c _2011 by Taejeong Kim 1

Linear Algebra
Introduction
mathematical spaces
The physical space around us is called the 3-dimensional
Euclidean space.
It is commonly dened by real coordinates.
It is equipped with algebraic operations, connectedness, con-
vergence, continuity, distance, length, angles, and orthogo-
nality(perpendicularity).
What if we throw away some of these facilities?
If we throw away all of them, we are left with only a set of
points.
c _2011 by Taejeong Kim 2
set
.
.
.
.
.
.
`
`
``
open
sets
algebraic
operations
vector
(linear)
_
norm
topo-
logical
_
metric
metric
/
/
/
/
/
algebraic operations
normed
linear
_
\
\
\
\
\1
inner
product
completeness
inner
product
\
\
\
\
\1

real
coordinates
Banach
_
inner
product
Hilbert
.
.
.
real
coordinates
Euclid-
ean
metric: generalized distance function
completeness: When x
n
and x
m
gets arbitrarily close as
m, n (Cauchy sequence), x
n
has a limit in the space.
c _2011 by Taejeong Kim 3
The set of rational numbers and an open set are not com-
plete.
We will focus on vector(linear) spaces and inner product spaces.
We will dene vectors and scalars using only sets and alge-
braic operations.
Forget about vectors you are accustomed to, with magni-
tude and direction, with arrow marks on the top, and with
coordinates.
They are vectors in the Euclidean space and only an example,
a special kind of vectors.
We want more general denitions that, to begin with, do not
require magnitude, direction, or orthogonality.
c _2011 by Taejeong Kim 4
According to the new denition,
a continuous function can be a vector;
an innite sequence can be a vector;
a matrix can be a vector.
So a vector will now be a more abstract, exible thing than
what you are used to.
So be ready to accept new concepts that seem at rst strange,
and you will feel comfortable with them later in this course.
c _2011 by Taejeong Kim 5
Field and vector space
eld: (F, +, ) such that
1. a, b F, a + b and a b are unique in F.
2. a, b, c F, the following hold:
F1 commutativity: a + b = b + a; a b = b a
F2 associativity: (a +b) +c = a +(b +c); (a b) c = a (b c)
F3 identity:
additive: 0 F such that a + 0 = a
multiplicative: 1 F such that a 1 = a
F4 inverse:
additive: d F such that a + d = 0
multiplicative: e F such that, for a ,= 0, a e = 1
F5 distributivity: a (b + c) = a b + a c
c _2011 by Taejeong Kim 6
Commonly, the operations + and become implicit, and the
set F is called the eld.
An element of a eld is called a scalar.
The additive inverse d of a is commonly denoted by a.
The multiplicative inverse e of a is commonly denoted by
a
1
.
example: eld
set of real numbers with the ordinary operations
set of rational numbers with the ordinary operations
set of complex numbers with the ordinary operations
0, 1 with the modulo-2 operations: prime-2 eld
0, 1, 2, 3, 4 with the modulo-5 operations: prime-5 eld
c _2011 by Taejeong Kim 7
prime-2 eld prime-5 eld
a b +
0 0 0 0
0 1 1 0
1 0 1 0
1 1 0 1
a a a
1
0 0
1 1 1
a a a
1
0 0
1 4 1
2 3 3
3 2 2
4 1 4
A eld with a nite number of elements are called a nite
eld, a very important concept in channel coding and cryp-
tography.
For a prime number n, 0, 1, , n 1 with modulo-n
operations is a prime(-n) eld.
How about the set of integers or the set of irrational num-
bers?
c _2011 by Taejeong Kim 8
vector space over a eld F: (V (F), +, ) such that
1. a F, x, y V , x +y and a x = ax are unique in V .
2. a, b F, x, y, z V , the following hold:
VS1 commutativity: x + y = y + x
VS2 associativity: (x + y) + z = x + (y + z)
VS3 identity: 0 V such that x + 0 = x
VS4 inverse: u V such that x + u = 0
VS5 1x = x
VS6 associativity: (ab)x = a(bx)
VS7 distributivity: a(x + y) = ax + ay
VS8 distributivity: (a + b)x = ax + bx
Commonly, the operations + and (and F) become implicit,
and the set V (F) (or V ) is called the vector space.
c _2011 by Taejeong Kim 9
An element of a vector space is called a vector.
The inverse u of x is commonly denoted by x.
Note that multiplication is always between a scalar and a
vector, so the inverse is always additive.
example: vector space
the set of n-tuples F
n
= (a
1
, , a
n
) : a
i
F
the set of mn matrices
M
mn
(F) =
_

