ES202 – Mathematics for Engineers

Murat Büyük, PhD

Determinants

For every square matrix we associate a value to this matrix called determinant. It is denoted as:
𝑎11 𝑎12 … 𝑎1𝑛
𝑎21 𝑎22 … …
det 𝐴 = |𝐴| = | ⋮ ⋮ ⋱ … |
𝑎𝑛1 … … 𝑎𝑛𝑛 𝑛𝑥𝑛

elements of the
Main diagonal of the determinant
determinant

n: order of determinant

if n=1, then first order determinant; |𝐴| = |𝑎11 | = 𝑎11

+ −
if n=2, then 2nd order determinant; |𝐴| = |𝑎11 𝑎12 | = 𝑎11 𝑎22 − 𝑎12 𝑎21 : Diagonal Expansion
𝑎21 𝑎22
Example:
3 4
Let, 𝐴 = [ ], find |𝐴|
1 −2

|𝐴| = |3 4 | = (3. −2) − (4.1) = −10

1 −2
If n=3, then 3rd order determinant; Diagonal Expansion
+ + + − − −
𝑎11 𝑎12 𝑎13 𝑎11 𝑎12 𝑎13
|𝐴| = |
𝑎21 𝑎22 𝑎23 | 𝑎21 𝑎22 𝑎23 |
𝑎31 𝑎32 𝑎33 𝑎31 𝑎32 𝑎33

= (𝑎11 𝑎22 𝑎33 ) + (𝑎12 𝑎23 𝑎31 ) + (𝑎13 𝑎21 𝑎32 ) − (𝑎11 𝑎23 𝑎32 ) − (𝑎12 𝑎22 𝑎23 ) − (𝑎13 𝑎22 𝑎31 )
Example:
0 2 3
|𝐴| = |4 1 0|, Expand the determinant with diagonal expansion
1 5 6
|𝐴| = 9

Note: Diagonal expansion cannot be used to find the determinant of a matrix for which n>3. It is only
valid for 𝑛 ≤ 3

1/ 10
Finding Determinant Using Minors and Cofactors

Minor: when ith row and jth column of a matrix is removed by matrix operations, the determinant of
the remaining (n-1) square matrix is called minor of A and shown by Mij.

𝑎11 𝑎12 … 𝑎1𝑛

𝑎21 𝑎22 … …
Let, 𝐴 = | ⋮ ⋮ ⋱ … | , for i=2 and j=1
𝑎𝑛1 … … 𝑎𝑛𝑛 𝑛𝑥𝑛
𝑎12 𝑎12 … 𝑎1𝑛
𝑎32 𝑎33 … 𝑎3𝑛
Minor: 𝑀21 =| ⋮ ⋮ ⋱ … |
𝑎𝑛2 … … 𝑎𝑛𝑛 (𝑛−1)𝑥(𝑛−1)

Example:
0 2 6
|𝐴| = |2 3 1|, Find the minors of |𝐴|.
4 1 2
3 1
𝑀11 = | | = (3.2) − (1.1) = 5
1 2
2 1
𝑀12 = | | = (2.2) − (4.1) = 0
4 2
2 3
𝑀13 = | | = (2.1) − (3.4) = −10
4 1
2 6
𝑀21 = | | = (2.2) − (6.1) = −2
1 2
0 6
𝑀22 = | | = (0.2) − (6.4) = −24
4 2
0 2
𝑀23 = | | = (0.1) − (2.4) = −8
4 1
2 6
𝑀31 = | | = (2.1) − (6.3) = −16
3 1
0 6
𝑀32 = | | = (0.1) − (6.2) = −12
2 1
0 2
𝑀33 = | | = (0.3) − (2.2) = −4
2 3

Cofactor:

𝐶𝑖𝑗 = (−1𝑖+𝑗 )𝑀𝑖𝑗

So, for the matrix that was given in the example above:

𝐶11 = (−11+1 )𝑀11 = (−12 )5 = 5

2/ 10
Methods for Expansion of Determinants (other than diagonal expansion)

1-Laplace Expansion

Since diagonal expansion gives incorrect results when applied to higher order determinants, we
utilize other methods.

