Module 3 - Subspace Properties

Syllabus: Row and column spaces - Rank and nullity - Bases for subspace -
Invertibility.

1 Row and Column Spaces

1.1 Definitions

Definition For an m × n matrix

 
a11 a12 · · · a1n
a
 21 a22 · · · a2n 
A =  ..

.. . . . ... 
 . . 
am1 am2 · · · amn
the vectors
r1 = [a11 a12 ··· a1n ]
r2 = [a21 a22 ··· a2n ]
..
.
rm = [am1 am2 ··· amn ]
in Rn formed from the rows of A are called the row vectors of A, and the
vectors      
a11 a12 a1n
a  a  a 
 21   22   2n 
c1 =  ..  , c2 =  ..  , . . . , cn =  .. 
 .   .   . 
am1 am2 amn
in Rm formed from the columns of A are called the column vectors of A.

Definition [Row Space] The row space of an m×n matrix A is the subspace
of Rn spanned by the row vectors of A.
Row(A) = span{r1 , r2 , . . . , rm }
where ri are the row vectors of A.
Definition [Column Space] The column space of an m × n matrix A is the
subspace of Rm spanned by the column vectors of A.
Col(A) = span{c1 , c2 , . . . , cn }

1
where cj are the column vectors of A. Also called the range or image of A.
Definition [Null Space] The null space of an m × n matrix A is the set
of all solutions to the homogeneous equation Ax = 0. Null(A) = {x ∈ Rn :
Ax = 0} The null space is a subspace of Rn . Also called the kernel of A.

1.2 Key Properties

Definition Two matrices A and B are row equivalent if and only if there
exists a finite sequence of elementary row operations that transforms A into
B.
Equivalently, A and B are row equivalent if
A = Ek Ek−1 · · · E1 B
where each Ei is an elementary matrix corresponding to an elementary row
operation.
That is,
A ∼ B ⇐⇒ A can be transformed into B by elementary row operations
Property 1 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 as well as for that of B.
Property 2 Row operations do not change the row space of a matrix. If B
is obtained from A by elementary row operations, then Row(A) = Row(B).
Property 3 The row space of A equals the row space of its row echelon
form (REF).
Property 4 The nonzero rows of the REF of A form a basis for Row(A).
Property 5 Col(A) = {Ax : x ∈ Rn }. The column space consists of all
possible linear combinations of the columns of A.
Property 6 For any matrix A:
Row(A) = Row(AT A)
Col(A) = Col(AAT )

1.3 Important Theorems

Theorem dim(Row(A)) = dim(Col(A)) = rank(A)

2
 Theorem If A is m × n, then:
Row(A) ⊆ Rn
Col(A) ⊆ Rm
dim(Row(A)) ≤ min(m, n)
dim(Col(A)) ≤ min(m, n)

2 Rank and Nullity

2.1 Definitions

Definition [Rank of a Matrix] The dimension of the row space (or equiva-
lently, the column space) of matrix A.
rank(A) = dim(Row(A)) = dim(Col(A))
• Equals the number of linearly independent rows (or columns)
• Equals the number of pivot positions in the REF(RREF)
Definition [Nullity of a Matrix] The dimension of the null space of matrix
A.
nullity(A) = dim(Null(A))
Definition [Full Rank] A matrix A is said to have full rank if:
rank(A) = min(m, n) where A is m × n
For square matrices: rank(A) = n (where A is n × n).

2.2 Fundamental Theorem

Theorem [Rank-Nullity Theorem] For an m × n matrix A:

rank(A) + nullity(A) = n

2.3 Properties of Rank

Property 1 0 ≤ rank(A) ≤ min(m, n) for an m × n matrix A.

Property 2 rank(A) = rank(AT )
Property 3 rank(AB) ≤ min(rank(A), rank(B))

3
 Property 4 If A is m × n and B is n × p, then:
rank(AB) ≤ rank(A)
rank(AB) ≤ rank(B)
Property 5 rank(A + B) ≤ rank(A) + rank(B)
Property 6 For any invertible matrices P and Q:
rank(P AQ) = rank(A)
Property 7 rank(AT A) = rank(A) = rank(AAT )

3 Bases for Subspaces

3.1 Definitions

Definition [Basis] A set of vectors {v1 , v2 , . . . , vk } is a basis for a subspace

V if:
1. The vectors are linearly independent
2. The vectors span V (every vector in V can be written as their linear
combination)
Definition [Orthogonal Basis] A basis where all vectors are mutually or-
thogonal.

Definition [Orthonormal Basis] An orthogonal basis where all vectors have

unit length.
Definition The standard basis for Rn is the ordered basis:
E = (e1 , e2 , . . . , en )
where each ei is the i-th standard unit vector:
     
1 0 0
0 1 0
     
e1 =   , e2 = 0 , . . . ,
0
en = 0
   
 ..   ..   .. 
. . .
0 0 1
In general, ei has 1 in the i-th position and 0 in all other positions.
The standard basis is:

4
 • Orthonormal (each vector has unit length and they are mutually orthog-
onal)
• The natural choice for coordinate representation in Rn
• The basis with respect to which coordinates are typically given unless
otherwise specified
Definition An ordered basis for a vector space V is a finite sequence
B = (v1 , v2 , . . . , vn ) of vectors in V such that:
1. The set {v1 , v2 , . . . , vn } is linearly independent
2. The set {v1 , v2 , . . . , vn } spans V
3. The order of the vectors matters for coordinate representation
Every vector v ∈ V can be uniquely written as:
v = c1 v1 + c2 v2 + · · · + cn vn
where the coefficients (c1 , c2 , . . . , cn ) are called the coordinates of v with
respect to the ordered basis B.

3.2 Key Theorems and Properties

Theorem: [Basis Characterization] Let V be a subspace of dimension n.

A set of n vectors in V is a basis if and only if the vectors are linearly
independent.
Theorem: [Dimension Theorem] All bases of a vector space have the same
number of vectors.
Theorem: [Basis Extension] Any linearly independent set in a finite-
dimensional vector space can be extended to a basis.
Theorem: [Basis Reduction] Any spanning set of a finite-dimensional
vector space contains a basis.

3.3 Finding Bases

For Row Space:

1. Reduce A to REF
2. The nonzero rows of REF form a basis for Row(A)

5
 For Column Space:
1. Reduce A to REF
2. Identify pivot columns in REF
3. The corresponding columns in the original matrix A form a basis for
Col(A)
For Null Space:
1. Reduce REF(A) (Use Backward Substitution method to find the solution)
[OR] RREF(A) (Continuation Process...)
2. Solve Ax = 0
3. Express the solution in parametric form
4. The parameter vectors form a basis for Null(A)
Lemma If A is m × n with rank r, then:
• Any basis for Row(A) has r vectors
• Any basis for Col(A) has r vectors
• Any basis for Null(A) has (n − r) vectors

4 Invertibility
4.1 Definition

Definition [Invertible Matrix] A square matrix A is invertible (or nonsingu-

lar) if there exists a matrix A−1 such that:
AA−1 = A−1 A = I

4.2 The Invertible Matrix Theorem

Theorem: Let A be an n × n matrix. The following statements are equiva-

lent:
1. A is invertible
2. A is row equivalent to the identity matrix I
3. A has n pivot positions

6
 4. The equation Ax = 0 has only the trivial solution
5. The columns of A form a linearly independent set
6. The columns of A span Rn
7. The columns of A form a basis for Rn
8. The rows of A form a linearly independent set
9. The rows of A span Rn
10. The rows of A form a basis for Rn
11. The equation Ax = b has exactly one solution for each b in Rn
12. det(A) ̸= 0
13. rank(A) = n
14. nullity(A) = 0
15. AT A is invertible
16. A can be expressed as a product of elementary matrices

4.3 Properties of Invertible Matrices

Property If A is invertible, then A−1 is unique.

Property If A is invertible, then (A−1 )−1 = A.
Property If A and B are invertible n × n matrices, then AB is invertible
and:
(AB)−1 = B −1 A−1
Property If A is invertible, then AT is invertible and:
(AT )−1 = (A−1 )T
Property If A is invertible and c ̸= 0, then cA is invertible and:
1
(cA)−1 = A−1
c

7
4.4 Methods to Find Inverse

Method 1 (Gauss-Jordan Elimination):

1. Form the augmented matrix [A|I]
2. Reduce to [I|A−1 ] using row operations
3. If reduction is possible, A−1 is the right half
Method 2 (Adjugate Method): For small matrices:
1
A−1 = × adj(A)
det(A)
Method 3 (LU Decomposition): If A = LU , then A−1 = U −1 L−1 .

5 Problem-Solving Tricks and Tips

5.1 Row and Column Space Tips

Tip 1 To find a basis for Row(A), always use REF. Never use the original
matrix rows.
Tip 2 To find a basis for Col(A), find pivot columns in REF, then use
corresponding columns from the original matrix A.
Tip 3 If asked whether a vector b is in Col(A), check if Ax = b has a
solution.
Tip 4 Remember: Row operations change column relationships but pre-
serve row relationships.

5.2 Rank and Nullity Tips

Tip 1 Use the Rank-Nullity Theorem as a check: rank + nullity should equal
the number of columns.
Tip 2 For quick rank calculation, count pivot positions in REF(RREF).
Tip 3 If A is m × n and rank(A) = m, then the rows are linearly indepen-
dent.
Tip 4 If A is m × n and rank(A) = n, then the columns are linearly
independent.
Tip 5 rank(AB) ≤ min(rank(A), rank(B)) - use this for theoretical prob-
lems.

8
5.3 Basis Tips

Tip 1 To check if vectors form a basis for Rn , form a matrix with these
vectors as columns and check if it’s invertible.
Tip 2 To extend a linearly independent set to a basis, add standard basis
vectors and remove dependent ones.

5.4 Invertibility Tips

 
a b
Tip 1 For 2 × 2 matrices: A = is invertible iff ad − bc ̸= 0.
c d
Tip 2 A matrix is invertible iff its RREF is the identity matrix.
Tip 3 If you need to solve multiple systems Ax = bi with the same A, find
−1
A first.
Tip 4 Use the determinant test for invertibility only when computing the
actual inverse isn’t needed.

5.5 Memory Aids

• Remember: “REF rows for row space, original columns for column space”
• Remember: “Rank + Nullity = Number of columns”
• Remember: “Square + Full rank = Invertible”
• Remember: “Pivot columns → basis for column space”

Problem 1

Determine the dimension and basis for the row space of the matrix:
 
2 −1 3
1 0 1 
A= 0 2 −1


1 1 4

Solution

To find the row space, we perform row reduction to obtain the row echelon
form (REF).

9
Step 1: Swap rows to get a leading 1 in the first position.
R1 ↔ R2
 
1 0 1
2 −1 3 
 
0 2 −1
1 1 4
Step 2: Eliminate entries below the first pivot.
R2 → R2 − 2R1
R4 → R4 − R1
 
1 0 1
0 −1 1 
 
0 2 −1
0 1 3
Step 3: Make the second pivot positive.
R2 → −R2
 
1 0 1
0 1 −1
 
0 2 −1
0 1 3
Step 4: Eliminate entries below the second pivot.
R3 → R3 − 2R2
R4 → R4 − R2
 
1 0 1
0 1 −1
 
0 0 1 
0 0 4
Step 5: Eliminate entries below the third pivot.
R4 → R4 − 4R3
 
1 0 1
0 1 −1
REF(A) =  
0 0 1 
0 0 0

10
Conclusion

Dimension: The row space has dimension 3, since there are 3 non-zero rows
in REF(A).
Basis: A basis for the row space is given by the non-zero rows of REF(A):
 T  T  T 
 1
 0 0  
B = 0 ,  1  , 0
 1 −1 1 
 

Or written as row vectors:

B = {[1, 0, 1], [0, 1, −1], [0, 0, 1]}
The row space is therefore a 3-dimensional subspace of R3 , which means
the row space is all of R3 .

Problem 2

Find bases for the row space, the column space, the null space of the matrix
A, dim(Row(A)), dim(Col(A)), dim(Null(A)). And hence, verify the Rank-
Nullity theorem.  
−2 −5 8 0 −17
 1 3 −5 1 5 
A=  3 11 −19 7 1 


1 7 −13 5 −3

Solution

We will use REF (Row Echelon Form) for finding the row and column space
bases, and RREF (Reduced Row Echelon Form) for finding the null space
basis.

PART I: Row Reduction to REF

Initial Matrix:  
−2 −5 8 0 −17
 1 3 −5 1 5 
A= 
 3 11 −19 7 1 
1 7 −13 5 −3

11
Step 1: R1 ↔ R2 (swap rows to get leading 1)
 
1 3 −5 1 5
−2 −5 8 0 −17
 
 3 11 −19 7 1 
1 7 −13 5 −3
Step 2: Eliminate below the first pivot
R2 → R2 + 2R1
R3 → R3 − 3R1
R4 → R4 − R1
Calculations:
R2 : [−2, −5, 8, 0, −17] + 2[1, 3, −5, 1, 5] = [0, 1, −2, 2, −7]
R3 : [3, 11, −19, 7, 1] − 3[1, 3, −5, 1, 5] = [0, 2, −4, 4, −14]
R4 : [1, 7, −13, 5, −3] − [1, 3, −5, 1, 5] = [0, 4, −8, 4, −8]
Result:  
1 3 −5 1 5
0
 1 −2 2 −7 

0 2 −4 4 −14
0 4 −8 4 −8
Step 3: Eliminate below the second pivot
R3 → R3 − 2R2
R4 → R4 − 4R2
Calculations:
R3 : [0, 2, −4, 4, −14] − 2[0, 1, −2, 2, −7] = [0, 0, 0, 0, 0]
R4 : [0, 4, −8, 4, −8] − 4[0, 1, −2, 2, −7] = [0, 0, 0, −4, 20]
Result:  
1 3 −5 1 5
0
 1 −2 2 −7
0 0 0 0 0
0 0 0 −4 20
Step 4: R4 → − 41 R4

12
 Calculation: − 41 [0, 0, 0, −4, 20] = [0, 0, 0, 1, −5]
Step 5: R3 ↔ R4 (move pivot up)
REF (Row Echelon Form):
 
1 3 −5 1 5
0 1 −2 2 −7
 
0 0 0 1 −5
0 0 0 0 0

PART II: Continue to RREF (OR simply apply the backward substitution method
from REF)

Step 6: Eliminate above the third pivot

R1 → R1 − R3
R2 → R2 − 2R3
Calculations:
R1 : [1, 3, −5, 1, 5] − [0, 0, 0, 1, −5] = [1, 3, −5, 0, 10]
R2 : [0, 1, −2, 2, −7] − 2[0, 0, 0, 1, −5] = [0, 1, −2, 0, 3]
Result:  
1 3 −5 0 10
0
 1 −2 0 3 
0 0 0 1 −5
0 0 0 0 0
Step 7: R1 → R1 − 3R2
Calculation: [1, 3, −5, 0, 10] − 3[0, 1, −2, 0, 3] = [1, 0, 1, 0, 1]
RREF (Reduced Row Echelon Form):
 
1 0 1 0 1
0 1 −2 0 3 
 
0 0 0 1 −5
0 0 0 0 0

PART III: Finding the Bases

1. Row Space Basis

Method: The basis for the row space consists of the non-zero rows of REF.

13
 From REF, the non-zero rows are:
Row 1: [1, 3, −5, 1, 5]
Row 2: [0, 1, −2, 2, −7]
Row 3: [0, 0, 0, 1, −5]

 T  T  T 


 1 0 0 


 3   1   0  

      

−5 , −2 ,  0  
Basis for Row Space =      


1 2 1  
 
 5 −7 −5 
 

dim(Row(A)) = 3

2. Column Space Basis

Method: The pivot columns are columns 1, 2, and 4 (from REF). We take
the corresponding columns from the original matrix.
From the original matrix A:
 
−2
1
Column 1: 3


1
 
−5
3
Column 2:  11 


7
 
0
1
Column 4: 7


5
     

 −2 −5 0 
     
1   3  1

Basis for Column Space =   ,   ,  


 3 11 7 

1 7 5
 

14
 dim(Col(A)) = 3

3. Null Space Basis

Method: From RREF, solve the homogeneous system Ax = 0.

From RREF, the system of equations is:
x1 + x3 + x5 = 0 (from row 1)
x2 − 2x3 + 3x5 = 0 (from row 2)
x4 − 5x5 = 0 (from row 3)
Identify variables:
• Pivot variables: x1 , x2 , x4 (columns 1, 2, 4)
• Free variables: x3 , x5 (columns 3, 5)
Solve for pivot variables:
x1 = −x3 − x5
x2 = 2x3 − 3x5
x4 = 5x5
Express general solution: Let x3 = s and x5 = t (free variables). Then:
  
−s − t
  
−1
 
x1 −1
x2  2s − 3t 2 −3
       
x= x 3
  
 =  s  = s 1  + t 0 
    
x4   5t  0 5
x5 t 0 1

15
 Basis vectors:
 
−1
2
 
1
v1 =   (corresponds to x3 = 1, x5 = 0)
0
0
 
−1
−3
 
0
v2 =   (corresponds to x3 = 0, x5 = 1)
5
1
   

 −1 −1 
 2  −3

   


 1  ,  0 
Basis for Null Space =    


 0   5  
 
0 1
 

dim(Null(A)) = 2

PART IV: Verification

Rank-Nullity Theorem: For any m × n matrix A:

rank(A) + nullity(A) = n
Our results:
• rank(A) = 3 (number of pivot columns)
• nullity(A) = 2 (dimension of null space)
• n = 5 (number of columns in A)
Check: 3 + 2 = 5 ✓
Additional verification:
• dim(Row(A)) = dim(Col(A)) = rank(A) = 3 ✓
• Number of free variables = n − rank(A) = 5 − 3 = 2 ✓

16
FINAL ANSWER

1. Row Space Basis:

{[1, 3, −5, 1, 5], [0, 1, −2, 2, −7], [0, 0, 0, 1, −5]}
dim(Row(A)) = 3
2. Column Space Basis:
     

 −2 −5 0 
     
 1  ,  3  , 1

 3   11  7
 
1 7 5
 

dim(Col(A)) = 3
3. Null Space Basis:    

 −1 −1 

   
 2  −3 

 1 , 0 
   


 0   5 

 
0 1
 

dim(Null(A)) = 2

Problem 3

For what value of b is the vector

 
1
2
B=
3


b
in the column space of the following matrix

 
2 3 3
0 −4 −5
A=
6 3 0 


1 1 3

17
Solution
Step 1: Set up the augmented matrix [A|B]

 
2 3 3 | 1
0 −4 −5 | 2
[A|B] = 
6 3 0 | 3


1 1 3 | b

Step 2: Row operations to achieve REF

R1 ↔ R4 : Get leading 1 in position (1, 1)

 
1 1 3 | b
0 −4 −5 | 2
 
6 3 0 | 3
2 3 3 | 1
R3 = R3 − 6R1 , R4 = R4 − 2R1 : Eliminate first column

 
1 1 3 | b
0 −4 −5 | 2 
 
0 −3 −18 | 3 − 6b
0 1 −3 | 1 − 2b
R2 ↔ R4 : Get leading 1 in position (2, 2)

 
1 1 3 | b
0 1 −3 | 1 − 2b
 
0 −3 −18 | 3 − 6b
0 −4 −5 | 2
R3 = R3 + 3R2 , R4 = R4 + 4R2 : Eliminate second column
For R3 : R3 + 3R2 = [0 − 3 − 18 | 3 − 6b] + 3[0 1 − 3 | 1 − 2b]
= [0 − 3 + 3 − 18 − 9 | 3 − 6b + 3 − 6b] = [0 0 − 27 | 6 − 12b]
For R4 : R4 + 4R2 = [0 − 4 − 5 | 2] + 4[0 1 − 3 | 1 − 2b]
= [0 − 4 + 4 − 5 − 12 | 2 + 4 − 8b] = [0 0 − 17 | 6 − 8b]

18
Step 3: Resulting REF form

 
1 1 3 | b
0
 1 −3 | 1 − 2b 
0 0 −27 | 6 − 12b
0 0 −17 | 6 − 8b
R3 R4
R3 = −27 , R4 = −17 :

1 3 |
 
1 b
0
 1 −3 | 1 − 2b

0 0 1 | 6−12b
−27

0 0 1 | 6−8b
−17

R4 = R4 − R3 :

1 3 |
 
1 b
0
 1 −3 | 1 − 2b 

6−12b
0 0 1 | −27

6−8b 6−12b
0 0 0 | −17 + 27

Step 4: Consistency condition

For the system to be consistent, row 4 of all entries must be zero:

That is, from row 4: 6−8b 6−12b
−17 + 27 = 0

12b − 6 8b − 6
=⇒ =
27 17
Simplify:
17(12b − 6) = 27(8b − 6)
204b − 102 = 216b − 162
204b − 216b = −162 + 102
−12b = −60
b=5

19
Step 5: Complete solution when b = 5

Substituting b = 5 into our REF form:

 
1 1 3 | 5
0
 1 −3 | −9

0 0 1 | 2
0 0 0 | 0
Final REF form:

 
1 1 3 | 5
0 1 −3 | −9
0 0 1 | 2
Back-substitution to find a particular solution:
From row 3: x3 = 2
From row 2: x2 − 3x3 = −9 ⇒ x2 − 3(2) = −9 ⇒ x2 = −3
From row 1: x 2 + 3x3 = 5 ⇒ x1 + (−3) + 3(2) = 5 ⇒ x1 = 2
1 + x
2
Solution: x = −3
2
Verification: A · x = B when b = 5

 
  2 · 2 + 3 · (−3) + 3 · 2
2 0 · 2 + (−4) · (−3) + (−5) · 2
A · −3 =  
 6 · 2 + 3 · (−3) + 0 · 2 
2
1 · 2 + 1 · (−3) + 3 · 2
 
4−9+6
0 + 12 − 10
= 12 − 9 + 0 


2−3+6
 
1
2
=3 ✓


20
Conclusion
 
1
2
The vector B = 
3 is in the column space of matrix A if and only if b = 5 .


Exercises

For the following matrices:

 
1 −3 4 −2 5 4
 2 −6 9 −1 8 2 
1. A =  
 2 −6 9 −1 9 7 
−1 3 −4 2 −5 −4


1 1 −3 7 9 −9

1 2 −4 10 13 −12
 
2. A = 
1 −1 −1 1 1 −3 

1 −3 1 −5 −7 3 
1 −2 0 0 −5 −4
Find bases for the row space, the column space and the null space of the
matrix. Also, find the Rank(A), Nullity(A) and Verify the Rank-Nullity
theorem.

21
 Bases for Subspaces

Procedure to solve the problem:

• V and W are subspaces of Rn

• α = {v1 , v2 , . . . , vk } is a basis for V
• β = {w1 , w2 , . . . , wℓ } is a basis for W

Step 1: Construct Matrix Q

1. Form the n × (k + ℓ) matrix Q by placing all basis vectors as columns:

Q = [v1 , v2 , v3 , . . . , vk , w1 , w2 , . . . , wℓ ]n×(k+ℓ)

2. Each vector is written in column form:

   
a1j b1j
a  b 
 2j   2j 
vj =  3j  and wj =  b3j  ∈ Rn
a   
 ..   .. 
 .   . 
anj bnj

Step 2: Find REF(Q) - Reduced Row Echelon Form

1. Apply elementary row operations to transform Q into its reduced row

echelon form REF(Q)

Step 3: Find Basis for V + W using C(Q)

1. Key Connection: C(Q) = V + W

• The column space of Q equals the sum of subspaces V and W
2. Extract basis for V + W:
• Identify pivot columns in REF(Q)
• The corresponding columns in the original matrix Q form a basis for
V +W
• These pivot columns are linearly independent and span V + W

22
Step 4: Find Basis for V ∩ W using N(Q)

1. Key Connection: N (Q) can be identified with V ∩ W

• The null space of Q corresponds to the intersection of subspaces V
and W
2. Find null space N(Q):
• From REF(Q), identify free variables
• Express basic variables in terms of free variables
• Write the general solution to Qx = 0
3. Convert null space vectors to intersection vectors:
• If x = (a1 , . . . , ak , b1 , . . . , bℓ ) ∈ N (Q), then:
Qx = a1 v1 + · · · + ak vk + b1 w1 + · · · + bℓ wℓ = 0

4. Extract vectors in V ∩ W:
• For each null space vector x = (a1 , . . . , ak , b1 , . . . , bℓ ), compute:
y = a1 v1 + · · · + ak vk = −(b1 w1 + · · · + bℓ wℓ )

5. Verify intersection property:

• y ∈ V (as linear combination of basis vectors in α)
• y ∈ W (as linear combination of basis vectors in β)
• Therefore, y ∈ V ∩ W

Step 5: Verification

1. Check dimensions:
• dim(V + W ) = number of pivot columns in REF(Q)
• dim(V ∩ W ) = number of free variables in REF(Q)
• Verify: dim(V ) + dim(W ) = dim(V + W ) + dim(V ∩ W )

23
Summary of Connections

• Column Space: C(Q) = V + W → basis from pivot columns of original

Q
• Null Space: N (Q) ↔ V ∩ W → basis from converting null vectors to
intersection vectors

Theorem: Prove that C(Q) = V + W .

