kLp →Lp ≤ kSN kL →Lp .
o p
Thus, we actually have kSN kLp→Lp = kSN kL →Lp .
(e) ⇒ (d). Suppose that (e) holds, and fix f ∈ Lp (T). Let
2N
X
t
g = SN f = −i sign(n) fb(n) e2πinx .
n=−2N
We have g ∈ L1 (T), and, by uniqueness,

(
−i sign(n) fb(n), |n| ≤ 2N,
gb(n) = .
0, |n| > 2N.
Therefore,
2N
X
t t t
SN SN f (x) = SN g(x) = −i sign(n) gb(n) e2πinx
n=−2N
2N
X
= −i sign(n) (−i) sign(n) fb(n) e2πinx
n=−2N
2N
X
= − fb(n) e2πinx + fb(0)
n=−2N
= −S2N f (x) + fb(0).
Rearranging,
t t
S2N f = −SN SN f + fb(0).
Also,
2N
X 2N
X
t
S2N f (x) + iSN f (x) = fb(n) e2πinx + sign(n) fb(n) e2πinx
n=−2N n=−2N
2N
X
= 2 fb(n) e2πinx + fb(0)
n=1
2N
X
= 2 fb(n) e2πinx − fb(0)
n=0
o
= 2 SN f (x) − 2 fb(0).
Therefore,
o t t t t
2 SN f = S2N f + iSN f + fb(0) = −SN SN f + iSN f + 2 fb(0).
Hence
o t t t
2 kSN f k ≤ kSN SN f kLp + kSN f kLp + kfb(0) · 1kLp
t t
≤ kSN k2Lp→Lp kf kLp + kSN kLp →Lp kf kLp + 2 |fb(0)| k1kLp
t t
≤ kSN k2Lp→Lp kf kLp + kSN kLp →Lp kf kLp + 2 kf kLp · 1

t t
= kSN k2Lp →Lp + kSN kLp →Lp + 2 kf kLp .
Thus,
o 1 t 1 t
kSN f k ≤ kS k2 p p + kSN kLp →Lp + 1.
2 N L →L 2
14.3 (a) We are given that f ∈ C(T), f (0) = 0, and f is differentiable at

x = 0. Then
f (x)
g(x) = −2πix
e −1
is continuous except possibly at x = 0. Since limx→0 g(x) has the form 0/0,
we can apply L’Hopital’s rule to compute that
f (x) f ′ (x) f ′ (0)
lim g(x) = lim = lim = − .
x→0 x→0 e−2πix − 1 x→0 −2πie2πix 2πi
Alternatively, writing e(x) = e2πix , we can compute directly that
f (x) f (x) − f (0) x−0

lim g(x) = lim = lim
x→0 x→0 e−2πix−1 x→0 x−0 e(x) − e(0)
1 f ′ (0)
= f ′ (0) = − .
e′ (0) 2πi
In any case, we see that g is continuous at x = 0, so g ∈ C(T) ⊆ L1 (T).

Since
f (x) = (e−2πix − 1) g(x) = M−1 g(x) − g(x),
we have
fb(k) = (M−1 g) (k) − gb(k) = T−1 b

∧
g(k) − gb(k) = gb(k + 1) − gb(k).
Therefore by the Riemann–Lebesgue Lemma,

N
X N
X
fb(k) = gb(k+1)−b
g(k) = gb(N +1)−b
g(−N ) → 0 as N → ∞.
k=−N k=−N
(b) Now we assume only that f ∈ C(T) is differentiable at some point a.

Let
g(x) = T−a f (x) − f (a) = f (x + a) − f (a).
Then g ∈ C(T),
g(0) = f (0 + a) − f (a) = 0,
and g is differentiable at x = 0, with
g ′ (0) = f ′ (a).
Further,
1(k) = (Ma fb)(k) − f (a) δ(k)

gb(k) = (T−a f ) (k) − f (a) b
∧
= e2πiak fb(k) − f (a) δ(k).
Applying part (a) to the function g, we therefore have that

N
X N
X
lim fb(k) e2πiak = lim gb(k) + f (a) δ(k)
N →∞ N →∞
k=−N k=−N
= g(0) + f (a) = f (a).
14.4 (a) ⇒. If g ∈ L1 (T) is real-valued, then

Z 1 Z 1
gb(n) = g(x) e−2πinx dx = g(x) e2πinx dx = b
g(−n).
0 0
⇐. Suppose that g ∈ L1 (T) satisfies gb(n) = gb(−n) for every n. Let

