1.

4 Convergence, Cauchy Sequence, Completeness 2S

R; (b) the set of all rational numbers onR; (c) the disks {z Ilzl< l}cC
and {z Ilzl~ l}cC.

12. (Space B[a, b]) Show that B[a, b], a < b, is not separable. (Cf.
1.2-2.)

13. Show that a metric space X is separable if and only if X has a

countable subset Y with the following property. For every E > 0 and
every x E X there is ayE Y such that d(x, y) < E.

14. (Continuous mapping) Show that a mapping T: X ---- Y is continu-

ous if and only if the inverse image of any closed set Me Y is a closed
set in X.

15. Show that the image of an open set under a continuous mapping need
not be open.

1.4 Convergence, Cauchy Sequence, Completeness

We know that sequences of real numbers play an important role in

calculus, and it is the metric on R which enables us to define the basic
concept of convergence of such a sequence. The same holds for
sequences of complex numbers; in this case we have to use the metric
on the complex plane. In an arbitrary metric space X = (X, d) the
situation is quite similar, that is, we may consider a sequence (x,.) of
elements Xl, X2, ••• of X and use the metric d to define convergence in
a fashion analogous to that in calculus:

1.4-1 Definition (Convergence of a sequence, limit). A sequence (x,.)

in a metric space X = (X, d) is said to converge or to be convergent if
there is an X E X such that

lim d(x,., x) = o.
n~=

x is called the limit of (xn) and we write

limx,.=x
n-->=

or, simply,
32 Metric Spaces

2. If (x,.) is Cauchy and has a convergent subsequence, say, x... --- x,

show that (x,.) is convergent with the limit x.

3. Show that x,. --- x if and only if for every neighborhood V of x there
is an integer no such that Xn E V for all n > no.

4. (Boundedness) Show that a Cauchy sequence is bounded.

5. Is boundedness of a sequence in a metric space sufficient for the

sequence to be Cauchy? Convergent?

6. If (x,.) and (Yn) are Cauchy sequences in a metric space (X, d), show
that (an), where an = d(x,., Yn), converges. Give illustrative examples.

7. Give an indirect proof of Lemma 1.4-2(b).

8. If d 1 and d 2 are metrics on the same set X and there are positive
numbers a and b such that for all x, YE X,

ad 1 (x, y);a d 2 (x, y);a bd 1 (x, Y),

show that the Cauchy sequences in (X, d 1 ) and (X, dz) are the same.

9. Using Prob. 8, show that the metric spaces in Probs. 13 to 15, Sec. 1.2,
have the same Cauchy sequences.

10. Using the completeness of R, prove completeness of C.

1.5 Examples. Completeness Proofs

In various applications a set X is given (for instance, a set of sequences

or a set of functions), and X is made into a metric space. This we do by
choosing a metric d on X. The remaining task is then to find out
whether (X, d) has the desirable property of being complete. To prove
completeness, we take an arbitrary Cauchy sequence (xn) in X and
show that it converges in X. For different spaces, such proofs may vary
in complexity, but they have approximately the same general pattern:
(i) Construct an element x (to be used as a limit).
(ii) Prove that x is in the space considered.
(iii) Prove convergence Xn ~ x (in the sense of the metric).

We shall present completeness proofs for some metric spaces

which occur quite frequently in theoretical and practical investigations.
1.5 Examples. Completeness Proofs 39

where a", = 1/2 + l/m. Hence for every x EX,

d(x,.., x) ~ f1x,..(t)- x(t)1 dt

= 11/2 Ix(t)1 dt+

ia~1x,..(t)-x(t)1 dt+
il 11-x(t)1 dt.
o 1/2 ~

Since the integrands are nonnegative, so is each integral on the right.

Hence d(x,.., x) ~ 0 would imply that each integral approaches zero
and, since x is continuous, we should have

x(t) = 0 if t E [0, t), x(t) = 1 if t E (!. 1].

But this is impossible for a continuous function. Hence (x,..) does not
converge, that is, does not have a limit in X. This proves that X is not
complete. •

Problems

1. Let a, bE R and a < b. Show that the open interval (a, b) is an

incomplete subspace of R, whereas the clOsed interval [a, b] is com-
plete.

2. Let X be the space of all ordered n-tuples x = (~h ... '~n) of real
numbers and

where y = ('1JJ. Show that (X, d) is complete.

3. Let Me too be the subspace consisting of all sequences x = (~j) with at

most finitely many nonzero terms. Find a Cauchy sequence in M which
does not converge in M, so that M is not complete.