Proof: To prove: C(Q) = V + W

Mathematical Tools: A = B ⇐⇒ A ⊆ B (∀x ∈ A =⇒ x ∈ B) and B ⊆ A

Step 1: Every vector in V + W is in C(Q) (i.e., V + W ⊆ C(Q))

Take any arbitrary vector z ∈ V + W . By definition:

z =v+w for some v ∈ V and w ∈ W
• Since v ∈ V , we can write v = a1 v1 + · · · + ak vk (using V ’s basis).
• Since w ∈ W , we can write w = b1 w1 + · · · + bm wm (using W ’s basis).
• Therefore:
z = a1 v1 + · · · + ak vk + b1 w1 + · · · + bm wm
But this is exactly a linear combination of Q’s columns! So z ∈ C(Q).
Therefore, V + W ⊆ C(Q)

Step 2: Every vector in C(Q) is in V + W (i.e., C(Q) ⊆ V + W )

Take any vector z ∈ C(Q). By definition:

z = c1 v1 + · · · + ck vk + d1 w1 + · · · + dm wm
We can group this as:
z = (c1 v1 + · · · + ck vk ) + (d1 w1 + · · · + dm wm )
• The first part is in V (linear combination of V ’s basis).
• The second part is in W (linear combination of W ’s basis).

24
 So z is the sum of a vector from V and a vector from W . Therefore
z ∈ V + W and hence, C(Q) ⊆ V + W .
Therefore, from step 1 & step 2 we get C(Q) = V + W .
Hence, the proof.

Linear Sum of Two Subspaces

Let W1 and W2 be two subspaces of a vector V (F ). Then the linear

sum of W1 and W2 is the set of all those elements each one of which is
expressible as the sum of an element of W1 and an element of W2 . The
linear sum of W1 and W2 can be written as W1 + W2 . That is
W1 + W2 = {α + β : α ∈ W1 , β ∈ W2 }
Theorem The linear sum of two subspaces of a vector space is also a
subspace.

Direct Sum of Vector Subspaces

Let W1 and W2 be two subspaces of a vector space V . Then V is said

to be the direct sum of W1 and W2 if each element of V can be uniquely
expressed as the sum of an element of W1 and an element of W2 . If V
is direct sum of W1 and W2 , then it can be written as V = W1 ⊕ W2 .
In general, if V is the direct sum of W1 , W2 , . . . , Wn , then
V = W1 ⊕ W2 ⊕ . . . ⊕ Wn
Here W1 , W2 , . . . , Wn are called complementary spaces.

Theorem The necessary and sufficient condition for a vector space V

to be the direct sum of two of its subspaces W1 and W2 are:
(i) V = W1 + W2
(ii) W1 ∩ W2 = {0}

25
Suppose W1 and W2 are two subspaces of a vector space R2 . Here are simple examples to explain the linear
sum of two subspaces with and without standard basis vectors.

Example 1: Using Standard Basis Vectors

Let’s define two subspaces of R2 :
W1 = span{(1, 0)} = {t(1, 0) : t ∈ R} = {(t, 0) : t ∈ R}
• This is the x-axis (all points of the form (x, 0))
W2 = span{(0, 1)} = {s(0, 1) : s ∈ R} = {(0, s) : s ∈ R}
• This is the y-axis (all points of the form (0, y))

Linear Sum W1 + W2 :

W1 + W2 = {(t, 0) + (0, s) : t, s ∈ R} (1)

= {(t, s) : t, s ∈ R} (2)
2
=R (3)

Verification:
• Any vector (a, b) ∈ R2 can be written as (a, b) = (a, 0) + (0, b)
• Where (a, 0) ∈ W1 and (0, b) ∈ W2
• So W1 + W2 = R2
Geometric Interpretation: The x-axis plus the y-axis gives us the entire plane.

Example 2: Without Standard Basis Vectors

Let’s define two different subspaces:
W1 = span{(1, 1)} = {t(1, 1) : t ∈ R} = {(t, t) : t ∈ R}
• This is the line y = x (diagonal line through origin)
W2 = span{(1, −1)} = {s(1, −1) : s ∈ R} = {(s, −s) : s ∈ R}
• This is the line y = −x (diagonal line with negative slope)

26
Linear Sum W1 + W2 :

W1 + W2 = {(t, t) + (s, −s) : t, s ∈ R} (4)

= {(t + s, t − s) : t, s ∈ R} (5)

To show this equals R2 :

For any vector (a, b) ∈ R2 , we need to find t, s such that:

t+s=a (6)
t−s=b (7)

Solving: t = a+b
2
and
s = a−b
2
a+b a+b
So (a, b) = 2 , 2 + a−b , − a−b


2 2
Therefore: W1 + W2 = R2

Specific Example:
Let’s find (3, 1) as a sum from W1 and W2 :
• t= 3+1
2
=2
• s= 3−1
2
=1
• Check: (2, 2) + (1, −1) = (3, 1) ✓
• Where (2, 2) ∈ W1 and (1, −1) ∈ W2

Example 3: Linear Sum That’s Not All of R2

W1 = span{(1, 0)} = x-axis
W2 = span{(2, 0)} = x-axis (same as W1 )

Linear Sum W1 + W2 :

W1 + W2 = {(t, 0) + (2s, 0) : t, s ∈ R} (8)

= {(t + 2s, 0) : t, s ∈ R} (9)
= {(u, 0) : u ∈ R} (10)
= x-axis (11)

So W1 + W2 = W1 = W2 (just the x-axis, not all of R2 )

Key Observations
1. Examples 1 & 2: Both give W1 + W2 = R2 because the subspaces span different directions
2. Example 3: Gives W1 + W2 ̸= R2 because both subspaces lie on the same line
3. Dimension:
• Examples 1 & 2: dim(W1 ) = 1, dim(W2 ) = 1, dim(W1 + W2 ) = 2
• Example 3: dim(W1 ) = 1, dim(W2 ) = 1, dim(W1 + W2 ) = 1
The key insight is that the linear sum depends on how the subspaces are oriented relative to each other, not
whether they use standard basis vectors or not.

27
Suppose W1 and W2 are two subspaces of a vector space R2 . Here are simple examples to explain the direct
sum of two subspaces with and without standard basis vectors.

Example 1: Direct Sum Using Standard Basis Vectors

Let’s define two subspaces of R2 :
W1 = span{(1, 0)} = {t(1, 0) : t ∈ R} = {(t, 0) : t ∈ R}
• This is the x-axis
W2 = span{(0, 1)} = {s(0, 1) : s ∈ R} = {(0, s) : s ∈ R}
• This is the y-axis

Check Direct Sum Conditions:

Condition 1: W1 + W2 = R2
• Any vector (a, b) = (a, 0) + (0, b) where (a, 0) ∈ W1 and (0, b) ∈ W2 ✓
Condition 2: W1 ∩ W2 = {(0, 0)}
• W1 ∩ W2 = {(t, 0) : t ∈ R} ∩ {(0, s) : s ∈ R} = {(0, 0)} ✓

Therefore: W1 ⊕ W2 = R2 (direct sum exists)

Unique Representation:
Every vector (a, b) ∈ R2 has a unique representation:

(a, b) = a(1, 0) + b(0, 1) (12)

Example: (3, 2) = 3(1, 0) + 2(0, 1) - this is the only way to write (3, 2) as a sum from W1 and W2 .

Example 2: Direct Sum Without Standard Basis Vectors

Let’s define two different subspaces:
W1 = span{(1, 1)} = {t(1, 1) : t ∈ R} = {(t, t) : t ∈ R}
• This is the line y = x
W2 = span{(1, −1)} = {s(1, −1) : s ∈ R} = {(s, −s) : s ∈ R}
• This is the line y = −x

Check Direct Sum Conditions:

Condition 1: W1 + W2 = R2
• For any (a, b) ∈ R2 , we need t, s such that:
• (a, b) = (t, t) + (s, −s) = (t + s, t − s)
• Solving: t = a+b
2
, s= a−b
2
• These always exist and are unique ✓
Condition 2: W1 ∩ W2 = {(0, 0)}
• W1 ∩ W2 = {(t, t) : t ∈ R} ∩ {(s, −s) : s ∈ R}
• For intersection: (t, t) = (s, −s), so t = s and t = −s
• This gives t = s = 0, so intersection is {(0, 0)} ✓

Therefore: W1 ⊕ W2 = R2 (direct sum exists)

28
Unique Representation:
Every vector (a, b) has a unique representation:

a+b a−b
(a, b) = (1, 1) + (1, −1) (13)
2 2
Example: (5, 1) = 3(1, 1) + 2(1, −1) = (3, 3) + (2, −2) = (5, 1)
• Check: t = 5+1
2
= 3, s = 5−1
2
=2✓

Example 3: NOT a Direct Sum (Counter-example)

W1 = span{(1, 0)} = x-axis
W2 = span{(2, 0)} = x-axis (same line)

Check Conditions:

Condition 1: W1 + W2 = x-axis ̸= R2 ×
Condition 2: W1 ∩ W2 = x-axis ̸= {(0, 0)} ×

Therefore: W1 ⊕ W2 does NOT exist

The vectors don’t have unique representations. For example:

(4, 0) = 4(1, 0) + 0(2, 0) = 0(1, 0) + 2(2, 0) (14)

(multiple representations)

Example 4: Another Non-Standard Basis Direct Sum

W1 = span{(2, 1)} = {t(2, 1) : t ∈ R}
W2 = span{(1, 3)} = {s(1, 3) : s ∈ R}

Check Direct Sum Conditions:

Condition 1: Do these span R2 ?

 
2 1
• Matrix has determinant = 2(3) − 1(1) = 5 ̸= 0
1 3
• So they span R2 ✓
Condition 2: W1 ∩ W2 = {(0, 0)}?
• If t(2, 1) = s(1, 3), then 2t = s and t = 3s
• From first: s = 2t, substituting: t = 3(2t) = 6t
• So t = 0, therefore s = 0
• Intersection is {(0, 0)} ✓

Therefore: W1 ⊕ W2 = R2

Example: Express (7, 10) uniquely:

(7, 10) = t(2, 1) + s(1, 3)
System of equations:

2t + s = 7 (15)
t + 3s = 10 (16)

29
 From second: t = 10 − 3s
Substitute: 2(10 − 3s) + s = 7

20 − 6s + s = 7 (17)
−5s = −13 (18)
 
13 13 11
s= , t = 10 − 3 = (19)
5 5 5

Check: 11 (2, 1) + 13 (1, 3) = 22 , 11 + 13 , 39 35 50

5 5 5 5 5 5
= 5
, 5 = (7, 10) ✓

6 Key Points
1. Direct sum requires TWO conditions:
• W1 + W2 = R2 (spans the whole space)
• W1 ∩ W2 = {0} (intersection is only zero vector)
2. Standard vs. Non-standard basis doesn’t matter - what matters is that the subspaces are
“linearly independent” (don’t overlap except at origin)
3. Unique representation is the key feature - every vector has exactly one way to be written as a
sum from the two subspaces
4. Geometric interpretation: Direct sum means the subspaces point in “different directions” and
together fill the entire space

Problem 1

Find the bases for V + W and V ∩ W

Subspaces of R5 :

v1 = (1, 3, −2, 2, 3),

V with basis: v2 = (1, 4, −3, 4, 2),

v3 = (1, 3, 0, 2, 3)


w1 = (2, 3, −1, −2, 9),

W with basis: w2 = (1, 5, −6, 6, 1),

w3 = (2, 4, 4, 2, 8)


30
Solution

Step 1: Construct Matrix Q

 
1 1 1 2 1 2
 3 4 3 3 5 4
 
Q = [v1 v2 v3 w1 w2 w3 ] = 
−2 −3 0 −1 −6 4

 2 4 2 −2 6 2
3 2 3 9 1 8

Step 2: Find REF(Q)

Row Operations:
Eliminate entries below using
R2 → R2 − 3R1 , R3 → R3 + 2R1 , R4 → R4 − 2R1 , R5 → R5 − 3R1 ,
which gives  
1 1 1 2 1 2
0 1 0 −3 2 −2
 
0 −1 2 3 −4 8  .
 
0 2 0 −6 4 −2
0 −1 0 3 −2 2
Eliminate entries below in column 2 using
R3 → R3 + R2 , R4 → R4 − 2R2 , R5 → R5 + R2 ,
which gives  
1 1 1 2 1 2
0
 1 0 −3 2 −2

0
 0 2 0 −2 6 
.
0 0 0 0 0 2
0 0 0 0 0 0

31
Row Echelon Form (REF)

Thus the REF of Q is

 
1 1 1 2 1 2
0
 1 0 −3 2 −2

0
 0 2 0 −2 6 

0 0 0 0 0 2
0 0 0 0 0 0

Step 3: Find Basis for V+W

Identification of pivot columns: Columns 1, 2, 3, 6 have pivots in REF(Q)

Basis for V + W : Corresponding columns from original Q
• Column 1: v1 = (1, 3, −2, 2, 3)
• Column 2: v2 = (1, 4, −3, 4, 2)
• Column 3: v3 = (1, 3, 0, 2, 3)
• Column 6: w3 = (2, 4, 4, 2, 8)
Therefore: Basis for V + W = {v1 , v2 , v3 , w3 }
dim(V + W ) = 4

Step 4: Find Basis for V ∩ W

In order to solve the homogeneous system

QX = 0.
From the row echelon form (REF) of Q, we have
 
1 1 1 2 1 2
0 1 0 −3 2 −2
 
REF(Q) =  0 0 2 0 −2 6 .

0 0 0 0 0 2 
0 0 0 0 0 0

32
  
x1
x2 
 
x3 
x4  .
Let X =  
 
x5 
x6

Step 1: System of equations

From REF(Q) we obtain:

x1 + x2 + x3 + 2x4 + x5 + 2x6 =0 (1)
x2 − 3x4 + 2x5 − 2x6 =0 (2)
2x3 − 2x5 + 6x6 =0 (3)
2x6 =0 (4)

Step 2: Back substitution

From (4):
x6 = 0
From (3):
x3 = x5
From (2):
x2 = 3x4 − 2x5
From (1):
x1 + (3x4 − 2x5 ) + x5 + 2x4 + x5 = 0
x1 + 5x4 = 0 ⇒ x1 = −5x4

Step 3: General solution with parameters

Let
x4 = t, x5 = s, t, s ∈ R.

33
 Then
−5t
 
3t − 2s
 
 s 
 t .
X= 
 
 s 
0
Equivalently,
−5
   
0
3 −2
   
0
 + s 1 ,
 
X = t
1 0 t, s ∈ R.
   
0 1
0 0

Final Answer

The null space of Q is

 
−5
   

 0 

3


 −2 


 
   
0
 + s 1 
   
N (Q) = t 
1 0 t, s ∈ R .

     



 0 1 



 

0 0
(OR)    

 −5 0 


 3  −2

    
 0   1 
N (Q) = span  ,  .
 1   0 
   
 


 0   1 



 

0 0
Basis vectors for null space:
• t = 1, s = 0: x1 = (−5, 3, 0, 1, 0, 0)
• t = 0, s = 1: x2 = (0, −2, 1, 0, 1, 0)

34
 Convert to vectors in V ∩ W :
For x1 = (−5, 3, 0, 1, 0, 0):
−5
 
3
 
0
 1 =0
Qx1 = 0 =⇒ [v1 v2 v3 w1 w2 w3 ]  
 
0
0

y1 = −5v1 + 3v2 + 0v3 = 1w1 + 0w2 + 0w3

y1 = −5(1, 3, −2, 2, 3) + 3(1, 4, −3, 4, 2)
y1 = (−5, −15, 10, −10, −15) + (3, 12, −9, 12, 6)
y1 = (−2, −3, 1, 2, −9)
For x2 = (0, −2, 1, 0, 1, 0):
 
0
−2
 
1
Qx2 = 0 =⇒ [v1 v2 v3 w1 w2 w3 ] 
 0 =0

 
1
0

y2 = 0v1 − 2v2 + 1v3 = 0w1 + 1w2 + 0w3

y2 = −2(1, 4, −3, 4, 2) + (1, 3, 0, 2, 3)
y2 = (−2, −8, 6, −8, −4) + (1, 3, 0, 2, 3)
y2 = (−1, −5, 6, −6, −1)
Therefore: Basis for V ∩ W = {(−2, −3, 1, 2, −9), (−1, −5, 6, −6, −1)}
dim(V ∩ W ) = 2

Step 5: Verification

Dimension formula check:

dim(V ) + dim(W ) = 3 + 3 = 6
dim(V + W ) + dim(V ∩ W ) = 4 + 2 = 6 ✓

35
Final Answer

• Basis for V +W : {(1, 3, −2, 2, 3), (1, 4, −3, 4, 2), (1, 3, 0, 2, 3), (2, 4, 4, 2, 8)}
• Basis for V ∩ W : {(−2, −3, 1, 2, −9), (−1, −5, 6, −6, −1)}

Problem 2

Given that
• W1 is the subspace generated by: S = {(1, 1, 0, −1), (1, 2, 3, 0), (2, 3, 3, −1)}
• W2 is the subspace generated by: T = {(1, 2, 2, −2), (2, 3, 2, −3), (1, 3, 4, −3)}
Find (i) dim(W1 + W2 ) and (ii) dim(W1 ∩ W2 )

Solution:

Step 1: Find dim(W1 ):

Place vectors from S as columns in matrix A1 :

 
1 1 2
1 2 3
A1 = 
0 3 3


−1 0 −1
Apply row operations to find REF:
R2 → R2 − R1 :  
1 1 2
0 1 1
 
0 3 3
−1 0 −1
R4 → R4 + R1 :  
1 1 2
0 1 1
 
0 3 3
0 1 1

36
 R3 → R3 − 3R2 and R4 → R4 − R2 :
 
1 1 2
0 1 1
 
0 0 0
0 0 0
Pivot columns: 1st and 2nd columns
Column rank = 2
Therefore: dim(W1 ) = 2

Step 2: Find dim(W2 ):

Place vectors from T as columns in matrix A2 :

 
1 2 1
2 3 3
A2 = 
2 2 4


−2 −3 −3
Apply row operations to find REF:
R2 → R2 − 2R1 , R3 → R3 − 2R1 , R4 → R4 + 2R1 :
 
1 2 1
0 −1 1 
 
0 −2 2 
0 1 −1
R2 → −R2 :  
1 2 1
0 1
 −1
0 −2 2
0 1 −1
R3 → R3 + 2R2 and R4 → R4 − R2 :
 
1 2 1
0 1
 −1
0 0 0
0 0 0
Pivot columns: 1st and 2nd columns
Column rank = 2
Therefore: dim(W2 ) = 2

37
Step 3: Find dim(W1 + W2 ):

Combine all vectors from both S and T as columns:

 
1 1 2 1 2 1
1 2 3 2 3 3
A3 = 
0

3 3 2 2 4
−1 0 −1 −2 −3 −3
Apply row operations to find REF:
R2 → R2 − R1 , R4 → R4 + R1 :
 
1 1 2 1 2 1
0 1 1 1 1 2
 
0 3 3 2 2 4
0 1 1 −1 −1 −2
R3 → R3 − 3R2 , R4 → R4 − R2 :
 
1 1 2 1 2 1
0 1 1 1 1 2 
 
0 0 0 −1 −1 −2
0 0 0 −2 −2 −4
R4 → R4 − 2R3 :  
1 1 2 1 2 1
0 1 1 1 1 2
 
0 0 0 −1 −1 −2
0 0 0 0 0 0
R3 → −R3 :  
1 1 2 1 2 1
0 1 1 1 1 2
 
0 0 0 1 1 2
0 0 0 0 0 0
Pivot columns: 1st, 2nd, and 4th columns
Column rank = 3
Therefore: dim(W1 + W2 ) = 3

Step 4: Apply Dimension Formula

Dimension Formula: dim(W1 )+dim(W2 ) = dim(W1 +W2 )+dim(W1 ∩W2 )

38
 Substituting our values:
2 + 2 = 3 + dim(W1 ∩ W2 )
4 = 3 + dim(W1 ∩ W2 )
dim(W1 ∩ W2 ) = 1

Final Answers

(i) dim(W1 + W2 ) = 3
(ii) dim(W1 ∩ W2 ) = 1
Problem 2 Let V and W be the subspaces of the vector space P3 (R)
spanned by

v1 (x)
 = 3 − x + 4x2 + x3 ,
v2 (x) = 5 + 5x2 + x3 ,

v3 (x) = 5 − 5x + 10x2 + 3x3 ,


and

2 3
w1 (x) = 9 − 3x + 3x + 2x ,

w2 (x) = 5 − x + 2x2 + x3 ,

w3 (x) = 6 + 4x2 + x3 ,


respectively. Find the dimensions and bases for V + W and V ∩ W .

Solution:

Finding Dimensions and Bases for V + W and V ∩ W

Use row echelon form to find the dimensions and bases for V + W and
V ∩ W.

Given Information

• V is spanned by: v1 (x) = 3 − x + 4x2 + x3 , v2 (x) = 5 + 5x2 + x3 ,

v3 (x) = 5 − 5x + 10x2 + 3x3
• W is spanned by: w1 (x) = 9 − 3x + 3x2 + 2x3 , w2 (x) = 5 − x + 2x2 + x3 ,
w3 (x) = 6 + 4x2 + x3

39
Step 1: Find V + W

V + W is spanned by {v1 , v2 , v3 , w1 , w2 , w3 }. We need to find which vectors

are linearly independent.
First, let’s represent each polynomial as a vector in the standard basis
{1, x, x2 , x3 }:
Matrix for V + W :
 
v1 = 3 −1 4 1 (20)
 
v2 = 5 0 5 1 (21)
 
v3 = 5 −5 10 3 (22)
 
w1 = 9 −3 3 2 (23)
 
w2 = 5 −1 2 1 (24)
 
w3 = 6 0 4 1 (25)
Setting up the matrix A:
−1
 
3 4 1
5 0 5 1
 
5 −5 10 3
A= 
9
 −3 3 2

5 −1 2 1
6 0 4 1
Row reduction to row echelon form:
Initial matrix:  
R1 : 3 −1 4 1
 
R2 : 5 0 5 1
 
R3 : 5 −5 10 3
 
R4 : 9 −3 3 2
 
R5 : 5 −1 2 1
 
R6 : 6 0 4 1
R2 − 35 R1 → R2 :
5
− 53 − 32
 
0 3
R3 − 53 R1 → R3 :
0 − 10 10 4
 
3 3 3

40
 R4 − 3R1 → R4 :  
0 0 −9 −1
R5 − 35 R1 → R5 :
0 − 32 − 14 2
 
3 − 3
R6 − 2R1 → R6 :  
0 2 −4 −1
Continuing the reduction:
3 −1 4 1
 
0 5 − 5 − 2 
 3 3 3
0 0 0 2 
 
0 0 −9 −1
 
0 0 −6 0 
0 0 −6 1
After complete row reduction:
−1
 
3 4 1
5
0
 3 − 53 − 23 

0
 0 −9 −1 
0 0 0 2 
 
0 0 0 0 
0 0 0 0
Result for V + W :
• Dimension of V + W = 4 (number of pivot columns)
n o
5x2 2x3
• Basis: 3 − x + 4x + x , 3 − 3 − 3 , −9x − x , 2x
2 3 5x 2 3 3

Using Dimension Formula

Step 1: Find dimensions of V and W separately

For V :  
3 −1 4 1
5 0 5 1
5 −5 10 3

41
 Row reducing:  
3 −1 4 1
0 5 − 5 − 2 
3 3 3
0 0 0 2
Dimension of V = 3
For W :  
9 −3 3 2
5 −1 2 1
6 0 4 1
Row reducing:  
9 −3 3 2
0 2 7 − 1 
9 3 9
0 0 83 13
Dimension of W = 3

Step 2: Use the dimension formula

dim(V + W ) = dim(V ) + dim(W ) − dim(V ∩ W )

From Step 1 of Method 1: dim(V +W ) = 4 Therefore: 4 = 3+3−dim(V ∩
W ) So: dim(V ∩ W ) = 2

Problem 2

Let
V = {(x, y, z, u) ∈ R4 : y + z + u = 0}, (26)
W = {(x, y, z, u) ∈ R4 : x + y = 0, z = 2u} (27)
be two subspaces of R4 . Find bases for V , W , V + W , and V ∩ W .

