In-Depth Notes: Linear Algebra

Chapter 4: Vector Spaces

A study guide for UC3M Aerospace Engineering

Based on ”Linear Algebra and its Applications” (5th Ed.) by David C. Lay,
Steven R. Lay, and Judi J. McDonald, and course notes.

November 4, 2025
Contents

1 Part 1: The Theory - Vector Spaces 3

1.1 Section 4.1: Vector Spaces and Subspaces . . . . . . . . . . . . . . . . 3
1.1.1 Introduction: Beyond Rn . . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 The Definition of a Vector Space . . . . . . . . . . . . . . . . . . 3
1.1.3 Subspaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.4 A Subspace Spanned by a Set . . . . . . . . . . . . . . . . . . . . 7
1.2 Section 4.2: Null Spaces, Column Spaces, and Linear Transformations 9
1.2.1 The Null Space of a Matrix . . . . . . . . . . . . . . . . . . . . . 9
1.2.2 The Column Space of a Matrix . . . . . . . . . . . . . . . . . . . 10
1.2.3 Linear Transformations: Kernel and Range . . . . . . . . . . . . 11
1.3 Section 4.3: Linearly Independent Sets; Bases . . . . . . . . . . . . . . 13
1.3.1 Linear Independence . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.2 Bases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.3 Finding a Basis: The Spanning Set Theorem . . . . . . . . . . . 14
1.3.4 Two Views of a Basis . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4 Section 4.4: Coordinate Systems . . . . . . . . . . . . . . . . . . . . . . 16
1.4.1 The Unique Representation Theorem . . . . . . . . . . . . . . . 16
1.4.2 Coordinate Systems in Rn . . . . . . . . . . . . . . . . . . . . . . 16
1.4.3 The Coordinate Mapping . . . . . . . . . . . . . . . . . . . . . . 17
1.5 Section 4.5: The Dimension of a Vector Space . . . . . . . . . . . . . . 19
1.6 Section 4.6: The Rank of a Matrix . . . . . . . . . . . . . . . . . . . . . . 21
1.6.1 The Row Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.6.2 The Rank Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.6.3 The Invertible Matrix Theorem (Continued) . . . . . . . . . . . 22
1.7 Section 4.7: Change of Basis . . . . . . . . . . . . . . . . . . . . . . . . 23

2 Part 2: Summary of Key Concepts and Formulas 25

2.1 Summary of Sections 4.1-4.4 . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.1 Section 4.1: Vector Spaces and Subspaces . . . . . . . . . . . . 25
2.1.2 Section 4.2: Null Spaces, Column Spaces, and Linear Trans-
formations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.1.3 Section 4.3: Linearly Independent Sets; Bases . . . . . . . . . . 26
2.1.4 Section 4.4: Coordinate Systems . . . . . . . . . . . . . . . . . . 27
2.1.5 Section 4.5: The Dimension of a Vector Space . . . . . . . . . . 27
2.1.6 Section 4.6: The Rank of a Matrix . . . . . . . . . . . . . . . . . 28
2.1.7 Section 4.7: Change of Basis . . . . . . . . . . . . . . . . . . . . 28

1
3 Part 3: Solutions to Teacher’s Exercises ([Link] & [Link]) 29
3.1 Exercises from [Link] . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 Exercises from [Link] . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4 Part 4: Solutions to Textbook Exercises (Lay, Chapter 4) 36

2
Chapter 1

Part 1: The Theory - Vector Spaces

This part elaborates on the theoretical concepts of vector spaces, drawing from
Chapter 4 of the textbook and the provided lecture slides. Each concept is defined
from the ground up, with explanatory examples.

1.1 Section 4.1: Vector Spaces and Subspaces

1.1.1 Introduction: Beyond Rn
So far in linear algebra, we have worked almost exclusively with Rn , the set of all
vectors (or ordered lists) with n entries. We learned how to add them (u + v) and
how to multiply them by scalars (cu). We saw that these operations follow familiar
algebraic rules, such as u + v = v + u and c(u + v) = cu + cv.
The central idea of ”Vector Spaces” is to generalize this. We will see that many
other sets of objects (such as polynomials, functions, or matrices) also behave
accordingto these same algebraic rules. When a set follows these rules, we call
it a vector space, and we can use all the tools of linear algebra (like span, linear
independence, and basis) to study it.

1.1.2 The Definition of a Vector Space

A vector space is a nonempty set V of objects, which we will call vectors, on which
two operations are defined:

• Vector Addition: A rule for adding two vectors u and v in V to produce a

new vector u + v which is also in V .

• Scalar Multiplication: A rule for multiplying a vector u in V by a real scalar

c to produce a new vector cu which is also in V .

These two operations must follow ten fundamental rules, called axioms. These
axioms must hold for all vectors u, v, w in V and for all scalars c, d.

Definition 1.1.1 (The 10 Vector Space Axioms). 1. (Closure under Addition) The
sum u + v is in V .

2. (Commutativity of Addition) u + v = v + u.

3
 3. (Associativity of Addition) (u + v) + w = u + (v + w).

4. (Existence of a Zero Vector) There is a zero vector 0 in V such that u+0 = u

for all u in V .

5. (Existence of an Additive Inverse) For each u in V , there is a vector −u in

V (called the negative of u) such that u + (−u) = 0.

6. (Closure under Scalar Multiplication) The scalar multiple cu is in V .

7. (Distributivity) c(u + v) = cu + cv.

8. (Distributivity) (c + d)u = cu + du.

9. (Associativity of Multiplication) c(du) = (cd)u.

10. (Identity for Multiplication) 1u = u.

Remark 1.1.2. These axioms are not just abstract rules. They are the ”rules of the
game” for algebra. Axioms 1 and 6 (the closure axioms) are the most important:
they ensure that when you add or scale vectors, you don’t leave the set V . Axioms
4 and 5 ensure we can ”solve” equations like x + u = v by writing x = v + (−u),
which we call v − u.

Example 1.1.3 (The Premier Example: Rn ). The set Rn (for any n ≥ 1) is a vector
space. We have been using this fact implicitly all along. For any two vectors in Rn ,
their sum is in Rn (Axiom 1). The zero vector is the vector of all zeros (Axiom 4).
The negative of u is (−1)u (Axiom 5). The other axioms are all just properties of
arithmetic in R.

Example 1.1.4 (The Space of Polynomials, Pn ). Let V be the set of all polynomials
of degree at most n, where n is a fixed non-negative integer. We denote this set
by Pn . A typical vector (polynomial) in Pn is:

p(t) = a0 + a1 t + a2 t2 + · · · + an tn

where a0 , . . . , an are real scalars and t is a real variable. Let’s check some axioms
to see why Pn is a vector space.

• Axiom 1 (Closure): Let p, q be in Pn .

p(t) = a0 + · · · + an tn and q(t) = b0 + · · · + bn tn

Their sum is:

(p + q)(t) = (a0 + b0 ) + (a1 + b1 )t + · · · + (an + bn )tn

This is another polynomial of degree at most n. So, p + q is in Pn .

• Axiom 4 (Zero Vector): The zero vector is the zero polynomial, 0(t) = 0,
which is in Pn (it has degree ≤ n).

(p + 0)(t) = (a0 + 0) + (a1 + 0)t + · · · = p(t)

4
 • Axiom 6 (Scalar Closure): cp(t) = (ca0 ) + (ca1 )t + · · · + (can )tn . This is also a
polynomial of degree at most n.

The other axioms also hold. Thus, Pn is a vector space.

Example 1.1.5 (The Space of Matrices, Mm×n ). Let V be the set of all m×n matrices
with real entries. We call this space Mm×n . We already know how to add two m × n
matrices (we add their corresponding entries) and how to multiply a matrix by
a scalar (we multiply every entry by that scalar). The 10 axioms are all satisfied,
thanks to the properties of matrix algebra (Theorem 1 in Section 2.1).

• The zero vector is the m × n zero matrix.

• The negative of a matrix A is the matrix (−1)A.

Example 1.1.6 (Space of Functions, C[a, b]). Let V be the set of all real-valued, con-
tinuous functions defined on a closed interval [a, b]. We call this space C[a, b]. Vec-
tors are functions (like f (t) = sin(t) on [0, 2π]).