4. Show that M in Prob. 3 is not complete by applying Theorem 1.4-7.

5. Show that the set X of all integers with metric d defined by

d(m, n) = 1m - nl is a complete metric space.
2.2 Normed Space. Banach Space 6S

IIXII.,= 1
IIxl~ = 1
IIXII2 = 1
IIxll1 = 1

Fig. 16. Unit spheres in Prob. 10

11. (Convex set, segment) A subset A of a vector space X is said to be

convex if x, YEA implies

M={ZEXlz=ax+(1-a)y, O~a~1}cA.

M is called a closed segment with boundary points x and y; any other

Z E M is called an interior point of M. Show that the closed unit ball

B(O; 1) ={x E X Illxll~ 1}

in a normed space X is convex.

(a) Convex (b) Not convex

FlI.17. Illustrative examples of convex and nonconvex sets (d. Prob. 11)
70 Normed Spaces. Banach Spaces

Similarly we define the product ax E X of a scalar a and x to be the

equivalence class for which (axn) is a representative. Again, this
definition is independent of the particular choice of a representative of
x. The zero element of X is the equivalence class containing all Cauchy
sequences which converge to zero. It is not difficult to see that those
two algebraic operations have all the properties required by the
definition, so that X is a vector space. From the definition it follows
that on W the operations of vector space induced from X agree with
those induced from X by means of A.
Furthermore, A induces on W a norm "·IIt. whose value at every
9 = Ax E W is 11911t = Ilxll. The corresponding metric on W is the restric-
tion of d to W since A is isometric. We can extend the norm 11·111 to X
by setting Ilxllz = d(O, x) for every X EX. In fact, it is obvious that II· liz
satisfies (Nl) and (N2) in Sec. 2.2, and the other two axioms (N3) and
(N4) follow from those for 11·111 by a limit process. •

Problems

1. Show that C c [00 is a vector subspace of [00 (cf. 1.5-3) and so is Co, the
space of all sequences of scalars converging to zero.

2. Show that Co in Prob. 1 is a closed subspace of [00, so that Co is complete

by 1.5-2 and 1.4-7.

3. In [00, let Y be the subset of all sequences with only finitely many
nonzero terms. Show that Y is a subspace of [00 but not a closed
subspace.

4. (Continuity of vector space operations) Show that in a normed space

X, vector addition and multiplication by scalars are continuous opera-
tions with respect to the norm; that is, the mappings defined by
(x, y) ~ x+y and (a, x) ~ ax are continuous.

5. Show that x" ~ x and Yn ~ Y implies Xn + Yn ~ X + y. Show that

an ~ a and Xn ~ x implies anx" ~ ax.

6. Show that the closure Y of a subspace Y of a normed space X is again

a vector subspace.
76 Normed Spaces. Banach Spaces

This theorem is of considerable practical importance. For instance,

it implies that convergence or divergence of a sequence in a finite
dimensional vector space does not depend on the particular choice of a
norm on that space.

Problems
1. Give examples of subspaces of [00 and e which are not closed.
2. What is the largest possible c in (1) if X = R2 and Xl = (1, 0),
X2 = (0, I)? If X = R3 and Xl = (1, 0, 0), X 2 = (0, 1,0), X3 = (0, 0, I)?

3. Show that in Def. 2.4-4 the axioms of an equivalence relation hold (d.
Al.4 in Appendix 1).

4. Show that equivalent norms on a vector space X induce the same

topology for X.

5. If II· I and II· lin are equivalent norms on X, show that the Cauchy
sequences in (X, I ·11) and (X, 11·110) are the same.

6. Theorem 2.4-5 implies that I . liz and 11·1100 in Prob. 8, Sec. 2.2, are
equivalent. Give a direct proof of this fact.

7. Let II· liz be as in Prob. 8, Sec. 2.2, and let 11·11 be any norm on that
vector space, call it X. Show directly (without using 2.4-5) that there is
°
a b > such that Ilxll ~ b Ilxllz for all x.

8. Show that the norms I ·111 and II· liz in Prob. 8, Sec. 2.2, satisfy

1
~ Ilxllt ~ 114 ~ Ilxlb·

9. If two norms II . II and II . 110 on a vector space X are equivalent, show

°
that (i) Ilxn - xll---- 0 implies (ii) Ilxn - xllo ---- (and vice versa, of
course).

10. Show that all complex m x n matrices A = (ajk) with fixed m and n
constitute an mn-dimensional vector space Z. Show that all norms on
Z are equivalent. What would be the analogues of II· III> I . 112 and I . 1100
in Prob. 8, Sec. 2.2, for the present space Z?
2.5 Compactness and Finite Dimension 81

Compact sets are important since they are "well-behaved": they

have several basic properties similar to those of finite sets and not
shared by noncompact sets. In connection with continuous mappings a
fundamental property is that compact sets have compact images, as
follows.