Solution:
Given Information

V = {(x, y, z, u) ∈ R4 : y + z + u = 0} (28)
W = {(x, y, z, u) ∈ R4 : x + y = 0, z = 2u} (29)

42
Step 1: Find a Basis for V

The subspace V is defined by the constraint y + z + u = 0, which means

y = −z − u.
We can parameterize V using free variables x, z, and u:
V = {(x, −z − u, z, u) : x, z, u ∈ R}
This can be written as:
V = x(1, 0, 0, 0) + z(0, −1, 1, 0) + u(0, −1, 0, 1)
Therefore, a basis for V is:
BV = {(1, 0, 0, 0), (0, −1, 1, 0), (0, −1, 0, 1)}
Verification using row echelon form: The constraint matrix for V
is [0 1 1 1], which is already in row echelon form. The null space has
dimension 4 − 1 = 3, confirming our basis has 3 vectors.

Step 2: Find a Basis for W

The subspace W is defined by two constraints:

x + y = 0 ⇒ y = −x (30)
z = 2u (31)
We can parameterize W using free variables x and u:
W = {(x, −x, 2u, u) : x, u ∈ R}
This can be written as:
W = x(1, −1, 0, 0) + u(0, 0, 2, 1)
Therefore, a basis for W is:
BW = {(1, −1, 0, 0), (0, 0, 2, 1)}
Verification using row echelon form: The constraint matrix for W is:
 
1 1 0 0
0 0 1 −2
This is already in row echelon form. The null space has dimension 4−2 = 2,
confirming our basis has 2 vectors.

43
Step 3: Construct Matrix Q and Find Basis for V + W

Matrix Q with columns as basis vectors of V followed by W :

 
1 0 0 1 0
0 −1 −1 −1 0
Q= 0 1 0

0 2
0 0 1 0 1
To find the basis of V + W , we need to find the column space of Q by row
reducing:
 
1 0 0 1 0
0 −1 −1 −1 0
 
 0 1 0 0 2
0 0 1 0 1
Row reducing:  
1 0 0 1 0
0 1 1 1 0
R2 ← −R2 : 0 1 0 0 2


0 0 1 0 1
 
1 0 0 1 0
0 1 1 1 0
R3 ← R3 − R2 : 
0 0 −1 −1 2


0 0 1 0 1
 
1 0 0 1 0
0 1 1 1 0 
R3 ← −R3 : 0 0 1 1 −2


0 0 1 0 1
 
1 0 0 1 0
0 1 1 1 0 
R4 ← R4 − R3 : 
0 0 1 1 −2


0 0 0 −1 3
The pivot columns are 1, 2, 3, and 4, so the basis for V + W consists of
the first 4 columns of Q:

BV +W = {(1, 0, 0, 0), (0, −1, 1, 0), (0, −1, 0, 1), (1, −1, 0, 0)}

44
Step 4: Find Basis for V ∩ W

To find V ∩ W , we need vectors that satisfy both sets of constraints:

y + z + u = 0 (from V ) (32)
x + y = 0 and z = 2u (from W ) (33)
Setting up the system:
 
  x  
0 1 1 1   0
1 1 0 0  y
z  = 0
   
0 0 1 −2 0
u
Row reducing this augmented matrix:
 
1 1 0 0 | 0
0 1 1 1 | 0
0 0 1 −2 | 0
From the third row: z = 2u
From the second row: y + z + u = 0 ⇒ y = −z − u = −2u − u = −3u
From the first row: x + y = 0 ⇒ x = −y = 3u
So the general solution is (3u, −3u, 2u, u) where u is free.
Therefore, a basis for V ∩ W is:
BV ∩W = {(3, −3, 2, 1)}

Summary of Results

• Basis for V : {(1, 0, 0, 0), (0, −1, 1, 0), (0, −1, 0, 1)} (dimension 3)
• Basis for W : {(1, −1, 0, 0), (0, 0, 2, 1)} (dimension 2)
• Basis for V + W : {(1, 0, 0, 0), (0, −1, 1, 0), (0, −1, 0, 1), (1, −1, 0, 0)} (di-
mension 4)
• Basis for V ∩ W : {(3, −3, 2, 1)} (dimension 1)
Verification: dim(V ) + dim(W ) = 3 + 2 = 5 = 4 + 1 = dim(V + W ) +
dim(V ∩ W ) ✓

45
 Invertibility

Standard Invertibility (Square Matrices Only)

True invertibility in the classical sense only applies to square ma-
trices. A matrix A is invertible if there exists a matrix A−1 such that:
• AA−1 = I
• A−1 A = I
This requires both products to equal the identity matrix, which is only
possible when A is square.

CASE I - Rectangular (non-square) matrix: If m ̸= n, then

(∀ Am×n (∃! Bn×m (Am×n · Bn×m = Im×m ))) (Bn×m - Right Inverse of A)
(∀ Am×n (∃! Bn×m (Bn×m · Am×n = In×n ))) (Bn×m - Left Inverse of A)
CASE II - Square matrix: If m = n, then
(∀ An×n (∃! Bn×n (An×n · Bn×n = In×n = Bn×n · An×n )))

Note: Right inverse & left inverse both are same in this case.

Examples of Matrix Invertibility

CASE I - Rectangular (Non-square) Matrices
Example 1: 2 × 3 Matrix
 
1 0 1
Let A =
0 1 0 2×3
Right Inverse: Find B3×2 such that A · B = I2×2
 
1 0
B = 0 1
0 0 3×2
 
  1 0  
1 0 1  1 0
Verification: A · B = 0 1 = = I2×2
0 1 0 0 1
0 0

47
 Left Inverse: Find C3×2 such that C · A = I3×3
No left inverse exists because C3×2 · A2×3 cannot equal I3×3

Example 2: 3 × 2 Matrix
 
1 0
Let A = 0 1
1 0 3×2
Left Inverse: Find B2×3 such that B · A = I2×2
 
1 0 0
B=
0 1 0 2×3
 
  1 0  
1 0 0  1 0
Verification: B · A = 0 1 = = I2×2
0 1 0 0 1
1 0
Right Inverse: Find C2×3 such that A · C = I3×3
No right inverse exists because A3×2 · C2×3 cannot equal I3×3

CASE II - Square Matrices

Example 1: 2 × 2 Matrix
 
2 1
Let A =
1 1 2×2
Find A such that A · A−1 = A−1 · A = I2×2
−1

Using the formula A−1 = det(A)

1
adj(A):

det(A) = 2(1) − 1(1) = 1

 
1 −1
adj(A) =
−1 2
   
1 1 −1 1 −1
A−1 = =
1 −1 2 −1 2
Verification:
    
2 1 1 −1 1 0
A · A−1 = = = I2×2
1 1 −1 2 0 1

48
Example 2: 3 × 3 Matrix
 
1 0 0
Let A = 0 2 0 (diagonal matrix)
0 0 3 3×3
For a diagonal matrix, the inverse is:
1   
1 0 0 1 0 0
−1
A = 0 21 0 = 0 0.5 0

0 0 13 0 0 13
Verification:
    
1 0 0 1 0 0 1 0 0
A · A−1 = 0 2 0 0 0.5 0 = 0 1 0 = I3×3
0 0 3 0 0 13 0 0 1

Key Observations

• CASE I (Rectangular): Only one-sided inverses exist

– m < n (wide matrix): Only right inverse exists
– m > n (tall matrix): Only left inverse exists
• CASE II (Square): Both left and right inverses exist and are identical
– A−1 exists if and only if det(A) ̸= 0
– A · A−1 = A−1 · A = In

Existence and Uniqueness Theorem

The following are the Existence and Uniqueness Theorems for a solution of a
system of linear equations Ax = b for an m×n matrix A and a vector b ∈ Rm .

Theorem 1 (Existence) Let A be an m × n matrix. Then the following

statements are equivalent.
1. For each b ∈ Rm , Ax = b has at least one solution x in Rn .
2. The column vectors of A span Rm , i.e., C(A) = Rm .
3. rank A = m, and hence m ≤ n.

49
 4. There exists an n × m right inverse B of A such that AB = Im .
Theorem 2 (Uniqueness) Let A be an m×n matrix. Then the following
statements are equivalent.
1. For each b ∈ Rm , Ax = b has at most one solution x in Rn .
2. The column vectors of A are linearly independent.
3. dimC(A) = rank A = n, and hence n ≤ m.
4. R(A) = Rn .
5. N (A) = {0}.
6. There exists an n × m left inverse C of A such that CA = In .
Remark: (1) Note that an m × n matrix A has a right inverse if and only
if rank A = m, and A has a left inverse if and only if rank A = n. In the
first theorem Ax = b always has a solution, and in the second theorem the
solution (if it exists) is unique. Therefore, if m ̸= n, A cannot have both left
and right inverses.

(2) Note that if m = n so that A is a square matrix, then A has a right

inverse (and a left inverse) if and only if rank A = m = n. Moreover, in this
case the inverses are the same. Therefore, a square matrix A has rank n if
and only if A is invertible. This means that for a square matrix ”Existence
= Uniqueness”. In particular, for the invertibility of a square matrix it is
enough to show the existence of a one-side inverse.

Theorem 3 For a square matrix A of order n, the following statements

are equivalent.
1. A is invertible.
2. det A ̸= 0.
3. A is row equivalent to In .
4. A is a product of elementary matrices.
5. Elimination can be completed: P A = LDU , with all di ̸= 0.
6. Ax = b has a solution for every b ∈ Rn .

50
 7. Ax = 0 has only a trivial solution, i.e., N (A) = {0}.
8. The columns of A are linearly independent.
9. The columns of A span Rn , i.e., C(A) = Rn .
10. A has a left inverse.
11. rank A = n.
12. The rows of A are linearly independent.
13. The rows of A span Rn , i.e., R(A) = Rn .
14. A has a right inverse.

51
 INTERPOLATION
Definition: Interpolating Polynomial

A polynomial which approximates the precise functional relationship between

input & output values is called an interpolating polynomial.

Digressions:

Consider (n + 1) distinct input values x0 , x1 , x2 , . . . , xn & the corresponding

output values y0 , y1 , y2 , . . . , yn

(i.e.) f (x0 ) = y0 , f (x1 ) = y1 , f (x2 ) = y2 , . . ., f (xn ) = yn

Our aim is to construct a polynomial P (x) of degree less than or equal to

n which interpolates f (x) at x0 , x1 , x2 , . . . , xn

(i.e.) P (xi ) = yi = f (xi ), i = 0, 1, . . . , n.

Notice that, if there is such a polynomial, it must be unique.

P (x) = a0 + a1 x + a2 x2 + · · · + an xn
P (xi ) = a0 + a1 xi + a2 x2i + · · · + an xni
= yi = f (xi ), i = 0, 1, . . . , n.

    
1 x0 x20 · · · xn0 a0 y0
1 x x2 · · · xn1   a1   y1 
   
1
⇒  .. .. ..1  ..  =  .. 
 
. . . . . . ...  .   . 
1 xn x2n · · · xn n
an yn

Ax = B
The coefficient matrix A is called as Vandermonde’s matrix.
Y
|A| = det(A) = (xj − xi )
0≤i<j≤n

48
 Since xi ’s are all distinct, |A| = ̸ 0 which implies that Ax = B has a unique
solution.
∴ It is the unique polynomial P (x) of degree ≤ n passing through the
(n + 1) points, (x0 , y0 ), (x1 , y1 ), . . . , (xn , yn ) in R2 .

Example 1

Given four points (0, 3), (1, 0), (−1, 2), (3, 6) in R2 . Construct a polynomial
P (x) which passes through the given points?

Solution:

let P (x) = a0 + a1 x + a2 x2 + a3 x3 be the polynomial passing through the

given points.
We know that P (xi ) = yi , i = 0, 1, 2, 3

P (x0 ) = y0 ⇒ P (0) = a0 = 3 (1)

P (x1 ) = y1 ⇒ P (1) = a0 + a1 + a2 + a3 = 0 (2)
P (x2 ) = y2 ⇒ P (−1) = a0 − a1 + a2 − a3 = 2 (3)
P (x3 ) = y3 ⇒ P (3) = a0 + 3a1 + 9a2 + 27a3 = 6 (4)
Given system:
Step 1: From equation (1), we know a0 = 3. Substituting into equations
(2), (3), and (4):

3 + a1 + a2 + a3 = 0 ⇒ a1 + a2 + a3 = −3
3 − a1 + a2 − a3 = 2 ⇒ −a1 + a2 − a3 = −1
3 + 3a1 + 9a2 + 27a3 = 6 ⇒ a1 + 3a2 + 9a3 = 1
Step 2: Set up the augmented matrix for the 3×3 system:
 
1 1 1 | −3
−1 1 −1 | −1
1 3 9 | 1
Step 3: Apply row operations to get REF:

49
 R2 = R2 + R1 and R3 = R3 − R1 :
 
1 1 1 | −3
0 2 0 | −4
0 2 8 | 4
R3 = R3 − R2 :  
1 1 1 | −3
0 2 0 | −4
0 0 8 | 8
Step 4: Backward substitution:
From row 3: 8a3 = 8 ⇒ a3 = 1
From row 2: 2a2 = −4 ⇒ a2 = −2
From row 1: a1 + a2 + a3 = −3 ⇒ a1 + (−2) + 1 = −3 ⇒ a1 = −2
Final Solution is:
a0 = 3, a1 = −2, a2 = −2, a3 = 1
Verification:
P (0) = 3 = 3 ✓
P (1) = 3 + (−2) + (−2) + 1 = 0 ✓
P (−1) = 3 − (−2) + (−2) − 1 = 2 ✓
P (3) = 3 + 3(−2) + 9(−2) + 27(1) = 6 ✓
Therefore, the interpolating polynomial is:

P (x) = 3 − 2x − 2x2 + x3

Exercise

1. Let f (x) = sin x, then at x = √0, π4 , π3 , 3π

4 ,π
1 3 1
the values of f are y = 0, √2 , 2 , √2 , 0.
Find the polynomial P (x) which passes through the given points.
2. Find a polynomial P (x) = a + bx + cx2 + dx3 which
satisfies P (0) = 1, P ′ (0) = 2, P (1) = 4 & P ′ (1) = 4.
3. Find the equation of a circle which passes through
the points (2, −2), (3, 5) & (−4, 6).

50
Hints for Matrix Formation
Problem 1: Sine Function Interpolation

Hint: You have 5 data points, so you need a polynomial of degree ≤ 4.

Let P (x) = a0 + a1 x + a2 x2 + a3 x3 + a4 x4
The 5 points are:
  √ !  
π 1 π 3 3π 1
(0, 0), ,√ , , , , √ , (π, 0)
4 2 3 2 4 2
Set up the Vandermonde matrix:
    
1 0 0 0 0 a0 0
1 π π 2 π 3 π 4  1 




4
 a1  √

 4 4
4
√2 
 π π 2 π 3 π 4  a  =  3 
1 3 3
3
3
  2
 2
 3π 3π 2 3π 3 4
3π  a3
1
1    √2 
4 4 4 4
1 π π2 π3 π4 a4 0

Problem 2: Polynomial with Derivative Conditions

Hint: You have conditions on both P (x) and P ′ (x).

Given: P (x) = a + bx + cx2 + dx3
Then: P ′ (x) = b + 2cx + 3dx2
The conditions are:
P (0) = 1 ⇒ a = 1
P ′ (0) = 2 ⇒ b = 2
P (1) = 4 ⇒ a + b + c + d = 4
P ′ (1) = 4 ⇒ b + 2c + 3d = 4
Set up the matrix (focusing on the last two equations since first two give
direct values):  
  a  
1 1 1 1   = 4
b 
0 1 2 3 c 4
d

51
Problem 3: Circle Through Three Points

Hint: General equation of a circle is x2 + y 2 + Dx + Ey + F = 0.

The three points (2, −2), (3, 5), (−4, 6) must satisfy this equation.
Substituting each point:
Point (2, −2) : 4 + 4 + 2D − 2E + F = 0 ⇒ 2D − 2E + F = −8
Point (3, 5) : 9 + 25 + 3D + 5E + F = 0 ⇒ 3D + 5E + F = −34
Point (−4, 6) : 16 + 36 − 4D + 6E + F = 0 ⇒ −4D + 6E + F = −52
Set up the matrix:
    
2 −2 1 D −8
 3 5 1 E  = −34
−4 6 1 F −52

52
 Linear Transformation
Syllabus
Linear transformations - Basic properties - Invertible linear transformation - Matrices of linear transfor-
mations - Vector space of linear transformations.

Definition: Linear Transformation

Let V and W be two vector spaces over the same field F. A function (mapping/transformation)

T : V (F) → W (F)

is called a linear transformation/map from V to W over the same field F, if the following conditions
hold:

(i) (∀ x, y ∈ V (T (x + y) = T (x) + T (y)))

(ii) (∀ x ∈ V, ∀ α ∈ F (T (αx) = αT (x)))

(OR)
(∀ x, y ∈ V, ∀ α, β ∈ F (T (αx + βy) = αT (x) + βT (y)))
The following diagram illustrates the vector addition and scalar multiplication property of linear
transformations:

x, y T
T (x), T (y)

+ +

x+y T (x) + T (y)

T ∥
T (x + y)

T
x T (x)

α∈F α∈F

αx αT (x)
T ∥
T (αx)

Note:
In particular:
A transformation (or function or mapping) T from Rn to Rm is a rule that assigns to each vector x in
Rn a vector T (x) in Rm . The set Rn is called the domain of T , and Rm is called the codomain of T .
The notation T : Rn → Rm indicates that the domain of T is Rn and the codomain is Rm . For x in Rn ,
the vector T (x) in Rm is called the image of x (under the action of T ). The set of all images T (x) is
called the range of T (See: Figure 1).

1
 Figure 1: Domain, codomain, and range of T : Rn → Rm

Theorem
If T : V → W is a linear transformation, then
(a) T (0) = 0.
(b) T (x − y) = T (x) − T (y) for all x and y in V.

(c) T (−x) = −T (x) for all x in V.

(d) For any x1 , x2 , . . . , xn ∈ V and scalars k1 , k2 , . . . , kn , T (k1 x1 + k2 x2 + · · · + kn xn ) = k1 T (x1 ) +
k2 T (x2 ) + · · · + kn T (xn ).

Note:
The identity transformation on a vector space V is the linear transformation

I:V →V

defined by
I(x) = x, for all x ∈ V.
The zero transformation on a vector space V is the linear transformation

Z:V →W

defined by
Z(x) = 0 (0 - zero vector in W), for all x ∈ V.

Examples of Linear Transformation

Problem 1
Let V = W = R and T : V → W be defined by

T (x) = 2x, ∀x.

Verify that T is a Linear Transformation.

2
Solution:
(i) ∀ x, y ∈ V ,
T (x + y) = 2(x + y) = 2x + 2y = T (x) + T (y).
∴ T (x + y) = T (x) + T (y).

(ii) ∀ x ∈ V, ∀ α ∈ F ,
T (αx) = 2(αx) = α(2x) = αT (x).
∴ T (αx) = αT (x).

Hence, T is a Linear Transformation.

Problem 2
Let T : R → R be defined by
T (x) = x2 − x.
Determine whether T is a linear transformation.

Solution. A map T is linear iff for all x, y ∈ R and all scalars α ∈ R we have

T (x + y) = T (x) + T (y) and T (αx) = αT (x).

(1) Additivity:
T (x + y) = (x + y)2 − (x + y)
= x2 + 2xy + y 2 − x − y
T (x) + T (y) = (x2 − x) + (y 2 − y).
Thus T (x + y) = T (x) + T (y) would require 2xy = 0 for all x, y, which is false.
For a concrete counterexample take x = 1, y = 1:

T (2) = 4 − 2 = 2, T (1) + T (1) = 0 + 0 = 0,

so T (2) ̸= T (1) + T (1). Hence additivity fails.

(2) Homogeneity:
T (αx) = (αx)2 − (αx) = α2 x2 − αx,
αT (x) = α(x2 − x) = αx2 − αx.
Therefore
T (αx) − αT (x) = (α2 − α)x2 ,
which is not zero for general α, x.
For example, α = 2, x = 1 gives

T (2) = 4 − 2 = 2, 2T (1) = 2 · 0 = 0,

so T (2) ̸= 2T (1). Homogeneity fails.

Therefore, T (x) = x2 − x is not a linear transformation.

Problem 3
Given: S : R2 → R2 defined by S(x, y) = (x − y, 2x). Check whether S is a linear transformation.

3
Solution
Additivity
S(u + v) = S(u) + S(v) for all u, v ∈ R2
Proof. Let u = (u1 , u2 ) and v = (v1 , v2 ) be arbitrary vectors in R2 .
Left side: S(u + v)

u + v = (u1 + v1 , u2 + v2 )
S(u + v) = S((u1 + v1 , u2 + v2 ))
= ((u1 + v1 ) − (u2 + v2 ), 2(u1 + v1 ))
= (u1 + v1 − u2 − v2 , 2u1 + 2v1 )
= (u1 − u2 + v1 − v2 , 2u1 + 2v1 )

Right side: S(u) + S(v)

S(u) = S((u1 , u2 )) = (u1 − u2 , 2u1 )

S(v) = S((v1 , v2 )) = (v1 − v2 , 2v1 )
S(u) + S(v) = (u1 − u2 , 2u1 ) + (v1 − v2 , 2v1 )
= (u1 − u2 + v1 − v2 , 2u1 + 2v1 )

Since S(u + v) = S(u) + S(v), is satisfied ✓

Scalar Multiplication (Homogeneity)

S(cu) = cS(u) for all c ∈ R and u ∈ R2

Proof. Let u = (u1 , u2 ) be an arbitrary vector in R2 and c ∈ R be an arbitrary scalar.

Left side: S(cu)

cu = c(u1 , u2 ) = (cu1 , cu2 )

S(cu) = S((cu1 , cu2 ))
= (cu1 − cu2 , 2cu1 )

Right side: cS(u)

S(u) = S((u1 , u2 )) = (u1 − u2 , 2u1 )

cS(u) = c(u1 − u2 , 2u1 )
= (c(u1 − u2 ), c(2u1 ))
= (cu1 − cu2 , 2cu1 )

Since S(cu) = cS(u), is satisfied ✓

Conclusion
Since both axioms are satisfied:
1. Additivity: S(u + v) = S(u) + S(v) ✓
2. Scalar multiplication: S(cu) = cS(u) ✓
Therefore, S : R2 → R2 defined by S(x, y) = (x − y, 2x) IS a linear transformation.

Problem 4
Given: k : R2 → R2 defined by k(x, y) = (xy, x2 + 1). Check whether k is a linear transformation by
verifying both axioms.