{wN }N ∈N be the Fejér kernel. Then since wN is real-valued, using the hy-
pothesis on g we have
∧
(g ∗ wN )∧ (n) = gb(n) w
d b(−n) w
N (n) = g dN (−n) = (g ∗ wN ) (−n).
Since the Inversion Formula applies to g ∗ wN , we therefore have that

X
(g ∗ wN )(x) = (g ∗ wN )∧ (n) e2πinx
n∈Z
X
(g ∗ wN ) (−n) e2πi(−n)x = (g ∗ wN )(x).
∧
=
n∈Z
Hence g ∗ wN is real-valued. Since g ∗ wN → g in L1 -norm, there is a subse-

quence that converges pointwise a.e., and hence we conclude that g is real-
valued a.e.
(b) Suppose that f ∈ A(T) is real-valued. Then
(Hf )∧ (n) = −i sign(n) fb(n) = i sign(n) fb(−n) = (Hf ) (−n).

∧
Therefore part (a) implies that Hf is real-valued.

14.5 Suppose that f ∈ A(T). Then f ∈ L1 (T) and fb ∈ ℓ1 (Z). By the
Inversion Formula, we therefore have
X
f (x) = fb(n) e2πinx ,
n∈Z
where this series converges absolutely in Lp -norm for every p. Consequently,

the series X
g(x) = −i sign(n) fb(n) e2πinx
n∈Z
p
converges absolutely in L -norm for every p. The function g belongs to A(T)
and is the conjugate function to f . The symmetric partial sums of the Fourier
series for g are the twisted partial sums of the Fourier series of f :
t
S2N g = SN f.
These partial sums converge to g in each Lp -norm.

14.6 (b) ⇒ (d). Suppose that Lp (T) admits conjugation, so every function
f ∈ Lp (T) has a conjugate function Hf ∈ Lp (T). We apply the Closed Graph
Theorem to show that H must be bounded. Suppose that fk → f in Lp (T)
and Hfk → g in Lp (T). Then we have
∧
kb
g − (Hfk ) k∞ ≤ kg − Hfk k1 ≤ kg − Hfk kp → 0.
Since (Hfk ) (n) = −i sign(n) fbk (n) for |n| ≤ 2N and 0 otherwise, we conclude
∧
that g = Hf . Therefore H is bounded by the Closed Graph Theorem.

(c) ⇒ (d). Assume that H is a bounded mapping of some dense subspace S
of Lp (T) into Lp (T). Then H has a unique extension to a bounded mapping of
Lp (T) into itself (Exercise 1.72), which we also call H. We must show that Hf
equals the conjugate function of f for each f ∈ Lp (T). To see this, choose
any fk ∈ S such that fk → f in Lp -norm. Then Hfk → Hf in Lp -norm, so
(Hfk ) (n) → (Hf )∧ (n) for each n ∈ Z. Since Hfk is the conjugate function
∧
of fk , it follows that Hf is indeed the conjugate function of f . Hence Lp (T)

admits conjugation.
(d) ⇒ (a). We fill in some details on this implication. First we check that
(Rf ) = fbχ[0,∞) . Letting 1 denote the constant function, we have
∧
f + iHf fb(0)
Rf = + · 1.
2 2
For n > 0 we therefore have that
fb(n) + i(Hf ) (n) fb(0) b

∧
∧
(Rf ) (n) = + · 1(n)
2 2
fb(n) + i(−i)fb(n)
= +0
2
= fb(n).
For n = 0 we have
fb(0) + i(Hf ) (0) fb(0) b

∧
∧
(Rf ) (0) = + · 1(0)
2 2
fb(0) + 0 fb(0)
= + ·1
2 2
= fb(0).
For n < 0 we have

fb(n) + i(Hf ) (n) fb(0) b

∧
∧
(Rf ) (n) = + · 1(n)
2 2
fb(n) + i(i)fb(n)
= +0
2
= 0.
o
Next we verify the claim that SN f = Rf − M2N +1 RM−2N −1 f :
∧ ∧ ∧
(Rf − M2N +1 RM−2N −1 f ) = (Rf ) − T2N +1 (RM−2N −1 f )

= fb · χ[0,∞) − T2N +1 (M−2N −1 f ) · χ[0,∞)
∧

= fb · χ[0,∞) − T2N +1 (T−2N −1 fb) · χ[0,∞)
= fb · χ[0,∞) − fb · χ[2N +1,∞)

o
= fb · χ[0,2N ] = SN f.
(d) ⇒ (b). This implication is trivial.