• Addition: (f + g)(t) = f(t) + g(t). From calculus, the sum of two continuous
functions is continuous.

• Scalar Multiplication: (cf)(t) = c · f(t). A scalar multiple of a continuous

function is continuous.

• Zero Vector: The zero vector 0 is the function 0(t) = 0 for all t in [a, b]. This is
a continuous function.

This space C[a, b] is a vector space.

1.1.3 Subspaces
Often, a vector space of interest is a subset of a larger, more familiar vector space.

Definition 1.1.7 (Subspace). A subspace of a vector space V is a subset H of V

that has three properties:

1. (Contains Zero) The zero vector of V is in H.

2. (Closed under Addition) For each u and v in H, the sum u + v is in H.

3. (Closed under Scalar Multiplication) For each u in H and each scalar c, the
vector cu is in H.

This definition is a ”shortcut.” If a subset H satisfies these three properties, it

is guaranteed to be a vector space itself, using the operations from V . The other 7
axioms are inherited from V for free.
  
 s 
3
Example 1.1.8 (A Plane in R ). Let H =  t : s, t ∈ R . This is the xy-plane in R3 .

0
 
Is H a subspace of R3 ?

5
 • Property 1: Is the zero vector of R3 in H? Yes. If we choose s = 0 and t = 0,
0
we get 0, which is in H.
0
   
s1 s2
• Property 2: Let u = t1 and v = t2  be two vectors in H. Their sum is
  
  0 0
s1 + s2
u + v = t1 + t2 . This new vector has a 0 in the third entry, so it is also in H.

0
H is closed under addition.
     
s cs cs
• Property 3: Let u =  t  be in H and c be a scalar. Then cu =  ct  =  ct .
0 c·0 0
This new vector is also in H. H is closed under scalar multiplication.

Since H satisfies all three properties, H is a subspace of R3 .

x2

u+v
u
H (the x1 x2 -plane)

x1

x3

Example 1.1.9 (A Set That Is  NOT a Subspace).

  Let H be the set of points on the
x
line y = 2x + 1 in R2 . So, H = : x ∈ R . Is H a subspace of R2 ?
2x + 1
 
0
• Property 1: We check for the zero vector 0 = . For this vector to be in H,
0
we would need to find an x such that x = 0 and 2x + 1 = 0. This means x = 0
and 1 = 0, which is impossible.

Since the zero vector is not in H, H is not a subspace. We don’t even need
 to check
1
the other two properties. (In fact, it fails those, too. For example, is in H, but
    3
1 2
2 = is not in H, because 6 6= 2(2) + 1.)
3 6

6
1.1.4 A Subspace Spanned by a Set
The most common way to describe a subspace is by constructing it as the set of
all linear combinations of some fixed vectors. This is the same as the ”Span” we
studied in Chapter 1.

Definition 1.1.10 (Span). Let v1 , . . . , vp be vectors in a vector space V . The span of

these vectors, denoted Span{v1 , . . . , vp }, is the set of all possible linear combina-
tions of these vectors.

Span{v1 , . . . , vp } = {c1 v1 + c2 v2 + · · · + cp vp : c1 , . . . , cp are scalars}

Theorem 1.1.11 (Span is a Subspace). If v1 , . . . , vp are in a vector space V , then

H = Span{v1 , . . . , vp } is a subspace of V .

Proof. We must show that H satisfies the three properties of a subspace.

1. (Contains Zero) The zero vector 0 is in H because

0 = 0v1 + 0v2 + · · · + 0vp

This is a linear combination of the vectors (with all weights as 0).

2. (Closed under Addition) Let u and w be two vectors in H. By definition, they

can be written as linear combinations:

u = c1 v1 + · · · + cp vp
w = d1 v1 + · · · + dp vp

Now, we add them:

u + w = (c1 v1 + · · · + cp vp ) + (d1 v1 + · · · + dp vp )
= (c1 + d1 )v1 + · · · + (cp + dp )vp (by Axioms 2, 3, 8)

This is a new linear combination of v1 , . . . , vp . So, u + w is in H.

3. (Closed under Scalar Multiplication) Let u be in H and k be a scalar.

u = c1 v1 + · · · + cp vp

Then,

ku = k(c1 v1 + · · · + cp vp ) = (kc1 )v1 + · · · + (kcp )vp (by Axioms 7, 9)

This is also a linear combination of v1 , . . . , vp . So, ku is in H.

Since H satisfies all three properties, it is a subspace of V .

We call Span{v1 , . . . , vp } the subspace spanned (or generated) by {v1 , . . . , vp }.

7
   

 a − b 

b − c
  
Example 1.1.12 (From teacher’s notes, week 6, Ex 5b). Is W =    : a, b, c ∈ R

 c − a 

b
 
a subspace of R4 ?
Solution: We can use Theorem 1.1.11. The typical vector in W can be re-written
by separating the variables a, b, c:
             
a−b a −b 0 1 −1 0
 b − c   0   b  −c 0 1 −1
c − a = −a +  0  +  c  = a −1 + b  0  + c  1 
             

b 0 b 0 0 1 0
     
1 −1 0
0 1 −1
Let v1 = 
−1, v2 =  0 , and v3 =  1 . Since any vector in W can be written
    

0 1 0
as av1 + bv2 + cv3 , we have

W = Span{v1 , v2 , v3 }

By Theorem 1.1.11, W is a subspace of R4 .

8
1.2 Section 4.2: Null Spaces, Column Spaces, and Lin-
ear Transformations
This section introduces two of the most important subspaces associated with a
matrix A.

1.2.1 The Null Space of a Matrix

Definition 1.2.1 (Null Space). The null space of an m×n matrix A, written as Nul A,
is the set of all solutions to the homogeneous equation Ax = 0. In set notation:

Nul A = {x ∈ Rn : Ax = 0}

Theorem 1.2.2. The null space of an m × n matrix A is a subspace of Rn .

Proof. We check the three properties of a subspace.

1. (Contains Zero) Is the zero vector 0 (of Rn ) in Nul A? Yes, because A0 = 0.

2. (Closed under Addition) Let u and v be in Nul A. This means Au = 0 and

Av = 0. Now we check their sum, u + v:

A(u + v) = Au + Av (by linearity of matrix multiplication)

=0+0=0
Thus, u + v is also a solution to Ax = 0, so u + v is in Nul A.

3. (Closed under Scalar Multiplication) Let u be in Nul A and c be a scalar. This

means Au = 0. We check the scalar multiple cu:

A(cu) = c(Au) (by linearity of matrix multiplication)

= c(0) = 0
Thus, cu is also in Nul A.

Since Nul A satisfies all three properties, it is a subspace of Rn .

Remark 1.2.3 (Implicit vs. Explicit Description). The definition of Nul A is implicit.
It gives a condition (Ax = 0) that vectors must pass to be in the set. It does not give
an explicit list or tell how to generate the vectors. The process of solving Ax = 0
(which we did in Chapter 1) is the process of finding an explicit description of
Nul A. This explicit description is a spanning set.

Example 1.2.4 (Finding a Spanning Set for Nul A). (Based on teacher’s notes, week
6, Ex 10a) Find an explicit description of Nul A (that is, a spanning set) for:
 
1 3 5 0
A=
0 1 4 −2

9
Solution: The null space is the set of all solutions to Ax = 0. The augmented
1 3 5 0 0
matrix is . This is already in echelon form. We find the reduced
0 1 4 −2 0
echelon form to get the general solution.
   
1 3 5 0 0 R1 −3R2 1 0 −7 6 0
−−−−→
0 1 4 −2 0 0 1 4 −2 0

The system of equations is:

x1 − 7x3 + 6x4 = 0
x2 + 4x3 − 2x4 = 0

The pivot variables (from pivot columns 1 and 2) are x1 and x2 . The free variables
are x3 and x4 . We solve for the basic variables in terms of the free variables:

x1 = 7x3 − 6x4
x2 = −4x3 + 2x4
x3 = x3 (free)
x4 = x4 (free)

This is the general solution. Now we write it in parametric vector form:

           
x1 7x3 − 6x4 7x3 −6x4 7 −6
x2  −4x3 + 2x4  −4x3   2x4  −4 2
x= x3  = 
  =
  x3  +  0  = x3  1  +x4  0 
      
x3
x4 x4 0 x4 0 1
| {z } | {z }
u v

This shows that Nul A = Span{u, v}. This is the explicit description of the null
space.