4
Solution:
Additivity
k(u + v) = k(u) + k(v) for all u, v ∈ R2
Given u = (u1 , u2 ) and v = (v1 , v2 ):
Left side:
k(u + v) = k((u1 + v1 , u2 + v2 ))
= ((u1 + v1 )(u2 + v2 ), (u1 + v1 )2 + 1)
= (u1 u2 + u1 v2 + v1 u2 + v1 v2 , u21 + 2u1 v1 + v12 + 1)
Right side:
k(u) + k(v) = (u1 u2 , u21 + 1) + (v1 v2 , v12 + 1)
= (u1 u2 + v1 v2 , u21 + v12 + 2)
Comparison:
• First component: u1 u2 + u1 v2 + v1 u2 + v1 v2 ̸= u1 u2 + v1 v2 (unless u1 v2 + v1 u2 = 0)
• Second component: u21 + 2u1 v1 + v12 + 1 ̸= u21 + v12 + 2 (unless u1 v1 = 12 )

Testing with Specific Vectors (Counter Example)

Let u = (1, 2) and v = (3, 4).
Left side: k(u + v)
u + v = (1, 2) + (3, 4) = (4, 6)
k(u + v) = k(4, 6) = (4 · 6, 42 + 1) = (24, 17)
Right side: k(u) + k(v)
k(u) = k(1, 2) = (1 · 2, 12 + 1) = (2, 2)
k(v) = k(3, 4) = (3 · 4, 32 + 1) = (12, 10)
k(u) + k(v) = (2, 2) + (12, 10) = (14, 12)
Result: k(u + v) = (24, 17) ̸= (14, 12) = k(u) + k(v)
Axiom 1 FAILS ×
With u = (1, 2) and v = (3, 4):
• k(u + v) = (24, 17)
• k(u) + k(v) = (14, 12)
• Since (24, 17) ̸= (14, 12), additivity fails

Scalar Multiplication (Homogeneity)

k(cu) = ck(u) for all c ∈ R and u ∈ R2
For a scalar c and vector u = (u1 , u2 ):
Left side:
k(cu) = k((cu1 , cu2 ))
= (cu1 · cu2 , (cu1 )2 + 1)
= (c2 u1 u2 , c2 u21 + 1)
Right side:
ck(u) = c(u1 u2 , u21 + 1)
= (cu1 u2 , cu21 + c)
Comparison:
• First component: c2 u1 u2 ̸= cu1 u2 (unless c = 1 or u1 u2 = 0)
• Second component: c2 u21 + 1 ̸= cu21 + c (unless specific conditions are met)

5
Testing with Specific Values (Counter Example)
Let c = 2 and u = (1, 3).
Left side: k(cu)

cu = 2(1, 3) = (2, 6)
k(cu) = k(2, 6) = (2 · 6, 22 + 1) = (12, 5)

Right side: ck(u)

k(u) = k(1, 3) = (1 · 3, 12 + 1) = (3, 2)

ck(u) = 2(3, 2) = (6, 4)

Result: k(cu) = (12, 5) ̸= (6, 4) = ck(u)

Axiom 2 FAILS ×
With c = 2 and u = (1, 3):

• k(cu) = (12, 5)
• ck(u) = (6, 4)
• Since (12, 5) ̸= (6, 4), homogeneity fails

Conclusion
k : R2 → R2 defined by k(x, y) = (xy, x2 + 1) is NOT a linear transformation.

Exercise Problems
1. Let T : R2 → R2 defined by T (x, y) = (x + y, 2x). Is this linear or not?
2. Let T : R2 → R2 defined by T (x, y) = (x − y, x + y). Is this linear or not?
3. Let T : R2 → R2 defined by T (x, y) = (x2 + 1, 2y). Is this linear or not?

4. Let T : R2 → R2 defined by T (x, y) = (x + y, 2xy). Is this linear or not?

5. Let T : R3 → R3 defined by T (x, y, z) = (x + y + z, 2x + 3y, y − 2z). Is this linear or not?
6. Let T : R2 → R2 defined by T (x, y) = (x2 + 2y, y 3 + 2). Is this linear or not?

6
 Kernel/Image of Linear Transformation
Kernel/Null space:
Let T : V → W be a linear transformation. The kernel of T , denoted by ker(T ), is defined as:

N (T ) = ker(T ) = {v ∈ V | T (v) = 0 (OR) 0W } (⊆) Subspace of V

where 0W is the zero vector in W .

Image/Range/Graph:
Let T : V → W be a linear transformation. The range (or image) of T , denoted by range(T ) or Im(T ),
is defined as:

R(T ) = range(T ) = {w ∈ W | ∃ v ∈ V such that T (v) = w} (⊆) Subspace of W

(OR)
Im(T ) = image(T ) = {T (v) | v ∈ V }

Figure 2: Kernel and range of T : V → W

Key Note:
Let T : V → W be a linear transformation where V, W - finite dimensional vector space (FDVS) over
the same field F , then dim(ker(T )) = dim(N (T )) = nullity(T ) and dim(range(T )) = dim(R(T )) =
dim(C(T )) = rank(T ). The dimension formula is

dim(V ) = nullity + rank = dim(N (T )) + dim(R(T ))

Theorem
Let T : V → W be a linear transformation. Then:
(i) ker(T ) is a subspace of V
(ii) Im(T ) is a subspace of W

7
Proof
Claim: ∀u, v ∈ ker(T ), α ∈ F
1. u + v ∈ ker(T )
2. αu ∈ ker(T )

Note: T : V → W linear
(i) T (u + v) = T (u) + T (v), ∀u, v ∈ V
(ii) T (αu) = αT (u) ∀u ∈ V, ∀α ∈ F

Part (i)(1): Let u, v ∈ ker(T ) =⇒ {u ∈ V | T (u) = 0} and {v ∈ V | T (v) = 0}. Then: (i.e.,)

T (u) = 0 and T (v) = 0

To prove: u + v ∈ ker(T )
We need to show: T (u + v) = 0
Since T is linear:

T (u + v) = T (u) + T (v)
=0+0
=0

Therefore, u + v ∈ ker(T ).

(2): Let u ∈ ker(T ), α ∈ F , then T (u) = 0.

To prove: αu ∈ ker(T )
We need to show: T (αu) = 0.

Since T is linear, T (αu) = αT (u), ∀u ∈ V

= α(0)
=0

∴ αu ∈ ker(T ).
Thus ker(T ) is a subspace of V .

Part (ii)(a): Let x, y ∈ Im(T ). Then ∃u, v ∈ V such that

x = T (u) and y = T (v).

To prove: x + y ∈ Im(T )
Since T is linear,

T (u + v) = T (u) + T (v)
=x+y

∴ x + y ∈ Im(T ).

(b) Let x ∈ Im(T ) and α ∈ F . Then ∃u ∈ V such that T (u) = x.

Using the property of linearity of T , we get

T (αu) = αT (u) = αx.

⇒ αx ∈ Im(T )
∴ Im(T ) is a subspace of W .

Basic Properties of Linear Transformations

T : V (F ) → W (F )

8
Properties of Kernel (Null Space)
• ker(T ) is always a subspace of the domain V

• 0 ∈ ker(T )
• If A is the matrix representation of T , then ker(T ) = null space of A

Properties of Range (Image/Column Space)

• Range(T ) is always a subspace of the codomain W
• If A is the matrix representation, Range(T ) = column space of A
• T is onto (surjective) if and only if Range(T ) = W

Matrix Connections
For an m × n matrix A representing linear transformation T :

Row Rank equals Column Rank

For any matrix A:
Row rank(A) = Column rank(A) = rank(A)
where:
• Row rank = dimension of row space
• Column rank = dimension of column space

REF and RREF Properties

• Number of pivot columns in REF/RREF = rank(A)
• Number of free variables = nullity(A) = n − rank(A)

• RREF directly shows basis for null space and column space

9
Function (Many-to-One)
”The Grade Assigner”

• Input: Students in your class

• Output: Letter grades (A, B, C, D, F)
Why it works: Multiple students can get the same grade, but each student gets exactly one grade.

One-to-One Function (Injective)

”The Register Number System”
• Input: Students in your college

• Output: Register numbers (like 24BCS0001, 24BCS0002, 24BCA0001...)

Why it works: Each student gets a unique register number - no duplicates allowed! If two students
had the same register number, the college database would crash during exam results.

Onto Function (Surjective)

”The Linear Algebra Faculty Assignment”
• Input: Students enrolled in Linear Algebra course
• Output: Faculty members teaching Linear Algebra (Prof. X, Dr. Y, Prof. Z)

Why it’s onto: Every faculty member who’s assigned to teach Linear Algebra gets at least some
students in their section. The department ensures no Linear Algebra professor is left without
students

Bijective Function (One-to-One AND Onto)

”The Perfect Seat Assignment”
• Input: Students in your exam hall
• Output: Seats in your exam hall

Why it’s bijective: Each student gets exactly one seat (one-to-one), and every seat in the room is
filled (onto). No student is left standing, no seat is empty.

10
Injectivity, Surjectivity, and Bijection
One-to-One (Injective) function
Let f : X → Y be a function. f is said to be one-to-one (injective) if
(∀x, y ∈ X(f (x) = f (y) in Y =⇒ x = y in X)).

Example 1
The following figure shows that the mapping T : Rn → Rm is not one-to-one and T is one-to-one.

Figure 3: Describes whether T : Rn → Rm is one-to-one function or not

Example 2
Test whether the function T (x) = x3 is injective.

Solution:
Suppose T (x) = T (y), that is,
x3 = y 3
Taking the cube root of both sides, √ p
3 3
x3 = y3
x=y
Therefore, T is one-to-one.

Example 3
Test whether the function T : R → R defined by T (x) = x2 is injective.

Solution:
Suppose T (x) = T (y), that is,
T (x) = T (y)
=⇒ x2 = y 2
=⇒ x2 − y 2 = 0
=⇒ (x − y)(x + y) = 0
=⇒ (x − y) = 0 or (x + y) = 0
=⇒ x = y or x = −y
=⇒ x = ±y
Therefore, x = y or x = −y.

∴ T is not one-to-one
Counter Eg: T (−2) = T (2), but − 2 ̸= 2

11
Remarks
For a linear transformation T with matrix representation A:

T is 1-1 ⇔ ker(T ) = {0} ⇔ Ax = 0 has only trivial solution

This is equivalent to: columns of A are linearly independent.

Onto (Surjective) function

Let f : X → Y be a function. f is said to be onto (surjective) if

(∀ y ∈ Y (∃ x ∈ X(f (x) = y))).

Example 1
The following figure shows that the mapping T : Rn → Rm is not onto and T is onto.

Figure 4: Describes whether T : Rn → Rm is onto function or not

Example 2
Test whether the function T : R → R defined by T (x) = x3 is onto.

Solution:
Let T (x) = y ∈ R, ∃x (To find x?),
√
Now, y = x3 ⇒ x = 3 y
√ √
∴ T (x) = T ( 3 y) = ( 3 y)3 = y✓
⇒ T is onto.

Example 3
Test whether the function T : R → R defined by T (x) = x2 is onto.

Solution:
Let T (x) = y, ∃x (To find x?),
√
Now, y = x2 ⇒ x = ± y ×
This is not true ∀ y ∈ R

Therefore, T is not onto.

Counter Eg:
∃ y = −2 (Take)
Then ∀x ∈ R
T (x) ̸= y (= −2)

12
Remarks
For a linear transformation T : Rn → Rm with matrix A:

T is onto ⇔ Range(T ) = Rm ⇔ rank(A) = m

This is equivalent to: every b ∈ Rm has a solution to Ax = b (i.e., the columns of A span Rm ).

rank(A) = dim(Range(T )) = dim(Col(A))

Bijective function (one-to-one correspondence)

Let f : X → Y be a function. f is bijective iff f is both one-to-one and onto.

Example: T : R → R by T (x) = x3 is bijective function. (i.e.,) (∀y ∈ Y (∃!x ∈ X(f (x) = y)))

Remarks
If T : Rn → Rm be a linear transformation and if A is an m × n matrix, here are three ways of viewing
the same subspace of Rn :

• Matrix view: the null space of A

• System view: the solution space of Ax = 0
• Transformation view: the kernel of TA

and here are three ways of viewing the same subspace of Rm :

• Matrix view: the column space of A

• System view: all b in Rm for which Ax = b is consistent
• Transformation view: the range of TA

13
 Invertible Linear Transformation
Definition: Let T : V → W be a linear transformation. Then T is said to be ”invertible” (inverse
exist/one-to-one correspondance between these two spaces) iff T is both one-to-one & onto. If
T is linear, then T −1 is also linear.
(OR)
A function f : X(̸= ϕ) → Y (̸= ϕ) is said to be invertible iff

(∀f (∃! g (inverse function of f ), g = f −1 : Y → X(f og = gof = Id)))

Figure 5: Inverse mapping

Procedure to Find Inverse Function

1. Start with the function y = f (x).
2. Interchange x and y:
x = f (y)

3. Solve for y in terms of x.

4. Rewrite y as the inverse function f −1 (x).

5. Verify by checking f (f −1 (x)) = x and f −1 (f (x)) = x (optional).

Example 1: Inverse of f (x) = 2x + 3

y = 2x + 3
x = 2y + 3 (interchange x and y)
x − 3 = 2y
x−3
y=
2
−1 x−3
f (x) =
2

14
 3x+2
Example 2: Inverse of f (x) = x−1
3x + 2
y=
x−1
3y + 2
x= (interchange x and y)
y−1
x(y − 1) = 3y + 2
xy − x = 3y + 2
xy − 3y = x + 2
y(x − 3) = x + 2
x+2
y=
x−3
x+2
f −1 (x) =
x−3

Properties of Invertible Linear Transformation

For a linear transformation T : V → V (same domain and codomain), the following are equivalent:
1. T is invertible
2. T is bijective
3. T is one-to-one
4. T is onto
5. ker(T ) = {0}
6. rank(T ) = dim(V )

Square Matrix Invertibility

For a square matrix A (of size n × n):

A is invertible ⇔ det(A) ̸= 0
⇔ rank(A) = n
⇔ null space of A = {0}

Determinant Connections
For a square matrix A:
• det(A) ̸= 0 ⇔ A is invertible
• det(A) = 0 ⇔ columns are linearly dependent
• det(A) = 0 ⇔ ker(A) ̸= {0}
• |A| = volume scaling factor of the transformation

Key Relationships
1. Rank-Nullity Theorem:
n = rank + nullity
connects dimensions of kernel and range.
2. REF/RREF computational tool:
• Pivot columns → basis for column space

15
 • Free variables → basis for null space
• Number of pivots = rank

3. Properties of square matrices:

Invertible ⇔ det ̸= 0 ⇔ full rank ⇔ trivial null space

4. Geometric interpretation:
• Kernel: vectors that get ”collapsed” to zero
• Range: all possible outputs (target set)
• Rank: effective dimensionality of output
• Determinant: volume scaling factor of the linear transformation

Example 1
Given T : R → R defined by T (x) = 2x. Find the dimensions of kernel and image of T .

Solution:
Kernel:- T (x) = 0 ⇒ 2x = 0 ⇒ x = 0
∴ ker(T ) = {0}
Image:- Im(T ) = R
Notice that here, dim(domain space) = 1, dim(ker(T )) = 0, dim(Im(T )) = 1

Verification: dim(domain space) = dim(ker(T )) + dim(Im(T ))

Therefore, 1 = 0 + 1.

Figure 6: Kernel and Images

Example 2
Given T : R → R defined by T (x) = 0. Find the dimensions of kernel and image of T .

Solution:
Kernel:- T (x) = 0 ⇒ y = 0 (entire x-axis)
∴ ker(T ) = R
Image:- Im(T ) = {0}
Notice that here, dim(domain space) = 1, dim(ker(T )) = 1, dim(Im(T )) = 0

Verification: dim(domain space) = dim(ker(T )) + dim(Im(T ))

Therefore, 1 = 1 + 0.

16
 Figure 7: Kernel and Images

Example 3
Given: T : R2 → R2 defined by T (x, y) = (2x + y, x − y). Find the dimensions of kernel and image of T .

Solution:
Finding the Kernel
The kernel consists of all vectors (x, y) such that T (x, y) = (0, 0).

T (x, y) = (0, 0)
(2x + y, x − y) = (0, 0)
This gives us the system of equations:
(
2x + y = 0
x−y =0

From the second equation: y = x

Substituting into the first equation:

2x + x = 0 =⇒ 3x = 0 =⇒ x = 0

Therefore: y = 0

Result:
ker(T ) = {(0, 0)}

Dimension:
dim(ker(T )) = 0

Finding the Image

Since ker(T ) = {(0, 0)}, this follows from the rank-nullity theorem: dim(R2 ) = dim(ker(T ))+dim(Im(T ))

2 = 0 + dim(Im(T )) =⇒ dim(Im(T )) = 2

Result:
Im(T ) = R2

Dimension:
dim(Im(T )) = 2

17
Verification using Rank-Nullity Theorem
dim(R2 ) = dim(ker(T )) + dim(Im(T ))
2=0+2=2 ✓

Example 4
Given: T : R3 → R3 defined by T (x, y, z) = (x + y, y + z, x + z). Find the dimensions of kernel and image
of T .

Solution:
Finding the Kernel
The kernel consists of all vectors (x, y, z) such that T (x, y, z) = (0, 0, 0).

T (x, y, z) = (0, 0, 0)
(x + y, y + z, x + z) = (0, 0, 0)
This gives us the system of equations:

x + y = 0 (1)

y + z = 0 (2)

x + z = 0 (3)


From equation (1): y = −x

From equation (2): z = −y = −(−x) = x
Substituting into equation (3):

x + x = 0 =⇒ 2x = 0 =⇒ x = 0

Therefore: y = 0 and z = 0

Result:
ker(T ) = {(0, 0, 0)}

Dimension:
dim(ker(T )) = 0

Finding the Image

Since ker(T ) = {(0, 0, 0)}, using the rank-nullity theorem:

dim(R3 ) = dim(ker(T )) + dim(Im(T ))

3 = 0 + dim(Im(T ))
dim(Im(T )) = 3

Result:
Im(T ) = R3

Dimension:
dim(Im(T )) = 3

18
Alternative Verification: Direct Computation
To verify Im(T ) = R3 , we show that any vector (a, b, c) ∈ R3 can be expressed as T (x, y, z) for some
(x, y, z).
We need to solve:
T (x, y, z) = (a, b, c)
This gives us the system: 
x + y = a

y+z =b

x+z =c


Adding all three equations:

2(x + y + z) = a + b + c
a+b+c
x+y+z =
2
From the system, we can solve:
a+c−b
x=
2
a+b−c
y=
2
b+c−a
z=
2

Verification:
 
a+c−b a+b−c b+c−a
T , ,
2 2 2
 
a+c−b a+b−c a+b−c b+c−a a+c−b b+c−a
= + , + , +
2 2 2 2 2 2
 
a+c−b+a+b−c a+b−c+b+c−a a+c−b+b+c−a
= , ,
2 2 2
 
2a 2b 2c
= , ,
2 2 2
= (a, b, c) ✓
This confirms that every vector in R3 is in the image of T .

Example 5
Consider the linear transformation T : R3 → R3 defined by

T (x1 , x2 , x3 ) = (2x1 + 3x2 + x3 , 3x1 + 3x2 + x3 , 2x1 + 4x2 + x3 )

1. Show that T is invertible, and find its inverse

2. Find the Ker(T )
3. Find the Range(T )
4. Find the basis for Range(T )

Solution:
Method 1
Step 1: Find the Matrix Representation of T
First, let’s express T as a matrix. We need to find how T acts on the standard basis vectors:
For e1 = (1, 0, 0):

T (1, 0, 0) = (2(1) + 3(0) + 0, 3(1) + 3(0) + 0, 2(1) + 4(0) + 0) = (2, 3, 2)

19
 For e2 = (0, 1, 0):

T (0, 1, 0) = (2(0) + 3(1) + 0, 3(0) + 3(1) + 0, 2(0) + 4(1) + 0) = (3, 3, 4)

For e3 = (0, 0, 1):

T (0, 0, 1) = (2(0) + 3(0) + 1, 3(0) + 3(0) + 1, 2(0) + 4(0) + 1) = (1, 1, 1)

Therefore, the matrix representation is:

 
2 3 1
A = 3 3 1
2 4 1

Part (i): Show that T is invertible and find its inverse

To show T is invertible, we need to show that the matrix A has rank 3 (full rank) (or) |A| ̸= 0.
Determinant:
2 3 1
3 3 1 = 2(3 − 4) − 3(3 − 2) + 1(12 − 6) = 1 ̸= 0
2 4 1
∵ |A| ̸= 0 and therefore T is invertible.
(OR)

Row Reduction to REF:

Starting matrix:  
2 3 1
3 3 1
2 4 1
Step 1: R1 ← 12 R1
3 1
 
1 2 2
3 3 1
2 4 1
Step 2: R2 ← R2 − 3R1 and R3 ← R3 − 2R1
3 1
 
1 2 2
0 − 3 − 21 
2
0 1 0

Step 3: R2 ← − 23 R2
3 1
 
1 2 2
0 1
1 3
0 1 0
Step 4: R3 ← R3 − R2
3 1
 
1 2 2
0 1 
1 3
0 0 − 13
From REF (A) we can see that A has rank 3 and is therefore invertible.

20
Finding the Inverse:
Matrix of cofactors:
 
(−1) × (+1) (1) × (−1) (6) × (+1)
C = (−1) × (−1) (0) × (+1) (2) × (−1)  (1)
(0) × (+1) (−1) × (−1) (−3) × (+1)
 
−1 −1 6
= 1 0 −2 (2)
0 1 −3

Take the Transpose (Adjugate Matrix)

 
−1 1 0
adj(A) = C T = −1 0 1 (3)
6 −2 −3

Apply the Formula A−1 = 1

det(A) × adj(A)
Since det(A) = 1:

 
−1 1 0
1 
A−1 = × −1 0 1 (4)
1
6 −2 −3
 
−1 1 0
= −1 0 1 (5)
6 −2 −3

(OR)
To find A−1 , we use the augmented matrix [A|I] and row reduce:
 
2 3 1 1 0 0
 3 3 1 0 1 0 
2 4 1 0 0 1
Following the same row operations as above:
After complete reduction:  
1 0 0 −1 1 0
 0 1 0 −1 0 1 
0 0 1 6 −2 −3
Therefore:  
−1 1 0
A−1 = −1 0 1
6 −2 −3
The inverse transformation is:

T −1 (y1 , y2 , y3 ) = (−y1 + y2 , −y1 + y3 , 6y1 − 2y2 − 3y3 )

Part (ii): Find Ker(T)

The kernel of T consists of all vectors x such that T (x) = 0.
This means solving Ax = 0:     
2 3 1 x1 0
3 3 1 x2  = 0
2 4 1 x3 0
Since we already showed that A is invertible (has full rank), the only solution to Ax = 0 is x = 0.
Therefore: Ker(T ) = {0}

21
Part (iii): Find Range(T)
Since T : R3 → R3 and T is invertible (as shown in part i), by the Rank-Nullity Theorem:

dim(R3 ) = dim(Ker(T )) + dim(Range(T ))

3 = 0 + dim(Range(T ))
dim(Range(T )) = 3
Since the range is a 3-dimensional subspace of R3 , we have:
Range(T ) = R3

Part (iv): Find a basis for Range(T)

Since Range(T ) = R3 , any basis for R3 will work.
The most natural choice is to use the columns of matrix A, since they span the range of T .
The columns of A are:
 
2
• v1 = 3
2
 
3
• v2 = 3
4
 
1
• v3 = 1
1
Since A has full rank, these three vectors are linearly independent.
A basis for Range(T ) is:      
 2 3 1 
3 , 3 , 1
2 4 1
 

Or the standard basis:

e1 = (1, 0, 0)
e2 = (0, 1, 0)
e3 = (0, 0, 1)

Method 2
Step 1: Find the Matrix Representation of T
Extract T as a matrix from the given

T (x1 , x2 , x3 ) = (2x1 + 3x2 + x3 , 3x1 + 3x2 + x3 , 2x1 + 4x2 + x3 )

The matrix representation is: x 1 x2 x3

 
2 3 1
A = 3 3 1
2 4 1

22
Part (i): Show that T is invertible and find its inverse
To show T is invertible, we need to show that the matrix A has rank 3 (full rank) (or) |A| ̸= 0.
Determinant:
2 3 1
3 3 1 = 2(3 − 4) − 3(3 − 2) + 1(12 − 6) = 1 ̸= 0
2 4 1
∵ |A| ̸= 0 and therefore T is invertible.