14.7 Fix 1 < p < ∞, and suppose that conjugation is bounded on Lp (T). It
is shown in the main part of the text that hHf, gi = −hf, Hgi for f, g ∈ A(T).
Fix f ∈ A(T). Then for g ∈ A(T) we have
|hHf, gi| = |hf, Hgi| ≤ kf kp′ kHgkp

≤ kf kp′ kHgkp
≤ kf kp′ kHkLp →Lp kgkp .
Suppose that g is an arbitrary function in Lp (T). Then we can find gk ∈ A(T)

such that gk → g in Lp -norm. Hence
|hHf, gi| = lim |hHf, gk i| ≤ lim sup kf kp′ kHkLp→Lp kgk kp

k→∞ k→∞
= kf kp′ kHkLp →Lp kgkp .
Consequently, by Hahn–Banach we have that
kHf kp′ = sup |hHf, gi| ≤ kHkLp→Lp kf kp′ .

kgkLp =1
′
Therefore H is bounded on a dense subset of Lp (T), and therefore is bounded
′
on all of Lp (T) by Theorem 14.7. Moreover, the above work shows that
kHkLp′ →Lp′ ≤ kHkLp→Lp .

′
Hence conjugation is bounded on Lp (T).
Detailed Solutions to Exercises from Appendix B
B.1 (c) Suppose that {xn }n∈N is a bounded sequence in X. Since T1 is com-
(1) (1)
pact, there exists a subsequence {xn }n∈N of {xn }n∈N such that {T1 xn }n∈N
(2)
converges. Then since T2 is compact, there exists a subsequence {xn }n∈N of
(1) (2) (2)
{xn }n∈N such that {T2 xn }n∈N converges (note that {T1 xn }n∈N also con-
verges). Continue to construct subsequences in this way.
(m) (k)
Note that xm is a member of the subsequence {xn }n∈N for each
(m)
k = 1, . . . , m. In particular, for every m we have that xm belongs to the
(1) (m)
subsequence {xn }n∈N . For m ≥ 2 we have that xm belongs to the sub-
(2) (m) (k)
sequence {xn }n∈N . In general, {xm }m≥k is a subsequence of {xn }n∈N .
(m) (m)
Therefore {Tk xm }m≥k converges. For simplicity, let ym = xm . Since our
original sequence is bounded, R = sup kym k < ∞.
Fix ε > 0. Then there exists some k such that kT − Tk k < ε. Since {Tk ym }
converges it is Cauchy, so there is some some M such that kTk ym − Tk yn k < ε
for all m, n ≥ M. Therefore for all m, n ≥ M we have
kT ym − T yn k ≤ kT ym − Tk ym k + kTk ym − Tk yn k + kTk yn − T yn k
≤ kT − Tk k kym k + ε + kTk − T k kynk
≤ Rε + ε + Rε.
Hence {T ym } is Cauchy, and therefore converges. Thus T is compact.

B.2 Choose any subsequence {fn }n∈N of {en }n∈N . Since T is compact, this
subsequence has a subsequence {gn }n∈N such that {T gn}n∈N converges in K,
say T gn → h. By Bessel’s Inequality,
X X
|hT gn , hi|2 = |hgn , T ∗ hi|2 ≤ kT ∗ hk2 < ∞.
n∈N n∈N
Using this and the continuity of the inner product, we therefore have
0 = lim hT gn , hi = hh, hi = khk2 .

n→∞
Hence h = 0, so T gn → h = 0. Since every subsequence has a subsequence for

which this holds, it follows from Exercise 1.6 that T en → 0.
B.3 The operator is
∞
X
Mλ f = λn hf, en i en ,
n=1
where {en }n∈N is an orthonormal basis for the separable Hilbert space H.
Since λ = (λn ) is bounded, we know that Mλ is bounded operator on H.
(a) ⇒. Suppose that λn → 0 as n → ∞. Define
N
X
TN f = λn hf, en i en .
n=1
This operator is linear, bounded, and has finite rank since range(TN ) ⊆
span{e1 , . . . , eN }. Further, TN is a good approximation to Mλ , because by
using the Plancherel Equality we can compute that
∞ 2
X
2
k(Mλ − TN )f k = λn hf, en i en

n=N +1
∞
X
= |λn |2 |hf, en i|2
n=N +1
∞
X
≤ sup |λn |2 |hf, en i|2
n>N n=N +1

≤ sup |λn |2 kf k2 .
n>N
It follows that TN converges to L in operator norm:

2 2
lim kL − TN k ≤ lim sup |λn | = lim sup |λn |2 = 0.
N →∞ N →∞ n>N N →∞
Since each TN is compact, we conclude that Mλ is compact as well.