1.2.2 The Column Space of a Matrix

Definition 1.2.5 (Column Space). The column space of an m × n matrix A, written
as Col A, is the set of all linear combinations of the columns of A. If A = [a1 . . . an ],
then:
Col A = Span{a1 , . . . , an }

Theorem 1.2.6. The column space of an m × n matrix A is a subspace of Rm .

Proof. This is a direct result of Theorem 1.1.11, since Col A is defined as the span
of a set of vectors.

Remark 1.2.7 (Implicit vs. Explicit Description). The definition of Col A is explicit.
It tells us exactly how to build the vectors in the set (by taking linear combinations
of the columns). However, testing if a vector b is in Col A is an implicit problem. It
requires solving the system Ax = b.

b ∈ Col A ⇐⇒ The equation Ax = b has a solution.

10
  
2
Example 1.2.8 (From teacher’s notes, week 6, Ex 12). Determine if w =  1  is in
  −2
−8 −2 −9
Col A, where A =  6 4 8 .
4 0 4
Solution: We need to check if the system Ax = w is consistent. We row reduce
the augmented matrix [A | w]:
     
−8 −2 −9 2 4 0 4 −2 2 0 2 −1
Swap R1, R3 Scale R1 by 1/2
6 4 8 1  −−−−−−−→  6 4 8 1  −−−−−−−−→  6 4 8 1
4 0 4 −2 −8 −2 −9 2 −8 −2 −9 2
   
2 0 2 −1 2 0 2 −1
R2 −3R1 ,R3 +4R1 R3 +(1/2)R2
−−−−−−−−−−→ 0 4 2 4  −−−−−−−→ 0 4 2 4 
0 −2 −1 −2 0 0 0 0
The system is consistent (there is no row of the form [0 0 0 | b] with b 6= 0). Therefore,
w is in Col A.
Remark 1.2.9 (Contrast between Nul A and Col A). Let A be an m × n matrix.

Null Space (Nul A) Column Space (Col A)

1. Is a subspace of Rn . 1. Is a subspace of Rm .
2. Defined implicitly: Ax = 0. 2. Defined explicitly: Span{a1 , . . . , a
3. To find Nul A, you must solve Ax = 0. 3. Finding Col A is easy; it’s the spa
4. Checking if v ∈ Nul A is easy: just compute Av. 4. Checking if b ∈ Col A is hard: mu
5. Nul A = {0} iff Ax = 0 has only the trivial solution. 5. Col A = Rm iff Ax = b has a solut
6. Nul A = {0} iff columns of A are linearly independent. 6. Col A = Rm iff columns of A span

1.2.3 Linear Transformations: Kernel and Range

We can generalize the concepts of Nul A and Col A to linear transformations on
abstract vector spaces.
Definition 1.2.10 (General Linear Transformation). A linear transformation T
from a vector space V into a vector space W is a rule that assigns to each vec-
tor x in V a unique vector T (x) in W , such that:
1. T (u + v) = T (u) + T (v) for all u, v in V . (Preserves addition)
2. T (cu) = cT (u) for all u in V and all scalars c. (Preserves scalar mult.)
For any linear transformation, we can define two special subspaces: the kernel
(which is the generalization of the null space) and the range (which is the general-
ization of the column space).
Definition 1.2.11 (Kernel and Range). Let T : V → W be a linear transformation.
• The kernel (or null space) of T is the set of all vectors v in V such that T (v) =
0 (the zero vector in W ).
ker(T ) = {v ∈ V : T (v) = 0W }

11
 • The range of T is the set of all vectors in W that are images of at least one
vector in V .
Range(T ) = {T (v) : v ∈ V }

Theorem 1.2.12. If T : V → W is a linear transformation, then:

• The kernel of T is a subspace of V .

• The range of T is a subspace of W .

Example 1.2.13 (From teacher’s notes, week 6, Ex 2d). Let V = Pn (polynomials

of degree ≤ n) and let W = R. Define a transformation T : Pn → R by the rule
T (p) = p(0). (This is an ”evaluation” transformation). Show it is linear and find its
kernel.
Solution:

• Show T is Linear: We must check the two properties.

1. Let p, q be in Pn .

T (p + q) = (p + q)(0) = p(0) + q(0) = T (p) + T (q)

2. Let p be in Pn and c be a scalar.

T (cp) = (cp)(0) = c · p(0) = cT (p)

Both properties hold, so T is a linear transformation.

• Find the Kernel: The kernel is the set of all vectors p in Pn such that T (p) = 0.

T (p) = p(0) = 0

A polynomial in Pn has the form p(t) = a0 +a1 t+a2 t2 +· · ·+an tn . The condition
p(0) = 0 means:

a0 + a1 (0) + a2 (0)2 + · · · + an (0)n = 0 =⇒ a0 = 0

So, the kernel of T is the set of all polynomials in Pn with a zero constant
term.
ker(T ) = {a1 t + a2 t2 + · · · + an tn } = Span{t, t2 , . . . , tn }
This is a subspace of Pn .

12
1.3 Section 4.3: Linearly Independent Sets; Bases
We now apply the concepts of linear independence and spanning sets to the gen-
eral case of vector spaces. The goal is to find the smallest possible set that ”builds”
an entire subspace.

1.3.1 Linear Independence

The definition of linear independence is the same as for Rn , but adapted for an
abstract vector space.
Definition 1.3.1 (Linear Independence). An indexed set of vectors {v1 , . . . , vp } in a
vector space V is said to be linearly independent if the vector equation
c1 v1 + c2 v2 + · · · + cp vp = 0
has only the trivial solution, c1 = 0, c2 = 0, . . . , cp = 0.
The set is linearly dependent if there exist weights c1 , . . . , cp , not all zero, such
that the equation holds. This equation is called a linear dependence relation.
All the properties of linear dependence from Chapter 1 still hold.
• A set of two vectors {v1 , v2 } is linearly dependent if and only if one is a mul-
tiple of the other (e.g., v1 = cv2 ).
• Any set containing the zero vector {0, v1 , . . . , vp } is linearly dependent (be-
cause 1 · 0 + 0v1 + · · · + 0vp = 0).
Theorem 1.3.2. An indexed set {v1 , . . . , vp } of two or more vectors (with v1 6= 0) is
linearly dependent if and only if some vector vj (with j > 1) is a linear combination
of the preceding vectors v1 , . . . , vj−1 .
Example 1.3.3 (Checking Linear Independence in Pn ). Determine if the set {p1 , p2 , p3 }
is linearly independent in P2 , where:
p1 (t) = 1, p2 (t) = t, p3 (t) = 4 − t
Solution: We test the equation c1 p1 + c2 p2 + c3 p3 = 0. The 0 in P2 is the zero
polynomial.
c1 (1) + c2 (t) + c3 (4 − t) = 0
c1 + c2 t + 4c3 − c3 t = 0
Regroup the terms by powers of t:
(c1 + 4c3 ) + (c2 − c3 )t = 0
A polynomial is the zero polynomial if and only if all its coefficients are zero. This
gives us a system of homogeneous equations:
c1 + 4c3 = 0 (constant term)
+ c2 − c3 = 0 (t term)
This system has c3 as a free variable. We can see a nontrivial solution. For example,
let c3 = 1. Then c2 = 1 and c1 = −4. We have found a linear dependence relation:
(−4)p1 + (1)p2 + (1)p3 = −4(1) + 1(t) + 1(4 − t) = −4 + t + 4 − t = 0
Since the weights are not all zero, the set is linearly dependent.

13
1.3.2 Bases
A basis is a set that is ”just right” for spanning a subspace. It is big enough to span
the whole space, but small enough to be linearly independent.
Definition 1.3.4 (Basis). Let H be a subspace of a vector space V . An indexed set
of vectors B = {b1 , . . . , bp } is a basis for H if:
1. B is a linearly independent set.
2. H = Span{b1 , . . . , bp } (i.e., B spans H).
Example 1.3.5 (The Standard Basis for Rn ). The set E = {e1 , . . . , en }, where ei are
n 3
the columns
   of the
 n× n identity matrix In , is the standard basis for R . For R ,
 1 0 0 
E= 0 , 1 , 0 .
   