2.5-6 Theorem (Continuous mapping). Let X and Y be metric spaces

and T: X - Ya continuous mapping (cf. 1.3-3). Then the image of
a compact subset M of X under T is compact.

Proof. By the definition of compactness it suffices to show that

every sequence (Yn) in the image T(M) c Y contains a subsequence
which converges in T(M). Since Yn E T(M), we have Yn = Tx,. for some
Xn EM Since M is compact, (xn) contains a subsequence (xn,J which
converges in M. The image of (x n.) is a subsequence of (Yn) which
converges in~(M) by 1.4-8 because T is continuous. Hence T(M) is
compact. I

From this theorem we conclude that the following property,

well-known from calculus for continuous functions, carries over to
metric spaces.

2.5-7 Corollary (Maximum and minimum). A continuous mapping T

of a compact subset M of a metric space X into R assumes a maximum
and a minimum at some points of M.

Proof. T(M) c R is compact by Theorem 2.5-6 and closed and

bounded by Lemma 2.5-2 [applied to T(M)], so that inf T(M)E T(M),
sup T(M) E T(M), and the inverse images of these two points consist of
points of M at which Tx is minimum or maximum, respectively. I

Problems
1. Show that R n and en are not compact.
2. Show that a discrete metric space X (cf. 1.1-8) consisting of infinitely
many points is not compact.

3. Give examples of compact and noncompact curves in the plane R2.

2.7 Bounded and Continuous Linear Operators 101

and let n ~ 00. Then TX n ~ Y= Tx. Since x ~ Ilxll defines a

continuous mapping (cf. Sec. 2.2), we thus obtain

IITxll~IITlIllxll.

Hence T is bounded and Iltll~IITII. Of course, IITII~IITII because the

nonn, being defined by a supremum, cannot decrease in an extension.
Together we have IltlL= IITII· I

Problems
1. Prove (7).

2. Let X and .y be nonned spaces. Show that a linear operator

T: X~ Y is bounded if and only if T'maps bounded sets in X into
bounded sets in Y.

3. If Tot- 0 is a bounded linear operator, show that for any x E~(T) such
that Ilxll < 1 we have the strict inequality IITxl1 < IITII.

4. Give a direct proof of 2.7-9(b), without using 2.7-9(a).

5. Show that the operator T: r ~ r defined by y= ('1Ji) = Tx, '1Ji = Q/ j,

x = (gj), is linear and bounded.

6. (Range) Show that the range meT) of a bounded linear operator

T: X ~ Y need not be closed in Y. Hint. Use Tin Prob. 5.

7. (Inverse operator) Let T be a bounded linear operator from a nonned

space X onto a nonned space Y. If there is a positive b such that

IITxll;;;;bllxll for all x EX,

show that then r1: Y ~ X exists and is bounded.

8. Show that the inverse r l : meT) ~ X of a bounded linear operator

T: X ~ Y need not be bounded. Hint. Use Tin Prob. 5.

9. Let T: C[O, 1] ~C[O, 1] be defined by

yet) = f X(T) dT.

Find meT) and r1: m(T)~ C[O, 1]. Is r 1 linear and bounded?
110 Normed Spaces. Banach Spaces

3. Find the norm of the linear functional f defined on C[-1,1] by

f(x) = r x(t) dt - f x(t) dt.

4. Show that
fl(x) = max x(t)
'eJ

J=[a, b]

f2(X) = min X(t)

tel '

define functionals on C[a, b]. Are they linear? Bounded?

5. Show that on any sequence space X we can define a linear functional f

by setting f(x) = ~n (n fixed), where x = (~i)' Is f bounded if X = ZOO?

6. (Space e'[a, b]) The space Cl[a, b] or C[a, b] is the normed space of all
continuously differentiable functions on J = [a, b] with norm defined by

Ilxll = max Ix(t)1 + max Ix'(t)l.

teJ teJ

Show that the axioms of a norm are satisfied. Show that f(x) = x'(c),
c = (a + b)/2, defines a bounded linear functional on C[ a, b]. Show
that f is not bounded, considered as a functional on the subspace of
C[ a, b] which consists of all continuously differentiable functions.
7. If f is a bounded linear functional on a complex normed space, is f
bounded? Linear? (The bar denotes the complex conjugate.)

8. (Null space) The. null space N(M*) of a set M* c X* is defined to be

the set of all x E X such that f(x) = 0 for all f E M*. Show that N(M*) is
a vector space.