⇐. If Mλ is compact, then Exercise B.2 implies that Mλ en → 0. However,
Mλ en = λn en , so
|λn | = kλn en k = kMλ en k → 0.
Hence λ ∈ c0 .
(b) Suppose that λ ∈ c0 and λn 6= 0 for every n. Exercise 1.66 shows that
range(Mλ ) is a dense but proper subspace of H. Let D = {x ∈ H : kxk ≤ 1},
and suppose that Mλ (D) was closed. Choose any y ∈ H. Since range(Mλ ) is
dense, we can find xn ∈ H such that Mλ xn → y. Since {xn }n∈N is bounded,
there is some r > 0 such that
Mλ xn ∈ Mλ (r D) ∈ r Mλ (D)
for every n. Since r Mλ (D) is closed, this implies that y ∈ r Mλ (D). Hence
y = Mλ x for some x with kxk ≤ r. Thus y ∈ range(Mλ ). Hence Mλ is
surjective, which is a contradiction. Therefore Mλ (D) cannot be closed.
B.4 (a) Let E = {en } be an orthonormal basis for H and let F = {fn } be
an orthonormal basis for K. Then, by the Plancherel Equality and Tonelli’s
Theorem for series,
X XX
kT emk2 = |hT em , fn i|2
m m n
XX X
= |hem , T ∗ fn i|2 = kT ∗ fn k2 .
n m n
These quantities may be infinite, but they are either all infinite, or all finite
and equal. THus, no matter what orthonormal bases we choose, we have
X X
kT em k2 = kT ∗ fn k2 .
m n
P
Fixing F, we see that kT kHS = kT emk isPindependent of the choice of
orthonormal basis for H. Likewise, kT ∗ kHS = kT fnk is independent of the
∗
choice of orthonormal basis for K, and furthermore kT P kHS = kT kHS.
Also, if x ∈ H then, since T is continuous, T x = hx, en i T en . Hence
X
kT xk ≤ |hx, en i| kT en k
n
X 1/2 X 1/2
≤ hx, en i|2 kT enk2 = kxk kT kHS,
n n
so kT k ≤ kT kHS.
(b) We have by definition that 0 ≤ kT kHS < ∞ for each T ∈ B2 (H, K).
If kT kHS = 0, then we must have kT enk = 0 for every element of an
orthonormal basis {en }n∈N . By linearity and continuity, we must then have
T f = 0 for every f ∈ H, so T = 0.
Homogeneity follows from the homogeneity of the Hilbert space and ℓ2
norms, i.e., kcT kHS = |c| kT kHS.
The Triangle inequality likewise follows from the corresponding inequali-
ties for H and for ℓ2 . Specifically, if {en }n∈N is any orthonormal basis for H,
then
X ∞ 1/2
kT + U kHS = kT en + U en k2
n=1
X
∞ 1/2
2
≤ kT enk + kU en k
n=1
X
∞ 1/2 X
∞ 1/2
2 2
≤ kT enk + kU en k
n=1 n=1
= kT kHS + kU kHS .
This shows that k · kHS is a norm. Now we must show that B2 (H, K) is
complete with respect to k·kHS. Suppose that {Tn }n∈N is a Cauchy sequence in
B2 (H, K). Then by part (a), kTm −Tnk ≤ kTm −Tn kHS , so {Tn }n∈N is Cauchy
in operator norm. Since B(H, K) is complete, there exists T ∈ B(H, K) such
that Tn → T in operator norm. We must show that Tn → T in Hilbert–
Schmidt norm.
If we fix ε > 0, then there exists an N such that kTm − Tn kHS ≤ ε for all
m, n > N. Choose any orthonormal basis {en }n∈N for H. Since Tn → T in
operator norm, for any fixed k we have Tn ek → T ek as n → ∞. Therefore, if
n > N then we have for each M > 0 that
M
X M
X
kT ek − Tn ek k2 = lim kTm ek − Tn ek k2
m→∞
k=1 k=1
M
X
= lim kTm ek − Tn ek k2
m→∞
k=1
∞
X
≤ lim sup kTm ek − Tn ek k2
m→∞
k=1
= lim sup kTm − Tn k2HS ≤ ε2 .

