Glossary for Linear Algebra

Elizabeth Meckes Mark Meckes

January 7, 2019
 1

Notation
∀: For all.

∃: There exists.

aij : The entry in the ith row and jth column of the matrix A.

Aij : For A ∈ Mn (F), Aij ∈ Mn−1 (F) is the result of removing the ith row and the jth
column of A.

A∗ : The conjugate transpose of the matrix A.

AT : The transpose of the matrix A.

kAkF :
The Frobenius (or Hilbert–Schmidt) norm of the matrix A.

kAkop :
The operator norm of the matrix A.

C: The set of all complex numbers.

c0 : The set of sequences (an )n≥1 ⊆ R with limn→∞ an = 0.

C(A):
The column space of the matrix A.

C[a, b]:
The set of continuous functions f : [a, b] → R.

D[a, b]:
The set of differentiable functions f : [a, b] → R (with one-sided derivatives at the
end points).

det A:
The determinant of the matrix A.

diag(d1 , . . . , dn ):
The n × n diagonal matrix with d1 , . . . , dn on the diagonal.

dim(V ):
The dimension of the vector space V .

Eigλ (T ), Eigλ (A):

The λ-eigenspace of T or A.

Fn : The set of n-dimensional column vectors over F.

 2

F∞ : The set of sequences with entries in F.

F2 : The field of two elements.

f (A):
The image of A under f : f (A) = {f (x) : x ∈ A}.

I, In :
The identity map or matrix.

ι: The identity permutation.

Im z:
The imaginary part of z ∈ C.

ker T , ker A:
The kernel of T or A.

L(V ):
The set of all linear maps from V to V .

L(V, W ):
The set of all linear maps from V to W .

Mm,n (F):
The set of all m × n matrices over F.

Mn (F):
The set of all n × n matrices over F.

N: The set of all natural numbers.

null T , null A:
The nullity of T or A.

Pn (F):
The set of polynomials with coefficients in F of degree at most n.

Pc,i,j :
The elementary matrix  
1 0 0
 ...
0 c

Pc,i,j :=  ,

..
 . 0
0 0 1
where the c is in the ith row and jth column.
 3

pA (x):
The characteristic polynomial of the matrix A.

p(T ), p(A):
For a polynomial p(x) = a0 + a1 x + · · · + ak xk , if T ∈ L(V ) is a linear map, p(T )
is the linear map a0 I + a1 T + · · · + ak T k . If A is a square matrix, p(A) is the matrix
p(A) = a0 I + a1 A + · · · + ak Ak .

Q: The set of all rational numbers.

Qc,i :
For c 6= 0, the elementary matrix
 
1 0 0
 ...
0


Qc,i :=  c ,
 
 ... 
 0
0 0 1
where the c is the ith diagonal entry.

R: The set of all real numbers.

Ri,j :
For i 6= j, the elementary matrix
 
1
 ..
.

 
0 1
 
 

Ri,j := 
 . ..

,


 1 0 


 ... 

1

where the off-diagonal ones are the (i, j) and (j, i) entries.

rank T , rank A:
The rank of T or A.

Re z:
The real part of z ∈ C.

Sn : The symmetric group on n letters.

 4

sgn(σ):
The sign of the permutation σ.

T ∗ : The adjoint of the operator T .

[T ]BV ,BW :
The matrix of T ∈ L(V, W ) with respect to the bases BV and BW .

tr A:
The trace of the matrix A.

hv1 , . . . , vk i:
The span of v1 , . . . , vk .

[v]B : The coordinate representation of the vector v in the basis B.

Z: The set of all integers.

|z|: The absolute value or modulus of z ∈ C.

z: The complex conjugate of z ∈ C.

Terms
Terms identified with an asterisk (*) appear in the book only in examples and exercises.

absolute value:
The absolute value or modulus of a complex number z ∈ C is
√
|z| = a2 + b2 ,

where z = a + ib for a, b ∈ R.

* addition mod p:
Let p ∈ N be a prime number. For a, b ∈ {0, 1, . . . , p − 1}, we write

a+b=r (mod p)

or
a+b=r (in Fp )
if r ∈ {0, 1, . . . , p − 1} is the remainder when a + b is divided by p.

addition (of functions):

If f : X → F and g : X → F are functions with the same domain X, taking values
in a field F, then the sum f + g : X → F is defined by (f + g)(x) = f (x) + g(x).
This is also called pointwise addition of functions.
 5

Fn ):  
addition (in 
a1 b1
 ..   .. 
Given  .  ,  .  ∈ Fn ,
an bn
     
a1 b1 a1 + b 1
 ..   ..   . 
 .  +  .  :=  ..  .
an bn an + b n

adjoint:
Let V and W be inner product spaces, and let T ∈ L(V, W ). A linear map S ∈
L(W, V ) is an adjoint of T if
hT v, wi = hv, Swi
for every v ∈ V and w ∈ W . We then write S = T ∗ .
additive:
Let V be a vector space over F. A function f on V is additive if f (v1 + v2 ) =
f (v1 ) + f (v2 ) for all v1 ,2 ∈ V .
adjugate:
The adjugate of A ∈ Mn (F) is the matrix adj A = CT , where C is the cofactor
matrix of A.
Also referred to as the classical adjoint of A.
affine subspace:
Let V be a vector space. An affine subspace of V is a subset of the form {v + u | u ∈ U },
where U is a subspace of V .
algebraically closed:
A field F is algebraically closed if every polynomial
p(x) = a0 + a1 x + · · · + an xn
with coefficients in F can be factored in the form
p(x) = b(x − c1 ) . . . (x − cn )
for some b, c1 , . . . , cn ∈ F.
alternating:
A multilinear function D : Mn (F) → F is called alternating if
 
| |
D a1 · · · an  = 0
| |
whenever there exist i 6= j such that ai = aj .
 6

* anti-Hermitian part:
The anti-Hermitian part of a matrix A ∈ Mn (C) is the matrix
1
Im A := (A − A∗ ).
2i

argument:
The argument of a complex number z ∈ C is the real number θ ∈ R, where z =
|z| eiθ = |z| (cos θ + i sin θ). Defined this way, the argument of z is not unique, since
if z = |z|eiθ , then also z = |z|ei(θ+2πk) for any integer k; the argument is therefore
sometimes required to lie in [0, 2π) or [−π, π).

associative:
An operation ∗ on a set S is associative if (s1 ∗ s2 ) ∗ s3 = s1 ∗ (s2 ∗ s3 ) for all
s1 , s2 , s3 ∈ S. Function composition is also considered to be associative, since if
f : X → Y , g : Y → Z and h : Z → W , then (h ◦ g) ◦ f = h ◦ (g ◦ f ). If
X = Y = Z = W , this is a special case of the definition above.

augmented matrix (of a linear system):

The augmented matrix of the m × n linear system

a11 x1 + · · · + a1n xn = b1
..
.
am1 x1 + · · · + amn xn = bm

is the m × (n + 1) matrix
 
a11 ··· a1n b1
 .. .. .. ..  .
 . . . . 
am1 ··· amn bm

base field:
If V is a vector space over the field F, then F is called the base field of V .