(OR)

Row Reduction to REF:

Starting matrix:  
2 3 1
3 3 1
2 4 1
Step 1: R1 ← 12 R1
3 1
 
1 2 2
3 3 1
2 4 1
Step 2: R2 ← R2 − 3R1 and R3 ← R3 − 2R1
3 1
 
1 2 2
0 − 3 − 21 
2
0 1 0

Step 3: R2 ← − 23 R2
3 1
 
1 2 2
0 1
1 3
0 1 0
Step 4: R3 ← R3 − R2
3 1
 
1 2 2
0 1 
1 3
0 0 − 13
From REF (A) we can see that A has rank 3 and is therefore invertible.

Finding the Inverse:

Matrix of cofactors:
 
(−1) × (+1) (1) × (−1) (6) × (+1)
C = (−1) × (−1) (0) × (+1) (2) × (−1)  (6)
(0) × (+1) (−1) × (−1) (−3) × (+1)
 
−1 −1 6
= 1 0 −2 (7)
0 1 −3

Take the Transpose (Adjugate Matrix)

 
−1 1 0
adj(A) = C T = −1 0 1 (8)
6 −2 −3

Apply the Formula A−1 = 1

det(A) × adj(A)
Since det(A) = 1:

23
  
−1 1 0
1 
A−1 = × −1 0 1 (9)
1
6 −2 −3
 
−1 1 0
= −1 0 1 (10)
6 −2 −3

(OR)
To find A−1 , we use the augmented matrix [A|I] and row reduce:
 
2 3 1 1 0 0
 3 3 1 0 1 0 
2 4 1 0 0 1
Following the same row operations as above:
After complete reduction:  
1 0 0 −1 1 0
 0 1 0 −1 0 1 
0 0 1 6 −2 −3
Therefore: y1 y2 y3
 
−1 1 0
A−1 = −1 0 1
6 −2 −3

The inverse transformation is:

T −1 (y1 , y2 , y3 ) = (−y1 + y2 , −y1 + y3 , 6y1 − 2y2 − 3y3 )

Part (ii): Find Ker(T)

The kernel of T consists of all vectors x such that T (x) = 0.
This means solving Ax = 0.
Since we already showed that A is invertible, this implies that T is invertible, and hence T is one-to-one.
Therefore: Ker(T ) = {0} = {(0, 0, 0)}

Part (iii): Find Range(T)

Since T : R3 → R3 and T is invertible (as shown in part i) and hence T is onto.
Range(T ) = R3

Part (iv): Find a basis for Range(T)

Since Range(T ) = R3 , any basis for R3 will work.

Use standard basis:

e1 = (1, 0, 0)
e2 = (0, 1, 0)
e3 = (0, 0, 1)

24
Summary
1. T is invertible because its matrix has full rank (3). The inverse is T −1 (y1 , y2 , y3 ) = (−y1 +
y2 , −y1 + y3 , 6y1 − 2y2 − 3y3 )
2. Ker(T ) = {0} (trivial kernel)
3. Range(T ) = R3 (entire codomain)
     
 2 3 1 
4. Basis for Range(T): 3 , 3 , 1
2 4 1
 
or

e1 = (1, 0, 0)
e2 = (0, 1, 0)
e3 = (0, 0, 1)

any other basis of R3

Note: If the linear transformation T : Rn → Rn , then for any n > 3 use Gauss-Jordan method to
find the inverse.

Example 6
Given a linear transformation T : V2 (R) → V3 (R) defined by:

T (a, b) = (a + b, a − b, b)

Find the range, rank, null-space, and nullity of T .

Solution
Step 1: Finding the Range of T
To find the range RT , we need to determine what vectors in V3 (R) can be obtained as outputs of T .
Since {(1, 0), (0, 1)} forms a basis for V2 (R), we can find the range by applying T to these basis vectors:
Applying T to basis vectors:

T (1, 0) = (1 + 0, 1 − 0, 0) = (1, 1, 0)
T (0, 1) = (0 + 1, 0 − 1, 1) = (1, −1, 1)

The matrix representation of the linear transformation T is

 
1 1
A = 1 −1
0 1

Since (1, 0) and (0, 1) generate V2 (R), the vectors T (1, 0) and T (0, 1) will generate T (V2 (R)) = RT .
Therefore:

RT = span{(1, 1, 0), (1, −1, 1)} = Column space (T) ⊆ subspace of V3 (R)

Step 2: Finding the Rank of T

To find the rank, we need to determine if the vectors (1, 1, 0) and (1, −1, 1) are linearly independent.
Testing linear independence:
Suppose a(1, 1, 0) + b(1, −1, 1) = (0, 0, 0) for some scalars a, b ∈ R.
This gives us:
(a + b, a − b, b) = (0, 0, 0)

25
 From this system of equations:

a+b=0
a−b=0
b=0

From the third equation: b = 0

Substituting into the first equation: a + 0 = 0 ⇒ a = 0
Since a = 0 and b = 0 is the only solution, the vectors (1, 1, 0) and (1, −1, 1) are linearly independent.
Hence, these two (1, 1, 0) and (1, −1, 1) vectors span a subspace of V3 (R) (range of T ) with dimension 2.
Therefore: rank(T ) = dim(RT ) = 2

Step 3: Finding the Nullity of T

Using the Rank-Nullity Theorem for linear transformations:

dim(V2 (R)) = rank(T ) + nullity(T )

We know:
• dim(V2 (R)) = 2
• rank(T ) = 2
Therefore:

2 = 2 + nullity(T )
nullity(T ) = 0

Step 4: Finding the Null-Space of T = Ker(T)

Since the nullity is 0, the null-space NT contains only the zero vector.
Verification: Let’s find all (a, b) ∈ V2 (R) such that T (a, b) = (0, 0, 0):

T (a, b) = (a + b, a − b, b) = (0, 0, 0)
This gives us the system:

a+b=0
a−b=0
b=0

From the third equation: b = 0

Substituting into the first: a + 0 = 0 ⇒ a = 0
Therefore: NT = {(0, 0)}

Final Answer
• Range: RT = span{(1, 1, 0), (1, −1, 1)}
• Rank: rank(T ) = 2
• Null-space: NT = {(0, 0)}
• Nullity: nullity(T ) = 0

Verification
The Rank-Nullity theorem is satisfied:

dim(V2 (R)) = 2 = 2 + 0 = rank(T ) + nullity(T ) ✓

26
Example 7
 
1 −1 3
The linear transformation T : R3 → R3 is defined by T (x) = Ax, where A = 5 6 −4.
7 4 2
Find a basis for the kernel of T and a basis for the range of T . Also find the rank and nullity of T .

Solution:

Step 1: Finding the Kernel of T

The kernel (null space) of T consists of all vectors x ∈ R3 such that T (x) = 0, or equivalently Ax = 0.
We need to solve the homogeneous system:
    
1 −1 3 x1 0
5 6 −4 x2  = 0
7 4 2 x3 0

Row reduction of the augmented matrix:

Starting matrix:  
1 −1 3 0
 5 6 −4 0 
7 4 2 0
R2 ← R2 − 5R1 :  
1 −1 3 0
 0 11 −19 0 
7 4 2 0
R3 ← R3 − 7R1 :  
1 −1 3 0
 0 11 −19 0 
0 11 −19 0
R3 ← R3 − R2 :  
1 −1 3 0
 0 11 −19 0 
0 0 0 0
1
R2 ← 11 R2 :  
1 −1 3 0
 0 1 − 19
11 0 
0 0 0 0
R1 ← R1 + R2 :
14
 
1 0 11 0
19
 0 1 − 11 0 
0 0 0 0
Solution: From the reduced form:
• x1 + 14
11 x3 = 0 ⇒ x1 = − 14
11 x3

• x2 − 19
11 x3 = 0 ⇒ x2 = 19
11 x3

• x3 is free
Setting x3 = 11t (to avoid fractions), we get:
   
 −14 
ker(T ) = t  19  : t ∈ R
11
 

Basis for the kernel:  

 −14 
 19 
11
 

27
Step 2: Finding the Range of T
The range (column space) of T is the span of the columns of A.
From our row reduction, we found that the matrix has rank 2 (two pivot columns). The first two
columns are linearly independent and form a basis for the column space.
Basis for the range:    
 1 −1 
5 ,  6 
7 4
 

Step 3: Finding the Rank and Nullity

Rank of T: The rank is the dimension of the range, which equals the number of pivot columns.

rank(T ) = 2

Nullity of T: The nullity is the dimension of the kernel.

nullity(T ) = 1

Verification: By the Rank-Nullity Theorem:

rank(T ) + nullity(T ) = dim(R3 ) = 3

2+1=3 ✓

Final Answer
 
 −14 
• Basis for kernel:  19 
11
 
    
 1 −1 
• Basis for range: 5 ,  6 
7 4
 

• Rank: 2
• Nullity: 1

Example 8
Let T : P3 → P2 be the mapping defined by T (a0 + a1 x + a2 x2 + a3 x3 ) = 5a0 + a3 x2 .

(a) Show that T is linear. (b) Find a basis for the kernel of T . (c) Find a basis for the range of T .

Solution:
Part (a): Show that T is linear.
To prove T is linear, we need to show:
1. T (u + v) = T (u) + T (v) for all u, v ∈ P3
2. T (cu) = cT (u) for all c ∈ R and u ∈ P3
Let u = a0 + a1 x + a2 x2 + a3 x3 and v = b0 + b1 x + b2 x2 + b3 x3 .

Additivity:

T (u + v) = T ((a0 + b0 ) + (a1 + b1 )x + (a2 + b2 )x2 + (a3 + b3 )x3 )

= 5(a0 + b0 ) + (a3 + b3 )x2
= 5a0 + 5b0 + a3 x2 + b3 x2
= (5a0 + a3 x2 ) + (5b0 + b3 x2 )
= T (u) + T (v)

28
Homogeneity:

T (cu) = T (ca0 + ca1 x + ca2 x2 + ca3 x3 )

= 5(ca0 ) + (ca3 )x2
= c(5a0 ) + c(a3 x2 )
= c(5a0 + a3 x2 )
= cT (u)

Therefore, T is linear.

Part (b): Find the matrix representation and use REF to find the kernel.
Using standard bases α = {1, x, x2 , x3 } for P3 and β = {1, x, x2 } for P2 :
Apply T to each basis vector:

T (1) = 5 · 1 + 0 · x2 = 5 = 5 · 1 + 0 · x + 0 · x2
T (x) = 5 · 0 + 0 · x2 = 0 = 0 · 1 + 0 · x + 0 · x2
T (x2 ) = 5 · 0 + 0 · x2 = 0 = 0 · 1 + 0 · x + 0 · x2
T (x3 ) = 5 · 0 + 1 · x2 = x2 = 0 · 1 + 0 · x + 1 · x2

The matrix representation is:  

5 0 0 0
[T ]βα = 0 0 0 0
0 0 0 1

Finding the kernel using REF:

To find ker(T ), we solve [T ]βα x = 0:
 
 a
0  0
  
5 0 0 0
0 a1   
0 0 0 
a2  = 0
0 0 0 1 0
a3

This gives us the system:

5a0 = 0 ⇒ a0 = 0
0 = 0 (always true)
a3 = 0

Therefore, a1 and a2 are free variables. The general solution is:

     
a0 0 0
a1 
  = a1 1 + a2 0
   
a2  0 1
a3 0 0

Basis for kernel: {x, x2 }

Verification: dim(ker(T )) = 2, and by rank-nullity theorem: dim(P3 ) = dim(ker(T ))+dim(range(T )),
so 4 = 2 + dim(range(T )), giving dim(range(T )) = 2.

Part (c): Find the range using REF.

The range of T is the column space of [T ]βα . We find the pivot columns by row reducing:
   
5 0 0 0 1 0 0 0
REF
0 0 0 0 −−−→ 0 0 0 1
0 0 0 1 0 0 0 0

29
  
5
The pivot columns are columns 1 and 4, corresponding to: - Column 1: 0 represents 5 · 1 + 0 · x +
0
 
0
0 · x2 = 5 - Column 4: 0 represents 0 · 1 + 0 · x + 1 · x2 = x2
1
Basis for range: {5, x2 } or equivalently {1, x2 }

Final Answers:
(a) T is linear (proven above)

(b) Basis for ker(T ): {x, x2 }

(c) Basis for range of T : {1, x2 }

Verification: dim(ker(T )) + dim(range(T )) = 2 + 2 = 4 = dim(P3 )

Theorem
Let V and W be vector spaces and α = {v1 , v2 , . . . , vn } be a basis for V and w1 , w2 , . . . , wm be any m
vectors (possibly repeated) in W . Then there exists a unique linear transformation T : V → W such that

T (vi ) = wj , ∀ i = 1, 2, . . . , n, ∀j = 1, 2, . . . , m

Find T and T −1 Problems

Example 1
Let β = {v1 = (1, 1, 1), v2 = (1, 1, 0), v3 = (1, 0, 0)} be a basis for R3 and T : R3 → R2 be a linear
transformation, defined by

T (v1 ) = w1 , T (v2 ) = w2 , and T (v3 ) = w3 ,

where w1 = (1, 0), w2 = (2, −1), w3 = (4, 3).

Find a formula for T (x1 , x2 , x3 ) and hence find T (2, −3, 5) and T (5, 4, 3).

Solution:
Let x = (x1 , x2 , x3 ) ∈ R3 then

x = av1 + bv2 + cv3

= a(1, 1, 1) + b(1, 1, 0) + c(1, 0, 0)
(x1 , x2 , x3 ) = (a, a, a) + (b, b, 0) + (c, 0, 0)
= (a + b + c, a + b, a)

a + b + c = x1

⇒ a + b = x2

a = x3


a = x3
b = x 2 − x3
c = x 1 − x2

⇒ a = x3 , b = x2 − x 3 , c = x1 − x 2
∴ x = x3 v1 + (x2 − x3 )v2 + (x1 − x2 )v3

30
Now,
T (x) = T (x3 v1 + (x2 − x3 )v2 + (x1 − x2 )v3 )
= x3 T (v1 ) + (x2 − x3 )T (v2 ) + (x1 − x2 )T (v3 ) [since T is linear]
= x3 w1 + (x2 − x3 )w2 + (x1 − x2 )w3
= x3 (1, 0) + (x2 − x3 )(2, −1) + (x1 − x2 )(4, 3)
= (x3 , 0) + (2x2 − 2x3 , −x2 + x3 ) + (4x1 − 4x2 , 3x1 − 3x2 )
= (4x1 − 2x2 − x3 , 3x1 − 4x2 + x3 )

T (x1 , x2 , x3 ) = (4x1 − 2x2 − x3 , 3x1 − 4x2 + x3 )

Hence, T (2, −3, 5) = (9, 23) and T (5, 4, 3) = (9, 2)

Verification:
Check T (v1 ) with v1 = (1, 1, 1):
   
4−2−1 1
T (v1 ) = = = w1 .
3−4+1 0

Check T (v2 ) with v2 = (1, 1, 0):

   
4−2−0 2
T (v2 ) = = = w2 .
3−4+0 −1

Check T (v3 ) with v3 = (1, 0, 0):

   
4−0−0 4
T (v3 ) = = = w3 .
3−0+0 3

Example 2
Find T −1 of T (x, y, z) = (2x, 4x − y, 2x + 3y − z) for T : R3 → R3 . Also, find T −1 (1, 2, 3).

Solution:
Let (r, s, t) = T (x, y, z) = (2x, 4x − y, 2x + 3y − z) eqn(1)
Then T −1 (r, s, t) = (x, y, z) =⇒ T −1 (2x, 4x − y, 2x + 3y − z) = (x, y, z)
Now from eqn(1), 
r = 2x

s = 4x − y

t = 2x + 3y − z


r
⇒x=
2
r
s = 4x − y = 4 − y = 2r − y =⇒ y = 2r − s
2

r
t = 2x + 3y − z = 2 + 3(2r − s) − z = r + 6r − 3s − z = 7r − 3s − z =⇒ z = 7r − 3s − t
2

Write the general inverse transformation T −1 (r, s, t) = (x, y, z)

Therefore: r 
T −1 (r, s, t) = , 2r − s, 7r − 3s − t
2

31
Find T −1 (1, 2, 3)
Substitute r = 1, s = 2, t = 3:
 
1
T −1 (1, 2, 3) = , 2(1) − 2, 7(1) − 3(2) − 3 (11)
2
 
1
= , 2 − 2, 7 − 6 − 3 (12)
2
 
1
= , 0, −2 (13)
2

Verification
Let’s verify by checking T 12 , 0, −2 :

         
1 1 1 1
T , 0, −2 = 2 ,4 − 0, 2 + 3(0) − (−2) (14)
2 2 2 2
= (1, 2, 1 + 0 + 2) (15)
= (1, 2, 3) ✓ (16)

Final Answer:
• General inverse: T −1 (r, s, t) = r


− s, 7r − 3s − t
2 , 2r

• Specific value: T −1 (1, 2, 3) = 21 , 0, −2

Exercise
1. Find T −1 of T (x, y, z) = (3x, x − y, 2x + y + z) for T : R3 → R3 and hence find T −1 (−1, 0, 1).

Answers: T −1 (r, s, t) = 3r , 3r − s, −r + s + t and T −1 (−1, 0, 1) = − 31 , − 13 , 2

2. Consider the linear transformation T : R3 → R3 defined by

T (x, y, z) = (2x + 3y + z, 3x + 3y + z, 2x + 4y + z).

Determine the invertibility of T . If T is invertible, then find T −1 (x, y, z).

3. Let T : V3 (R) → V3 (R) be a linear transformation defined by

T (a, b, c) = (3a, a − b, 2a + b + c), ∀ a, b, c ∈ R.

Prove that T is invertible and find T −1 . Also prove that (T 2 − I)(T − 3I) = 0.
4. Let T : R3 → R3 be a linear transformation such that

T (1, 0, 0) = (2, −1, 4)

T (0, 1, 0) = (1, 5, −2)
T (0, 0, 1) = (0, 3, 1)

Find the formula for T (x, y, z) and compute T (2, 3, −2).

32
 Isomorphism
A linear transformation T : V → W from a vector space V to a vector space W is called an isomorphism,
if it is invertible (or one-to-one and onto). In this case, we say V and W are isomorphic to each other.

Note: If T is an isomorphism, then its inverse T −1 is also an isomorphism with (T −1 )−1 = T . There-
fore, if V and W are isomorphic to each other, then it means that they look the same as vector spaces.

Remarks: If T : V → W and S : W → Z are linear transformations, then it is quite easy to show

that their composition (S ◦ T )(v) = S(T (v)) is also a linear transformation from V to Z.

In particular, if two linear transformations are given by matrices A : Rn → Rm and B : Rm → Rk ,

then their composition is nothing but the matrix multiplication BA of them, i.e., (B ◦ A)(x) = B(Ax) =
(BA)x.

Hence, if a linear transformation is given by an invertible n × n square matrix A : Rn → Rn , then the

inverse matrix A−1 plays the inverse linear transformation, so that it is an isomorphism of Rn . That is,
a linear transformation given by an n × n square matrix A : Rn → Rn is an isomorphism if and only if
rank A = n.

Lemma
Suppose that S and T are linear transformations whose composition S ◦ T is well-defined. Then, the
following are true:
1. if S ◦ T is one-to-one, then T is one-to-one,

2. if S ◦ T is onto, then S is onto,

3. if S and T are isomorphisms, then S ◦ T is also isomorphism,
4. if A and B are two n × n matrices of rank n, then AB also has rank n.

Theorem
Two vector spaces V and W are isomorphic if and only if dim V = dim W .

Remark
Let T : V → W be a linear transformation. Then, the following are true:
1. T is one-to-one if and only if Ker(T ) = {0},
2. if V = W , then T is one-to-one if and only if T is onto.

Corollary
Any n-dimensional vector space V is isomorphic to the n-space Rn .

Problems
Example 1: Show that T is an Isomorphism on R2
.
Define,
T : R2 → R2 , by T (x, y) = (x + y, x − y).
−1
and hence, find T .

33
Solution
Step 1: Linearity
A transformation T is linear if for all vectors u, v ∈ R2 and scalar c ∈ R,

T (u + v) = T (u) + T (v), T (cu) = cT (u).

Let u = (x1 , y1 ) and v = (x2 , y2 ).

T (u + v) = T (x1 + x2 , y1 + y2 ) = (x1 + x2 + y1 + y2 , x1 + x2 − y1 − y2 ).
On the other hand,

T (u) + T (v) = (x1 + y1 , x1 − y1 ) + (x2 + y2 , x2 − y2 ) = (x1 + x2 + y1 + y2 , x1 + x2 − y1 − y2 ).

So T (u + v) = T (u) + T (v).
Also,
T (cu) = T (cx1 , cy1 ) = (cx1 + cy1 , cx1 − cy1 ) = c(x1 + y1 , x1 − y1 ) = cT (u).
Hence T is linear.

Step 2: Standard Matrix of T

We evaluate T on the standard basis vectors:

T (1, 0) = (1, 1), T (0, 1) = (1, −1).

So the standard matrix of T is  
1 1
A= .
1 −1

Step 3: Check Isomorphism

A linear transformation is an isomorphism if and only if its matrix is invertible.

det(A) = (1)(−1) − (1)(1) = −1 − 1 = −2 ̸= 0.

Therefore T is invertible, hence an isomorphism.

Step 4: Inverse Transformation

The inverse matrix is
     
1 −1 −1 1 −1 −1 1 1 1
A−1 = = = .
det(A) −1 1 −2 −1 1 2 1 −1

Thus for (u, v) ∈ R2 ,

T −1 (u, v) = u+v u−v


2 , 2 .

Example 2: Show that T is an Isomorphism between P2 (R) and

R3
Let
T : P2 (R) → R3 , T (a + bx + cx2 ) = (a + b, b + c, a + c).
and hence compute T −1 .

34
Solution
Step 1: Linearity
Each component of T is a linear combination of a, b, c. Therefore T is linear. (Try!)

Step 2: Standard Matrix

We compute T on the standard basis {1, x, x2 } of P2 (R):

T (1) = (1, 0, 1), T (x) = (1, 1, 0), T (x2 ) = (0, 1, 1).

So the matrix is  
1 1 0
A = 0 1 1 .
1 0 1

Step 3: Isomorphism Check

det(A) = 1 · (1 · 1 − 1 · 0) − 1 · (0 · 1 − 1 · 1) + 0 = 1(1) − 1(−1) = 1 + 1 = 2 ̸= 0.

So T is invertible, hence an isomorphism.

Step 4: Inverse Transformation

The inverse matrix is  
1 −1 1
1
A−1 = 1 1 −1 .
2
−1 1 1
So for (u, v, w) ∈ R3 ,      
a u u−v+w
1
 b  = A−1  v  =  u + v − w  .
2
c w −u + v + w
Thus
u−v+w u+v−w −u + v + w 2
T −1 (u, v, w) = + x+ x .
2 2 2

Step 5: Verification Example

Take p(x) = 2 + 3x + x2 . Then a = 2, b = 3, c = 1.

T (p) = (2 + 3, 3 + 1, 2 + 1) = (5, 4, 3).

Now apply the inverse:

a′ = 5−4+3
2 = 2, b′ = 5+4−3
2 = 3, c′ = −5+4+3
2 = 1.

Recovered coefficients (2, 3, 1) match the original. Verified.

Example 3: General Isomorphism Pn (R) → Rn+1

Define
T : Pn (R) → Rn+1 , T (a0 + a1 x + · · · + an xn ) = (v0 , . . . , vn )T ,
where
v0 = a0 , vi = ai + ai−1 , (i = 1, . . . , n).
and hence determine T −1 .