_
_
_
_
_
_
_
_
_
_
_
_
a
11
a
1n
.
.
.
.
.
.
a
m1
a
mn
_
_
_
_
_
_
_
_
_
_
_
: a
ij
F
_

_
the set of functions T(S, F) = f : SF
c _2011 by Taejeong Kim 10
the set of polynomials
P(F) = a
0
+ a
1
x + + a
n
x
n
: n = 0, 1, 2, , a
i
F
the set of innite sequences F

= (a
1
, a
2
, a
3
, ) : a
i
F
example: not a vector space
(I R
2
, +, ) where + and are dened by
(a
1
, a
2
) + (b
1
, b
2
) = (a
1
+ b
1
, a
2
b
2
); c(a
1
, a
2
) = (ca
1
, ca
2
)
not commutative [VS1], not associative [VS2]
(I R
2
, +, ) where + and are dened by
(a
1
, a
2
) + (b
1
, b
2
) = (0, a
2
+ b
2
); c(a
1
, a
2
) = (0, ca
2
)
no identity [VS3], no inverse [VS4], 1(a
1
, a
2
) ,= (a
1
, a
2
) [VS5]
c _2011 by Taejeong Kim 11
Theorem 1.1 (cancellation law):
x, y, z V ; x + z = y + z x = y
proof: Let w = z. Then
x = x + 0 = x + (z + w) = (x + z) + w
= (y + z) + w = y + (z + w) = y + 0 = y.
Corollary 1.1.1: 0 is unique in V.
proof: If both 0 and are identities,
x + 0 = x = x + 0 = [cancellation law]
c _2011 by Taejeong Kim 12
Corollary 1.1.2 and Theorem 1.2: x V and a F,
1. The additive inverse is unique in V.
2. 0x = 0
3. (a)x = (ax) = a(x)
4. a0 = 0
It is customary to express n-tuples as column vectors when
vector-matrix arithmetic is involved.
transpose A
t
of a matrix A: (A
t
)
ij
= A
ji
_
_
_
_
_
_
a b c
d e f
_
_
_
_
_
_
t
=
_
_
_
_
_
_
_
_
_
_
_
a d
b e
c f
_
_
_
_
_
_
_
_
_
_
_
, (a, b, c, d)
t
=
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
a
b
c
d
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
c _2011 by Taejeong Kim 13
Subspace
subspace: a vector space W in another vector space V
That is, (V (F), +, ) is a vector space; W V ; and (W(F), +, )
is also a vector space, for the same + and .
For any vector space V ,
V is a subspace.
0 is a subspace.
is not a subspace.
Theorem 1.3: V is a vector space and W V . Then W is a
subspace of V if and only if
1. x, y W x + y W and
2. a F, x W ax W.
c _2011 by Taejeong Kim 14
0 W is not necessary because 2 implies 0x = 0 W.
The necessary and sucient condition can be replaced by
a, b F, x, y W ax + by W or
a F, x, y W ax + y W.
a
i
F, u
i
W a
1
u
1
+ + a
k
u
k
W
example: subspace
V = F
n
= (a
1
, , a
n
) : a
i
F
W = (a
1
, a
2
, 0, , 0) : a
i
F
.
.
.
.
.
.
.
.
.
.
.
.
.
_
p
p
p
p
p
'
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
a
1
a
2
a
3
W
. .p
p
p
p
p
p
p
p
p
p
.
.
.
.
.
.
.
p p p p p p p p p p _
p
p
p
p
p
p
p
p
p
p
'
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/

a
1
a
2
a
3
W
V = M
nn
(F)
W = symmetric matrices
V = T(I R, I R)
W = continuous functions
c _2011 by Taejeong Kim 15
V = P(F), W = P
n
(F) = polynomials of degree n or less
V = I R