basic logical connectives:

The five basic logical connectives are

• not
• and
• or
• if..., then ...
• if and only if
 7

basis:
A list of vectors (v1 , . . . , vn ) in a finite-dimensional vector space V is a basis for V
if (v1 , . . . , vn ) is linearly independent and V ⊆ hv1 , . . . , vn i.
The plural of basis is bases.
See also Perspectives: Bases at the end of chapter 3.

biconditional statement:
A biconditional statement is a statement with the connective “if and only if”.

bijective:
A function f is bijective or is a one-to-one correspondence if f is both injective
and surjective.

binary code:
A binary code is a subset C ⊆ Fn2 for some n.

binary linear code:

A binary linear code is a subspace C ⊆ Fn2 for some n.

bit: An element of F2 = {0, 1}.

bounded set:
Let V be a normed space. A subset S ⊆ V is bounded if there is a constant C > 0
such that
ksk ≤ C
for each s ∈ S.

change of basis matrix:

Let B and B0 be two different bases of a finite-dimensional vector space V . The
change of basis matrix from B to B0 is the matrix S = [I]B,B0 . It has the defining
property that S [v]B = [v]B0 for all v ∈ V .

characteristic polynomial:
The characteristic polynomial of A ∈ Mn (F) is the polynomial pA (x) = det(A −
xIn ).

* Cholesky decomposition:
A Cholesky decomposition of a positive definite matrix A ∈ Mn (C) is a factoriza-
tion A = X∗ X, where X ∈ Mn (F) is upper triangular.
 8

* circulant matrix:
A circulant matrix is a matrix of the form
 
a1 a2 a3 · · · an−1 an
 an a1 a2 · · · an−2 an−1 
.. 
 
..
an−1 an a1 . . 

 . . . . . .. 
 .. .. .. .. .. . 
 

 a3 a4 . ..

a1 a2 
a2 a3 · · · · · · an a1
where a1 , . . . , an ∈ C.
classical adjoint:
See adjugate.
closed under addition:
Let V be a vector space. A subset U ⊆ V is closed under addition if whenever
u1 , u2 ∈ U , then u1 + u2 ∈ U .
closed under scalar multiplication:
Let V be a vector space over F. A subset U ⊆ V is closed under scalar multiplica-
tion if whenever u ∈ U and c ∈ F, then cu ∈ U .
closed (set):
Let V be a normed space. A subset S ⊆ V is closed if S contains its own boundary;
that is, if every convergent sequence in S has its limit point in S as well.
codomain:
The codomain of a function f : X → Y is the set Y .
coefficient matrix:
The coefficient matrix of the m × n linear system
a11 x1 + · · · + a1n xn = b1
..
.
am1 x1 + · · · + amn xn = bm
is the m × n matrix  
a11 ··· a1n
 .. .. ..  .
A= . . . 
am1 ··· amn

cofactor matrix:
The cofactor matrix of A ∈ Mn (F) is the matrix C ∈ Mn (F) with cij = (−1)i+j det Aij ,
where Aij is the result of removing the ith row and jth column from A.
 9

collinear:
Let V be a vector space over F. Two vectors v, w ∈ V are called collinear if v = aw
or w = av for some a ∈ F.

column space:
The column
 C(A) of a matrix A is the span of the columns of A. That is, if
space 
| |
A = a1 · · · an , then C(A) := ha1 , . . . , an i.
| |
column vector:
An n-dimensional column vector over a field F is an n × 1 matrix over F.

* companion matrix:
The companion matrix of a polynomial p(x) = c0 + c1 x + · · · + cn−1 xn−1 + xn with
coefficients in F is the matrix
 
0 0 · · · 0 −c0
1 0 · · · 0 −c1 
 
0 1 · · · 0 −c2 
 .
 .. .. . . .. .. 
. . . . . 
0 0 · · · 1 −cn−1

complex conjugate:
The complex conjugate of a complex number z ∈ C is the complex number z =
a − ib, where z = a + ib for a, b ∈ R.

complex vector space:

A complex vector space is a vector space over the field C of complex numbers.

composition:
Let X, Y , and Z be sets and let f : X → Y and g : Y → Z be functions. The
composition g ◦ f : X → Z of g and f is defined by


g ◦ f (x) = g f (x) .

* condition number:
The condition number of an invertible matrix A ∈ Mn (C) is κ(A) = kAkop kA−1 kop .

conditional statement:
A conditional statement is a statement using the logical connective “if..., then...”.

conjugate transpose:
The conjugate transpose of B ∈ Mm,n (C) is the matrix B∗ ∈ Mn,m (C) such that
[B∗ ]jk = bkj .
 10

consistent:
A system of equations is consistent if there exists at least one solution of the system.

contraction:
Let V and W be normed spaces. A linear map T ∈ L(V, W ) is a contraction if
kT vk ≤ kvk for all v ∈ V ; equivalently, if kT kop ≤ 1.

contrapositive:
The contrapositive of the conditional statement “A ⇒ B” is the conditional statement
“(not B) ⇒ (not A)”.

coordinate representation:
Let B be a basis of a vector space V over F. The coordinate representation of
v ∈ V with respect to B is the column vector
 
a1
 .. 
[v]B :=  .  ∈ Fn ,
an

where
v = a1 v1 + · · · + an vn .

coordinates:
Let B be a basis of a vector space V . The coordinates of v ∈ V with respect to B
are the (unique ordered list of) scalars a1 , . . . , an such that

v = a1 v1 + · · · + an vn .

counterexample:
A counterexample to a statement which claims that something is always be true is
an example in which the statement is false.

determinant:
The determinant is the unique alternating multilinear function det : Mn (F) → F
which satisfies det(In ) = 1. The determinant of a matrix A ∈ Mn (F) is the value
det(A).
See also Perspectives: Determinants at the end of chapter 6.

determinant function:
A determinant function on Mn (F ) is an alternating multilinear function Dn : Mn (F) →
F such that Dn (In ) = 1.

determinant of a map:
Let V be a finite-dimensional vector space. The determinant of a linear map T ∈
L(V ) is det T = det [T ]B , where B is any basis of V .
 11

diagonal entries:
The diagonal entries of a matrix A ∈ Mn (F) are the (i, i) entries for i = 1, . . . , n.
diagonal matrix:
A matrix A ∈ Mn (F) is diagonal if aij = 0 whenever i 6= j.
diagonalizable linear map:
A linear map T ∈ L(V ) is diagonalizable if there exists a basis B of V such that
[T ]B is a diagonal matrix.
diagonalizable matrix:
A matrix A ∈ Mn (F) is diagonalizable if it is similar to a diagonal matrix.
diagonalize:
A basis B of a finite-dimensional vector space V diagonalizes T ∈ L(V ) if the
matrix [T ]B of T with respect to B is diagonal.
dimension:
Let V be a finite-dimensional vector space. If V =6 {0}, then the dimension of V ,
written dim V , is the length of any basis of V . If V = {0}, then we define the
dimension of V to be 0.
If dim V = n, we say that V is n-dimensional
* discrete Fourier transform:
The n × n discrete Fourier transform (or DFT matrix) is the matrix F ∈ Mn (C)
with entries
1
fjk = √ ω jk ,
n
where ω = e2πi/n = cos(2π/n) + i sin(2π/n).
division:
Let F be a field and let a, b ∈ F with b 6= 0. Then we define a
b
= ab−1 .
domain:
The domain of a function f : X → Y is the set X.
dot product:
The dot product of x, y ∈ Rn is denoted hx, yi and is defined by
n
X
hx, yi = xj yj ,
j=1
   