35
Solution
Step 1: Linearity
Each vi is a linear combination of coefficients aj . Hence T is linear. (Try!)

Step 2: Matrix of T
With basis {1, x, . . . , xn } of Pn (R) and the standard basis of Rn+1 , the matrix is
 
1 0 0 ··· 0
1
 1 0 · · · 0 
0 1 1 · · · 0
. .
 
A= ..
0
 0 1 . .. 

. .. . . ..
 ..

. . . 0
0 0 ··· 1 1

This is a lower-bidiagonal matrix with 1’s on the diagonal and 1’s just below the diagonal.

Step 3: Isomorphism
Since A is lower triangular with diagonal entries all equal to 1, we have

det(A) = 1 · 1 · · · 1 = 1 ̸= 0.

Thus A is invertible, so T is an isomorphism.

Step 4: Inverse Transformation

From the definition,
v0 = a0 =⇒ a0 = v0 .
For i ≥ 1,
vi = ai + ai−1 =⇒ ai = vi − ai−1 .
Thus recursively,

a0 = v0 , a1 = v1 − v0 , a2 = v2 − a1 , a3 = v3 − a2 , . . .

In closed form,
i
X
ai = (−1)i−k vk .
k=0

So
i
X
T −1 (v0 , . . . , vn ) = a0 + a1 x + · · · + an xn , ai = (−1)i−k vk .
k=0

Step 5: Verification Example (n = 3)

Take p(x) = 2 + 3x − 1x2 + 4x3 . Then a0 = 2, a1 = 3, a2 = −1, a3 = 4.
Apply T :
v0 = 2, v1 = 3 + 2 = 5, v2 = −1 + 3 = 2, v3 = 4 + (−1) = 3.
So T (p) = (2, 5, 2, 3).
Now recover using recursion:

a0 = v0 = 2, a1 = v1 − a0 = 5 − 2 = 3, a2 = v2 − a1 = 2 − 3 = −1, a3 = v3 − a2 = 3 − (−1) = 4.

Recovered coefficients (2, 3, −1, 4) match the original polynomial. Verified.

36
 Matrix of a Linear Transformation
Matrix Basics
Row Picture/Column Picture for the System of Linear Equations:
Example 1
Draw the row and column picture for the systme of linear equation of the form Ax = b.

2x − 3y = 1 (= Row Equation 1) (1)

x + 2y = −3 (= Row Equation 2) (2)

Solution:
Write the given system of linear equations in the form Ax = b.
    
2 −3 x 1
= (3)
1 2 y −3
     
2 −3 1
=⇒ x (= Column 1) + y (= Column 2) = (= Column 3) (4)
1 2 −3

Row Picture from Eqns.(1) and (2)

Figure 1: Row Picture

1
Column Picture from Eqn.(4)

Figure 2: Column Picture

Example 2
Draw the row and column picture for the systme of linear equation of the form Ax = b.

x+y+z =6 (5)
2x − y + z = 3 (6)
x + 2y − z = 2 (7)

Solution:
Write the given system of linear equations in the form Ax = b.
    
1 1 1 x 6
2 −1 1  y  = 3 (8)
1 2 −1 z 2
       
1 1 1 6
=⇒ x 2 (= Column 1) + y −1 (= Column 2) + z  1  (= Column 3) = 3 (= Column 4) (9)
1 2 −1 2

Row Picture from Eqns.(5) - (7)

2
 Figure 3: Row Picture

Column Picture from Eqn.(9)

Figure 4: Column Picture

3
 Introduction
For an m × n matrix A, the equation Ax = b means that to every vector x = [x1 x2 · · · xn ]T in Rn
the matrix multiplication Ax assigns a vector b ( = Ax ) in Rm . That is, the matrix A transforms every
vector x in Rn into a vector b in Rm by the matrix multiplication Ax = b. Moreover, the distributive
law
A(x + ky) = Ax + kAy
for k ∈ R and x, y ∈ Rn , of matrix multiplication means that A preserves the sum of vectors and scalar
multiplication.

Example 1
     
1 0 1 0
The columns of I2 = are e1 = and e2 = .
0 1 0 1
Suppose T : R2 → R3 is a linear transformation such that
   
5 −3
T (e1 ) = −7 and T (e2 ) =  8 
2 0

With no additional information, find a formula for the image T (x) of an arbitrary x ∈ R2 .
Solution: Write      
x 1 0
x = 1 = x1 + x2 = x1 e1 + x2 e2 (10)
x2 0 1
Since T is a linear transformation,

T (x) = x1 T (e1 ) + x2 T (e2 ) (11)

      
5 −3 5x1 − 3x2 5 −3     
x  x1
=⇒ T (x) = x1 −7 + x2  8  = −7x1 + 8x2  = −7 8  1 = T (e1 ) T (e2 ) = Ax
x2 x2
2 0 2x1 + 0 2 0

The step from equation (1) to equation (2) explains why knowledge of T (e1 ) and T (e2 ) is sufficient
to determine T (x) for any x. Moreover, since (2) expresses T (x) as a linear combination of vectors, we
can put these vectors into the columns of a matrix A and write (2) as
 
  x1
T (x) = T (e1 ) T (e2 ) = Ax
x2

Example 2
       
1 0 0 1 0 0
The columns of I3 = 0 1 0 are e1 = 0, e2 = 1 and e3 = 0.
0 0 1 0 0 1
Suppose T : R → R2
3
is a linear transformation such that
     
a a a
T (e1 ) = 11 , T (e2 ) = 12 and T (e2 ) = 13
a21 a22 a23

With no additional information, find a formula for the image T (x) of an arbitrary x ∈ R3 .
Solution: Write
       
x1 1 0 0
x = x2  = x1 0 + x2 1 + x3 0 = x1 e1 + x2 e2 + x3 e3
x3 0 0 1
Since T is a linear transformation,

T (x) = x1 T (e1 ) + x2 T (e2 ) + x3 T (e3 )

4
  
          x1
a a a a x + a12 x2 + a13 x3 a a12 a13  
=⇒ T (x) = x1 11 + x2 12 + x3 13 = 11 1 = 11 x2
a21 a22 a23 a21 x1 + a22 x2 + a23 x3 a21 a22 a23
x3
 
 x1
   
 y1 2
= T (e1 ) T (e2 ) T (e3 ) x2  = Ax =b= ∈R
y2
x3

Figure 5: Visual representation of T act as a matrix A (i.e., T (ej ) = Cj of A2×3 , ∀ 1 ≤ j ≤ 3)

Theorem 1
Let T : Rn → Rm be a linear transformation such that, for the standard basis vectors ei of Rn .
     
a11 a12 a1n
 a21   a22   a2n 
T (e1 ) =  .  , T (e2 ) =  .  , . . . , T (en ) =  .  .
     
 ..   ..   .. 
am1 am2 amn

Then there exists a unique matrix A such that T (x) = Ax, ∀ x ∈ Rn (A is called the standard matrix
for T )

5
 In fact, A is the m × n matrix whose jth column is the vector T (ej ), where ej is the jth column of
the identity matrix in Rn :
 
A = T (e1 ) ··· T (en )

Figure 6: Pictorial representation of T (ej ), ∀ 1 ≤ j ≤ n act as a coefficient matrix A

6
 Geometric Linear Transformations of R2
Note: Since the transformations are linear, they are determined completely by what they do to the
columns of I2 , and what a transformation does to the unit square.

Figure 7: The unit square

Example 1 (Dilation)
Find the standard matrix A for the dilation (scale factor> 1 [Enlargement], < 1 [Reduction]) transfor-
mation T (x) = 3x, ∀ x ∈ R2 .

Solution:
   
3 0
T (e1 ) = 3e1 = and T (e2 ) = 3e2 =
0 3
 
  3 0
A = T (e1 ) T (e2 ) =
0 3
In general,

Figure 8: Dilation/Scaling

Example 2 (Rotation)
Let T : R2 → R2 be the transformation that rotates (rotation) each point in R2 about the origin through
an angle φ, with counterclockwise rotation for a positive angle. We could show geometrically that such
a transformation is linear. Find the standard matrix A of this transformation.

7
Solution:
       
1 cos φ 0 − sin φ
rotates into , and rotates into . (See Figure)
0 sin φ 1 cos φ
By Theorem 1,  
cos φ − sin φ
A=
sin φ cos φ

Figure 9: A rotation transformation

In general,

Figure 10: Rotation

Example 3 (Understand the Linear map T (or) determine T )

Let T : V (F) → W (F) be the linear transformation. Then the linear map T acts as a standard matrix
(coefficient matrix) A shown in the following picture.
 
a11 a12
1. T : R2 → R3 and its associated standard matrix A = a21 a22 
a31 a32
 
3 2 a11 a12 a13
2. T : R → R and its associated standard matrix A =
a21 a22 a23
 
a a12
3. T : R2 → R2 and its associated standard matrix A = 11
a21 a22

8
Solution:

Figure 11: liner map T and its associated standard matrix A

 
a11 a12 ··· a1n
 a21 a22 ··· a2n 
In general, if T : Rn → Rm be a linear map, then its associated standard matrix is A =  .
 
.. .. .. 
 .. . . . 
am1 am2 ··· amn

Figure 12: liner map T and its associated standard matrix A

9
1. Transformation: Reflections
I - Reflection through the x1 -axis:
The standard matrix is given by  
1 0
A=
0 −1
and the image of the unit square will be

Figure 13: A reflection transformation through the x1 -axis

II - Reflection through the x2 -axis:

The standard matrix is given by  
−1 0
A=
0 1
and the image of the unit square will be

Figure 14: A reflection transformation through the x2 -axis

10
III - Reflection through the x2 = x1 :
The standard matrix is given by  
0 1
A=
1 0
and the image of the unit square will be

Figure 15: A reflection transformation through the x2 = x1

IV - Reflection through the x2 = −x1 :

The standard matrix is given by  
0 −1
A=
−1 0
and the image of the unit square will be

Figure 16: A reflection transformation through the x2 = −x1

11
V - Reflection through the origin:
The standard matrix is given by  
−1 0
A=
0 −1
and the image of the unit square will be

Figure 17: A reflection transformation through the origin

2. Transformation: Contractions and Expansions

Alternative names for contractions: Reduction, Shrinkage, Compression, Tightening, Shortening.
Alternative names for expansions: Increase, Enlargement, Stretching, Dilation.

I - Horizontal contraction and expansion:

The standard matrix is given by
 
k 0
A= , ∀ 0 < k < 1 (Horizontal Contraction), ∀ k > 1 (Horizontal Expansion)
0 1

and the image of the unit square will be

Figure 18: Horizontal contraction and expansion

12
II - Vertical contraction and expansion:
The standard matrix is given by
 
1 0
A= , ∀ 0 < k < 1 (Vertical Contraction), ∀ k > 1 (Vertical Expansion)
0 k

and the image of the unit square will be

Figure 19: Vertical contraction and expansion

3. Transformation: Shears/Scalings
I - Horizontal shear:
The standard matrix is given by
 
1 k
A= , ∀ k < 0 (leftward shear), ∀ k > 0 (rightward shear)
0 1

and the image of the unit square will be

Figure 20: Horizontal shear

13
II - Vertical shear:
The standard matrix is given by
 
1 0
A= , ∀ k < 0 (downward shear), ∀ k > 0 (upward shear)
k 0

and the image of the unit square will be

Figure 21: Vertical shear

4. Transformation: Projections
I - Projection onto the x1 -axis:
The standard matrix is given by  
1 0
A=
0 0
and the image of the unit square will be

Figure 22: Projection onto the x1 -axis

14
II - Projection onto the x2 -axis:
The standard matrix is given by  
0 0
A=
0 1
and the image of the unit square will be

Figure 23: Projection onto the x2 -axis

15
 Matrix Representation of a Linear
Transformation
Ordered basis/Coordinate vector
An ordered basis for a vector space is a basis endowed with a specific order. Let V be a vector space
of dimension n with an ordered basis α = {v1 , . . . , vn }. Let β = {e1 , . . . , en } be the standard basis for
Rn in this order. Then clearly the linear transformation Φ defined by Φ(vi ) = ei is an isomorphism
Pn from
V to Rn , called the natural isomorphism with respect to the basis α. Now for any x = i=1 ai vi ∈ V ,
the image of x under this natural isomorphism is written as
 
n n a1
X X  .. 
Φ(x) = ai Φ(vi ) = ai ei = (a1 , . . . , an ) =  .  ∈ Rn ,
i=1 i=1 an
which is called the coordinate vector of x with respect to the basis α, and is denoted by [x]α (= Φ(x)).
Clearly [vi ]α = ei .

Theorem
Let α = {v1 , v2 , . . . , vn } and β = {w1 , w2 , . . . , wm } be the bases for the vector spaces V and W respec-
tively, with dim V = n and dim W = m.
Let T : V → W be a linear transformation defined by

T (vi ) = wj , for i = 1, 2, . . . , n and j = 1, 2, . . . , m

Then,

T (v1 ) = a11 w1 + a21 w2 + a31 w3 + · · · + am1 wm

T (v2 ) = a12 w1 + a22 w2 + a32 w3 + · · · + am2 wm
..
.
T (vn ) = a1n w1 + a2n w2 + a3n w3 + · · · + amn wm

 
a11 a12 ··· a1n
 a21 a22 ··· a2n 
[T ]βα =  .
 
.. .. .. 
 .. . . . 
am1 am2 ··· amn

ASSOCIATED MATRIX from α to β = [T ]βα

If V = W , then [T ]βα = [T ]α
α = [T ]α

The following commutative diagram illustrates the concept of matrix representation of linear trans-
formations:

Suppose that V is an n-dimensional vector space, that W is an m-dimensional vector space, and that
T : V → W is a linear transformation. Suppose further that B is a basis for V , that B ′ is a basis for W ,
and that for each vector x in V , the coordinate vectors for x and T (x) are [x]B and [T (x)]B ′ , respectively.

16
 Figure 24: Commutative Diagram - 1

Figure 25: Commutative Diagram - 2

Example 1
Let T : P1 (R) → P2 (R) be a linear transformation
defined by
T (p(x)) = xp(x)

Find the associated matrix [T ]βα with respect to the bases

α = {1, x} & β = {1, x, x2 } for P1 (R) & P2 (R) respectively.

Solution:
Soln:
Given T (p(x)) = x · p(x)
Since we know that T : V → W by T (vi ) = wi
T (1) = x · 1 = x = 0 · 1 + 1 · x + 0 · x2
T (x) = x · x = x2 = 0 · 1 + 0 · x + 1 · x2
∴ The associated matrix is
 
0 0
[T ]βα = 1 0
0 1

17
Example 2
L.T.
Given T : R2 −−−→ R2 by

T (1, 1) = (0, 1) & T (−1, 1) = (2, 3)

.

• (i) Find a matrix representation [T ]α w.r.t the standard basis.

• (ii) Find a matrix representation [T ]β w.r.t the basis β = {v1 = (0, 1), v2 = (2, 3)}.

Solution:
(i) To find [T ]α w.r.t the standard basis:

Since, we know that the standard basis in R2 is α = {e1 = (1, 0), e2 = (0, 1)}

Method-I
Given that,

v1 = (1, 1) v2 = (−1, 1)
w1 = (0, 1) w2 = (2, 3)

Finding (x, y) in R2 :

(x, y) = av1 + bv2

= (a, a) + (−b, b)
(x, y) = (a − b, a + b)

From this system:

a−b=x ...(1)
a+b=y ...(2)

Adding equations (1) + (2):

2a = x + y
x+y
a=
2
Substituting in (1):
x+y
−b=x
2
−2b = 2x − (x + y)
−2b = x − y
y−x
b=
2
Therefore:
   
x+y y−x
(x, y) = v1 + v2
2 2

18
Hence, the formula for T (x, y) is:
   
x+y y−x
T (x, y) = T (v1 ) + T (v2 )
2 2
   
x+y y−x
= w1 + w2
2 2
   
x+y y−x
= (0, 1) + (2, 3)
2 2
   
x+y y−x y−x
= 0, + · 2, ·3
2 2 2
   
x+y 3(y − x)
= 0, + y − x,
2 2
 
x + y + 3(y − x)
= y − x,
2
 
x + y + 3y − 3x
= y − x,
2
 
−2x + 4y
= y − x,
2
= (y − x, −x + 2y)

=⇒ T (x, y) = (y − x, −x + 2y) (∗)

Now,

T (e1 ) = T (1, 0) = (0 − 1, −1 + 2(0)) = (−1, −1) = (−1) · e1 + (−1) · e2

T (e2 ) = T (0, 1) = (1 − 0, −1(0) + 2(1)) = (1, 2) = (1) · e1 + (2) · e2
 
−1 1
∴ [T ]α =
−1 2

Method-II (Alternatively)
Let
(1, 1) = 1(1, 0) + 1(0, 1) = 1 · e1 + 1 · e2 = e1 + e2
Now, apply T on both side,
T (1, 1) = T (e1 + e2 ) [∵ T is linear]
=⇒ (0, 1) = T (e1 ) + T (e2 )
=⇒ e2 = T (e1 ) + T (e2 ) (1)
Let
(−1, 1) = −1(1, 0) + 1(0, 1) = −1 · e1 + 1 · e2 = −e1 + e2
Now, apply T on both side,
T (−1, 1) = T (−e1 + e2 ) [∵ T is linear]
=⇒ (2, 3) = −T (e1 ) + T (e2 )
=⇒ −T (e1 ) + T (e2 ) = 2e1 + 3e2 (2)
Solving (1) & (2), we obtain

T (e1 ) = −e1 − e2
T (e2 ) = e1 + 2e2
 
−1 1
∴ [T ]α =
−1 2

19
 (ii) To find [T ]β w.r.t the basis β = {v1 = (0, 1), v2 = (2, 3)}:

Given that,

v1 = (0, 1), v2 = (2, 3)

w1 = T (v1 ) = T (0, 1) = (1, 2) = [T (v1 )]α , w2 = T (v2 ) = T (2, 3) = (1, 4) = [T (v2 )]α from Eq.(*)

Now,:

(1, 2) = av1 + bv2 = a(0, 1) + b(2, 3)

= (0, a) + (2b, 3b)
(1, 2) = (2b, a + 3b) (I)

similarly,

(1, 4) = cv1 + dv2 = c(0, 1) + d(2, 3)

= (0, c) + (2d, 3d)
(1, 4) = (2d, a + 3d) (II)

from Eqs.(I) & (II):

2b = 1,
a + 3b = 2,
2d = 1,
c + 3d = 4
Solving for a, b, c and d, we obtain
=⇒ b = 1/2
a = 2 − 3/2 = 1/2
d = 1/2
c = 4 − 3/2 = 5/2
Now,        
a 1 1 c 1 5
[T (v1 )]β = = , and [T (v2 )]β = = .
b 2 1 d 2 1
  1 5  
a c 2 2
1 1 5
Therefore, [T ]β = [T (v1 ) T (v2 )]β = = 1 1 = .
b d 2 2 2 1 1

Example 3
Let T : P3 → R3 be the linear transformation given by

T (ax3 + bx2 + cx + d) = (4a − b + 3c + 3d, a + 3b − c + 5d, −2a − 7b + 5c − d).

Find the matrix for T = [T ]βα with respect to the standard bases α = {1, x, x2 , x3 } for P3 and β =
{e1 , e2 , e3 } for R3 .

20
Solution:
Given that:
• Domain: P3 (polynomials of degree ≤ 3)
• Codomain: R3
• Transformation: T (ax3 + bx2 + cx + d) = (4a − b + 3c + 3d, a + 3b − c + 5d, −2a − 7b + 5c − d)
• Basis for P3 : α = {1, x, x2 , x3 } (standard basis)
• Basis for R3 : β = {e1 , e2 , e3 } (standard basis)
Step 1: Apply T to Each Basis Element of α
To find the matrix representation, we apply T to each basis element and express the results in terms
of the basis β.
T (1) - First basis element:
For the polynomial 1: a = 0, b = 0, c = 0, d = 1

T (1) = T (0 · x3 + 0 · x2 + 0 · x + 1)
= (4(0) − 0 + 3(0) + 3(1), 0 + 3(0) − 0 + 5(1), −2(0) − 7(0) + 5(0) − 1)
= (3, 5, −1)

T (x) - Second basis element:

For the polynomial x: a = 0, b = 0, c = 1, d = 0

T (x) = T (0 · x3 + 0 · x2 + 1 · x + 0)
= (4(0) − 0 + 3(1) + 3(0), 0 + 3(0) − 1 + 5(0), −2(0) − 7(0) + 5(1) − 0)
= (3, −1, 5)

T (x2 ) - Third basis element:

For the polynomial x2 : a = 0, b = 1, c = 0, d = 0

T (x2 ) = T (0 · x3 + 1 · x2 + 0 · x + 0)
= (4(0) − 1 + 3(0) + 3(0), 0 + 3(1) − 0 + 5(0), −2(0) − 7(1) + 5(0) − 0)
= (−1, 3, −7)

T (x3 ) - Fourth basis element:

For the polynomial x3 : a = 1, b = 0, c = 0, d = 0

T (x3 ) = T (1 · x3 + 0 · x2 + 0 · x + 0)
= (4(1) − 0 + 3(0) + 3(0), 1 + 3(0) − 0 + 5(0), −2(1) − 7(0) + 5(0) − 0)
= (4, 1, −2)

Step 2: Express Results in Terms of Basis β

Since β = {e1 , e2 , e3 } is the standard basis for R3 , we have:

e1 = (1, 0, 0)
e2 = (0, 1, 0)
e3 = (0, 0, 1)

The coordinates are already in standard form:

T (1) = (3, 5, −1) = 3e1 + 5e2 + (−1)e3

T (x) = (3, −1, 5) = 3e1 + (−1)e2 + 5e3
T (x2 ) = (−1, 3, −7) = (−1)e1 + 3e2 + (−7)e3
T (x3 ) = (4, 1, −2) = 4e1 + 1e2 + (−2)e3

Step 3: Construct the Matrix

21
 The matrix representation [T ]βα has columns formed by the coordinate vectors of T applied to each
basis element of α.

Column 1: Coordinates of T (1) = [3, 5, −1]T

Column 2: Coordinates of T (x) = [3, −1, 5]T
Column 3: Coordinates of T (x2 ) = [−1, 3, −7]T
Column 4: Coordinates of T (x3 ) = [4, 1, −2]T

Final Answer:
The matrix representation of T with respect to the given bases is:
 
3 3 −1 4
[T ]βα =  5 −1 3 1
−1 5 −7 −2

Example 4
Consider the vector space M2×2 (R) (all 2 × 2 matrices over the field of real numbers R).  
1 1
Let T : M2×2 (R) → M2×2 (R) be the linear transformation defined by T (X) = AX, where A = .
1 1
Find the matrix of T with respect to the ordered basis B = {E1 , E2 , E3 , E4 } for M2×2 (R) where:
       
1 0 0 1 0 0 0 0
E1 = , E2 = , E3 = , E4 =
0 0 0 0 1 0 0 1

Solution:
 
1 1
Given: T : M2×2 (R) → M2×2 (R) defined by T (X) = AX where A = .
1 1
To find the matrix representation of T , we need to compute T (Ei ) for each basis element and express
the result as a linear combination of the basis elements.

