MICHAEL J.

NEELY, UNIVERSITY OF SOUTHERN CALIFORNIA, FALL 2007 1

EE 441: General vector spaces over a field

I. A XIOMS FOR A V ECTOR S PACE V OVER A F IELD F
Let F be a general field. Let V be a set of objects called “vectors.” We say that V is a vector space over the field F if there
are rules for vector addition and scalar multiplication such that V is closed with respect to these operations, that is:
• v 1 + v 2 ∈ V for any two vectors v 1 , v 2 ∈ V.

• αv ∈ V for any vector v ∈ V and any scalar α ∈ F.

and such that the following additional six properties hold:
1) (Commutativity)

v 1 + v 2 = v 2 + v 1 (for all v 1 , v 2 ∈ V)
2) (Associativity)

(v 1 + v 2 ) + v 3 = v 1 + (v 2 + v 3 ) (for all v 1 , v 2 , v 3 ∈ V)
(αβ)v = α(βv) (for all α, β, ∈ F and v ∈ V)
3) (Distributive Properties)

α(v 1 + v 2 ) = αv 1 + αv 2 (for all v 1 , v 2 ∈ V, and α ∈ F)

(α + β)v = αv + βv (for all v ∈ V, and α, β ∈ F)
4) (Additive Identity) There exists a vector 0 ∈ V such that:
v + 0 = v for all v ∈ V
5) (Additive Inverse) For every v ∈ V, there exists a vector −v ∈ V such that:

v + −v = 0
6) (Multiplicative Identity) 1v = v for every vector v ∈ V.
Note that the vector space Rn over the field R satisfies all these properties. Similarly, the set Fn is a vector space over F,
where vector addition and scalar multiplication are defined entrywise via the arithmetic of F. Other examples of vector spaces:
• The vector space V over the field F (where F is any general field), consisting of all countably infinite tuples (x1 , x2 , x3 , . . .),
where xi ∈ F for all entries i ∈ {1, 2, . . . , }, and where arithmetic is defined entrywise using arithmetic in F.

• The vector space V over the field R, where V is the set of all continuous functions of time t for t ∈ (−∞, ∞).

• The vector space V over the field R, where V is the set of all polynomial functions f (t) of degree less than or equal to
n. That is, V = {f (t) | f (t) = α0 + α1 t + α2 t2 + . . . + αn tn , αi ∈ R}.

II. S IMPLE L EMMAS FOR V ECTOR S PACES

Let V be a vector space over a field F.
Lemma 1: (Uniqueness of 0) The vector 0 ∈ V is the unique additive identity.
Proof: Suppose that w satisfies v + w = v for any v ∈ V. Then adding −v to both sides of the equation v + w = v yields:
−v + v + w = −v + v
and hence: 0 + w = 0. Therefore, w = 0.
Lemma 2: (Uniqueness of −v) For any vector v ∈ V, if there is a vector w ∈ V such that v + w = 0, then w = −v.
Proof: Suppose that v + w = 0. Adding −v to both sides yields 0 + w = −v, and hence w = −v.
Lemma 3: 0v = 0 for any v ∈ V.
Proof: Take any vector v ∈ V. Then:
v = 1v = (1 + 0)v = 1v + 0v
and hence v = v + 0v. Adding −v to both sides yields 0 = 0v, proving the result.
MICHAEL J. NEELY, UNIVERSITY OF SOUTHERN CALIFORNIA, FALL 2007 2

Lemma 4: α0 = 0 for any α ∈ F.

Proof: Take any α ∈ F and any vector v ∈ V. Then:
αv = α(v + 0)
= αv + α0
Thus, we have αv = αv + α0. Adding −(αv) to both sides yields 0 = α0, proving the result.

Lemma 5: −v = (−1)v for any v ∈ V.

Proof: Take any vector v ∈ V. Then:
0 = 0v = (1 + −1)v = 1v + (−1)v = v + (−1)v
Thus, 0 = v + (−1)v. Adding −v to both sides yields −v = (−1)v, proving the result.

III. S UBSPACES
Definition 1: Let V be a vector space over a field F. Let S ⊂ V be a subset of V. We say that S is a subspace if:
v1 + v2 ∈ S for all v 1 , v 2 ∈ S
αv ∈ S for all v ∈ V, α ∈ F
where addition and scalar multiplication are the same in S as they are in V.
It is easy to prove that if S is a subspace of vector space V over field F, then S is itself a vector space over field F. (It is
important to note that −v = (−1)v in proving this...why?).

IV. L INEAR C OMBINATIONS AND L INEAR I NDEPENDENCE

Definition 2: Let {x1 , . . . , xk } be a collection of vectors in a vector space V over a field F. We say that a vector v ∈ V is
a linear combination of {x1 , . . . , xk } if it can be written: v = α1 x1 + . . . + αk xk for some scalars αi ∈ F for i ∈ {1, . . . , k}.