x1 y1
 ..   .. 
where x =  .  and y =  . .
xn yn
 12

eigenspace:
Let V be a vector space over F. If λ ∈ F is an eigenvalue of T ∈ L(V ), then the
λ-eigenspace of T is the subspace
Eigλ (T ) := {v ∈ V | T (v) = λv}
of V .
If λ ∈ F is an eigenvalue of A ∈ Mn (F), the λ-eigenspace of A, denoted Eigλ (A)
is the λ-eigenspace of the linear map on Fn given by A.
eigenvalue:
Let V be a vector space over F. A scalar λ ∈ F is an eigenvalue of T ∈ L(V ) if
there exists a nonzero v ∈ V with T v = λv.
A scalar λ ∈ F is an eigenvalue of A ∈ Mn (F) if there exists a non-zero v ∈ Fn
with Av = λv.
See also Perspectives: Eigenvalues at the ends of chapters 3 and 6.
eigenvector:
Let V be a vector space over F. A nonzero vector v ∈ V is an eigenvector of
T ∈ L(V ) if T v = λv for some λ ∈ F.
A nonzero vector v ∈ Fn is an eigenvector of A ∈ Mn (F) if Av = λv for some
λ ∈ F.
element:
See set.
elementary matrices:
The elementary matrices in Mm (F) are the matrices of the following three types,
for 1 ≤ i, j ≤ m and c ∈ F.
• For i 6= j,  
1 0 0
 ...
0 c

Pc,i,j :=  .

 . . . 0
0 0 1
The (i, j) entry is c, all diagonal entries are 1, and all off-diagonal entries other
than the (i, j) entry are 0.
• For c 6= 0, then  
1 0 0
 ...
0


Qc,i :=  c .
 

 . 
. . 0
0 0 1
 13

The (i, i) entry is c, all other diagonal entries are 1, and all the off-diagonal
entries are 0.
• For i 6= j,  
1
 ..
.

 
0 1
 
 
Ri,j

:=  . ..

.
 

 1 0 

 .. 
 . 
1
The (i, j) and (j, i) entries are both 1, all the other off-diagonal entries are 0,
the i, i and j, j entries are both 0, and all the other diagonal entries are 1.

encoding function:
An encoding function is a bijective function

f : Fk2 → C,

where C ⊆ Fn2 is a binary code.

encoding matrix:
An encoding matrix is a matrix A ∈ Mn,k (F2 ) which represents a linear encoding
function.

entry:
See matrix.

equivalent (equations or systems of equations):

Two equations, or two systems of equations, are equivalent if they have the same
sets of solutions.

error-correcting:
An error-detecting linear encoding with encoding matrix A ∈ Mn,k (F2 ) is error-
correcting if, whenever z ∈ Fn2 differs in at most 1 entry from some y ∈ C(A),
there is a unique such y ∈ C(A).

error-detecting:
A linear encoding with encoding matrix A ∈ Mn,k (F2 ) is error-detecting if, when-
ever y ∈ C(A) and z ∈ Fn2 differ in exactly 1 entry, z ∈
/ C(A).

essentially disjoint:
Two rectangles in Rn are essentially disjoint if their intersection has volume 0.
 14

field:
A field consists of a set F with operations + and ·, and distinct elements 0, 1 ∈ F,
such that all of the following properties hold:

• For each a, b ∈ F, a + b ∈ F.
• For each a, b ∈ F, ab = a · b ∈ F.
• For each a, b ∈ F, a + b = b + a.
• For each a, b, c ∈ F, a + (b + c) = (a + b) + c.
• For each a ∈ F, a + 0 = 0 + a = a.
• For each a ∈ F, there is an element b ∈ F such that a + b = b + a = 0.
• For each a, b ∈ F, ab = ba.
• For each a, b, c ∈ F, a(bc) = (ab)c.
• For each a ∈ F, a1 = 1a = a.
• For each a ∈ F such that a 6= 0, there is an element c ∈ F such that ac = ca =
1.
• For each a, b, c ∈ F, a(b + c) = ab + ac.

finite-dimensional:
A vector space V is finite-dimensional if there exists a finite list of vectors (v1 , . . . , vn )
in V such that V ⊆ hv1 , . . . , vn i.

fixed point:
A fixed point of a function f : X → X is an element x ∈ X such that f (x) = x.

four fundamental subspaces:

Let V and W be finite-dimensional inner product spaces. The four fundamental
subspaces associated to a linear map T ∈ L(V, W ) are range T , ker T , range T ∗ ,
and ker T ∗ .

free variable:
A variable in a linear system is called a free variable if the corresponding column in
the RREF of the augmented matrix of the system does not contain a pivot.

Frobenius inner product:

The Frobenius inner product of A, B ∈ Mm,n (C) is
m X
X n
hA, BiF = tr AB∗ = ajk bjk .
j=1 k=1
 15

Frobenius norm:
The Frobenius norm of A ∈ Mm,n (C) is
v
√ u m X
n
uX
∗
kAkF = tr AA = t |ajk |2 .
j=1 k=1

function:
Let X and Y be sets. A function f from X to Y , written f : X → Y , is a rule that
assigns an output f (x) ∈ Y to each element x ∈ X.

function space:
A function space is a vector space whose elements are functions.

* functional calculus:
Let A ∈ Mn (C) be Hermitian with spectral decomposition A = U diag(λ1 , . . . , λn )U∗ .
The functional calculus defines the matrix f (A) ∈ Mn (C) by

f (A) = U diag(f (λ1 ), . . . , f (λn ))U∗

for any function f : R → R. More generally, f need only be defined on the eigen-
values λ1 , . . . , λn of A.

Gaussian elimination:
Gaussian elimination, or Gauss–Jordan elimination, is the algorithm presented in
the proof of Theorem 1.1 to put a linear system into reduced row-echelon form via a
specific sequence of row operations.

* geometric multiplicity:
Let λ ∈ F be an eigenvalue of A ∈ Mn (F). The geometric multiplicity of λ as an
eigenvalue of A is null(A − λIn ) = dim Eigλ (A).