m→∞
Since this is true for every M, we conclude that

∞
X
kT − Tn k2HS = kT ek − Tn ek k2 ≤ ε2
k=1
for every n > N. Hence T ∈ B2 (H, K) and kT − Tn kHS → 0, so B2 (H, K) is

complete.
Now fix T, U ∈ B2 (H, K) and any orthonormal basis {en } for H. Then we
have
X X
|hT en , U en i| ≤ kT enk kU en k
n n
X 1/2 X 1/2
2 2
≤ kT enk kU en k
n n
≤ kT kHS kU kHS,
P
so hT, U iHS = hT en , U en i is well defined. Further, if {fn } is any orthonor-
mal basis for K then, applying Fubini’s Theorem, we have
X XX
hT em , U em i = hT em , fn i hfn , U em i
m m n
XX
= hem , T ∗ fn i hU ∗ fn , em i
m n
XX
= hU ∗ fn , em i hem , T ∗ fn i
n m
X
= hU ∗ fn , T ∗ fn i.
n
Fubini is justified since

XX
|hT em , fn i hfn , U em i|
m n
X X 1/2 X X 1/2
≤ |hT em , fn i|2 |hfn , U em i|2
m n m n
X 1/2 X 1/2
= kT emk2 kU em k2
m m
≤ kT kHS kU kHS .
In any case, hT, U iHS is well defined and independent of the choice of or-
thonormal basis {en }.
Finally, h·, ·iHS satisfies all the properties of an inner product, so since we
have already proved that B2 (H, K) is complete, it is a Hilbert space.
(c) If T ∈ B2 (H, K) and B ∈ B(K) then
∞
X ∞
X
kBT k2HS = kBT en k2 ≤ kBk2 kT en k2 = kBk2 kT k2HS.
n=1 n=1
If A ∈ B(H) then, since T ∗ ∈ B2 (H), we have by the above work that A∗ T ∗ ∈

B2 (K, H), and
kT AkHS = k(T A)∗ kHS = kA∗ T ∗ kHS ≤ kA∗ k kT ∗kHS = kAk kT kHS.
(d) Suppose that L ∈ B(H, K) has finite rank. Then M = range(L) is

finite dimensional and hence closed. Therefore we can find an orthonormal
basis {en }n∈N for K such that {e1 , . . . , ed } is an orthonormal basis for M,
where d = dim(M ). Then
X d
X d
X
kLen k2 = kLen k2 ≤ kLk2 ken k2 = d kLk2 < ∞.
n n=1 n=1
Hence L is Hilbert–Schmidt.
Now suppose that T ∈ B2 (H, K) is given, and let {en } be an orthonormal
basis for H. For each N ∈ N define
N
X
TN x = hx, en i T en , x ∈ H.
n=1
Then TN is continuous and has finite rank, so is compact by Theorem B.5.

Further, (
T em , m = 1, . . . , N,
T N em =
0, m > N.
Therefore, since T is a Hilbert–Schmidt operator.
∞
X ∞
X
k(T − TN )k2HS = kT en − TN en k2 = kT en k2 → 0 as N → ∞.
n=1 n=N +1
Since the operator norm is dominated by the Hilbert–Schmidt norm, we con-

clude that the compact operators TN converge to T in both operator norm
and Hilbert–Schmidt norm. Therefore T is compact by Theorem B.5, and
B00 (H, K) is dense in B2 (H, K) with respect to both operator norm and
Hilbert–Schmidt norm.
B.5 (a) ⇒. If T has rank one then range(T ) = span{y} for some unit vector y.
Then T x = hT x, yi y = hx, T ∗ yi y = (T ∗ y ⊗ y)(x).
(b) ⇒. Suppose T has finite rank, so range(T ) is a finite-dimensional sub-
space of K. Every finite-dimensional subspace is closed, so we can find a finite
orthonormal basis {yk }N k=1 for range(T ). Therefore, if x ∈ H then we can
express T x in terms of this orthonormal basis:
N
X N
X
Tx = hT x, yk i yk = hx, T ∗ yk i yk
k=1 k=1
N
X N
X
= hx, xk i yk = (xk ⊗ yk )(x),
k=1 k=1
where xk = T ∗ yk .
⇐. If T has this form, then range(T ) ⊆ span{y1 , . . . , yN }, so T has finite
rank.
(c) By part (b),
B00 (H, K) = span{x ⊗ y : x ∈ H, y ∈ K}.
Since B00 (H, K) is dense in B2 (H, K) we conclude that {x⊗y : x ∈ H, y ∈ K}

is complete.
(d) Fix an orthonormal basis for H. Then by the Parseval Equality,
X

hx1 ⊗ y1 , x2 ⊗ y2 iHS = (x1 ⊗ y1 )(en ), (x2 ⊗ y2 )(en )
n
X

= hen , x1 i y1 , hen , x2 i y2
n
X
= hen , x1 i hx2 , en i hy1 , y2 i
n
X
= hy1 , y2 i hx2 , en i hen , x1 i
n
= hx2 , x1 i hy1 , y2 i.
(f) By part (e) we have