0 0 1
 

• The set is linearly independent (because In x = 0 has only the trivial solution).
• The set spans Rn (because any x in Rn can be written as x1 e1 + · · · + xn en ).
Example 1.3.6 (The Standard Basis for Pn ). The set S = {1, t, t2 , . . . , tn } is the stan-
dard basis for Pn .
• S spans Pn by the very definition of a polynomial.
• S is linearly independent because the only way c0 (1) + c1 t + · · · + cn tn = 0 (the
zero polynomial) is if all the coefficients c0 , . . . , cn are zero.

1.3.3 Finding a Basis: The Spanning Set Theorem

How do we find a basis? We can start with a set that we know spans the space, and
then ”throw away” any vectors that are redundant (i.e., are linear combinations of
the others). This is the idea of the Spanning Set Theorem.
Theorem 1.3.7 (The Spanning Set Theorem). Let S = {v1 , . . . , vp } be a set in a
vector space V , and let H = Span{S}.
1. If one of the vectors in S—say vk —is a linear combination of the remaining
vectors in S, then the set formed from S by removing vk still spans H.
2. If H 6= {0}, some subset of S is a basis for H.
This theorem gives us a practical way to find a basis for Col A and Nul A.

Basis for Nul A

The method we used in Example 1.2 (page 9) to find a spanning set for Nul A always
produces a basis. When we solve Ax = 0 and write the solution in parametric
vector form:
x = xk u + xj v + . . .
The vectors {u, v, . . . } (one for each free variable) are automatically linearly inde-
pendent. Therefore, the method from Section 4.2 for finding the spanning set
of Nul A always produces a basis for Nul A.

14
Basis for Col A

The columns of A span Col A by definition, but they are not always linearly inde-
pendent. We need to find a subset of the columns that is a basis.

Theorem 1.3.8 (Basis for Column Space). The pivot columns of a matrix A form
a basis for its column space Col A.

Justification. Let B be the reduced echelon form of A.

1. The pivot columns of A are linearly independent because any linear depen-
dence relation between them would also be a dependence relation for the
pivot columns of B. But the pivot columns of B are standard basis vectors
(like e1 , e2 , . . . ), which are linearly independent.

2. The Spanning Set Theorem says we can discard non-pivot columns. Why?
Because a non-pivot column is a linear combination of the pivot columns to
its left. The row operations that create B from A do not change the linear
dependence relations. So, any non-pivot column in A is a linear combination
of the pivot columns of A. By the Spanning Set Theorem, we can discard all
non-pivot columns, and the remaining pivot columns will still span Col A.

Since the pivot columns are linearly independent and span Col A, they form a basis.

Remark 1.3.9 (Warning!). You must use the pivot columns from the original ma-
trix A, not from its reduced echelon form B. The column space of B is often dif-
ferent from the column space of A.
 
−2 4 −2 −4
Example 1.3.10 (Basis for Col A from teacher’s notes, week 7, Ex 2). Let A =  2 −6 −3 1 .
  −3 8 2 −3
1 0 6 5
We are given that A is row equivalent to B = 0 2 5 3. Find a basis for Col A.
0 0 0 0
Solution:

1. Look at the echelon form B to find the pivot columns. The pivots are in col-
umn 1 and column 2.

2. The basis for Col A is the set of pivot columns from the original matrix A.
   
 −2 4 
So, a basis for Col A is  2 , −6 .
 
−3 8
 

1.3.4 Two Views of a Basis

A basis is a spanning set that is as small as possible. (If you remove any vector,
it won’t span anymore.) A basis is a linearly independent set that is as large as
possible. (If you add any other vector from the space, it will become linearly de-
pendent.)

15
1.4 Section 4.4: Coordinate Systems
The main idea of this section is that a basis B in a vector space V acts like a ”coor-
dinate system.” It allows us to translate any vector x in V into a normal vector of
coordinates [x]B in Rn , and vice-versa. This lets us work with abstract spaces like
Pn as if they were just Rn+1 .

1.4.1 The Unique Representation Theorem

Theorem 1.4.1 (The Unique Representation Theorem). Let B = {b1 , . . . , bn } be a
basis for a vector space V . Then for each vector x in V , there exists a unique set
of scalars c1 , . . . , cn such that:
x = c1 b 1 + c2 b 2 + · · · + cn b n
Proof. • Existence: Since B spans V (by definition of a basis), at least one set
of scalars exists.
• Uniqueness: Suppose x also has another representation: x = d1 b1 + · · · +
dn bn . Subtracting the two equations gives:
0 = x − x = (c1 − d1 )b1 + · · · + (cn − dn )bn
Since B is linearly independent (by definition of a basis), all the weights in this
linear dependence relation must be zero.
(c1 − d1 ) = 0, ..., (cn − dn ) = 0
This implies ci = di for all i. Thus, the representation is unique.

Definition 1.4.2 (Coordinates of x relative to B). Suppose B = {b1 , . . . , bn } is a basis

for V and x is in V . The unique scalars c1 , . . . , cn from the theorem are called the
coordinates of x relative to the basis B. The vector in Rn :
 
c1
 .. 
[x]B =  . 
cn
is called the coordinate vector of x (relative to B) or the B-coordinate vector of
x.

1.4.2 Coordinate Systems in Rn

When we work in Rn , we can use a non-standard basis B to create a new coordi-
nate system. Let B = {b1 , . . . , bn } be a basis for Rn . We can form the change-of-
coordinates matrix PB from B to the standard basis E:
 
| |
PB = b1 . . . bn 
| |
Let x ∈ Rn . We have two ways to ”name” this vector:

16
 1. The standard coordinate vector, which is just x itself (or [x]E ).
 
c1
 .. 
2. The B-coordinate vector, [x]B =  . .
cn

The relationship between these two ”names” is given by the equation:

x = c1 b1 + · · · + cn bn = PB [x]B

This equation converts B-coordinates into standard coordinates. To go the other

way (from standard to B-coordinates), we just multiply by PB−1 :

PB−1 x = [x]B

Example 1.4.3 (From teacher’s notes, week 7, Ex 8a). Find the coordinate vector
[x]B of x relative to the basis B = {b1 , b2 }, where
     
1 5 4
b1 = , b2 = , x=
−2 −6 0

Solution: We need to find scalars c1 , c2 such that

     
1 5 4
c1 + c2 =
−2 −6 0
 
1 5
This is the vector equation PB c = x, where PB = and c = [x]B . We can
−2 −6
solve this by row reducing the augmented matrix [PB | x]:
       
1 5 4 R2 +2R1 1 5 4 R2 /4 1 5 4 R1 −5R2 1 0 −6
−−−−→ −−−→ −−−−→
−2 −6 0 0 4 8 0 1 2 0 1 2
 
−6
So, c1 = −6 and c2 = 2. The B-coordinate vector is [x]B = .
       2  
1 5 −6 10 4
Check: −6b1 + 2b2 = −6 +2 = + = = x.
−2 −6 12 −12 0

1.4.3 The Coordinate Mapping

Theorem 8 states that the ”coordinate mapping” T : V → Rn defined by T (x) = [x]B
is a one-to-one linear transformation from V onto Rn . This mapping is called an
isomorphism. It means that any vector space V with a basis of n vectors ”looks
and acts” exactly like Rn . This is an incredibly powerful idea. It means we can
”translate” a problem from an abstract space (like P2 ) into a concrete problem in
R3 , solve it there, and then translate the answer back.