Hamming code:
The Hamming code is the binary linear code with encoding and parity check matri-
ces  
1 0 0 0
0 1 0 0
   
0 0 1 0 1 1 1 0 1 0 0
 
0 0 0 1
A= and B = 1 1 0 1 0 1 0 ,

1 1 1 0 1 0 1 1 0 0 1
 
1 1 0 1
1 0 1 1
respectively.
 16

Hermitian matrix:
A matrix A ∈ Mn (C) is Hermitian if A∗ = A.
* Hermitian part:
The Hermitian part of a matrix A ∈ Mn (C) is the matrix
1
Re A := (A + A∗ ).
2
Homogeneous (function):
Let V be a vector space over F. A function f on V is homogeneous if f (cv) = cf (v)
for all c ∈ F and all v ∈ V .
homogeneous linear system:
A linear system
a11 x1 + · · · + a1n xn = b1
..
.
am1 x1 + · · · + amn xn = bm
is called homogeneous if bi = 0 for each i = 1, . . . , m.
* Householder matrix:
A Householder matrix is a matrix of the form In − 2xx∗ , where x ∈ Cn is a unit
vector.
identity function:
The identity function on a set X is the function ι : X → X such that for each
x ∈ X, ι(x) = x.
identity matrix:
The n × n identity matrix In ∈ Mn (F) is the n × n matrix with ones on the diagonal
and zeroes elsewhere.
identity operator:
Let V ge a vector space. The identity operator on V is the function I : V → V
such that for each v ∈ V ,
I(v) = v.

image:
Let f : X → Y be a function and A ⊆ X. The image of A under f is the set

f (A) := {f (x) | x ∈ A} .

imaginary part:
Let z = a + ib for a, b ∈ R. Then the imaginary part of z is Im z = b.
 17

* inclusion map:
Suppose that X ⊆ Y . The inclusion map i is the function i : X → Y given by
i(x) = x for each x ∈ X.
inconsistent (system of equations):
A system of equations is inconsistent if there exist no solutions to the system.
induced norm:
See operator norm.
induction hypothesis:
In a proof by (strong) induction that a family of statements P (n) are true for each
value of n ∈ N, the induction hypothesis is the assumption that P (n) is true for
every n ≤ N ; this is then used to prove that P (N + 1) is also true.
infinite-dimensional:
A vector space V is infinite-dimensional if there exists no finite list of vectors in V
which spans V .
injective:
A function f : X → Y is injective or one-to-one if, whenever f (x1 ) = f (x2 ), it
holds that x1 = x2 .
inner product:
Let F be either R or C, and let V be a vector space over F. An inner product is an
operation on V , written hv, wi, such that the following properties hold:
• For each v, w ∈ V , hv, wi ∈ F.
• For each u, v, w ∈ V , hu + v, wi = hu, wi + hv, wi.
• For each v, w ∈ V and a ∈ F, hav, wi = a hv, wi.
• For each v, w ∈ V , hv, wi = hw, vi.
• For each v ∈ V , hv, vi ≥ 0.
• If hv, vi = 0, then v = 0.
inner product space:
An inner product space is a real or complex vector space together with an inner
product.
* (integral) kernel:
See integral operator.
* integral operator:
Let k : R2 → R be continuous. An operator Tk : C[a, b] → C[a, b] defined by
Z b
Tk f (x) := k(x, y)f (y) dy
a
 18

is called an integral operator with kernel k. The kernel k is often referred to as an

integral kernel to distinguish it from the completely different notion of the kernel of
a linear map.
intersection:
The intersection of sets S1 and S2 is the collection S1 ∩ S2 of every element which
is in both S1 and S2 . That is,
S1 ∩ S2 = {s | s ∈ S1 and s ∈ S2 } .

invariant (of a matrix):

An invariant on Mn (F) is any property of a matrix in Mn (F) which is preserved by
similarity.
inverse (matrix):
Let A ∈ Mn (F). The matrix B ∈ Mn (F) is an inverse (matrix) of A if
AB = BA = In .
In that case we write B = A−1 .
inverse function:
A function g : Y → X is called an inverse function (or simply inverse) of f : X →
Y if g ◦ f is the identity function on X and f ◦ g is the identity function on Y . That
is, for each x ∈ X,
g ◦ f (x) = x,
and for each y ∈ Y ,
f ◦ g(y) = y.
−1
In that case we write g = f .
invertible function:
A function f : X → Y is invertible if there exists an inverse function to f .
invertible matrix:
A matrix A ∈ Mn (F) is invertible or nonsingular if there exists an inverse matrix
to A.
isometric:
Normed spaces V and W are isometric if there exists an isometry T : V → W .
isometry:
Let V and W be normed spaces. A linear map T : V → W is an isometry if T is
surjective
kT vk = kvk
for every v ∈ V .
See also Perspectives: Isometries at the end of chapter 4.
 19

isomorphic:
Vector spaces V and W are isomorphic if there exists an isomorphism between V
and W .
isomorphism:
Let V and W be vector spaces. An isomorphism between V and W is a bijective
linear map T : V → W .
See also Perspectives: Isomorphisms at the ends of chapters 3 and 6.
isoscopic:
A function f : Mn (F) → F is called isoscopic if f (A) = 0 whenever A has two
identical columns.
(This is not a standard term. It is introduced in this book for expositional conve-
nience. When f is also assumed to be a multilinear function, the term alternating is
used instead.)
Jordan measurable:
A set Ω ⊆ Rn is Jordan measurable if, whenever Ω is approximated by a sequence
of simple sets, the volumes of the simple sets actually do approach a limit, which is
independent of the sequence.
kernel:
The kernel or null space of a linear map T ∈ L(V, W ) is the set

ker T := {v ∈ V | T (v) = 0} .

The kernel or null space of a matrix A ∈ Mm,n (F) is the kernel of the linear map
from Fn → Fm represented by A. That is, it is the set

ker A := {x ∈ Fn | Ax = 0} .

kernel, integral:
See integral operator.
* `1 norm:
The `1 norm on Rn or Cn is defined by
n
X
kxk1 := |xj | .
j=1

* `∞ norm:
The `∞ norm on Rn or Cn is defined by

kxk∞ := max |xi | .

1≤j≤n
 20

* L1 norm:
The L1 norm on the space C([a, b]) is defined by
Z b
kf k1 := |f (x)| dx.
a

* LDU decomposition:
An LDU decomposition of A ∈ Mm,n (F) is a factorization A = LDU, where:
• L ∈ Mm (F) is lower triangular with `i,i = 1 for each i,
• D ∈ Mm (F) is diagonal, and
• U ∈ Mm,n (F) is upper triangular with ui,i = 1 for each i.
left-invertible:
A matrix A ∈ Mm,n (F) is left-invertible if there exists B ∈ Mn,m (F) such that
BA = In .
left singular vectors of a map:
See singular vectors.
left singular vectors of a matrix:
Let A ∈ Mm,n (F) and suppose that A = UΣV∗ is a singular value decomposition
of A; the columns of U are called left singular vectors of A.
length:
See norm.
linear combination:
Let V be a vector space over F. A linear combination of vectors v1 , v2 , . . . , vk ∈ V
is a vector of the form
k
X
ci v i = c1 v 1 + c2 v 2 + · · · + ck v k ,
i=1

where c1 , c2 , . . . , ck ∈ F.
linear encoding:
An linear encoding is a bijective function
T : Fk2 → C,
where C ⊆ Fn2 is a binary linear code.
linear map:
Let V and W be vector spaces. A function T : V → W is called a linear map
(or linear transformation, linear operator, or simply operator) if it additive and
homogeneous; i.e.,
 21

• For each u, v ∈ V , T (u + v) = T (u) + T (v).