X
T = hT, em ⊗ fn i (em ⊗ fn ).
m,n
However, the proof of part (e) also shows that

X
T = hT em , fn i (em ⊗ fn ),
m,n
so we conclude that hT, em ⊗ fn i = hT em , fn i.

B.6 (a) We have

kx ⊗ yk2HS = x ⊗ y, x ⊗ y HS = hx, xi hy, yi = kxk2 kyk2 .
(b) For u ∈ H we have

(ax) ⊗ y (u) = hu, axi y = ā hu, xi y = ā (x ⊗ y)(u).
(c) For u ∈ H we have

(x+w)⊗y (u) = hu, x+wi y = hu, xi y+hu, wi y = (x⊗y)(u)+(w⊗y)(u).
(d) We have
kx ⊗ y − w ⊗ zkHS ≤ kx ⊗ y − w ⊗ ykHS + kw ⊗ y − w ⊗ zkHS

= k(x − w) ⊗ ykHS + kw ⊗ (y − z)kHS
= kx − wk kyk + kwk ky − zk.
B.7 Fix x ∈ H and y ∈ K. If z ∈ K and w ∈ H then

(x⊗y)(w), z = hw, xi y, z = hw, xi hy, zi = w, hz, yi x = w, y⊗x .
Thus the adjoint of x ⊗ y is y ⊗ x.

If T is compact then there exist finite-rank operators P
TN such that TN → T
M
in operator norm. By Theorem B.10 we have that TN = k=1 xk ⊗yk for some
P M
xk ∈ H and yk ∈ K. Hence TN∗ = k=1 yk ⊗xk has finite rank. Also, TN∗ → T ∗
since the operator norm is preserved by adjoints. Therefore T ∗ is compact,
and the converse is symmetrical.
B.8 Since
Z Z
hem ⊗ fn , ej ⊗ fk i = (em ⊗ fn )(x, y) (ej ⊗ fk )(x, y) dx dy
F E
Z Z
= em (x) fn (y) ej (x) fk (x) dx dy
F E
Z Z
= ej (x) em (x) fn (y) fk (y) dy dx
E F
Z
= ej (x) em (x) hfn , fk i dx
E
= hej , em i hfn , fk i = δmj δkn ,
we have that {em ⊗ fn }m,n∈N is an orthonormal system in L2 (E × F ). We

must show that this orthonormal system is complete.
First Proof. Suppose that F ∈ L2 (R2 ) is such that hF, em ⊗ fn i = 0 for
every m and n. Then
Z Z
0 = hF, em ⊗ fn i = F (x, y) (em ⊗ fn )(x, y) dx dy
F E
Z Z
= F (x, y) em (x) fn (y) dx dy
F E
Z Z
= F (x, y) em (x) dx fn (y) dy
F E
Z
= hm (y) fn (y) dy = hfn , hm i,
F
where Z
hm (y) = F (x, y) em (x) dx.
E
The function hm belongs to L2 (F ) because, by the Cauchy–Bunyakovski–

Schwarz inequality,
Z Z Z 2

2
khm kL2 = 2
|hm (y)| dy =
F (x, y) em (x) dx dy
F F E
Z Z Z
≤ |F (x, y)|2 dx |em (x)|2 dx dy
F E E
Z Z
= |F (x, y)| dx kem k2 dy
2
F E
Z Z
= |F (x, y)|2 dx dy = kF k2L2 < ∞.
F E
Since hm is orthogonal to every function en and {en }n∈N is complete in L2 (E),

this implies
RR that hm = 0 a.e.
Since |F (x, y)|2 dx dy < ∞, Fubini’s Theorem implies that F y (x) =
F (x, y) is a measurable, square-integrable function of x for almost every y.
For these y, we have
Z
hF y , em i = F (x, y) em (y) dy = hm (y).
Since hm = 0 a.e., we conclude that hF y , em i = 0 for almost every y. As