Definition 3: Let {x1 , . . . , xk } be a collection of vectors in a vector space V over a field F. We define Span{x1 , . . . , xk }
as the set of all linear combinations of {x1 , . . . , xk }. Note that Span{x1 , . . . , xk } ⊂ V.

Definition 4: Let S be a subspace of a vector space V over a field F. We say that a collection of vectors {x1 , . . . , xk } span
S if Span{x1 , . . . , xk } = S.

Definition 5: We say that a collection of vectors {x1 , . . . , xk } in a vector space V (over a field F) are linearly independent
if the equation α1 v 1 + α2 v 2 + . . . + αk v k = 0 can only be true if αi = 0 for all i ∈ {1, . . . , k}.

Definition 6: A collection of vectors {x1 , . . . , xk } is a basis for a subspace S if the collection {x1 , . . . , xk } is linearly
independent in S and spans S.

The following lemmas have proofs that are identical (or nearly identical) to the corresponding lemmas proven in class for
the vector space Rn . The proofs are left as an exercise.

Lemma 6: A collection of vectors {x1 , . . . , xk } are linearly independent if and only if none of the vectors can be written
as a linear combination of the others.

Lemma 7: If {x1 , . . . , xk } are linearly independent in a vector space V, and if w ∈ V and w ∈

/ Span{x1 , . . . , xk }, then
{x1 , x2 , . . . , xk , w} are linearly independent.
MICHAEL J. NEELY, UNIVERSITY OF SOUTHERN CALIFORNIA, FALL 2007 3

Lemma 8: (k ≤ m) Let {x1 , . . . , xk } be a collection of vectors that are linearly independent in a subspace S. Let
{y 1 , . . . , y m } be a collection of vectors that span S. Then k ≤ m.

Lemma 9: Any two bases of a subspace S have the same size, defined as the dimension of the subspace.

Lemma 10: The dimension of Fn is n.

Lemma 11: Let S be a subspace of a vector space V, where V has dimension n. Then S has a finite basis, and the dimension
of S is less than or equal to n.

Note that the standard basis for Fn is given by {e1 , . . . , en }, where ei is a n-tuple with all entries equal to 0 except for
entry i, which is equal to 1.
Note that the collection {1, t, t2 , . . . , tn } is a basis for the vector space V over the field R, where V is the space of all
polynomial functions of degree less than or equal to n (why is this true?). Thus, this vector space has dimension n + 1. Note
also that, for any n, this vector space is a subspace of the vector space over R defined by all continuous functions. Thus, the
dimension of the vector space of all continuous functions is infinite (as it contains subspaces of dimension n for arbitrarily
large n).

V. M ATRICES
Let A be a m × n matrix with elements in F. Note that the equation Ax = 0 (where 0 ∈ Fm and x ∈ Fn ) has only the
trivial solution x = 0 ∈ Fn if and only if the columns of A are linearly independent.

Lemma 12: A square n×n matrix A (with elements in F) is non-singular if and only if its columns are linearly independent,
if and only if Ax = 0 has only the trivial solution.
The above lemma follows from the fact that if A is non-singular, it has a single unique solution to Ax = b for all b ∈ Fn
(which is true by Gaussian Elimination), and if it is singular it does not have a solution for some vectors b ∈ Fn and it has
multiple solutions for the remaining b ∈ Fn .

Lemma 13: Let A, B be square n × n matrices. If AB = I, then both A and B are invertible, and A−1 = B and B −1 = A.
The above lemma follows from the fact that AB = I implies A is non-singular (why?) and hence invertible.

Lemma 14: A square n × n matrix A (with elements in F) has linearly independent columns (and hence is invertible) if
and only if its transpose AT has linearly independent columns (and hence is invertible). Thus, a square invertible matrix A
has both linearly independent rows and linearly independent columns.
The above lemma follows from the fact that AA−1 = I, and hence (A−1 )T AT = I.

VI. BASIC P ROBABILITY

The next several lectures on erasure coding will use the following simple but important probability facts:
• If a probability experiment has K equally likely outcomes, then the probability of each individual outcome is 1/K.
• Let E1 , E2 , . . . , Em be a set of independent events (say, from m independent probability experiments). Then:

P r[E1 ∩ E2 ∩ · · · ∩ Em ] = P r[E1 ]P r[E2 ] · · · P r[Em ]

That is, the probability that all independent events occur is equal to the product of the individual event probabilities.
• (Union Bound) Let E1 , E2 , . . . , Em be a collection of m events (possibly not independent). Then:
m
X
P r[E1 ∪ E2 ∪ · · · ∪ Em ] ≤ P r[Ei ]
i=1

That is, the probability that at least one of the events occurs is less than or equal to the sum of the individual event
probabilities.