Step 1: Compute T (E1 ) = AE1

    
1 1 1 0 1 0
T (E1 ) = AE1 = =
1 1 0 0 1 0
Express as linear combination of basis elements:
         
1 0 1 0 0 1 0 0 0 0
T (E1 ) = =1· +0· +1· +0·
1 0 0 0 0 0 1 0 0 1

T (E1 ) = 1 · E1 + 0 · E2 + 1 · E3 + 0 · E4

Step 2: Compute T (E2 ) = AE2

    
1 1 0 1 0 1
T (E2 ) = AE2 = =
1 1 0 0 0 1
Express as linear combination of basis elements:
         
0 1 1 0 0 1 0 0 0 0
T (E2 ) = =0· +1· +0· +1·
0 1 0 0 0 0 1 0 0 1

T (E2 ) = 0 · E1 + 1 · E2 + 0 · E3 + 1 · E4

22
Step 3: Compute T (E3 ) = AE3
    
1 1 0 0 1 0
T (E3 ) = AE3 = =
1 1 1 0 1 0
Express as linear combination of basis elements:
         
1 0 1 0 0 1 0 0 0 0
T (E3 ) = =1· +0· +1· +0·
1 0 0 0 0 0 1 0 0 1

T (E3 ) = 1 · E1 + 0 · E2 + 1 · E3 + 0 · E4

Step 4: Compute T (E4 ) = AE4

    
1 1 0 0 0 1
T (E4 ) = AE4 = =
1 1 0 1 0 1
Express as linear combination of basis elements:
         
0 1 1 0 0 1 0 0 0 0
T (E4 ) = =0· +1· +0· +1·
0 1 0 0 0 0 1 0 0 1

T (E4 ) = 0 · E1 + 1 · E2 + 0 · E3 + 1 · E4

Step 5: Construct the Matrix Representation

The matrix representation [T ]B has columns that are the coordinate vectors of T (Ei ) with respect to the
basis B:
From our calculations:
• T (E1 ) = 1 · E1 + 0 · E2 + 1 · E3 + 0 · E4 → Column 1: (1, 0, 1, 0)T
• T (E2 ) = 0 · E1 + 1 · E2 + 0 · E3 + 1 · E4 → Column 2: (0, 1, 0, 1)T

• T (E3 ) = 1 · E1 + 0 · E2 + 1 · E3 + 0 · E4 → Column 3: (1, 0, 1, 0)T

• T (E4 ) = 0 · E1 + 1 · E2 + 0 · E3 + 1 · E4 → Column 4: (0, 1, 0, 1)T
Therefore, the matrix representation is:
 
1 0 1 0
0 1 0 1
[T ]B = 
1

0 1 0
0 1 0 1

Final Answer
The matrix of the linear transformation T with respect to the ordered basis B = {E1 , E2 , E3 , E4 } is:
 
1 0 1 0
0 1 0 1
[T ]B = 
1 0

1 0
0 1 0 1

Example 5
Consider the vector space M2×2 (R) (all 2 × 2 matrices over the field of real numbers R).  
1 1
Let T : M2×2 (R) → M2×2 (R) be the linear transformation defined by T (X) = AX, where A = .
1 1
Find the matrix of T with respect to the ordered basis B = {S1 , S2 , S3 , S4 } for M2×2 (R) where:
       
2 0 1 4 0 1 0 1
S1 = , S2 = , S3 = , S4 =
0 3 0 1 3 2 2 0

23
Solution:
Step 1: Compute T (S1 ) = AS1
    
1 1 2 0 2 3
T (S1 ) = AS1 = =
1 1 0 3 2 3
Express as linear combination of basis elements:
         
2 3 2 0 1 4 0 1 0 1
T (S1 ) = = c1 + c2 + c3 + c4
2 3 0 3 0 1 3 2 2 0
This gives us the system:
• 2c1 + c2 = 2
• 4c2 + c3 + c4 = 3
• 3c3 + 2c4 = 2
• 3c1 + c2 + 2c3 = 3
Solving: c1 = 1, c2 = 0, c3 = 0, c4 = 1

T (S1 ) = 1 · S1 + 0 · S2 + 0 · S3 + 1 · S4

Step 2: Compute T (S2 ) = AS2

    
1 1 1 4 1 5
T (S2 ) = AS2 = =
1 1 0 1 1 5
Express as linear combination of basis elements:
         
1 5 2 0 1 4 0 1 0 1
T (S2 ) = = c1 + c2 + c3 + c4
1 5 0 3 0 1 3 2 2 0
Solving: c1 = 0, c2 = 1, c3 = 0, c4 = 1

T (S2 ) = 0 · S1 + 1 · S2 + 0 · S3 + 1 · S4

Step 3: Compute T (S3 ) = AS3

    
1 1 0 1 3 3
T (S3 ) = AS3 = =
1 1 3 2 3 3
Express as linear combination of basis elements:
Solving: c1 = 1, c2 = 0, c3 = 1, c4 = 0

T (S3 ) = 1 · S1 + 0 · S2 + 1 · S3 + 0 · S4

Step 4: Compute T (S4 ) = AS4

    
1 1 0 1 2 1
T (S4 ) = AS4 = =
1 1 2 0 2 1
Express as linear combination of basis elements:
Solving: c1 = 1, c2 = 0, c3 = 0, c4 = 0.5

T (S4 ) = 1 · S1 + 0 · S2 + 0 · S3 + 0.5 · S4

Matrix of T with respect to basis B:

 
1 0 1 1
0 1 0 0
[T ]B = 
0

0 1 0
1 1 0 0.5

24
Example 6
Find the matrix representation of the L.T. T : R3 → R3 defined by

T (x, y, z) = (2x − 3y + 4z, 5x − y + 2z, 4x + 7y)

w.r.t the bases

α = {e1 , e2 , e3 } & β = {e3 , e2 , e1 }
.
Find [T ]α , [T ]β & [T ]βα .

Solution:

Given Information
• Linear transformation T : R3 → R3
• T (x, y, z) = (2x − 3y + 4z, 5x − y + 2z, 4x + 7y)

• Standard basis: α = {e1 , e2 , e3 }

• Different (ordered) basis: β = {e3 , e2 , e1 }

Step 1: Find [T ]α (Matrix with respect to basis α)

Apply T to each vector in α = {e1 , e2 , e3 }:
T (e1 ) = T (1, 0, 0) = (2, 5, 4) = [T (e1 )]α
T (e2 ) = T (0, 1, 0) = (−3, −1, 7) = [T (e2 )]α
T (e3 ) = T (0, 0, 1) = (4, 2, 0) = [T (e3 )]α
 
  2 −3 4
[T ]α = [T (e1 )]α [T (e2 )]α [T (e3 )]α = 5 −1 2
4 7 0

Step 2: Find [T ]β (Matrix with respect to basis β)

Apply T to each vector in β = {e3 , e2 , e1 }:
T (e3 ) = T (0, 0, 1) = (4, 2, 0)
T (e2 ) = T (0, 1, 0) = (−3, −1, 7)
T (e1 ) = T (1, 0, 0) = (2, 5, 4)
Now express these results as coordinates with respect to β = {e3 , e2 , e1 }:
For T (e3 ) = (4, 2, 0) = 0 · e3 + 2 · e2 + 4 · e1 → [T (e3 )]β = (0, 2, 4)
For T (e2 ) = (−3, −1, 7) = 7 · e3 + (−1) · e2 + (−3) · e1 → [T (e2 )]β = (7, −1, −3)
For T (e1 ) = (2, 5, 4) = 4 · e3 + 5 · e2 + 2 · e1 → [T (e1 )]β = (4, 5, 2)
 
  0 7 4
[T ]β = [T (e3 )]β [T (e2 )]β [T (e1 )]β = 2 −1 5
4 −3 2

Step 3: Find [T ]βα (Matrix from α to β)

This represents T with input basis α and output basis β.
Apply T to each vector in α = {e1 , e2 , e3 } and express results in β coordinates:
T (e1 ) = (2, 5, 4) expressed in β = {e3 , e2 , e1 }: (2, 5, 4) = 4 · e3 + 5 · e2 + 2 · e1 → [T (e1 )]β = (4, 5, 2)
T (e2 ) = (−3, −1, 7) expressed in β = {e3 , e2 , e1 }: (−3, −1, 7) = 7·e3 +(−1)·e2 +(−3)·e1 → [T (e2 )]β =
(7, −1, −3)
T (e3 ) = (4, 2, 0) expressed in β = {e3 , e2 , e1 }: (4, 2, 0) = 0 · e3 + 2 · e2 + 4 · e1 → [T (e3 )]β = (0, 2, 4)
 
4 7 0
[T ]βα = [T (e1 )]β [T (e2 )]β [T (e3 )]β = 5 −1 2
 
2 −3 4

25
Final Answer
     
2 −3 4 0 7 4 4 7 0
[T ]α = 5 −1 2 [T ]β = 2 −1 5 [T ]βα = 5 −1 2
4 7 0 4 −3 2 2 −3 4
The key difference is that β has a different ordering of the standard basis vectors, which affects how
we express the transformation results as coordinate vectors.

Exercises
1. Let T : R2 → R3 be the linear transformation defined by
T (x, y) = (x + 2y, 0, 2x + 3y) ∀x, y ∈ R.
Find the associated matrix for T w.r.t the standard bases.
Ans:  
1 2
[T ]βα = 0 0
2 3
2. Find the matrix representations [T ]α , [T ]β and [T ]βα of each of the following linear transformations T
of R3 with respect to the standard basis α = {e1 , e2 , e3 }, and β = {e3 , e2 , e1 }:

(1) T (x, y, z) = (2x − 3y + 4z, 5x − y + 2z, 4x + 7y),

(2) T (x, y, z) = (2y + z, x − 4y, 3x).

3. Let T : R4 → R3 be the linear transformation defined by

T (x, y, z, u) = (x + 2y, x − 3z + u, 2y + 3z + 4u).

Let α and β be the standard bases for R4 and R3 , respectively. Find [T ]βα .
4. Let T : R3 → R2 be defined by T (x, y, z) = (3x + 2y − 4z, x − 5y + 3z).
Find [T ]βα for α = {(1, 1, 1), (1, 1, 0), (1, 0, 0)} &
β = {(1, 3), (2, 5)}.

26
 Problem (Analogue to 6.1.8)

Let

2 −1 3 0 5
B=[ ]
1 0 −2 4 1

Define S : R5 → R2 by S(x) = Bx.

. Compute S(2, 1, −1, 0, 3).

. Compute the preimage under S of (9, 2); that is, find all x ∈ R5 such that S(x) = (9, 2).

Solution

1. Compute S(2, 1, −1, 0, 3):

Matrix multiplication:

2
1
2 −1 3 0 5
S(2, 1, −1, 0, 3) = [ ] −1
1 0 −2 4 1
0
3

First row:

2 ⋅ 2 + (−1) ⋅ 1 + 3 ⋅ (−1) + 0 ⋅ 0 + 5 ⋅ 3 = 4 − 1 − 3 + 0 + 15 = 15

Second row:

1 ⋅ 2 + 0 ⋅ 1 + (−2) ⋅ (−1) + 4 ⋅ 0 + 1 ⋅ 3 = 2 + 0 + 2 + 0 + 3 = 7
So,

S(2, 1, −1, 0, 3) = (15, 7)

2. Compute all x = (x1 , x2 , x3 , x4 , x5 ) such that S(x) = (9, 2):

Set up the system:

2x1 − x2 + 3x3 + 0x4 + 5x5 = 9

{
x1 + 0x2 − 2x3 + 4x4 + x5 = 2

Write as augmented matrix:

2 −1 3 0 5 ∣ 9
[ ]
1 0 −2 4 1 ∣ 2

Let's solve:
Let x2 = a, x4 = b, x5 = c be free variables.
Express x1 and x3 in terms of a, b, c:

From equation 2:

x1 − 2x3 + 4b + c = 2 ⟹ x1 = 2 + 2x3 − 4b − c

From equation 1:

2x1 − a + 3x3 + 5c = 9

Substitute x1 from above:

 2(2 + 2x3 − 4b − c) − a + 3x3 + 5c = 9

4 + 4x3 − 8b − 2c − a + 3x3 + 5c = 9

4 + 7x3 − 8b + 3c − a = 9

7x3 − 8b + 3c − a = 5
5 + 8b − 3c + a
x3 =
7
Now back to x1 :

x1 = 2 + 2x3 − 4b − c

Substitute for x3 :

5 + 8b − 3c + a 10 + 16b − 6c + 2a
x1 = 2 + 2 ( ) − 4b − c = 2 + − 4b − c
7 7

Combine:

14 10 + 16b − 6c + 2a 24 + 16b − 6c + 2a
x1 = + − 4b − c = − 4b − c
7 7 7
So the general solution is:
 24 + 16b − 6c + 2a
x1 = − 4b − c
7
x2 =a
5 + 8b − 3c + a
x3 =
7
x4 =b
x5 =c

where a, b, c ∈ R.

Final Answer (boxed for clarity)

Any x ∈ R5 such that:

24 + 16b − 6c + 2a
x1 = − 4b − c
7
x2 =a
5 + 8b − 3c + a
x3 =
7
x4 =b
x5 =c

where a, b, c ∈ R, is a solution to S(x) = (9, 2).

 Vector space of a Linear Transformation
Syllabus
Vector space of linear transformation - Change of bases - Similarity of
linear transformation
Definition
Let V and W be vector spaces and L(V ; W ) denote the set of all linear transformations from V to W ,

L(V ; W ) = {T : T is a Linear Transformation from V to W }

Let S, T ∈ L(V ; W ) and α ∈ R, then

(S + T )(v) = S(v) + T (v)

(αS)(v) = αS(v) for any v ∈ V

Note that: L(V ; W ) is a vector space.

Definition
Let dim V = n and dim W = m with fixed ordered bases α and β respectively. Then L(V ; W ) can be
identified with the vector space Mm×n (R) of all (m × n) matrices, so that

dim L(V ; W ) = dim Mm×n (R) = mn = dim V · dim W.

Thoerem 1
Let V and W be vector spaces with ordered bases α and β, respectively, and let S, T : V → W be linear.
Then we have
[S + T ]βα = [S]βα + [T ]βα and [kS]βα = k[S]βα .

Theorem 2
Let V , W and Z be vector spaces and S : V → W , T : W → Z are linear transformations, then the
composition (T ◦ S) : V → Z is also linear.

Theorem 3
Let V , W and Z be V.S. with ordered bases α, β and γ respectively. Suppose that S : V → W and
T : W → Z are linear transformations then

[T ◦ S]γα = [T ]γβ · [S]βα .

Theorem 4
Let V and W be vector spaces with ordered bases α and β, respectively, and let T : V → W be an
isomorphism. Then
[T −1 ]α β −1
β = ([T ]α ) .

Example 1
Let α be the standard basis for R3 , and let S, T : R3 → R3 be two linear transformations given by

S(e1 ) = (2, 2, 1), S(e2 ) = (0, 1, 2), S(e3 ) = (−1, 2, 1),

T (e1 ) = (1, 0, 1), T (e2 ) = (0, 1, 1), T (e3 ) = (1, 1, 2).

Compute [S + T ]α , [2T − S]α and [T ◦ S]α .

1
Solution:
Step 1: Find the matrix representations [S]α and [T ]α .
Since α is the standard basis for R3 , the columns of [S]α are the images of the standard basis vectors
under S:  
2 0 −1
[S]α = 2 1 2 
1 2 1
Similarly, the columns of [T ]α are the images of the standard basis vectors under T :
 
1 0 1
[T ]α = 0 1 1
1 1 2

Step 2: Compute [S + T ]α .
By Theorem 1, [S + T ]α = [S]α + [T ]α :
     
2 0 −1 1 0 1 3 0 0
[S + T ]α = 2 1
 2  + 0 1 1 = 2 2 3
1 2 1 1 1 2 2 3 3

Step 3: Compute [2T − S]α .

By Theorem 1, [2T − S]α = 2[T ]α − [S]α :
   
1 0 1 2 0 −1
[2T − S]α = 2 0 1 1 − 2 1 2
1 1 2 1 2 1
     
2 0 2 2 0 −1 0 0 3
= 0 2 2 − 2 1 2  = −2 1 0
2 2 4 1 2 1 1 0 3
Step 4: Compute [T ◦ S]α .
By the composition theorem, [T ◦ S]α = [T ]α · [S]α :
  
1 0 1 2 0 −1
[T ◦ S]α = 0 1 1 2 1 2
1 1 2 1 2 1

Therefore:  
3 2 0
[T ◦ S]α = 3 3 3
6 5 3

Example 2
Let S, T : R2 → R2 be defined by

S(x, y) = (x + 3y, 2x)

T (x, y) = (y, x + 2y)

Let α = {(1, 1), (1, 2)} be any basis for R2 , then find [S + T ]α , [S ◦ T ]α and [2T − 3S]α .

Solution:
Step 1: Find the images of the basis vectors under S and T .
For S:

S(1, 1) = (1 + 3(1), 2(1)) = (4, 2)

S(1, 2) = (1 + 3(2), 2(1)) = (7, 2)

2
 For T :
T (1, 1) = (1, 1 + 2(1)) = (1, 3)
T (1, 2) = (2, 1 + 2(2)) = (2, 5)
Step 2: Express the images as linear combinations of the basis vectors.
We need to express each image vector as c1 (1, 1) + c2 (1, 2) = (c1 + c2 , c1 + 2c2 ).
For S(1, 1) = (4, 2):
c1 + c2 = 4
c1 + 2c2 = 2
Subtracting: −c2 = 2, so c2 = −2 and c1 = 6. Therefore, S(1, 1) = 6(1, 1) − 2(1, 2).
For S(1, 2) = (7, 2):
c1 + c2 = 7
c1 + 2c2 = 2
Subtracting: −c2 = 5, so c2 = −5 and c1 = 12. Therefore, S(1, 2) = 12(1, 1) − 5(1, 2).
For T (1, 1) = (1, 3):
c1 + c2 = 1
c1 + 2c2 = 3
Subtracting: c2 = 2 and c1 = −1. Therefore, T (1, 1) = −1(1, 1) + 2(1, 2).
For T (1, 2) = (2, 5):
c1 + c2 = 2
c1 + 2c2 = 5
Subtracting: c2 = 3 and c1 = −1. Therefore, T (1, 2) = −1(1, 1) + 3(1, 2).
Step 3: Construct the matrix representations.
   
6 12 −1 −1
[S]α = , [T ]α =
−2 −5 2 3
Step 4: Compute [S + T ]α .
By Theorem 1, [S + T ]α = [S]α + [T ]α :
     
6 12 −1 −1 5 11
[S + T ]α = + =
−2 −5 2 3 0 −2
Step 5: Compute [S ◦ T ]α .
By the composition theorem, [S ◦ T ]α = [S]α · [T ]α :
  
6 12 −1 −1
[S ◦ T ]α =
−2 −5 2 3
Therefore:  
18 30
[S ◦ T ]α =
−8 −13
Step 6: Compute [2T − 3S]α .
By Theorem 1, [2T − 3S]α = 2[T ]α − 3[S]α :
   
−1 −1 6 12
[2T − 3S]α = 2 −3
2 3 −2 −5
     
−2 −2 18 36 −20 −38
= − =
4 6 −6 −15 10 21
Final Answers:
 
5 11
[S + T ]α =
0 −2
 
18 30
[S ◦ T ]α =
−8 −13
 
−20 −38
[2T − 3S]α =
10 21

3
Example 3
Let T : P2 (R) → P2 (R) be the linear transformation defined by T (f ) = (3+x)f ′ +2f , and S : P2 (R) → R3
defined by S(a + bx + cx2 ) = (a − b, a + b, c). For a basis α = {1, x, x2 } for P2 (R) and the standard basis
β = {e1 , e2 , e3 } for R3 , compute [S]βα , [T ]α , and [S ◦ T ]βα .

Solution:
Step 1: Compute [S]βα .
We need to find S(1), S(x), and S(x2 ):

S(1) = S(1 + 0x + 0x2 ) = (1 − 0, 1 + 0, 0) = (1, 1, 0) = 1e1 + 1e2 + 0e3

S(x) = S(0 + 1x + 0x2 ) = (0 − 1, 0 + 1, 0) = (−1, 1, 0) = −1e1 + 1e2 + 0e3
S(x2 ) = S(0 + 0x + 1x2 ) = (0 − 0, 0 + 0, 1) = (0, 0, 1) = 0e1 + 0e2 + 1e3

Therefore:  
1 −1 0
[S]βα = 1 1 0
0 0 1
Step 2: Compute [T ]α .
We need to find T (1), T (x), and T (x2 ):

T (1) = (3 + x)(1)′ + 2(1) = (3 + x)(0) + 2 = 2 = 2 · 1 + 0 · x + 0 · x2

T (x) = (3 + x)(x)′ + 2(x) = (3 + x)(1) + 2x = 3 + x + 2x = 3 + 3x = 3 · 1 + 3 · x + 0 · x2
T (x2 ) = (3 + x)(x2 )′ + 2(x2 ) = (3 + x)(2x) + 2x2 = 6x + 2x2 + 2x2 = 6x + 4x2 = 0 · 1 + 6 · x + 4 · x2

Therefore:  
2 3 0
[T ]α = 0 3 6
0 0 4
Step 3: Compute [S ◦ T ]βα .
By the composition theorem, [S ◦ T ]βα = [S]βα · [T ]α :
  
1 −1 0 2 3 0
[S ◦ T ]βα = 1 1 0 0 3 6
0 0 1 0 0 4
Therefore:  
2 0 −6
[S ◦ T ]βα = 2 6 6
0 0 4

Example 4
For the vector spaces P1 (R) and R2 , choose the bases α = {1, x} for P1 (R) and β = {e1 , e2 } for R2 ,
respectively. Let T : P1 (R) → R2 be the linear transformation defined by T (a + bx) = (a, a + b).
1. Show that T is invertible.
2. Find [T ]βα and [T −1 ]α
β.

Solution:
Part (1): Show that T is invertible
To show that T is invertible, we first find its matrix representation [T ]βα and check if its determinant
is non-zero.
Step 1: Find [T ]βα .
We compute T applied to each basis vector in α:

T (1) = T (1 + 0x) = (1, 1 + 0) = (1, 1) = 1e1 + 1e2

T (x) = T (0 + 1x) = (0, 0 + 1) = (0, 1) = 0e1 + 1e2

4
 Therefore:  
1 0
[T ]βα =
1 1
Step 2: Check if T is invertible using determinant.
A linear transformation is invertible if and only if its matrix representation has a non-zero determinant.
 
β 1 0
det([T ]α ) = det = (1)(1) − (0)(1) = 1 ̸= 0
1 1
Since det([T ]βα ) = 1 ̸= 0, the transformation T is invertible.

Part (2): Find [T ]βα and [T −1 ]α

β
From Part (1), we already found:  
1 0
[T ]βα =
1 1
Finding [T −1 ]α
β using adjugate method:
By Theorem 4, [T −1 ]α
β = ([T ]βα )−1 .
 
a b
For a 2 × 2 matrix A = , the inverse is:
c d
1
A−1 = · adj(A)
det(A)
 
d −b
where adj(A) = is the adjugate matrix.
−c a
Step 1: We already found det([T ]βα ) = 1.
β
Step 2: Find  the adjugate of [T ]α .
1 0
For [T ]βα = :
1 1
 
1 0
adj([T ]βα ) =
−1 1
Step 3: Calculate the inverse.
   
1 1 0 1 0
[T −1 ]α β −1
β = ([T ]α ) = · =
1 −1 1 −1 1
Verification:
We can verify our answer by checking that [T ]βα · [T −1 ]α
β = I2 :
    
1 0 1 0 1 0
= = I2 ✓
1 1 −1 1 0 1
Final Answers:    
1 0 1 0
[T ]βα = , [T −1 ]α
β =
1 1 −1 1

Exercises
1. Let S, T : R3 → R3 and S(x, y, z) = (2x + 3y, x + y + z, 2z), T (x, y, z) = (y − 2z, x + 3z, 3x − 2y) with
α is a standard basis. Find [S + T ]α , [T ◦ S]α and [2S − T ]α .
   
2 3 0 0 1 −2
Solution: [S]α = 1 1 1, [T ]α = 1 0 3
0 0 2 3 −2 0
2. Let T : R3 → R3 is a linear transformation defined by T (x, y, z) = (2y + x, x − 4y, 3x + y). Find
[T ]α and [T ]β of α = {e1 , e2 , e3 } and β = {(1, 1, 1), (1, 1, 0), (1, 0, 0)}, where e1 , e2 , e3 are standard basis
in R3 .    
1 2 0 4 4 3
Solution: [T ]α = 1 −4 0, [T ]β = −7 −7 −2
3 1 0 6 6 3

5
 Change of Bases
Definition: Transition Matrix (Coordinate Change Matrix)
Let V be a finite-dimensional vector space with two ordered bases α = {v1 , v2 , . . . , vn } and β =
{w1 , w2 , . . . , wn }. The transition matrix (or coordinate change matrix) from basis α to basis β, de-
noted Pα→β , is the n × n matrix such that for any vector v ∈ V :

[v]β = Pα→β [v]α

where [v]α and [v]β are the coordinate vectors of v with respect to bases α and β, respectively.