• For each v ∈ V and a ∈ F, T (av) = aT (v).

linear operator:
See linear map.

linear system of equations:

Let F be a field. A linear system over F of m equations in n variables (or m × n
linear system over F) is a set of equations of the form

a11 x1 + · · · + a1n xn = b1 ,
..
.
am1 x1 + · · · + amn xn = bm .

Here, ai,j ∈ F for each pair (i, j) with 1 ≤ i ≤ m and 1 ≤ j ≤ n, bi ∈ F for each
1 ≤ i ≤ m, and x1 , . . . , xn are the n variables of the system.

linear transformation:
See linear map.

linearly dependent:
Let V be a vector space over F. A list of vectors (v1 , . . . , vn ) in V is linearly depen-
dent if there exist scalars a1 , . . . , an ∈ F which are not all 0 such that
n
X
aj vj = 0.
j=1

linearly independent:
Let V be a vector space over F. A list of vectors (v1 , . . . , vn ) in V is linearly inde-
pendent if it is not linearly dependent; i.e., (v1 , . . . , vn ) in V is linearly independent
if, whenever
Xn
aj vj = 0
j=1

for scalars a1 , . . . , an ∈ F, we must have ai = 0 for each i.

lower triangular:
A matrix A ∈ Mm,n (F) is lower triangular if aij = 0 whenever i < j.
(This term is most commonly applied in the case that m = n.)

LU decomposition:
An LU decomposition of A ∈ Mm,n (F) is a factorization A = LU, where L ∈
Mm (F) is lower triangular and U ∈ Mm,n (F) is upper triangular.
 22

LUP decomposition:
An LUP decomposition of A ∈ Mm,n (F) (or an LU decomposition with pivoting)
is a factorization PA = LU, where:
• P ∈ Mm (F) is a permutation matrix (that is, a product of elementary matrices
of type Ri,j ),
• L ∈ Mm (F) is lower triangular, and
• U ∈ Mm,n (F) is upper triangular.
magnitude:
See norm.
matrix:
Let m and n be natural numbers and let F be a field. An m × n matrix A over F is
a doubly-indexed collection of elements of F
 
a11 a12 · · · a1n
   a21 a22 · · · a2n 
A = aij 1≤i≤m =  .. ..  .
 
.. . .
1≤j≤n  . . . . 
am1 am2 · · · amn

That is, for each ordered pair of integers (i, j) with i ∈ {1, 2, . . . , m} and j ∈
{1, 2, . . . , n}, there is a corresponding element aij ∈ F. The
 number aij is called
the (i, j) entry of A. We denote the (i, j) entry of A by A ij .

* matrix exponential:
The matrix exponential of A ∈ Mn (C) is the matrix
n
A
X 1 k
e = A .
k=0
k!

matrix form (of a linear system):

The matrix form of the linear system
a11 x1 + · · · + a1n xn = b1
..
.
am1 x1 + · · · + amn xn = bm
over F is the matrix equation
Ax = b,
   
x1 b1
   ..   .. 
where A := aij 1≤i≤m , x :=  . , and b :=  . .
1≤j≤n
xn bm
 23

matrix multiplication:
Given A ∈ Mm,n (F) and B ∈ Mn,p (F), the product AB ∈ Mm,p (F) is the unique
matrix such that for every v ∈ Fp ,


A (Bv) = AB v.

The entries of AB are given by

n
X
 
AB ij = aik bkj .
k=1

See also Perspectives: Matrix Multiplication at the end of chapter 2.

matrix of T with respect to B:

See matrix of T with respect to BV and BW .

matrix of T with respect to BV and BW :

Let BV = (v1 , . . . , vn ) be a basis of V and BW = (w1 , . . . , wm ) a basis of W , and
let T ∈ L(V, W ). For each 1 ≤ j ≤ n, define the scalars aij for 1 ≤ i ≤ m to be the
coordinates of T vj with respect to BW ; i.e.,

T vj = a1j w1 + · · · + amj wm .

The matrix of T with respect to BV and BW is the matrix [T ]BV ,BW = A =

 
aij 1≤i≤m with these entries. It has the defining property that for all v ∈ V ,
1≤j≤n

[T ]BV ,BW [v]BV = [T v]BW .

When V = W and BV = BW = B, this is called the matrix of T with respect to B

and denoted [T ]B .

matrix of T :
Let T ∈ L(Fn , Fm ). The matrix of T is the unique matrix A ∈ Mm,n (F) such that
for every v ∈ Fn ,
T (v) = Av.

The matrix of T ∈ L(Fn , Fm ) coincides with the matrix of T with respect to the
standard bases of Fn and Fm .

matrix-vector multiplication:  
v1
   .. 
Let A = aij 1≤i≤m ∈ Mm,n (F) and v =  .  ∈ Fn . The product Av ∈ Fm is
1≤j≤n
vn
 24

defined by
n
 
X
 a1j vj 
 j=1 
 
Av :=  ..
.
 
 n . 
X 
 amj vj 
j=1

modulus:
See absolute value.
multilinear:
A function D : Mn (F) → F is multilinear if, for every a1 , . . . , an and b1 , . . . , bn ∈
Fn and c ∈ F, for each j = 1, . . . , n,
 
| | |
D a1 · · · aj + bj · · · an 
| | |
   
| | | | | |
= D a1 · · · aj · · · an  + D a1 · · · bj · · · an 
| | | | | |
and
   
| | | | | |
D a1 · · · caj · · · an  = cD a1 · · · aj · · · an  .
| | | | | |

That is, D(A) is linear when thought of as a function of one of the columns of A,
with the other columns held fixed.
* multiplication mod p:
Let p ∈ N be a prime number. For a, b ∈ {0, 1, . . . , p − 1}, ab = r (mod p)
(sometimes written ab = r in Fp ) if r ∈ {0, 1, . . . , p − 1} is the remainder when ab
is divided by p.
multiplicity (of a root of a polynomial):
Let p(x) be a polynomial over F such that p(x) = b(x − c1 )k1 · · · (x − cm )km with
c1 , . . . , cm ∈ F distinct. Then c1 , . . . , cm are the roots of p and the multiplicity of
the root cj is kj .
multiplicity (of an eigenvalue):
Let V be a vector space, T ∈ L(V ), and let λ be an eigenvalue of T . The multiplic-
ity of the eigenvalue λ is its multiplicity as a root of the characteristic polynomial of
T.
 25

n-dimensional vector over F:

Let n ∈ N and let F be a field. An n-dimensional vector
 over F is an ordered list of
v1
 v2 
n elements of F arranged as a column; i.e., v =  .. .
 