{em }m∈N is complete, this implies that for almost every y we have F y = 0 a.e.
Hence F = 0 a.e.
Second Proof. Instead of showing that only the zero function is orthogonal
to every function emn , we will show that the finite linear span of the emn is
dense.
Choose any function F ∈ L2 (R2 ). By Lemma A.29, we can approximate
F by a step function
XN
H = ck χEk ×Fk
k=1
where Ek and Fk have finite measure. Since χEk , χFk ∈ L2 (R), we can write
X X
χE k = amk em and χFk = bnk fn
m∈N n∈N
for some scalars amk and bnk , where the series converge in L2 -norm. Hence,
N
X
H(x, y) = ck χEk (x) χFk (y)
k=1
N
X X X
= ck amk em (x) bnk fn (y)
k=1 m n
X XX
N
= ck amk bnk em (x) fn (y)
m n k=1
XX
= dmn (em ⊗ fn )(x, y),
m n
PN
where dmn = k=1 ck amk bnk are some new scalars (note that this is a finite
sum, so is well defined). The partial sums of this series converge to H and lie
in span{em ⊗ fn }m,n∈N . Therefore F can be approximated arbitrarily closely
by functions in this span. Therefore {em ⊗ fn }m,n∈N is complete.
A third alternative is to show that the Plancherel Equality holds. This
approach is similar to what is done in the proof of Exercise 8.36.
B.9 (a) Fix U ∈ B(H) and V ∈ B(K), and choose g ∈ H and h ∈ K. Since
g ⊗ h ∈ B(H, K) is a bounded rank one operator, we know that V (g ⊗ h)U ∗ ∈
B(H, K) is a well-defined bounded operator, and given f ∈ H we have

V (g ⊗ h)U ∗ f = V (g ⊗ h)(U ∗ f )

= V hU ∗ f, gi h
= hf, U gi V h
= (U g ⊗ V h)(f ).
Hence U (g ⊗ h)V ∗ = U g ⊗ V h.
(b) Define T (F ) = V F U ∗ for F ∈ B2 (H, K). Then T (F ) ∈ B2 (H, K) by
Theorem B.8(c), and that theorem also implies that
kT (F )kHS = kV F U ∗ kHS ≤ kV k kF kHS kU k.
Therefore T is a bounded mapping of H ⊗ K = B2 (H, K) into itself, and

kT k ≤ kV k kU k.
Suppose that S was another bounded operator that satisfied S(g ⊗ h) =
U g ⊗ V h for all g ∈ H and h ∈ K. Then S and T are bounded linear operators
that agree on {g ⊗ h : g ∈ H, h ∈ K}. Since this set is complete, S and T
must agree on all of H ⊗ K. Hence T is unique.
(c) Part (b) shows that there exist unique operators U1 ⊗ V1 , U2 ⊗ V2 , and
U1 U2 ⊗ V1 V2 . Given g ∈ H and h ∈ K we have

(U1 ⊗ V1 )(U2 ⊗ V2 ) (g ⊗ h) = (U1 ⊗ V1 ) (U2 ⊗ V2 )(g ⊗ h)
= (U1 ⊗ V1 )(U2 g ⊗ V2 h)
= (U1 U2 g ⊗ V1 V2 h)
= (U1 U2 ⊗ V1 V2 )(g ⊗ h).
By the uniqueness property established in part (b), we conclude that
(U1 ⊗ V1 )(U2 ⊗ V2 ) = U1 U2 ⊗ V1 V2 .
(d) Suppose that U and V are topological isomorphisms. Then by part (c),
(U −1 ⊗ V −1 )(U ⊗ V ) = U −1 U ⊗ V −1 V = I ⊗ I = I,
and similarly (U ⊗ V )(U −1 ⊗ V −1 ) = I. Hence U ⊗ V has a continuous inverse,

so is a topological isomorphism.
(e) Let L(k) = Lk be the isometric isomorphism that maps L2 (E × F )
onto L2 (E) ⊗ L2 (F ). Given U ∈ B(L2 (E)) and V ∈ B(L2 (F )), let
S(k) = L−1 (U ⊗ V ) = L−1 (V Lk U ∗ ).

This operator has the required properties. In particular,

S(g ⊗ h) = L−1 V (Lg⊗h )U ∗

= L−1 V (g ⊗ h)U ∗

= L−1 U g ⊗ V h = U g ⊗ V h.
http://www.springer.com/978-0-8176-4686-8

A Basis Theory Primer Expanded Edition Solutions Manual - Heil Problems

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Basis Theory Primer Expanded Edition Solutions Manual - Heil Problems

Uploaded by

Copyright:

Available Formats

Christopher Heil

A Basis Theory Primer:

These are my solutions to the exercises from “A Basis Theory Primer.” Of

Detailed Solutions to Exercises from Chapter 1

1.1 Since k · k is a norm, we must have λ = k1k 6= 0. Then given any x ∈ F,

kxn k = kxn − xN + xN k ≤ kxn − xk + kxN k ≤ 1 + kxN k.