Example 1.4.4 (From teacher’s notes, week 7, Ex 10). The set B = {1 + t2 , t + t2 , 1 +

2t + t2 } is a basis for P2 . Find the coordinate vector of p(t) = 1 + 4t + 7t2 relative to
B.
Solution:

17
1. Translate to R3 : We use the standard basis E = {1, t, t2 } for P2 . The coordi-
nate vectors for B are:
     
1 0 1
[b1 ]E = 0 , [b2 ]E = 1 , [b3 ]E = 2
1 1 1

The coordinate vector for p is:

 
1
[p]E = 4

7

2. Solve in R3 : We need to find c1 , c2 , c3 such that:

       
1 0 1 1
c1 0 + c2 1 + c3 2 = 4
      
1 1 1 7

This is a standard R3 problem. We solve it by row reducing the augmented

matrix:
     
1 0 1 1 1 0 1 1 1 0 1 1
R3 −R1 R −R2
0 1 2 4 − −−−→ 0 1 2 4 −−3−−→ 0 1 2 4
1 1 1 7 0 1 0 6 0 0 −2 2
   
1 0 1 1 1 0 0 2
R3 /−2 R1 −R3 ,R2 −2R3
−−−−→ 0 1 2 4  −−−−−−−−−→ 0 1 0 6 
0 0 1 −1 0 0 1 −1
So, c1 = 2, c2 = 6, and c3 = −1.

3. Translate back to P2 : The coordinate vector of p relative to B is

 
2
[p]B = 6 

−1

Check: 2(1 + t2 ) + 6(t + t2 ) − 1(1 + 2t + t2 ) = (2 + 2t2 ) + (6t + 6t2 ) − (1 + 2t + t2 ) =

(2 − 1) + (6 − 2)t + (2 + 6 − 1)t2 = 1 + 4t + 7t2 = p(t).

18
1.5 Section 4.5: The Dimension of a Vector Space
We observed
  that
 a subspace
  can
have many different bases. For example, in
1 0 1 1
R2 , both , and , are bases. Notice both bases have the same
0 1 0 1
number of vectors: two. This is not a coincidence.

Theorem 1.5.1. If a vector space V has a basis B = {b1 , . . . , bn }, then any set in V
containing more than n vectors must be linearly dependent.

Proof. Let S = {u1 , . . . , up } be a set in V with p > n. We can use the coordinate
mapping to ”translate” this set into Rn . The coordinate vectors [u1 ]B , . . . , [up ]B are in
Rn . Since p > n, this set of p vectors in Rn must be linearly dependent (by Theorem
8 in Section 1.7). This means there are scalars c1 , . . . , cp , not all zero, such that:

c1 [u1 ]B + · · · + cp [up ]B = 0 (the zero vector in Rn )

Because the coordinate mapping is a linear transformation, we can ”translate” this

equation back to V :
[c1 u1 + · · · + cp up ]B = 0
This equation says that the B-coordinates of the vector w = c1 u1 + · · · + cp up are all
zero. The only vector in V with this property is the zero vector 0V . Thus:

c1 u1 + · · · + cp up = 0V

Since not all the scalars ci are zero, this is a linear dependence relation. Therefore,
the set S is linearly dependent.

Theorem 1.5.2. If a vector space V has a basis of n vectors, then every basis of V
must consist of exactly n vectors.

Proof. Let B1 be a basis with n vectors and B2 be a basis with p vectors.

• Since B1 is a basis (spanning) and B2 is linearly independent, the previous

theorem says that B2 cannot have more vectors than B1 . So, p ≤ n.

• Since B2 is a basis (spanning) and B1 is linearly independent, the same theo-

rem says that B1 cannot have more vectors than B2 . So, n ≤ p.

Because both p ≤ n and n ≤ p are true, we must have p = n.

This theorem allows us to make a formal definition of dimension.

Definition 1.5.3 (Dimension). If V is spanned by a finite set, then V is finite-

dimensional, and the dimension of V , written dim V , is the number of vectors
in any basis for V . The dimension of the zero vector space {0} is defined to be 0.
If V is not spanned by a finite set, it is infinite-dimensional.

Example 1.5.4. • dim Rn = n (because the standard basis {e1 , . . . , en } has n

vectors).

• dim Pn = n + 1 (because the standard basis {1, t, t2 , . . . , tn } has n + 1 vectors).

19
 • dim R3 = 3. A plane through the origin in R3 is a 2-dimensional subspace. A
line through the origin is a 1-dimensional subspace.

Theorem 1.5.5 (The Basis Theorem). Let V be a p-dimensional vector space (p ≥ 1).

1. Any linearly independent set of exactly p vectors in V is automatically a basis

for V .

2. Any set of exactly p vectors that spans V is automatically a basis for V .

Remark 1.5.6. This theorem is extremely useful. It gives us a shortcut. If we know

dim V = p, we don’t have to check both conditions for a basis.

• If we have p vectors and show they are linearly independent, we know they
must also span V .

• If we have p vectors and show they span V , we know they must also be linearly
independent.

Dimensions of Nul A and Col A

This new language allows us to re-state what we already know about Nul A and
Col A.

• The dimension of Col A is the number of pivot columns in A.

• The dimension of Nul A is the number of free variables in the equation Ax =

0.

Example 1.5.7 (From slide 4.5-14). Find the dimensions of Nul A and Col A for
 
−3 6 −1 1 −7
A =  1 −2 2 3 −1
2 −4 5 8 −4

Solution: Row reduce A to an echelon form:

     
1 −2 2 3 −1 1 −2 2 3 −1 1 −2 2 3 −1
A∼ 0 0 5 10 −10 ∼ 0 0 5 10 −10 ∼ 0 0 5 10 −10
−3 6 −1 1 −7 0 0 5 10 −10 0 0 0 0 0

The pivot columns are 1 and 3.

• dim(Col A) = number of pivots = 2.

• The free variables are x2 , x4 , x5 .

• dim(Nul A) = number of free variables = 3.

20
1.6 Section 4.6: The Rank of a Matrix
This section connects the dimensions of the column space, null space, and a new
subspace: the row space.

1.6.1 The Row Space

The columns of AT are the rows of A. This gives us a new subspace to study.
Definition 1.6.1 (The Row Space). The row space of an m × n matrix A, written
Row A, is the set of all linear combinations of the rows of A. It is a subspace of Rn .

Row A = Col AT

Theorem 1.6.2. If two matrices A and B are row equivalent, then their row spaces
are the same. If B is in echelon form, the nonzero rows of B form a basis for the
row space of A.
Example 1.6.3 (Finding a Basis for Row A). Find a basis for the row space of the
matrix A from Section 4.5, Example 5.
   
−3 6 −1 1 −7 1 −2 2 3 −1
A =  1 −2 2 3 −1 ∼ B = 0 0 5 10 −10
2 −4 5 8 −4 0 0 0 0 0

Solution: The nonzero rows of the echelon form B form the basis.
( )
(1, −2, 2, 3, −1),
Basis for Row A =
(0, 0, 5, 10, −10)

Remark 1.6.4 (Comparing Bases for Nul, Col, and Row Spaces). Let A be an m × n
matrix, and B be an echelon form of A.
• Basis for Col A: Use the pivot columns of the original matrix A.

• Basis for Row A: Use the nonzero rows of the echelon matrix B.

• Basis for Nul A: Use the vectors from the parametric vector form of the so-
lution to Ax = 0, which is found from the reduced echelon form.

1.6.2 The Rank Theorem

We can now define rank.
Definition 1.6.5 (Rank). The rank of a matrix A, denoted rank A, is the dimension
of the column space of A.
rank A = dim(Col A)
From our work above, we know that dim(Col A) is just the number of pivot po-
sitions. We also just learned that dim(Row A) is the number of nonzero rows in
an echelon form, which is also the number of pivot positions. This gives us a deep
and important theorem.

21
Theorem 1.6.6 (The Rank Theorem). The dimensions of the column space and the
row space of an m × n matrix A are equal. This common dimension is the rank of
A. Furthermore, the dimensions of the null space and column space are related
by:
rank A + dim(Nul A) = n
(where n is the number of columns of A).

Justification. • rank A = dim(Col A) = number of pivot columns.

• dim(Row A) = number of pivot rows. These are the same.

• dim(Nul A) = number of free variables = number of non-pivot columns.

Therefore:

rank A+dim(Nul A) = (num. of pivots)+(num. of non-pivots) = (total num. of columns) = n

Example 1.6.7 (From slide 4.6-15). If A is a 7 × 9 matrix with a 2-dimensional null

space, what is the rank of A? Solution: Here n = 9 and dim(Nul A) = 2. By the
Rank Theorem:
rank A + dim(Nul A) = 9
rank A + 2 = 9
rank A = 7
Since rank A = dim(Row A), we also know that dim(Row A) = 7.