.
vn
negation:
The negation of the logical statment “A” is “not A”; the statement “not A” is true if
and only if the statement “A” is false.
nonsingular:
A matrix A ∈ Mn (F) is nonsingular if and only if it is invertible.
normal matrix:
A matrix A ∈ Mn (C) is normal if A∗ A = AA∗ .
normal operator:
Let V be an inner product space and let T ∈ L(V ); T is a normal operator if
T ∗T = T T ∗.
normed space:
A normed space is a vector space V together with a norm k · k : V → R.
norm:
Let V be a vector space over R or C. A norm on V is a real-valued function k·k on
V such that the following properties hold:
• For each v ∈ V , kvk ≥ 0.
• If kvk = 0, then v = 0.
• For each v ∈ V and a ∈ F, kavk = |a| kvk.
• For each v, w ∈ V , kv + wk ≤ kvk + kwk.
If V is an inner product space, then the norm of v ∈ V is
p
kvk := hv, vi.
This defines a norm in the above sense.
null space:
See kernel.
nullity:
Let V and W be vector spaces and let T ∈ L(V, W ). The nullity of T is the
dimension of the kernel of T . For a matrix A ∈ Mm,n (F), the nullity of A is the
dimension of the kernel of A.
 26

one-to-one correspondence:
See bijective.

one-to-one:
A function f : X → Y is one-to-one if whenever f (x1 ) = f (x2 ), x1 = x2 .
Also known as injective.

onto:
A function f : X → Y is onto if for every y ∈ Y , there is an x ∈ X so that f (x) = y.
Also known as surjective.

operator:
See linear map.

operator norm:
Let V and W be inner product spaces and T ∈ L(V, W ). The operator norm of T
is
kT kop := max kT vk .
v∈V,
kvk=1

If A ∈ Mm,n (F), then the operator norm of A is the operator norm of the linear
map in L(Fn , Fm ) whose matrix is A.
Also known as the spectral norm.

orthogonal:
Two vectors v, w ∈ V are orthogonal if hv, wi = 0. Also known as perpendicular.
A list of vectors (v1 , . . . , vn ) in V is called orthogonal if hvj , vk i = 0 whenever
j 6= k.

orthogonal complement:
Let U be a subspace of an inner product space V . The orthogonal complement of
U is
U ⊥ = {v ∈ V | hu, vi = 0 for every u ∈ U } .

orthogonal direct sum:

Let U1 , . . . , Um be subspaces of an inner product space V ; V is the orthogonal direct
sum of U1 , . . . , Um (written V = U1 ⊕ · · · ⊕ Um ) if:

• every vector v ∈ V can be written as

v = u1 + · · · + um

for some u1 ∈ U1 , . . . , um ∈ Um
• whenever uj ∈ Uj and uk ∈ Uk for j 6= k, uj and uk are orthogonal.
 27

orthogonal matrix:
A matrix A ∈ Mn (R) is orthogonal if AT A = In
See also Perspectives: Isometries at the end of chapter 4.

orthogonal projection:
Let U be a subspace of V . The orthogonal projection PU : V → V is defined by
PU (v) = u, where v = u+w is the unique representation of v as a sum of an element
of U and an element of U ⊥ .

orthonormal:
Let V be an inner product space. A finite or countable set {ei } of vectors in V is
orthonormal if (
0 if j 6= k,
hej , ek i =
1 if j = k.

orthonormal basis:
Let V be a finite-dimensional inner product space. An orthonormal basis of V is a
basis B = (e1 , . . . , en ) which is also an orthonormal list.

overdetermined:
An m × n linear system is called overdetermined if m > n.

parity bit code:

The parity bit code is the encoding of Fk2 in Fk+1
2 given by x 7→ y, with
(
xi , i = 1, . . . , k;
yi =
x1 + · · · + xk , i = k + 1.

parity-check matrix:
For a linear encoding with encoding matrix A ∈ Mn,k (F2 ), a matrix B ∈ Mm,n (F2 )
is called a parity-check matrix for A if ker B = C(A).

permanent:
The permanent of a matrix A ∈ Mn (F) is defined by
X
perA = a1σ(1) · · · anσ(n) ,
σ∈Sn

where Sn denotes the set of permutations on n letters.

permutation:
A permutation of the set S is a bijection from S to itself.
 28

permutation matrix:
Let σ be a permutation of the set {1, . . . , n}. The permutation matrix associated to
σ is the matrix Aσ ∈ Mn (R) with entries
(
1, σ(i) = j;
aij =
0, otherwise.

A matrix A ∈ Mn (R) is the permutation matrix associated to some permutation σ if

and only if it has exactly one 1 in each row and in each column, and zeros otherwise.

permutation (on n letters):

A permutation on n letters is a bijective function {1, . . . , n} → {1, . . . , n}.

perpendicular:
Two vectors v, w in an inner product space are perpendicular if hv, wi = 0. Also
known as orthogonal.

pivot variable:
A variable in a linear system is called a pivot variable if the corresponding column
in the RREF of the augmented matrix of the system contains a pivot.

pivot:
Let A ∈ Mm,n (F) be in row-echelon form. An entry of A which is the first nonzero
entry in its row (and is therefore a 1) is called a pivot.

pointwise addition (of functions):

See addition (of functions).

polarization identities:
The polarization identity in a real inner product space V is the fact that
1
kv + wk2 − kv − wk2


hv, wi =
4
for each v, w ∈ V .
The polarization identity in a complex inner product space V is the fact that

1
kv + wk2 − kv − wk2 + i kv + iwk2 − i kv − iwk2


hv, wi =
4
for each v, w ∈ V .

positive definite:
A matrix A ∈ Mn (C) is positive definite if it is Hermitian (i.e., A = A∗ ) and for all
nonzero x ∈ Cn , hAx, xi > 0.
 29

positive homogeneity:
Let V be a vector space over R or C. A scalar-valued function f : V → F is
positively homogeneous if f (cv) = |c|f (v) for all scalars c.

* positive semidefinite:
A matrix A ∈ Mn (C) is positive semidefinite if it is Hermitian (i.e., A = A∗ ) and
for all x ∈ Cn , hAx, xi ≥ 0.

proper subset:
A set A is a proper subset of the set B if A is a subset of B and there is at least one
b ∈ B with b ∈/ A.

* pseudoinverse:
Let A ∈ Mm,n (C) with singular value decomposition A = UΣV∗ . The pseudoin-
verse A† ∈ Mn,m (C) of A is
A† = VΣ† U∗ ,
where Σ† is an n × m matrix with 1
σj
in the (j, j) entry for 1 ≤ j ≤ rank A and
zeroes otherwise.