, W = convergent sequences
Theorem 1.4: Any intersection of subspaces of a vector space
V is a subspace of V .
proof: Let U
1
, , U
k
be subspaces of V and W =
k
i=1
U
i
.
x, y W x, y U
i
, i = 1, , k
x + y U
i
, i = 1, , k [subspace]
x + y W
a F, x W x U
i
, i = 1, , k
ax U
i
, i = 1, , k [subspace]
ax W
c _2011 by Taejeong Kim 16
The union of two subspaces of a vector space V is not a subspace
in general.
example: V = I R
3
,
U
1
= (a
1
, 0, a
3
) : a
1
, a
3
I R, U
2
= (a
1
, a
2
, 0) : a
1
, a
2
I R
are subspaces.
U
1
U
2
= (a
1
, 0, 0) : a
1
I R is a subspace.
U
1
U
2
is not a subspace because
.
.
.
.
.
.
.
.
.
.
.
.
.
.
_
'
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

a
1
a
2
a
3
U
1
U
2
(a
1
, 0, a
3
) + (a
1
, a
2
, 0) , U
1
U
2
.
c _2011 by Taejeong Kim 17
Linear combination
linear combination of u
1
, , u
k
V :
a
1
u
1
+ + a
k
u
k
, a
i
F
a
1
u
1
+ a
2
u
2
+ is not a linear combination.
a
1
u
1
+ a
2
is not a linear combination.
0 is a linear combination of any set of vectors.
A subspace is closed under linear combination.
Whether v is a linear combination of u
1
, , u
k
is determined
by solving the equation v = a
1
u
1
+ +a
k
u
k
for a
1
, , a
k
.
example:
V = I R
3
, 0
_
_
_
_
_
_
_
_
_
_
_
3
2
1
_
_
_
_
_
_
_
_
_
_
_
+ 1
_
_
_
_
_
_
_
_
_
_
_
0
1
2
_
_
_
_
_
_
_
_
_
_
_
2
_
_
_
_
_
_
_
_
_
_
_
0
3
0
_
_
_
_
_
_
_
_
_
_
_
=
_
_
_
_
_
_
_
_
_
_
_
0
5
2
_
_
_
_
_
_
_
_
_
_
_
c _2011 by Taejeong Kim 18
V = P(I R) or P
3
(I R),
4(x
3
2x
2
5x 3) + 2(3x
3
5x
2
4x 9)
= 2x
3
2x
2
+ 12x 6
span of S: the set of all linear combinations of the vectors in
S
notation: span(S)
span()=0 for convenience
example:

p
p
p
p
p
`
.
.
_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
a
1
a
2
a
3
span
V = I R
3
,
span((1, 0, 0), (0, 1, 0)) = (a
1
, a
2
, 0) : a
1
, a
2
I R
V = P(I R), span(1, x
2
) = a
0
+ a
2
x
2
: a
0
, a
2
I R
c _2011 by Taejeong Kim 19
V = T(I R, I R),
span(cos x, sin x) = a
1
cos x + a
2
sin x : a
1
, a
2
I R
= a cos(x + b) : a I R, b [0, 2)
Theorem 1.5: V is a vector space; W is its subspace; and
S V . Then
1. span(S) is a subspace of V , and
2. S W span(S) W.
proof: 1: Any span is closed under + and .
2: Assume S W and w span(S)
w = a
1
u
1
+ + a
k
u
k
, a
i
F, u
i
S [span]
w W [u
i
W, subspace]
span(S) W
c _2011 by Taejeong Kim 20
The span operation can make a set S a lot bigger, but it
cannot make it bigger than any subspace or vector space
containing S.
In order to show span(S) W, we only need to show S
W.
generating set S for V : span(S) = V
S is a generating/spanning set for V .
S generates/spans V .
The elements of S generate/span V .
There are many generating sets for V .
c _2011 by Taejeong Kim 21
example:
I R
3
= span((1, 0, 0), (0, 1, 0), (0, 0, 1))
I R
3
= span((1, 0, 0), (0, 1, 0), (0, 0, 1), (1, 1, 1))
I R
3
= span((1, 1, 0), (1, 0, 1), (0, 1, 1))
P
2
(I R) = span(1, x, x
2
)
P
2
(I R) = span(x
2
+3x2, 2x
2
3, 5x, x+1, x
2
4x+4)
It will be convenient to nd a smallest generating set because it
will be a tiniest seed from which we can grow the whole vector
space. (smallest linearly independent basis)
By smallest, we mean smallest in cardinality.
0 < 1 < 2 < < nite < countably innite (denumerable)
< uncountably innite
c _2011 by Taejeong Kim 22
Linear dependence and linear independence
linearly dependent S V (F):
u
1
, , u
k
S, and a
1
, , a
k
F, not all zero, such that
a
1
u
1
+ + a
k
u
k
= 0 .
a
i
,= 0 u
i
= a
1
i
(a
1
)u
1
+ + a
1
i
(a
i1
)u
i1
+a
1
i
(a
i+1
)u
i+1
+ + a
1
i
(a
k
)u
k
0 S S is linearly dependent. [a0 = 0 ]
linearly independent S V (F): not linearly dependent
equivalent denition: u
1
, , u
k
S,
a
1
u
1
+ + a
k
u
k
= 0 a
1
= = a
k
= 0
is linearly independent.
A set containing a single nonzero vector is linearly indep.
c _2011 by Taejeong Kim 23
example: V = T(I R, I R)
cos x, sin x is linearly independent.
cos x, sin x, cos(x +

4
) is linearly dependent.
cos x, sin x, cos 2x, cos 3x is linearly independent.
Theorem 1.6 and Corollary 1.6: R S V
1. R is linearly dependent S is linearly dependent.
2. S is linearly independent R is linearly independent.
Theorem 1.7: S V ; S is linearly independent; and v V S.
Then, S v is linearly dependent v span(S).
proof: :

a
i
u
i
+ bv = 0 b ,= 0 [u
i
are lin indep]
v =

(a
i
b
1
)u
i
: Obvious.
c _2011 by Taejeong Kim 24
Basis and dimension
basis for V : linearly independent generating set for V
A basis is a smallest generating set for V .
Z
Z
Z
Z
`
`
`
``
linearly independent sets generating sets
sets
smaller
than
a basis
bases
sets
larger
than
a basis
c _2011 by Taejeong Kim 25
example:
(1, 0, 0), (0, 1, 0), (0, 0, 1) is the standard basis for I R
3
.
(1, 0, 0), (0, 1, 0), (0, 0, 1), (1, 1, 1) is not a basis for I R
3
.
(1, 1, 0), (1, 0, 1), (0, 1, 1) is a basis for I R
3
.
1, x, x
2
is the standard basis for P
2
(I R).
x
2
+ 3x 2, 2x
2
3, 5x, x + 1, x
2
4x + 4
is not a basis for P
2
(I R).
x
2
+ 3x 2, 2x
2
3, x + 1 is a basis for P
2
(I R).
Lagrange polynomials f
i
(x) =

n
j=0,j,=i
xc
j
c
i
c
j
, i = 0, , n,
where c
i
,= c
j
for i ,= j, form a basis for P
n
(I R).
c _2011 by Taejeong Kim 26
Theorem 1.8: = u
1
, , u
n
is a basis for V
v V , unique (a
1
, , a
n
) such that
v = a
1
u
1
+ + a
n
u
n
.
proof: : Let is a basis for V .
V = span() [basis]
v V , v = a
1
u
1
+ + a
n
u
n
for some a
1
, , a
n
Show uniqueness:
Let also v = b
1
u
1
+ + b
n
u
n
.
0 = (a
1
b
1
)u
1
+ + (a
n
b
n
)u
n
a
1
= b
1
, , a
n
= b
n
[ is lin indep]
c _2011 by Taejeong Kim 27
: Assume v V , unique a
1
, , a
n
such that
v = a
1
u
1
+ + a
n
u
n
= u
1
, , u
n
generates V .
Show linear independence:
Let 0 = c
1
u
1
+ + c
n
u
n
.
c
1
= 0, , c
n
= 0 [0u
1
+ + 0u
n
= 0 , uniqueness]
This theorem means that given a vector space V (F) and its
basis , whatever kind of vector space it may be, each vector v
in V is uniquely represented by (a
1
, , a
n
).
Given a basis, there is a one-to-one correspondence between V
and F
n
.
[v]