1.6.3 The Invertible Matrix Theorem (Continued)

The Rank Theorem allows us to add the final, crucial statements to the Invertible
Matrix Theorem. For an n × n matrix A, all of the following are equivalent:

• A is an invertible matrix.

• …(all previous statements) …

• The columns of A form a basis for Rn . (This is (m) in the text)

• Col A = Rn .

• dim(Col A) = n.

• rank A = n.

• Nul A = {0}.

• dim(Nul A) = 0.

22
1.7 Section 4.7: Change of Basis
We often have one vector x but two or more different ”coordinate systems” (bases)
to describe it. This section shows how to translate between them.

• Let B = {b1 , . . . , bn } be a basis for V .

• Let C = {c1 , . . . , cn } be another basis for V .

• Let x be a vector in V .

We have two coordinate vectors: [x]B (the B-coordinates) and [x]C (the C-coordinates).
How are they related?
The vector x is the same in either basis:

x = [b1 . . . bn ][x]B and x = [c1 . . . cn ][x]C

If we are in Rn , we can write this using the change-of-coordinates matrices PB and

PC :
PB [x]B = x and PC [x]C = x
This gives PC [x]C = PB [x]B . To solve for [x]C , we multiply by PC−1 :

[x]C = (PC )−1 PB [x]B

The matrix PC←B = (PC )−1 PB is the change-of-coordinates matrix from B to C.

There is a more general way to think about this that also works for abstract
vector spaces. Let’s find the C-coordinates of the B basis vectors.

[x]C = [c1 b1 + · · · + cn bn ]C = c1 [b1 ]C + · · · + cn [bn ]C

 
c1

| |
[x]C = [b1 ]C . . . [bn ]C   ... 
 
| | cn

Theorem 1.7.1 (Change of Basis). Let B = {b1 , . . . , bn } and C = {c1 , . . . , cn } be

bases for V . There is a unique n × n matrix PC←B such that

[x]C = PC←B [x]B

The columns of this matrix are the C-coordinate vectors of the B basis vectors:
 
PC←B = [b1 ]C [b2 ]C . . . [bn ]C

Furthermore, this matrix is invertible and its inverse is the matrix that converts
from C to B:
(PC←B )−1 = PB←C

23
Procedure for Finding PC←B in Rn

To find the C-coordinates of the B-vectors (like [b1 ]C ), we must solve PC y = b1 ,

PC y = b2 , and so on. We can solve all of these at once with a single row reduction.
 
| | | |
PC PB =  c1 . . . cn b1 . . . bn  ∼ I (PC )−1 PB = I PC←B
     

| | | |

Example 1.7.2 (From slide 4.7-10). Let B = {b1 , b2 } and C = {c1 , c2 } be bases for
R2 , where        
−9 −5 1 3
b1 = , b2 = , c1 = , c2 =
1 −1 −4 −5
Find the change-of-coordinates matrix from B to C.
Solution: We need to find PC←B . We form the augmented matrix [PC | PB ] and
row reduce.
   
1 3 −9 −5 R2 +4R1 1 3 −9 −5
[c1 c2 | b1 b2 ] = −−−−→
−4 −5 1 −1 0 7 −35 −21
   
R2 /7 1 3 −9 −5 R1 −3R2 1 0 6 4
−−−→ −−−−→ = [I | PC←B ]
0 1 −5 −3 0 1 −5 −3
 
6 4
Thus, the change-of-coordinates matrix is PC←B = .
−5 −3

24
Chapter 2

Part 2: Summary of Key Concepts

and Formulas

This part is a concise summary of all major definitions, theorems, and procedures
from Chapter 4, intended for quick review.

2.1 Summary of Sections 4.1-4.4

2.1.1 Section 4.1: Vector Spaces and Subspaces
• Vector Space: A set V of objects (vectors) with two operations (addition and
scalar multiplication) that satisfies 10 axioms (closure under addition, com-
mutativity, associativity, zero vector, additive inverse, closure under scalar
multiplication, two distributivity laws, scalar associativity, and identity 1u =
u).

• Key Examples: Rn ; Pn (polynomials of degree at most n); Mm×n (matrices);

C[a, b] (continuous functions on [a, b]).

• Subspace: A subset H of a vector space V that is itself a vector space.

• Subspace Test (Shortcut): A subset H is a subspace of V if it satisfies three

properties:

1. 0 is in H.
2. H is closed under addition: u + v is in H for all u, v in H.
3. H is closed under scalar multiplication: cu is in H for all u in H, c in R.

• Spanning Set: Span{v1 , . . . , vp } is the set of all linear combinations of the

vectors.

• Theorem: Span{v1 , . . . , vp } is always a subspace of V .

25
2.1.2 Section 4.2: Null Spaces, Column Spaces, and Linear Trans-
formations
• Null Space (Nul A): The set of all solutions to the homogeneous equation
Ax = 0.

– Nul A is a subspace of Rn (where n is the number of columns of A).

– It is defined implicitly. You must solve Ax = 0 to find it.

• Column Space (Col A): The set of all linear combinations of the columns of
A. (This is the same as the range of the transformation x 7→ Ax).

– Col A is a subspace of Rm (where m is the number of rows of A).

– It is defined explicitly.
– b is in Col A if and only if Ax = b is consistent.

• Linear Transformation (General): A mapping T : V → W that preserves

addition (T (u + v) = T (u) + T (v)) and scalar multiplication (T (cu) = cT (u)).

• Kernel (ker(T)): The set of v in V such that T (v) = 0. This is a subspace of V .

(Generalizes Nul A).

• Range (Range(T)): The set of all images T (v). This is a subspace of W . (Gen-
eralizes Col A).

2.1.3 Section 4.3: Linearly Independent Sets; Bases

• Linearly Independent: A set {v1 , . . . , vp } is linearly independent if c1 v1 +· · ·+
cp vp = 0 has only the trivial solution (c1 = · · · = cp = 0).

• Basis: A set B = {b1 , . . . , bp } is a basis for a subspace H if:

1. B is a linearly independent set.

2. B spans H (i.e., Span{B} = H).

• Standard Bases: For Rn , E = {e1 , . . . , en } (columns of In ). For Pn , S = {1, t, t2 , . . . , tn }.

• Spanning Set Theorem: A tool to get a basis. If H = Span{v1 , . . . , vp }, you

can get a basis for H by removing any vectors that are linear combinations
of the others.

• How to find a Basis for Nul A:

1. Solve Ax = 0 by row reduction.

2. Write the general solution in parametric vector form.
3. The vectors {u, v, . . . } that appear in the parametric form are a basis for
Nul A.

• How to find a Basis for Col A:

1. Row reduce A to an echelon form (any echelon form will do).

26
 2. Identify the pivot columns.
3. The basis for Col A is the set of the pivot columns from the original
matrix A.

2.1.4 Section 4.4: Coordinate Systems

• Unique Representation Theorem: If B = {b1 , . . . , bn } is a basis for V , every
x in V can be written as a linear combination x = c1 b1 + · · · + cn bn in exactly
one way.

• B-Coordinate Vector: The unique

  weights c1 , . . . , cn are the coordinates of x
c1
 .. 
relative to B, written [x]B =  . .
cn

• Coordinate Systems in Rn :

– Let B = {b1 , . . . , bn } be a basis for Rn .

– Change-of-Coordinates Matrix: PB = [b1 . . . bn ].
– Standard-to-B Coordinates: [x]B = PB−1 x. (This is found by solving
PB c = x).
– B-to-Standard Coordinates: x = PB [x]B .

• Isomorphism: The coordinate mapping T (x) = [x]B is a one-to-one linear

transformation from V onto Rn . This allows us to ”translate” any problem
from V (like Pn ) into Rn+1 .

2.1.5 Section 4.5: The Dimension of a Vector Space

• Theorem: All bases for a vector space V have the same number of vectors.

• Dimension: The dimension of a (finite-dimensional) vector space V , dim V ,

is the number of vectors in any basis for V . dim{0} = 0.

• Dimensions of Subspaces: If H is a subspace of V , then dim H ≤ dim V .

• The Basis Theorem: Let V be a p-dimensional vector space.