9. Let f¥- 0 be any linear functional on a vector space X and Xo any fixed
element of X - .N'(f), where .N'(f) is the null space of f. Show that any
x E X has a unique representation x = axo + y, where y E .N'(!).

10. Show that in Prob. 9, two elements Xl> X2 E X belong to the same
element of the quotient space X/.N'(f) if and only if f(Xl) = f(X2); show
that codim.N'(f) = 1. (Cf. Sec. 2.1, Prob. 14.)
.
11. Show that two linear functionals fl ¥- 0 and f2 ¥- 0 which are defined on
the same vector space and have the same null space are proportional.
3.1 Inner Product Space. Hilbert Space 135

Problems
1. Prove (4).

2. (Pythagorean theorem) If x.L y in an inner product space X, show

that (Fig. 24)

Extend the formula to m mutually orthogonal vectors.

Fig. 24. Illustration of the Pythagorean theorem in the plane

3. If X in Prob. 2 is real, show that, conversely, the given relation implies

that x.L y. Show that this may not hold if X is complex. Give examples.

4. If an inner product space X is real, show that the condition Ilxll = Ilyll
implies (x + y, x - y) = O. What does this mean geometrically if X = R2?
What does the condition imply if X is complex?

5. (Appolonius' identity) Verify by direct calculation that for any ele-

ments in an inner product space,

Show that this identity can also be obtained from the parallelogram
equality.

6. Let x~ 0 and y~ O. (a) If x.L y, show that {x, y} is a linearly independ-

ent set. (b) Extend the result to mutually orthogonal nonzero vectors

7. If in an inner product space, (x, u)=(x, v) for all x, show that u = v.

8. Prove (9).

9. Prove (10).
140 Inner Product Spaces. Hilbert Spaces

isometry T: H - - H. Reasoning as in the case of A, we conclude

that T must be an isomorphism of the Hilbert space H onto the
Hilbert space H. •

A subspace Y of an inner product space X is defined to be a

vector subspace of X (cf. Sec. 2.1) taken with the inner product on X
restricted to Y x Y. \"
Similarly, a subspace Y of a Hilbert space H is defined to be a
subspace of H, regarded as an inner product space. Note that Y need
not be a Hilbert space because Y may not be complete. In fact, from
Theorems 2.3-1 and 2.4-2 we immediately have the statements (a) and
(b) in the following theorem.

3.2-4 Theorem (Subspace). Let Y be a subspace of a Hilbert space H.

Then:
(a) Y is complete if and only if Y is closed in H.

(b) If Y is finite dimensional, then Y is complete.

(c) If H is separable, so is Y. More generally, every subset of a

separable inner product space is separable.

The simple proof of (c) is left to the reader.

Problems

1. What is the Schwarz inequality in a 2 or a 3 ? Give another proof of it in

these cases.

2. Give examples of subspaces of f.

3. Let X be the inner product space consisting of the polynomial x = 0

(cf. the remark in Prob. 9, Sec. 2.9) and all real polynomials in t, of
degree not exceeding 2, considered for real t E [a, b], with inner product
defined by (7), Sec. 3.1. Show that X is complete. Let Y consist of all
x E X such that x(a) = O. Is Y a subspace of X? Do all x E X of degree
2 form a subspace of X?

4. Show that y.Lx" and x" - - x together imply x.Ly.

S. Show that for a sequence (x,,) in an inner product space the conditions
Ilx"II--llxll and (x"' x) - - (x, x) imply convergence x. - - x.
150 Inner Product Spaces. Hilbert Spaces

(b) Conversely, suppose that M.L= {O}. If x.L V, then

x.L M, so that x E M.L and x= O. Hence V.L ={O}. Noting that V is a
subspace of H, we thus obtain V = H from 3.3-4 with Y = V. •

Problems
1. Let H be a Hilbert space, Me H a convex subset, and (xn) a sequence
in M such that Ilxnll- d, where d = xEM
inf Ilxll. Show that (xn) converges
in H. Give an illustrative example in R2 or R3.

2. Show that the subset M = {y = ('TIt) IL 'TIj = 1} of complex space en (cf.

3.1-4) is complete and convex. Find the vector of minimum norm in M.

3. (a) Show that the vector space X of all real-valued continuous func-
tions on [-1, 1] is the direct sum of the set of all even continuous
functions and the set of all odd continuous functions on [-1, 1].
(b) Give examples of representations of R3 as a direct sum (i) of a
subspace and its orthogonal complement, (ii) of any complementary
pair of subspaces.