Construction of Transition Matrix:

The transition matrix Pα→β can be constructed in two ways:

Method 1: Direct Method

The columns of Pα→β are the coordinate vectors of the basis vectors in α expressed with respect to basis
β. That is:  
Pα→β = [v1 ]β [v2 ]β · · · [vn ]β
To find each column [vi ]β , solve the system:

vi = c1i w1 + c2i w2 + · · · + cni wn

where (c1i , c2i , . . . , cni )T = [vi ]β .

Method 2: Alternative Method (Matrix Formula)

If the vector space is embedded in Rn (or can be identified with it), then:

Pα→β = B −1 A

Notice that: To find B −1 ,

adj(B)
Use: either Gauss-Jordan method for n > 3 (or) B −1 = |B|
where:
• A is the n × n matrix with basis vectors of α as columns
• B is the n × n matrix with basis vectors of β as columns

Properties:
1. Pα→β is always invertible (non-singular)
2. (Pα→β )−1 = Pβ→α

3. Pα→α = In (identity matrix)

4. If γ is another basis, then Pα→γ = Pβ→γ · Pα→β

Example 1
Find the coordinate change matrix (transition matrix) from a basis α = {(2, 3, 1), (1, 2, 0), (2, 0, 3)} to
another basis β = {(1, 0, 1), (1, 1, 0), (0, 1, 1)}.

6
Solution:

Method - I
Given: α = {(2, 3, 1), (1, 2, 0), (2, 0, 3)} and β = {(1, 0, 1), (1, 1, 0), (0, 1, 1)}.
We need to find the transition matrix Pα→β such that for any vector v:
[v]β = Pα→β [v]α
Step 1: Express each vector in α as a linear combination of vectors in β.
We need to find coefficients such that:
(2, 3, 1) = c1 (1, 0, 1) + c2 (1, 1, 0) + c3 (0, 1, 1)
(1, 2, 0) = d1 (1, 0, 1) + d2 (1, 1, 0) + d3 (0, 1, 1)
(2, 0, 3) = e1 (1, 0, 1) + e2 (1, 1, 0) + e3 (0, 1, 1)
For (2, 3, 1):
c1 (1, 0, 1) + c2 (1, 1, 0) + c3 (0, 1, 1) = (2, 3, 1)
(c1 + c2 , c2 + c3 , c1 + c3 ) = (2, 3, 1)
This gives us the system:
c1 + c 2 = 2 (1)
c2 + c 3 = 3 (2)
c1 + c 3 = 1 (3)
From equations (1) and (3): c2 − c3 = 1. Combined with equation (2): c2 + c3 = 3. Adding:
2c2 = 4 ⇒ c2 = 2. Therefore: c3 = 1 and c1 = 0.
So (2, 3, 1) = 0(1, 0, 1) + 2(1, 1, 0) + 1(0, 1, 1).
For (1, 2, 0):
d1 (1, 0, 1) + d2 (1, 1, 0) + d3 (0, 1, 1) = (1, 2, 0)
(d1 + d2 , d2 + d3 , d1 + d3 ) = (1, 2, 0)
This gives us:
d1 + d2 = 1 (4)
d2 + d3 = 2 (5)
d1 + d3 = 0 (6)
From equations (4) and (6): d2 − d3 = 1. Combined with equation (5): d2 + d3 = 2. Adding:
2d2 = 3 ⇒ d2 = 23 . Therefore: d3 = 12 and d1 = − 12 .
So (1, 2, 0) = − 12 (1, 0, 1) + 32 (1, 1, 0) + 12 (0, 1, 1).
For (2, 0, 3):
e1 (1, 0, 1) + e2 (1, 1, 0) + e3 (0, 1, 1) = (2, 0, 3)
(e1 + e2 , e2 + e3 , e1 + e3 ) = (2, 0, 3)
This gives us:
e1 + e 2 = 2 (7)
e2 + e 3 = 0 (8)
e1 + e 3 = 3 (9)
From equations (7) and (9): e2 − e3 = −1. Combined with equation (8): e2 + e3 = 0. Adding:
2e2 = −1 ⇒ e2 = − 21 . Therefore: e3 = 12 and e1 = 25 .
So (2, 0, 3) = 52 (1, 0, 1) + (− 12 )(1, 1, 0) + 12 (0, 1, 1).
Step 2: Construct the transition matrix.
The transition matrix Pα→β has columns that are the coordinate representations of the α basis vectors
with respect to β:

0 − 12 5
 
2
3
Pα→β = 2 2 − 21 
1 1
1 2 2

7
Method - II
Given: α = {(2, 3, 1), (1, 2, 0), (2, 0, 3)} and β = {(1, 0, 1), (1, 1, 0), (0, 1, 1)}.
The transition matrix from α to β is given by Pα→β = B −1 A, where:
• A is the matrix with α basis vectors as columns
• B is the matrix with β basis vectors as columns
Step 1: Form matrices A and B.
   
2 1 2 1 1 0
A = 3 2 0 , B = 0 1 1
1 0 3 1 0 1
Find B −1using the adjoint method.
Step 2: 
1 1 0
For B = 0 1 1:
1 0 1
Determinant: Using the first row expansion:

1 1 0 1 0 1
det(B) = 1 −1 +0
0 1 1 1 1 0

= 1(1) − 1(−1) + 0 = 1 + 1 = 2
Cofactor Matrix:
 
1 1 0 1 0 1
 0 −
1 1 1 1 0   

 1
 1 1 −1
0 1 0 1 1 
 = −1
− 0
Cof(B) = 
1 1 1
−
1 0  1 1

 1
 1 −1 1
0 1 0 1 1 
−
1 1 0 1 0 1

Adjoint Matrix:  
1 −1 1
adj(B) = [Cof(B)]T =  1 1 −1
−1 1 1
Inverse Matrix:
  1
− 12 1
 
1 −1 1
1 1 2 2
B −1 = · adj(B) =  1 1 −1 =  21 1
2 − 12 
det(B) 2
−1 1 1 − 12 1
2
1
2

Step 3: Calculate Pα→β = B −1 A.

1
− 12 1
  
2 2 2 1 2
1 1
Pα→β = 
2 2 − 21  3 2 0
− 12 1
2
1
2
1 0 3
Final Answer:
− 12 5
 
0 2
3
Pα→β = 2 2 − 21 
1 1
1 2 2

Example 2
Find the transition matrix from a basis α = {(1, 0, 1), (1, 1, 0), (0, 1, 1)} to another basis β = {(2, 3, 1), (1, 2, 0), (2, 0, 3)}.

8
Solution:
The transition matrix from α to β is given by Pα→β = B −1 A, where:

• A is the matrix with α basis vectors as columns

• B is the matrix with β basis vectors as columns
Step 1: Form matrices A and B.
   
1 1 0 2 1 2
A = 0 1 1 , B = 3 2 0
1 0 1 1 0 3
Find B −1using the adjoint method.
Step 2: 
2 1 2
For B = 3 2 0:
1 0 3
Determinant: Using the first row expansion:

2 0 3 0 3 2
det(B) = 2 −1 +2
0 3 1 3 1 0

= 2(6) − 1(9) + 2(−2) = 12 − 9 − 4 = −1

Cofactor Matrix:
 
2 0 3 0 3 2
 0 −
3 1 3 1 0   

 1
 6 −9 −2
2 2 2 2 1 
 = −3
− 0
Cof(B) = 
3 1 3
−
1 0  4 1

 1
 −4 6 1
2 2 2 2 1 
−
2 0 3 0 3 2

Adjoint Matrix:  
6 −3 −4
adj(B) = [Cof(B)]T = −9 4 6
−2 1 1
Inverse Matrix:
   
6 −3 −4 −6 3 4
1 1 
B −1 = · adj(B) = −9 4 6 = 9 −4 −6
det(B) −1
−2 1 1 2 −1 −1

Step 3: Calculate Pα→β = B −1 A.

  
−6 3 4 1 1 0
Pα→β = 9 −4 −6 0 1 1
2 −1 −1 1 0 1
Final Answer:  
−2 −3 7
Pα→β = 3 5 −10
1 1 −2

Example 3
Let T : R2 → R2 be defined by T (x, y) = (3x + y, x − 2y). Find the transition matrix from the standard
basis α = {(1, 0), (0, 1)} to the basis β = {(1, 1), (2, −1)} for the matrix representation of T .

9
Solution:
To compute: Pα→β = B −1 A:
   
1 0 1 2
A= , B=
0 1 1 −1
Find B −1 :
det(B) = (1)(−1) − (2)(1) = −3
  1 2

1 −1 −2
B −1 = = 31 3
−3 −1 1 3 − 13
Therefore: 1 2
  1 2

3 3 1 0
Pα→β = 1 = 13 3
3 − 13 0 1 3 − 31

Example 4
In P2 (R), find the transition matrix from basis α = {1, x, x2 } to basis β = {1 + x, x + x2 , 1 + x2 }.

Solution:
Given:    
1 0 0 1 0 1
A = 0 1 0 , B = 1 1 0
0 0 1 0 1 1
Find B −1 using adjoint method:

1 0 1 1
det(B) = 1 −0+1 = 1(1) + 1(1) = 2
1 1 0 1

Cofactor matrix:  
1 −1 1
Cof(B) =  1 1 −1
−1 1 1
  1 1
− 21
 
1 1 −1
1 2 2
B −1 = −1 1 1 = −2
  1 1
2
1
2

2 1
1 −1 1 2 − 21 1
2

Therefore:
1 1
− 21
 
2 2
Pα→β = − 1 1 1 
2 2 2
1
2 − 12 1
2

Example 5
In M2×2 (R), find the transition matrix from the standard basis
       
1 0 0 1 0 0 0 0
α= , , ,
0 0 0 0 1 0 0 1

to basis        
1 1 1 0 0 1 0 0
β= , , ,
0 0 1 0 1 0 1 1

10
Solution:
Converting matrices to vectors using the standard vectorization (row-wise):
   
1 0 0 0 1 1 0 0
0 1 0 0 1 0 1 0
A= 0 0 1 0 , B = 0 1 1 0
  

0 0 0 1 0 0 1 1
Find B −1 :
0 1 0 1 1 0
det(B) = 1 1 1 0 −1 0 1 0 = 1(−1) − 1(1) = −2
0 1 1 0 1 1
Use Gauss-Jordan method to compute inverse:

0 − 21 1
 
2 0
−1 1 1
0
B −1 =  2 2 
1 0 −1 0
1
0 2 − 21 1

Therefore:
− 21 1
 
0 2 0
−1 1 1
−1 2 2 0
Pα→β =B A= 
1 0 −1 0
1
0 2 − 12 1

Example 6
Consider the linear transformation T : P2 (R) → R3 defined by T (a + bx + cx2 ) = (a + b, b + c, a + c). Find
the transition matrix for changing from basis α = {1, x, x2 } in P2 (R) to basis β = {1 + x, x − x2 , 1 + x2 }
in P2 (R) for the domain, while using the standard basis in R3 for the codomain.

Solution:
For the domain transformation from α to β in P2 (R):
   
1 0 0 1 0 1
A = 0 1 0 , B = 1 1 0
0 0 1 0 −1 1
Find B −1 :
1 0 1 1
det(B) = 1 −0+1 = 1(1) + 1(−1) = 0 + 2 = 2
−1 1 0 −1
  1
− 21 − 21
 
1 −1 −1
1 2
B −1 = −1 1 1  = − 12 1
2
1 
2
2 1 1 1
−1 1 1 −2 2 2

The domain transition matrix is:

1
− 21 − 12
 
2
domain
Pα→β = B −1 A = − 1
2
1
2
1
2

− 12 1
2
1
2

This matrix transforms coordinate vectors in the α basis to coordinate vectors in the β basis for
polynomials in P2 (R).

11
Exercises
1. In R3 , find the transition matrix from basis α = {(1, 2, 1), (0, 1, 2), (2, 0, 1)} to basis β = {(1, 1, 0), (0, 1, 1), (1, 0, 2)}.

2. In the vector space P3 (R) of polynomials of degree at most 3, find the transition matrix from the
standard basis α = {1, x, x2 , x3 } to the basis β = {1 + x, 1 − x, x2 + x3 , x2 − x3 }.

3. Consider the vector space M2×2 (R) of all 2 × 2 matrices. Find the transition matrix from the
standard basis        
1 0 0 1 0 0 0 0
α= , , ,
0 0 0 0 1 0 0 1
to the basis        
1 1 1 0 0 1 1 1
β= , , ,
1 0 0 1 1 1 0 1

Solutions to Exercise Problems

Solution 1:
For α = {(1, 2, 1), (0, 1, 2), (2, 0, 1)} and β = {(1, 1, 0), (0, 1, 1), (1, 0, 2)}:
   
1 0 2 1 0 1
A = 2 1 0 , B = 1 1 0
1 2 1 0 1 2
0 1 1 1
det(B) = 1 −0+1 =2+1=3
2 1 0 1
 
2 1 −1
1
B −1 = −2 2 1
3
1 −1 1
    
2 1 −1 1 0 2 3 −1 3
1 1
Pα→β = B −1 A = −2 2 1  2 1 0 = 3 4 −3
3 3
1 −1 1 1 2 1 0 1 3

1 − 13
 
1
4
Pα→β = 1

3 −1
1
0 3 1

Solution 2:
For α = {1, x, x2 , x3 } and β = {1 + x, 1 − x, x2 + x3 , x2 − x3 }:
Using coordinate representations:
   
1 0 0 0 1 1 0 0
0 1 0 0 1 −1 0 0
A= 0 0 1 0 , B = 0 0
  
1 1
0 0 0 1 0 0 1 −1

1 1 1 1
det(B) = · = (−2)(−2) = 4
1 −1 1 −1
  1
− 2 − 12
 
−2 −2 0 0 0 0
1−2  1 1
B −1
= 
 2 0 0  = − 2 2 0 0 
4 0 0 −2 −2   0 0 − 2 − 12 
1

0 0 −2 2 0 0 − 21 1
2
 1
− 21

−2 0 0
− 1 1
0 0 
Pα→β = 2 2 
 0 0 − 12 − 21 
0 0 − 12 1
2

12
Solution 3:
Converting matrices to vectors (row-wise vectorization):
   
1 0 0 0 1 1 1 0
0 1 0 0 1 0 0 1
A= 0 0 1 0 , B = 0
  
1 1 1
0 0 0 1 1 1 0 1

det(B) = −2
   
0 0 2 0 0 0 −1 0
1 0 −2 0 2 0 1 0 −1
B −1 =−  = 
2 2
 0 −2 0 −1 0 1 0
−2 2 0 0 1 −1 0 0
 
0 0 −1 0
0 1 0 −1
Pα→β =
−1

0 1 0
1 −1 0 0

13
 Similarity of Linear Transformation
Definition
Let A and B be two square matrices. Then we say that A is similar to B if there exists a non-singular
matrix P such that
B = P −1 AP.

Note:

[T ]α = [Id]α β
β [T ]β [Id]α

[T ]β = [Id]βα [T ]α [Id]α
β

= P −1 [T ]α P

Example 1
If T : R3 → R3 be the linear transformation defined by T (x1 , x2 , x3 ) = (2x1 + x2 , x1 + x2 + 3x3 , −x2 )
and α = (e1 , e2 , e3 ) is the standard basis and β = {(−1, 0, 0), (2, 1, 0), (1, 1, 1)} is any ordered basis. Find
[T ]α and hence [T ]β = P −1 [T ]α P .

Solution:
Step 1: Find [T ]α (matrix representation with respect to standard basis).
Apply T to each standard basis vector:

T (e1 ) = T (1, 0, 0) = (2(1) + 0, 1 + 0 + 3(0), −0) = (2, 1, 0)

T (e2 ) = T (0, 1, 0) = (2(0) + 1, 0 + 1 + 3(0), −1) = (1, 1, −1)
T (e3 ) = T (0, 0, 1) = (2(0) + 0, 0 + 0 + 3(1), −0) = (0, 3, 0)

Therefore:  
2 1 0
[T ]α = 1 1 3
0 −1 0
Step 2: Find the transition matrix P from β to α.
The transition matrix P has the basis vectors of β as its columns:
 
−1 2 1
P = 0 1 1
0 0 1

Step 3: Find P −1 using the adjoint method.

Determinant:
1 1 0 1 0 1
det(P ) = −1 −2 +1 = −1(1) − 2(0) + 1(0) = −1
0 1 0 1 0 0

Cofactor Matrix:  
1 0 0
Cof(P ) = −2 −1 0
1 1 −1
Adjoint Matrix:  
1 −2 1
adj(P ) = [Cof(P )]T = 0 −1 1
0 0 −1
Inverse Matrix:
   
1 −2 1 −1 2 −1
1 1 
P −1 = · adj(P ) = 0 −1 1 = 0 1 −1
det(P ) −1
0 0 −1 0 0 1

14
 Step 4: Calculate [T ]β = P −1 [T ]α P .
First, compute [T ]α P :
    
2 1 0 −1 2 1 −2 5 3
[T ]α P = 1 1 3  0 1 1 = −1 3 5
0 −1 0 0 0 1 0 −1 −1

Then, compute P −1 ([T ]α P ):

    
−1 2 −1 −2 5 3 0 2 8
[T ]β =  0 1 −1 −1 3 5  = −1 4 6
0 0 1 0 −1 −1 0 −1 −1

Final Answer:  
2 1 0
[T ]α = 1 1 3
0 −1 0
 
0 2 8
[T ]β = P −1 [T ]α P = −1 4 6
0 −1 −1
Verification of Similarity: The matrices [T ]α and [T ]β are similar because [T ]β = P −1 [T ]α P , where
P is the non-singular transition matrix from basis β to the standard basis α.

Example 2
Let β = {v1 , v2 , v3 } be a basis for R3 consisting of v1 = (1, 1, 0), v2 = (1, 0, 1) and v3 = (0, 1, 1). Let T
be the linear transformation on R3 given by the matrix
 
2 1 −1
[T ]β =  1 2 3  .
−1 1 1

Let α = {e1 , e2 , e3 } be the standard basis. Find the transition matrix [Id]βα (or Pα→β ) and [T ]α .

Solution:
Using the similarity relationship between matrix representations in different bases:

[T ]α = P [T ]β P −1

where P = [Id]αβ (or Pβ→α ) is the transition matrix from basis β to standard basis α.
Step 1: Find the transition matrix P = [Id]α β.
The transition matrix P has the basis vectors of β as its columns:
 
1 1 0
P = [Id]α
β = 1 0 1
 
0 1 1

The transition matrix [Id]βα (or Pα→β ) = P −1 .

Step 2: Find P −1
Determinant:
0 1 1 1
det(P ) = 1 −1 + 0 = 1(−1) − 1(1) = −2
1 1 0 1
Cofactor Matrix:
 
0 1 1 1 1 0
 1 −
1 0 1 0 1   

 1
 −1 −1 1
0 1 0 1 1 
 = −1
− 1
Cof(P ) =  − 1 −1
1 0 1 0 1 

 1
 1 −1 −1
0 1 0 1 1 
−
0 1 1 1 1 0

15
 Adjoint Matrix:  
−1 −1 1
adj(P ) = [Cof(P )]T = −1 1 −1
1 −1 −1
Inverse Matrix:
  1 1
− 12
 
−1 −1 1
1 1  2 2
P −1 = · adj(P ) = −1 1 −1 =  12 − 21 1
2

det(P ) −2
1 −1 −1 − 12 1
2
1
2

Therefore:
1 1
− 12
 
2 2
Pα→β = [Id]βα = P −1 =  1
2 − 21 1
2

− 21 1
2
1
2

Step 3: Calculate [T ]α = P [T ]β P −1 .
First, compute [T ]β P −1 :
 1 1
− 21
 
2 1 −1 2 2
[T ]β P −1 = 1 2 3   12 − 12 1
2

−1 1 1 − 12 1
2
1
2
 
2 0 −1
= 0 1 2
− 21 − 12 3
2

Then, compute P ([T ]β P −1 ):

  
1 1 0 2 0 −1
[T ]α = 1 0 1  0 1 2
0 1 1 − 12 − 12 3
2
 
4 2 2
1
[T ]α = 3 −1 1
2
−1 1 7
Final Answers:
1 1
− 12
 
2 2
Pα→β = [Id]βα =  1
2 − 21 1
2

− 12 1
2
1
2
 
4 2 2
1
[T ]α =  3 −1 1
2
−1 1 7

Example 3
Let D be the differential operator on the vector space P2 (R). Given two ordered bases α = {1, x, x2 } and
β = {1, 2x, 4x2 − 2} for P2 (R). Find [D]α and [D]β .

Solution:
Step 1: Find [D]α (matrix representation with respect to standard basis α).
Apply the differential operator D to each basis vector in α:

D(1) = 0 = 0 · 1 + 0 · x + 0 · x2
D(x) = 1 = 1 · 1 + 0 · x + 0 · x2
D(x2 ) = 2x = 0 · 1 + 2 · x + 0 · x2

16
 Therefore:  
0 1 0
[D]α = 0 0 2
0 0 0
Step 2: Find [D]β (matrix representation with respect to basis β).
Apply the differential operator D to each basis vector in β:
D(1) = 0 = 0 · 1 + 0 · 2x + 0 · (4x2 − 2)
D(2x) = 2 = 2 · 1 + 0 · 2x + 0 · (4x2 − 2)
D(4x2 − 2) = 8x = 0 · 1 + 4 · 2x + 0 · (4x2 − 2)
Therefore:  
0 2 0
[D]β = 0 0 4
0 0 0
Step 3: Find the transition matrix Q from β to α and verify the similarity relationship.
To find the transition matrix Q = [Id]α
β , we express each vector in β in terms of the basis α:

1 = 1 · 1 + 0 · x + 0 · x2
2x = 0 · 1 + 2 · x + 0 · x2
4x2 − 2 = −2 · 1 + 0 · x + 4 · x2
Therefore:  
1 0 −2
Q = [Id]α
β = 0 2 0
0 0 4
Step 4: Find Q−1 using the adjoint method.
Since Q is an upper triangular matrix, we can find its inverse more easily. For this particular matrix:
Determinant:
det(Q) = 1 · 2 · 4 = 8
For upper triangular matrices, the inverse can be computed directly:
1 0 12
   
4 0 2
1
Q−1 = [Id]βα = 0 2 0 = 0 12 0 
4
0 0 1 0 0 14

Step 5: Verify the similarity relationship [D]β = Q−1 [D]α Q.

First, compute [D]α Q:
    
0 1 0 1 0 −2 0 2 0
[D]α Q = 0 0 2 0 2 0  = 0 0 8
0 0 0 0 0 4 0 0 0

Then, compute Q−1 ([D]α Q):

1
    
1 0 2 0 2 0 0 2 0
Q−1 ([D]α Q) = 0 1
2 0  0 0 8 = 0 0 4 = [D]β ✓
1
0 0 4
0 0 0 0 0 0
Final Answers:
 
0 1 0
[D]α = 0 0 2
0 0 0
 
0 2 0
[D]β = 0 0 4
0 0 0
Verification: The similarity relationship [D]β = Q−1 [D]α Q is confirmed, showing that both matrices
represent the same differential operator D with respect to different bases.

17
Exercises
1. Let T : R3 → R3 be the linear transformation defined by

T (x1 , x2 , x3 ) = (x1 + 2x2 + x3 , −x2 , x1 + 4x3 ).

Let α be the standard basis, and let β = {v1 , v2 , v3 } be another ordered basis consisting of v1 =
(1, 0, 0), v2 = (1, 1, 0), and v3 = (1, 1, 1) for R3 . Find the associated matrix of T with respect to α and
the associated matrix of T with respect to β. Are they similar?

18