– Any linearly independent set of exactly p vectors is a basis for V .

– Any set of p vectors that spans V is a basis for V .

• Dimensions of Nul A and Col A:

– dim(Col A) = number of pivot columns in A.

– dim(Nul A) = number of free variables = (number of columns)−(number of pivots).

27
2.1.6 Section 4.6: The Rank of a Matrix
• Row Space (Row A): The subspace of Rn spanned by the rows of A. Row A =
Col AT .
• Basis for Row A: The set of nonzero rows in any echelon form of A is a basis
for Row A.
• Rank: The rank of A is dim(Col A).
• The Rank Theorem:
– rank A = dim(Col A) = dim(Row A).
– rank A + dim(Nul A) = n (where n is the number of columns).
• Invertible Matrix Theorem (IMT) Additions: For an n × n matrix A, the fol-
lowing are equivalent to A being invertible:
– The columns of A form a basis for Rn .
– Col A = Rn .
– dim(Col A) = n.
– rank A = n.
– Nul A = {0}.
– dim(Nul A) = 0.

2.1.7 Section 4.7: Change of Basis

• Let B = {b1 , . . . , bn } and C = {c1 , . . . , cn } be two bases for V .
• Change-of-Coordinates Matrix from B to C: The matrix PC←B that converts
B-coordinates to C-coordinates:

[x]C = PC←B [x]B

• Formula for PC←B (General): The columns of PC←B are the C-coordinate vec-
tors of the B-basis.
 
PC←B = [b1 ]C [b2 ]C . . . [bn ]C

• Formula for PC←B (in Rn ):

PC←B = (PC )−1 PB

where PC = [c1 . . . cn ] and PB = [b1 . . . bn ].

• Shortcut for finding PC←B (in Rn ):
   
PC PB ∼ I PC←B

• Inverse: (PC←B )−1 = PB←C .

28
Chapter 3

Part 3: Solutions to Teacher’s

Exercises ([Link] & [Link])

This part contains the detailed, step-by-step solutions to all exercises provided by
the teacher in the [Link] and [Link] files.

3.1 Exercises from [Link]

Example 3.1.1 (Week 6, Exercise 1). Let H be a set of points inside and on the unit
circle in the xy-plane, i.e. H = {(x, y) : x2 + y 2 ≤ 1}. Is H a subspace of R2 ? Why?

Solution. To determine if H is a subspace of R2 , we must check the three properties

of a subspace.
 
2 0
1. Contains Zero? The zero vector of R is 0 = . We check if it satisfies the
0
condition x2 + y 2 ≤ 1:
02 + 02 = 0 ≤ 1
The zero vector is in H.

2. Closed
 under Addition? Let’s pick two vectors u and  v that are in H. Let
1 0
u= . u is in H because 12 + 02 = 1 ≤ 1. Let v = . v is in H because
0 1
02 + 12 = 1 ≤ 1. Now we check if their sum, u + v, is in H:
     
1 0 1
u+v= + =
0 1 1

We check the condition for this new vector: 12 +12 = 2. Since 2 6≤ 1, the vector
u + v is not in H.

3. Closed under Scalar Multiplication? (We already know H is  not a subspace,
1
but we can check this property for completeness.) Let u = , which is in H.
    0
1 3
Let c = 3. Then cu = 3 = . We check the condition: 32 + 02 = 9. Since
0 0
9 6≤ 1, the vector cu is not in H.

29
Conclusion: H is not a subspace of R2 because it is not closed under addition
(and also not closed under scalar multiplication).

Example 3.1.2 (Week 6, Exercise 2). Determine if the given set is a subspace of Pn
for an appropriate value of n.

Solution. We check the three properties for each set. The ”parent” vector space V
is Pn for some n, and its zero vector is the zero polynomial 0(t) = 0.
(a) all polynomials of the form p(t) = at2 , a ∈ R. (This is a subset of P2 ).

1. Zero? If a = 0, we get p(t) = 0t2 = 0, which is the zero polynomial. Yes, 0 is in

the set.

2. Addition? Let p(t) = at2 and q(t) = bt2 be in the set. (p + q)(t) = at2 + bt2 =
(a + b)t2 . Since a + b is a real scalar, the sum is in the set. Yes.

3. Scalar Mult.? Let p(t) = at2 be in the set and c be a scalar. (cp)(t) = c(at2 ) =
(ca)t2 . Since ca is a real scalar, the new polynomial is in the set. Yes.

Conclusion: Yes, this is a subspace. (It is Span{t2 }).

(b) all polynomials of the form p(t) = a + t2 , a ∈ R. (This is a subset of P2 ).

1. Zero? We check if the zero polynomial 0(t) = 0 is in the set. Is it possible to

find a scalar a such that a + t2 = 0 for all t? No. If t = 1, a = −1. If t = 2, a = −4.
There is no single value of a that makes this the zero polynomial.

Conclusion: No, the zero vector is not in the set, so it is not a subspace.
(c) all polynomials of degree at most 3, with integer coefficients. (This is a
subset of P3 ).

1. Zero? Yes, p(t) = 0 has the integer coefficient 0.

2. Addition? Let p(t) = a0 + a1 t + . . . and q(t) = b0 + b1 t + . . . where all ai , bi are

integers. Their sum (p + q)(t) = (a0 + b0 ) + (a1 + b1 )t + . . . also has integer
coefficients (since ai + bi is an integer). Yes.

3. Scalar Mult.? The scalars in a vector space are (by default) real numbers. Let
p(t) = 2t2 . p is in the set (its coefficients 2, 0, 0 are integers). Let c = 0.5. Then
√
(cp)(t) = 0.5(2t2 )√= 1t2 . This is in the set.
√ Let’s try another scalar. Let c = 2.
Then (cp)(t) = 2 2t2 . The coefficient 2 2 is not an integer.

Conclusion: No, the set is not closed under scalar multiplication.

(d) all polynomials in Pn such that p(0) = 0.

1. Zero? Let p = 0, the zero polynomial. Then p(t) = 0 for all t. Thus p(0) = 0.
Yes, 0 is in the set.

2. Addition? Let p, q be in the set. This means p(0) = 0 and q(0) = 0. We check
their sum p + q: (p + q)(0) = p(0) + q(0) = 0 + 0 = 0. So p + q is in the set. Yes.

3. Scalar Mult.? Let p be in the set, so p(0) = 0. Let c be a scalar. We check cp:
(cp)(0) = c · p(0) = c · 0 = 0. So cp is in the set. Yes.

30
Conclusion: Yes, this is a subspace of Pn . (It is the kernel of the ’evaluation at 0’
transformation).
(e) {p(t) : p(t) ∈ P3 , p(2) = 1}

1. Zero? Let p = 0, the zero polynomial. Then p(2) = 0. This is not equal to 1.

Conclusion: No, the zero vector is not in the set, so it is not a subspace.
(f) {tp(t) : p(t) ∈ P2 } This set is the set of all polynomials of the form t(a0 + a1 t +
a2 t ) = a0 t + a1 t2 + a2 t3 . This is the set of all polynomials in P3 with a zero constant
2

term.
W = {a0 t + a1 t2 + a2 t3 : a0 , a1 , a2 ∈ R}
This is a subspace of P3 . We can prove it with the subspace test:

1. Zero? If a0 = a1 = a2 = 0, we get the zero polynomial. Yes.

2. Addition? (a0 t+a1 t2 +a2 t3 )+(b0 t+b1 t2 +b2 t3 ) = (a0 +b0 )t+(a1 +b1 )t2 +(a2 +b2 )t3 .
This is in the set. Yes.

3. Scalar Mult.? c(a0 t + a1 t2 + a2 t3 ) = (ca0 )t + (ca1 )t2 + (ca2 )t3 . This is in the set.
Yes.

Conclusion: Yes, this is a subspace of P3 . (It is Span{t, t2 , t3 }).

Example 3.1.3 (Week 6, Exercise 3). Determine if the given set is a subspace of
M2×2 .