4. (a) Show that the conclusion of Theorem 3.3-1 also holds if X is a

Hilbert space and Me X is a closed subspace. (b) How could we use
Appolonius' identity (Sec. 3.1, Prob. 5) in the proof of Theorem 3.3-1?

5. Let X=R2. Find M-'- if M is (a) {x}, where X=(gl,g2)~0, (b) a

linearly independent set {Xl' X2} c x.

6. Show that Y = {x I x = (Q) E e, g2n = 0, n E N} is a closed subspace of e

and find y-'-. What is Y-'- if Y=span{el,·· ., en}c e, where ej = (8jk )?

7. Let A and B ~A be nonempty subsets of an inner product space X.

Show that

8. Show that the annihilator M-'- of a set M ~ 0 in an inner product space

X is a closed subspace of X.

9. Show that a subspace Y of a Hilbert space H is closed in H if and only

if Y= yll.

10. If M ~ 0 is any subset of a Hilbert space H, show that Mil is the

smallest closed subspace of H which contains M, that is, Mil is
contained in any closed subspace Y c H such that Y ~ M.
3.7 Legendre, Hermite and Laguerre Polynomials 175

Problems
1. If F is an orthonormal basis in an inner product space X, can we
represent every x E X as a linear combination of elements of F? (By
definition, a linear combination consists of finitely many terms.)

2. Show that if the orthogonal dimension of a Hilbert space H is finite, it

equals the dimension of H regarded as a vector space; conversely, if
the latter is finite, show that so is the former.

3. From what theorem of elementary geometry does (3) follow in the case
of Euclidean n-space?

4. Derive from (3) the following formula (which is often called the
Parseval relation).
(x, y) = L (x, ek)(Y, ek).
k

s. Show that an orthonormal family (e.), K E I, in a Hilbert space H is

total if and only if the relation in Prob. 4 holds for every x and Y in H.

6. Let H 'be a separable Hilbert space and M a countable dense subset of

H. Show that H contains a total orthonormal sequence which can be
obtained from M by. the Gram-Schmidt process.

7. Show that if a Hilbert space H is separable, the existence of a total

orthonormal set in H can be proved without the use of Zorn's lemma.

8. Show that for any orthonormal sequence F in a separable Hilbert space

H there is a total orthonormal sequence P which contains F.

9. Let M be a total set in an inner product space X. If (v, x) = (w, x) for

all x E M, show that v = w.

'10. Let M be a subset of a Hilbert space H, and let v, w E H. Suppose that

(v, x) = (w, x) for all x E M implies v = w. If this holds for all v, wE H,
show that M is total in H.

3.7 Legendre, Hermite and Laguerre Polynomials

The theory of Hilbert spaces has applications to various solid topics in
analysis. In the present section we discuss some total orthogonal and
orthonormal sequences which are used quite frequently in connection
3.10 Self-Adjoint, Unitary and Normal Operators 207

Problems
1. If Sand T are bounded self-adjoint linear operators on a Hilbert space
H and a and f3 are real, show that f = as + f3T is self-adjoint.

2. How could we use Theorem 3.10-3 to prove Theorem 3.10-5 for a

complex Hilbert space H?

3. Show that if T: H ---- H is a bounded self-adjoint linear operator, so

is Tn, where n is a positive integer.

4. Show that for any bounded linear operator T on H, the operators

1 1
Tl =2: (T+ T*) and T =-(T-T*)
2 2i

are self-adjoint. Show that

Show uniqueness, that is, Tl + iT2 = SI + iS2 implies SI = Tl and

.82 = T 2 ; here, SI and S2 are self-adjoint by assumption.

5. On e 2 (cf. 3.1-4) let the operator T: e 2 ____ e 2 be defined by

Tx = (gl + ig2, gl - ig2), where x = (gl, g2)' Find T*. Show that we
have T*T= TT* = 2I. Find TI and T2 as defined in Prob. 4.

6. If T: H ---- H is a bounded self-adjoint linear operator and T'" 0,

then Tn", O. Prove this (a) for n = 2,4,8, 16, ... , (b) for every n EN.

7. Show that the column vectors of a unitary matrix constitute an or-

thonormal set with respect to the inner product on en.

8. Show that an isometric linear operator T: H ---- H satisfies T* T = [,

where [ is the identity operator on H.

9. Show that an isometric linear operator T: H ---- H which is not

unitary maps the Hilbert space H onto a proper closed subspace of H.

10. Let X be an inner product space and T: X ---- X an isometric linear

operator. If dim X < 00, show that T is unitary.

11. (Unitary equivalence) Let Sand T be linear operators on a Hilbert

space H. The operator S is said to be unitarily equivalent to T if there