QR decomposition:
Let F be either R or C, and let A ∈ Mn (F) be invertible. If A = QR for Q ∈ Mn (F)
unitary (if F = C) or orthogonal (if F = R) and R ∈ Mn (F) upper triangular, then
A = QR is called a QR decomposition of A.

quantifiers:
The two basic logical quantifiers are “there exists” (written ∃) and “for all” (written
∀).

range:
Let f : X → Y be a function. The range of f is the set {f (x) : x ∈ X}.

rank (of a linear map):

Let V and W be vector spaces and let T ∈ L(V, W ). The rank of T is the dimension
of range T .

rank (of a matrix):

Let A ∈ Mn (F). The rank of A is the dimension of the column space of A; equiva-
lently, rank A is the dimension of the range of the linear map in L(Fn , Fm ) given by
multiplication by A.

real part:
Let z ∈ C. If z = a + ib with a, b ∈ R, then a is the real part of z. Equivalently, the
real part of z is z+z
2
. Denoted Re(z).
 30

real vector space:

A real vector space is a vector space with base field R.

rectangle:
Let n ∈ N and a1 , . . . , an , b1 , . . . , bn ∈ R. A set of the form

R = {(x1 , . . . , xn ) : a1 ≤ xi ≤ bi , i = 1, . . . , n}

is called a rectangle in Rn .

reduced row-echelon form (RREF):

A matrix A ∈ Mm,n (F) is in reduced row-echelon form (RREF) if:

1. A is in row-echelon form, and

2. any column which contains a pivot contains no other nonzero entries.

right-invertible:
A matrix A ∈ Mm,n (F) is called right-invertible if it has a right inverse; that is, if
there is a matrix B ∈ Mn,m (F) such that AB = Im .

right singular vectors of a map:

See singular vectors.

right singular vectors of a matrix:

Let A ∈ Mm,n (F) and suppose that A = UΣV∗ is a singular value decomposition
of A; the columns of V are called right singular vectors of A.

row operations:
The three elementary row operations that can be performed on a matrix are:

R1: Add a multiple of one row to another row.

R2: Multiply a row by a nonzero constant.
R3: Switch any two rows.

row rank:
Let A ∈ Mm,n (F). The row rank of A is the dimension of the row space of A.

row space:
Let A ∈ Mm,n (F). The row space of A is the span of the rows of A. The row space
is usually thought of as a subspace of Fn by identifying the rows of A with their
transposes.

row vector:
A row vector is a matrix v ∈ M1,m (F) for some m ∈ N.
 31

row-echelon form (REF):

A matrix A ∈ Mm,n (F) is in row-echelon form (REF) if:
1. every row containing any nonzero entries is above every row which contains
only zeroes,
2. the first nonzero entry in each row is 1, and
3. if i < j then the first nonzero entry in the ith row is strictly farther to the left
than the first nonzero entry in the j th row.
scalar:
Let V be a vector space over a field F. Elements of F are called scalars.
scalar multiple:
Let V be a vector space over a field F and let v ∈ V . A scalar multiple of v is a
vector of the form cv for some c ∈ F.
scalar multiplication:
Scalar multiplication is one of the two fundamental operations of a vector space,
which maps the pair (c, v) ∈ F × V to the vector cv. Scalar multiplication is required
to satisfy the associative law
a(bv) = (ab)v
for all a, b ∈ F and v ∈ V , and the two distributive laws
a(v + w) = av + aw (a + b)v = av + bv
for all a, b ∈ F and v, w ∈ V .
n
scalar multiplication
 (in F ):
a1
 .. 
Given v =  .  ∈ Fn and c ∈ F, the scalar multiplication of v by c is defined by
a
  n
ca1
 .. 
cv :=  .  .
can
self-adjoint:
Let V be an inner product space. A linear map T ∈ L(V ) is self-adjoint if T ∗ = T .
set: A set is a collection of distinct objects, called elements of the set.
sign (of a permutation):
Let σ ∈ Sn , and let Aσ be the associated permutation matrix; i.e., Aσ has entries
(
1, σ(i) = j;
aij =
0, otherwise.
 32

The sign of σ is sgn(σ) = det Aσ .

similar:
A matrix A ∈ Mn (F) is similar to B ∈ Mn (F) if there is an invertible matrix S ∈
Mn (F) such that B = SAS−1 .
simple linear regression:
For a set of points {(xi , yi )}ni=1 ⊆ R2 , simple linear regression is a method of
choosing a line y = mx + b which is the best fit to these points, in the sense that
these values of m and b minimize the expression
n
X
(mxi + b − yi )2 .
i=1

simple set:
A set S ⊆ Rn is called a simple set if it is the union of finitely many rectangles.
singular:
A matrix A ∈ Mn (F) is singular if it is not invertible. A linear map T is singular if
ker T 6= {0}.
singular value decomposition (SVD) of a map:
Let V and W be finite-dimensional inner product spaces and let T ∈ L(V, W ) with
rank T = r. A singular value decomposition of T is a pair of orthonormal bases
(e1 , . . . , en ) of V and (f1 , . . . , fm ) of W together with numbers σ1 ≥ · · · ≥ σr > 0
such that (
σj fj , j = 1, . . . , r;
T ej =
0, j = r + 1, . . . , n.

singular value decomposition (SVD) of a matrix:

Let A ∈ Mm,n (F) with rank A = r. A singular value decomposition of A is
a factorization A = UΣV∗ such that U ∈ Mm (F) and V ∈ Mn (F) are unitary
(orthogonal if F = R), Σ ∈ Mm,n (R) has (j, j)-entry σj for 1 ≤ j ≤ r, and all other
entries 0, such that σ1 ≥ · · · ≥ σr > 0.
singular values of a map:
Let V and W be finite-dimensional inner product spaces and let T ∈ L(V, W ) with
rank T = r. If the orthonormal bases (e1 , . . . , en ) of V and (f1 , . . . , fm ) of W
together with numbers σ1 ≥ · · · ≥ σr > 0 are an SVD of T , then the numbers σj
for 1 ≤ j ≤ r together with σj := 0 for r + 1 ≤ j ≤ p = min{m, n} are called the
singular values of T .
singular values of a matrix:
Let A ∈ Mm,n (F) have SVD A = UΣV∗ . Let p = min{m, n}. The j-j entries
σ1 ≥ · · · ≥ σp of Σ are called the singular values of A.
 33

singular vectors of a map:

Let V and W be finite-dimensional inner product spaces and let T ∈ L(V, W ) with
rank T = r. If the orthonormal bases (e1 , . . . , en ) of V and (f1 , . . . , fm ) of W
together with numbers σ1 ≥ · · · ≥ σr > 0 are an SVD of T , then the vectors
(e1 , . . . , en ) are called the right singular vectors of T and the vectors (f1 , . . . , fm )
are called the left singular vectors of T .
singular vectors of a matrix:
Let A ∈ Mm,n (C), and suppose that A = UΣV∗ is an SVD for A. The columns
of V are called right singular vectors of A and the columns of U are called left
singular vectors of A.
solution (to a linear system):
A solution to the linear system
a11 x1 + · · · + a1n xn = b1 ,
.. (1)
.
am1 x1 + · · · + amn xn = bm .
is a set of c1 , . . . , cn such that
a11 c1 + · · · + a1n cn = b1 ,
..
.
am1 c1 + · · · + amn cn = bm .

span:
n., Let V be a vector space and v1 , . . . , vk ∈ V . The span of v1 , . . . , vk is the set of
all linear combinations of v1 , . . . , vk :
( k  )
X 
hv1 , v2 , . . . , vk i := ci vi  c1 , c2 , . . . , ck ∈ F .


i=1

v., The vectors v1 , . . . , vk ∈ V span V if V = hv1 , . . . , vk i.

spectral decomposition:
For A ∈ Mn (C), a factorization A = UDU∗ for D diagonal and U unitary is called
a spectral decomposition of A. Such a decomposition exists if and only if A is a
normal matrix.
spectral norm:
See operator norm.
spectrum:
The spectrum of a linear operator or a matrix is its set of eigenvalues.
 34

* stable rank:
Let A ∈ Mm,n (C) be nonzero. The stable rankof A is

kAk2F
srank A := .
kAk2op

standard basis:
In Fn , the standard basis is E = (e1 , . . . , en ), where ej ∈ Fn has a 1 in the jth entry
and zeros in all other entries.

standard basis vectors:

In Fn , the standard basis vectors are the vectors e1 , . . . , en , where ej ∈ Fn has a 1
in the jth entry and zeros in all other entries.

standard inner product:

In Rn , the standard inner product is given by
   
* a1 b1 + X n
 ..   .. 
 . , .  = aj b j .
an bn j=1

In Cn , the standard inner product is given by

   
* a1 b1 + Xn
 ..   .. 
 . , .  = aj b j .
an bn j=1

* strictly convex norm:

V is called strictly convex if for every v, w ∈ V with
A norm on a real vector space
1
kvk = kwk = 1, 2 (v + w) < 1.