= (a
1
, , a
n
)
t
is called the (n-tuple) representation of
v in or relative to .
c _2011 by Taejeong Kim 28
example:
= 1, x, x
2
, a
0
+ a
1
x + a
2
x
2
(a
0
, a
1
, a
2
)
= 1, 1 + x, 1 + x + x
2
,
a
0
+ a
1
x + a
2
x
2
(a
0
a
1
, a
1
a
2
, a
2
)
=
_

_
_
_
_
_
_
_
1 0
0 0
_
_
_
_
_
_
,
_
_
_
_
_
_
0 1
0 0
_
_
_
_
_
_
,
_
_
_
_
_
_
0 0
1 0
_
_
_
_
_
_
,
_
_
_
_
_
_
0 0
0 1
_
_
_
_
_
_
_

_
,
_
_
_
_
_
_
a b
c d
_
_
_
_
_
_
(a, b, c, d)
= e
i2kf
0
t
: k = , 1, 0, 1, ,
f(t) =

k=
a
k
e
i2kf
0
t
( , a
1
, a
0
, a
1
, )
periodic function with frequency f
0
Fourier coecients
c _2011 by Taejeong Kim 29
Theorem 1.9: A nite generating set S for V can be reduced
to a basis for V .
proof: (i) If S = or S = 0
V = 0 and = is considered a basis for V .
(ii) If S has non-zero vectors, let = u
1
, , u
k
be a largest
linearly independent subset of S.
span() V = span(S) [generating set]
v S v or v ,
v , v is lin dep. [ is a largest lin indep subset]
v span() [Thm 1.7]
S span()
span(S) span() [span() is a subsp; Thm 1.5]
c _2011 by Taejeong Kim 30
example:
x
2
+3x 2, 2x
2
3, 5x, x +1, x
2
4x +4 spans P
2
(I R).
x
2
+ 3x 2, 2x
2
3, 5x is a basis for P
2
(I R).
(1, 0, 0), (0, 1, 0), (0, 0, 1), (1, 1, 1) spans I R
3
.
(0, 1, 0), (0, 0, 1), (1, 1, 1) is a basis for I R
3
.
Theorem 1.10 (replacement theorem):
G V is a generating set for V and has n elements;
L V is a linearly independent set and has m elements. Then
1. m n, and
2. L can be extended to a generating set for V by adding nm
elements from G.
c _2011 by Taejeong Kim 31
G
L
/
H
/
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
L v
m+1
H
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p
p

`
`
`
`
`
`
`-
n
m1 n m + 1
m n m
proof by induction: Let the set of the n m elements be H.
(i) If [L[ = m = 0, then
1. 0 n and
2. H = G generates V .
(ii) Assume the theorem holds for [L[ = m. That is, assume
1. m n, and
2. H G such that [H[ = n m and L H generates V .
(iii) Let L = v
1
, , v
m+1
and L
/
= v
1
, , v
m
.
c _2011 by Taejeong Kim 32
L
/
is linearly independent. [Thm 1.6]
m n and H
/
= u
1
, , u
nm
such that
L
/
H
/
generates V . [assumption]
v
m+1
= a
1
v
1
+ + a
m
v
m
+ b
1
u
1
+ + b
nm
u
nm
If m = n
v
m+1
= a
1
v
1
+ + a
m
v
m
: contradiction [L is lin indep]
m < n [assumption: m n]
m + 1 n: 1
b
i
,= 0 for some i [L is lin indep], and we can let i = 1 by
rearranging so that b
1
,= 0.
u
1
= (b
1
1
a
1
)v
1
+ + (b
1
1
a
m
)v
m
+ (b
1
1
)v
m+1
+(b
1
1
b
2
)u
2
+ + (b
1
1
b
nm
)u
nm
: (1)
c _2011 by Taejeong Kim 33
Let H = u
2
, , u
nm
.
L H = v
1
, , v
m
, v
m+1
, u
2
, , u
nm