Hence for any n we have

kxk = k(x − y) + yk ≤ kx − yk + kyk,

so kxk − kyk ≤ kx − yk.

(e) If xn → x and yn → y, then

k(x + y) − (xn + yn )k = k(x − xn ) + (y − yn )k ≤ kx − xn k + ky − yn k → 0.

(f) Suppose xn → x and cn → c. Then C = sup |cn | < ∞, so

kcx − cn xn k ≤ kcx − cn xk + kcn x − cn xn k

1.3 (a) Suppose first that 1 ≤ p < ∞ and q = ∞. Given x = (x1 , . . . , xd ) ∈

Conversely, kxk∞ = |xk | for some particular k, so we have

Hence | · |p and | · |∞ are equivalent norms on Fd .

|ck (xm ) − ck (xn )| = |ck (xm − xn )| ≤ kxm − xn k.

Hence {xn } is Cauchy.

∀ ε > 0, ∃ N > 0, ∀ m, n ≥ N, d(fm , fn ) < ε.

Hence, if x, y ∈ ℓp (I), then

This establishes the Triangle Inequality.

For the nonnormalized case, if x, y 6= 0, equality holds in Hölder’s Inequality if

If either x = 0 or y = 0 then equality holds trivially in Hölder’s Inequality, so

Lemma. If E ⊆ R is measurable and 0 < |E| < ∞, then there exists a

A similar argument shows that |Et | is a continuous function of t. Therefore

/ Lq (E). On the other hand,

where the limit exists because kxkq is finite and nonzero.

countable sets is countable, at most countably many components of x can be

some x ∈ X such that xn → x. However, M is closed, so this implies that

inf{M : f (x) ≤ M a.e.} = inf{M : f (x) ≤ M for every x} = sup |f (x)|,

so the uniform and L∞ norms agree for functions in Cb (R).

so {yN }N ∈N is a Cauchy sequence of scalars and therefore converges, say to y.

Since ε is arbitrary, we conclude that y = limk→∞ x(k), so x ∈ c. Thus c is

Then xN ∈ c00 , and

lim kx − xN kℓ∞ = lim sup |x(k)| = lim sup |x(k)| = 0.

By part (b) we know thatP kx − xN kℓ∞ → 0 as N → ∞. Since P the xN is the

so kf − fn k∞ ≤ ε for all n > N. Also, kf k∞ ≤ kf − fn k∞ + kfn k∞ , so f is

Each gN belongs to Cc (R), and

f (x + 1) = lim fn (x + 1) = lim fn (x) = f (x).

Hence f is 1-periodic, so f ∈ C(T) and therefore C(T) is closed in Cb (R).

Letting m → ∞, we conclude that

This is valid for x, y ∈ R as long as n > N.

Therefore f is differentiable and f ′ (x) = 0 for every x, so f is constant.

|f (x) − f (y)| = |f ′ (c)| |x − y| ≤ kf ′ k∞ |x − y|.

If f ′ is bounded, then it follows that f is Lipschitz.

Homogeneity and the Triangle Inequality follow directly, so k · kC α is a

|fn (x) − fn (y)|

Then we have both

n=1 x6=y |x − y|α

(a) Given {xn } ∈ C, we have lim kxn − xn k = 0, so {xn } ∼ {xn }.

kxn − zn k ≤ kxn − yn k + kyn − zn k,

we have {xn } ∼ {zn }.

lim sup kA − Ym kXe = lim sup lim kxn − xm k = 0.

YM = T (yM ) = [{yM , yM , . . . }].

Then, rewriting a line from above, we have

Hence for N ≥ N0 we have

e Also, if N ≥ N0 and N ≥ 1/ε, then

(e) Suppose that Y is a Banach space and U : X → Y is a linear isometry

V (f + h) = lim U (xn + hn ) = lim U (xn ) + U (hn ) = V (f ) + V (h).

Similarly, V (cf ) = cV (f ), so V is linear.

kf k = lim kT (xn )k = lim kU (xn )k = kV (f )k.

is countable. Suppose f ∈ C0 (R) and ε > 0 is fixed. Then there exists an M

Now, applying the Mean Value Theorem to the function xk , given h ∈ R