* strictly upper triangular:

A matrix A ∈ Mn (F) with entries aij is strictly upper triangular if aij = 0 when-
ever i ≥ j.

* subfield:
Let F be a field. A subset F0 ⊆ F is a subfield of F if F0 is itself a field, with the field
operations inherited from F. This is equivalent to the requirement that F0 contain
0 and 1, be closed under addition and multiplication, that every element of F0 has
an additive inverse in F0 , and that every nonzero element of F0 has a multiplicative
inverse in F0 .

subset:
A set A is a subset of a set B if every element of A is also an element of B.
 35

subspace:
Let V be a vector space. A subset U ⊆ V is a subspace if U is itself a vector space,
with the vector space operations inherited from V . This is equivalent to the require-
ment that U contain 0 and be closed under the formation of linear combinations.
sum (of vectors in Fn ):
The sum of two vectors in Fn is defined by
     
a1 b1 a1 + b 1
 ..   ..   .. 
 .  +  .  =  . .
an bn an + b n
supremum norm:
Let f : A → C be a continuous function on a subset A ⊆ C. The supremum norm
of f is kf k∞ := maxa∈A |f (a)|.
surjective:
A function f : X → Y is surjective if for every y ∈ Y , there is an x ∈ X so that
f (x) = y.
Also known as onto.
symmetric group on n letters:
The symmetric group on n letters the set of all permutations of {1, . . . , n}
symmetric matrix:
A matrix A ∈ Mn (R) is symmetric if AT = A.
trace of a matrix:
Let A ∈ Mn (F). The trace of A is the sum of its diagonal entries:
n
X
tr A = aii .
i=1

trace of an operator:
Let T ∈ L(V ). The trace of T is the trace of any matrix representation of T .
transpose:  
n., Let A ∈ Mm,n (F) with A = aij 1≤i≤m . The transpose AT ∈ Mn,m (F) of A is
1≤j≤n
the matrix whose (i, j) entry is [AT ]ij := aji .
v., To take the transpose of a matrix.
* transposition:
A transposition is a permutation of {1, . . . , n} which exchanges two elements and
fixes all others; i.e. there are distinct i, j ∈ {1, . . . , n} such that σ(i) = j and
σ(j) = i, and σ(k) = k for all k 6= i, j.
 36

trigonometric system: n  o
The set of functions T := √12π , √1π sin(nθ), √1π cos(nθ)  n ∈ N is called the trigono-

metric system.

triple repetition code:

The triple repetition code is the encoding of F2 in F32 given by
 
x
x 7→ x .
x

underdetermined:
An m × n linear system is called underdetermined if m < n.

union:
The union of a family of sets {Xα }α∈A is written ∪α∈A Xα and is the collection of all
elements in any of the Xα ; i.e., x ∈ ∪α∈A Xα if and only if x ∈ Xα for some α ∈ A.

unit vector:
Let V be a normed space. A vector v ∈ V is a unit vector if kvk = 1.

unitarily diagonalizable:
A matrix A ∈ Mn (C) is unitarily diagonalizable if there is a diagonal matrix D
and a unitary matrix U such that A = UDU∗ .

unitarily invariant norm:

A norm k·k on Mn (C) is unitarily invariant if kAk = kU AV k for all A ∈ Mn (C)
and all unitary U, V.

unitary matrix:
A matrix U ∈ Mn (C) is unitary if U∗ U = In .
See also Perspectives: Isometries at the end of chapter 4.

upper triangular linear system:

An n × n linear system is upper triangular if its coefficient matrix is upper triangu-
lar.

upper triangular matrix:

A matrix A ∈ Mm,n (F) is upper triangular if aij = 0 whenever i > j.
(This term is most commonly applied when m = n.)
 37

* Vandermode determinant:
A Vandermonde determinant is the determinant of a matrix of the form
 
1 x0 x20 · · · xn0
1 x1 x2 · · · xn 
1 1
V :=  .. .. ..  ∈ Mn+1 (F).

..
. . . .
1 xn xn · · · xnn
2

It can be shown that for such a matrix V,

Y
det V = (xj − xi ).
0≤i<j≤n

vector:
A vector is an element of a vector space.

(vector) addition:
Vector addition is one of the two fundamental operations of a vector space, which
maps the pair (v1 , v2 ) ∈ V × V to the vector v1 + v2 . Vector addition is required to
be commutative and associative, and to satisfy the two distributive laws

a(v + w) = av + aw (a + b)v = av + bv

for all a, b ∈ F and v, w ∈ V .

vector space:
A vector space over a field F consists of a set V together with operations + and ·
and an element 0 ∈ V such that all of the following properties hold:

• For each v, w ∈ V , v + w ∈ V.
• For each c ∈ F and v ∈ V , c · v ∈ V.
• For each v, w ∈ V , v + w = w + v.
• For each u, v, w ∈ V , u + (v + w) = (u + v) + w.
• For each v ∈ V , v + 0 = 0 + v = v.
• For each v ∈ V , there is an element w ∈ V such that v + w = w + v = 0.
• For each v ∈ V , 1 · v = v.
• For each a, b ∈ F and v ∈ V , a · (b · v) = (ab) · v.
• For each a ∈ F and v, w ∈ V , a · (v + w) = a · v + a · w.
• For each a, b ∈ F and v ∈ V , (a + b) · v = a · v + b · v.
 38

well-defined:
A concept is well-defined if it is determined uniquely by its definition. This is an
issue when a definition appears to depend on an arbitrary choice; e.g., the dimension
of a vector space is the length of any basis of that space. For this to be a well-defined
concept, it must be verified that different choices of basis will result in the same value
for the dimension.

with multiplicity:
A set of objects which have a notion of multiplicity are counted with multiplicity if
the number of times each object is counted is given by its multiplicity. For example,
the polynomial p(x) = x3 (x − 1)2 has two root (x = 0 and x = 1), but five roots if
roots are counted with multiplicity, since x = 0 is a root of multiplicity 3 and x = 1
is a root of multiplicity 2.

zero map:
Let V and W be vector spaces. The map T ∈ L(V, W ) with T (v) = 0 for all v ∈ V
is called the zero map.