L
/
H
/
= v
1
, , v
m
, u
1
, u
2
, , u
nm

V = span(L
/
H
/
) span(LH) [assumption, (1)]: 2
example:
G = x
2
+3x2, 2x
2
3, 5x, x+1, x
2
4x+4 P
2
(I R).
L = 1, x H = 5x, x + 1, x
2
4x + 4
G = (1, 0, 0), (0, 1, 0), (0, 0, 1), (1, 1, 1) generates I R
3
.
L = (1, 1, 0) H = (1, 0, 0), (0, 1, 0), (0, 0, 1)
In these examples, H can be any three elements from G.
c _2011 by Taejeong Kim 34
Corollary 1.10.1: If V has a nite basis, then every basis for V
contains the same number of vectors.
proof: Let and be bases for V .
[[ [[. [ is lin indep; spans V ; Thm 1.10]
[[ [[. [ is lin indep; spans V ; Thm 1.10]
dimension: the number of vectors in a basis
The dimension can be either nite or innite.
This denition of dimension applies to all spaces with alge-
braic structure vector, normed linear, inner product, Ba-
nach, Hilbert, and Euclidean spaces.
There are other kinds of dimensions, but they are not our
concern.
c _2011 by Taejeong Kim 35
example:
dim(0 ) = 0
dim(F
n
) = n
dim(M
mn
) = mn
dim(P
n
) = n + 1
The space of complex numbers can be
1. a vector space of dimension 1 when the eld is complex.
2. a vector space of dimension 2 when the eld is real.
c _2011 by Taejeong Kim 36
Corollary 1.10.2, expanded: Let dim(V ) = n.
1. S generates V .
[S[ n
S can be reduced to a basis for V .
2. S generates V and [S[ = n.
S is a basis for V .
3. S is linearly independent.
[S[ n
S can be extended to a basis for V .
4. S is linearly independent and [S[ = n.
S is a basis for V .
c _2011 by Taejeong Kim 37
Z
Z
Z
Z
`
`
`
``
linearly independent sets generating sets
sets
smaller
than
a basis
bases
sets
larger
than
a basis
Note that
[S[ n , S generates V .
[S[ n , S is linearly independent.
[S[ = n , S is a basis for V .
c _2011 by Taejeong Kim 38
Theorem 1.11: dim(V ) < ; W is a subspace of V . Then
1. dim(W) dim(V )
2. dim(W) = dim(V ) W = V .
proof: 1: Let be a basis for W.
dim(W) = [[ dim(V ). [ is lin indep; Corol 1.10.2]
2: If [[ = dim(V ),
is a basis for V . [ is lin indep; Corol 1.10.2]
Corollary 1.11: dim(V ) < ; W is a subspace of V . Then any
basis for W can be extended to a basis for V .
proof: Let be a basis for W.
is linearly independent.
can be extended to a basis for V . [Corol 1.10.2]
c _2011 by Taejeong Kim 39
So the best way to describe a vector space V and its subspace
W is to nd a basis u
1
, , u
m
for W and extend it to a
basis u
1
, , u
m
, u
m+1
, , u
n
for V .
example:
p
p
p
p
p
p
p
p
p
p

.
p p p p p p p p p p
p
p
p
p
p
p
p
p
p
p
`
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/
/

/
//

'
u
1
u
2
u
3
W V = I R
3
n = 3
m = 2
Though our discussion considers mainly nite-dimensional vec-
tor spaces, the discussion can be generalized to innite-dimen-
sional vector spaces.