Solution. The vector spaceis M2×2  , the set of all 2 × 2 matrices. The zero vector is
0 0
the 2 × 2 zero matrix, O = .
  0 0
a b
(a) : a, b, c ∈ R (The set of upper triangular 2 × 2 matrices)
0 c
 
0 0
1. Zero? O = . This is in the set (by choosing a = 0, b = 0, c = 0). Yes.
0 0
   
a1 b 1 a2 b2
2. Addition? Let A = and B = be in the set.
0 c1 0 c2
 
a1 + a2 b 1 + b 2
A+B =
0 c1 + c2

This sum is also an upper triangular matrix, so it is in the set. Yes.

3. Scalar Mult.? Let k be a scalar.

   
a1 b 1 ka1 kb1
kA = k =
0 c1 0 kc1

This is also an upper triangular matrix. Yes.

Conclusion: Yes, this is a subspace of M2×2 .

(b) {A : A ∈ M2×2 , AT = A} (The set of symmetric 2 × 2 matrices)

31
  T  
T 0 0 0 0
1. Zero? Is O = O? Yes, = . 0 is in the set.
0 0 0 0

2. Addition? Let A, B be in the set. This means AT = A and B T = B. We check

their sum A + B. Using properties of transposes:

(A + B)T = AT + B T = A + B

So A + B is also symmetric and is in the set. Yes.

3. Scalar Mult.? Let A be in the set (AT = A) and c be a scalar. We check cA.
Using properties of transposes:

(cA)T = c(AT ) = cA

So cA is also symmetric and is in the set. Yes.

Conclusion: Yes, this is a subspace of M2×2 .
(c) {A : A ∈ M2×2 , AB = 0} where B is fixed
1. Zero? We check the zero matrix O. O · B = 0. Yes, 0 is in the set.

2. Addition? Let A1 , A2 be in the set. This means A1 B = 0 and A2 B = 0. We

check their sum A1 + A2 :

(A1 + A2 )B = A1 B + A2 B (Distributive property)

=0+0=0
So A1 + A2 is in the set. Yes.

3. Scalar Mult.? Let A1 be in the set (A1 B = 0) and c be a scalar. We check cA1 :

(cA1 )B = c(A1 B) = c(0) = 0

So cA1 is in the set. Yes.

Conclusion: Yes, this is a subspace of M2×2 . (It is the kernel of the linear transfor-
mation T (A) = AB).
(d) {A : A ∈ M2×2 , A2 = A} (The set of idempotent 2 × 2 matrices)
    
2 0 0 0 0 0 0
1. Zero? Is O = O? Yes, = . Yes, 0 is in the set.
0 0 0 0 0 0
 
1 0
2. Addition? Let’s test this. The identity matrix I = is in the set because
0 1
I 2 = I. Let A
 = I and B = I. Both are in the set. Their sum is A + B =
2 0
I + I = 2I = . Now we check if the sum, 2I, is in the set. We check the
0 2
condition A2 = A:
(2I)2 = (2I)(2I) = 4I 2 = 4I
But the matrix itself is 2I. Since 4I 6= 2I, the sum A + B is not in the set.
Conclusion: No, the set is not closed under addition, so it is not a subspace.

32
Example 3.1.4 (Week 6, Exercise 4). Determine if the given set is a subspace of
F[0, 1] (the space of all functions defined on [0, 1]).

Solution. The vector space is V = F[0, 1], the set of all real-valued functions on [0, 1].
The zero vector 0 is the function 0(x) = 0 for all x ∈ [0, 1].
(a) {f : f (0) = 0}

1. Zero? 0(0) = 0. Yes, the zero function is in the set.

2. Addition? Let f, g be in the set, so f (0) = 0 and g(0) = 0. Check f + g:

(f + g)(0) = f (0) + g(0) = 0 + 0 = 0. So f + g is in the set. Yes.

3. Scalar Mult.? Let f be in the set, so f (0) = 0. Let c be a scalar. Check cf :

(cf )(0) = c · f (0) = c · 0 = 0. So cf is in the set. Yes.

Conclusion: Yes, this is a subspace.

(b) {f : f (0) = 1}

1. Zero? 0(0) = 0. This is not 1.

Conclusion: No, the zero vector is not in the set, so it is not a subspace.
(c) {f : f (0) = f (1)}

1. Zero? 0(0) = 0 and 0(1) = 0. Since 0 = 0, the zero function is in the set. Yes.

2. Addition? Let f, g be in the set. So, f (0) = f (1) and g(0) = g(1). Check f + g:
(f + g)(0) = f (0) + g(0) = f (1) + g(1) = (f + g)(1). So f + g is in the set. Yes.

3. Scalar Mult.? Let f be in the set, so f (0) = f (1). Let c be a scalar. Check cf :
(cf )(0) = c · f (0) = c · f (1) = (cf )(1). So cf is in the set. Yes.

Conclusion: Yes, this is a subspace.

R1
(d) {f : f is integrable and 0 f (x)dx = 0}
R1
1. Zero? The zero function 0(x) = 0 is integrable, and 0 0dx = 0. Yes.
R1 R1
2. Addition? Let f, g be in the set. 0 f (x)dx = 0 and 0 g(x)dx = 0. Check f + g.
From calculus, the sum of integrable functions is integrable, and:
Z 1 Z 1 Z 1
(f (x) + g(x))dx = f (x)dx + g(x)dx = 0 + 0 = 0
0 0 0

So f + g is in the set. Yes.

R1
3. Scalar Mult.? Let f be in the set ( 0
f (x)dx = 0) and c be a scalar. Check cf :
Z 1 Z 1
(cf (x))dx = c f (x)dx = c · 0 = 0
0 0

So cf is in the set. Yes.

Conclusion: Yes, this is a subspace.

Example 3.1.5 (Week 6, Exercise 5). Determine whether W is a subspace of Rn .

33
   
 −a + 1 
Solution. (a) W =  a − 6b  : a, b, c ∈ R We check the three properties for this
2b + ac
 
subset of R3 .
 
0
1. Zero? Is 0 in W ? We would need to find a, b, c such that:

0
• −a + 1 = 0 =⇒ a = 1
• a − 6b = 0 =⇒ 1 − 6b = 0 =⇒ b = 1/6
• 2b + ac = 0 =⇒ 2(1/6) + (1)c = 0 =⇒ 1/3 + c = 0 =⇒ c = −1/3

Since we found values for a, b, c that produce the zero vector, 0 is in W .

2. Closed under Addition? Let’s pick two simple vectors in W .

 
1
• Let a = 0, b = 0, c = 0. Then u = 0 is in W .
0
 
0
• Let a = 1, b = 0, c = 0. Then v = 1 is in W .

0
 
1
Now we check their sum: u + v = 1. Is this sum in W ? We must find a, b, c
0
such that:

• −a + 1 = 1 =⇒ a = 0
• a − 6b = 1 =⇒ 0 − 6b = 1 =⇒ b = −1/6
• 2b + ac = 0 =⇒ 2(−1/6) + (0)c = 0 =⇒ −1/3 = 0.

This is a contradiction. We cannot find a, b, c that produce the vector u + v.

Conclusion: No,
 the  set is not closed
 under addition, so it is not a subspace.

 a − b 

b − c
  
4
(b) W =  
 : a, b, c ∈ R This is a subset of R . As shown in Example


 c − a 
b
 
1.1.11 (from the theory section), we can write any vector in W as:
     
1 −1 0
0 1 −1
x = a −1 + b  0  + c  1 
    

0 1 0

This shows that W = Span{v1 , v2 , v3 }, where v1 , v2 , v3 are the three vectors above.
Conclusion: By Theorem 1.1.11, any span of vectors is a subspace. Yes, this is a
subspace of R4 .

34
   

 a + 3b 

a − b
  
(c) W =   : a, b ∈ R This is a subset of R4 . We rewrite the vector:

 2a − b  

4b
 

      
a 3b 1 3
 a  −b 1 −1
x=
2a + −b = a 2 + b −1
      

0 4b 0 4
   
1 3
1 −1
This shows that W = Span{v1 , v2 }, where v1 = 
2 and v2 = −1. Conclusion:
  

0 4
4
By Theorem 1.1.11, W is a subspace of R .

3.2 Exercises from [Link]

(Solutions for Part 3 to be generated in a future session)

35
Chapter 4

Part 4: Solutions to Textbook

Exercises (Lay, Chapter 4)

(Solutions for Part 4 to be generated in a future session)

36