Column-wise expansion
|𝐴| = ∑𝑛𝑖=1 𝑎𝑖𝑘 𝑐𝑖𝑘 , (1 ≤ 𝑘 ≤ 𝑛), k is the column number for which expansion is done

Row-wise expansion
|𝐴| = ∑𝑛𝑗=1 𝑎𝑘𝑗 𝑐𝑘𝑗 , (1 ≤ 𝑘 ≤ 𝑛), k is the row number for which expansion is done

Example:
0 2 6
|𝐴| = |2 3 1|, (we have found the minors before)
4 1 2
First row expansion: (k=1), j is from 1 to 3 for a 3x3 matrix, n=3

|𝐴| = 𝑎11 . 𝐶11 + 𝑎12 . 𝐶12 + 𝑎13 . 𝐶13

𝐶11 = (−11+1 )𝑀11 = (−12 )5 = 5

𝐶12 = (−11+2 )𝑀12 = −1.0 = 0

𝐶13 = (−11+3 )𝑀13 = 1. (−10) = −10
|𝐴| = 0(5) + 2.0 + 6. (−10) = −60

Example:
1 3 0
|𝐴| = | 2 6 4|
−1 0 2
First row expansion: |𝐴| = 1((−12 ). (12 − 0)) + 3((−13 ). (4 + 4)) + 0((−14 ). (0 + 6))

= 12 − 24 + 0 = −12

Third row expansion: |𝐴| = −1((−14 ). (12 − 0)) + 0 ((−15 ). (4 − 0)) + 2((−16 ). (6 − 6))

= −12 − 0 + 0 = −12
Example:
1 2 3 4
|𝐴| = |4 3 2 1
| , Apply 3rd row expansion
0 −1 2 3
1 6 4 −2 4𝑥4
𝑛

∑ 𝑎3𝑘 𝑐3𝑘 = 𝑎31 . 𝐶31 + 𝑎32 . 𝐶32 + 𝑎33 . 𝐶33 + 𝑎34 . 𝐶34
𝑗=1

𝑎31 = 0

3/ 10
 1 3 4
𝐶32 = (−13+2 ) |4 2 1 | = −75
1 4 −2
1 2 4
𝐶33 = (−13+3 ) |4 3 1 | = 90
1 6 −2
1 2 3
𝐶34 = (−13+4 ) |4 3 2| = −35
1 6 4
|𝐴| = −1(−75) + 2(90) + 3(−35) = 150

Properties of Determinants

A is an nxn matrix:

1-If any row or column of A has only zero elements, then |𝐴| = 0

2- |𝐴| = |𝐴|𝑇 Matrix and its transpose have the same determinant

3- Determinant of triangular matrices (upper or lower) are : |𝐴| = 𝑎11. 𝑎22 . 𝑎33 … . 𝑎𝑛𝑛
(multiplication of diagonal elements)

4- If two rows or columns of matrix A are equal then, |𝐴| = 0

5- If two different rows (or column) of a determinant (matrix) are interchanged, the sign of the
determinant is also changed.

𝑅𝑖 ->𝑅𝑗 𝐶𝑖 ->𝐶𝑗 𝑤ℎ𝑒𝑟𝑒 𝑖 ≠ 𝑗

or
𝑅𝑗 ->𝑅𝑖 𝐶𝑗 ->𝐶𝑖 |𝐴| = −|𝐴|
6- if a single row (or column) is multiplied by a scalar (k), the determinant is multiplied by this scalar.
𝑘𝑎11 𝑘𝑎12 … 𝑘𝑎1𝑛 𝑎11 𝑘𝑎21 … 𝑎1𝑛
𝑎 𝑎22 … 𝑎2𝑛 𝑎21 𝑘𝑎22 … 𝑎2𝑛
𝑘|𝐴| = | 21 | =| |
⋮ ⋮ ⋱ … ⋮ ⋮ ⋱ …
𝑎𝑛1 … … 𝑎𝑛𝑛 𝑛𝑥𝑛 𝑎𝑛1 𝑘𝑎𝑛2 … 𝑎𝑛𝑛 𝑛𝑥𝑛

7- if a multiple of one row (or column) is added to a different row (or column) the determinant does
not change. Unlike matrices, we can do elementary column operations on determinants, similar to
elementary row operation.

Example:
2 1 3 1 0 0 1 0 0
|𝐴| = |0 1 6| ~𝑅1 → 𝑅3 ~ − |0 1 6| ~ − 2𝑅1 + 𝑅3 → 𝑅3 ~ − |0 1 6|
1 0 0 2 1 3 0 1 3
1 6
First row expansion: |𝐴| = 1 | |=6−3=3
1 3
Or by proceeding with upper triangular form
1 0 0
~ − 𝑅2 + 𝑅3 → 𝑅3 ~ − |0 1 6|
0 0 −3
|𝐴| = −(1.1. −3) = 3

4/ 10
8- |𝐴𝐵| = |𝐴||𝐵| = |𝐵𝐴|, where |𝐴|𝑛𝑥𝑛 and |𝐵|𝑛𝑥𝑛 (same size)

Recall, 𝐴𝐵 ≠ 𝐵𝐴

9- If A is a square matrix of nxn size, it has a rank of ‘n’ if and only if |𝐴| ≠ 0

An mxn matrix 𝐴 = [𝑎𝑖𝑗 ] has rank ≥ 1 , if and only if A has an rxr submatrix with nonzero
determinant.

2-Upper Triangular Form

Considering 3rd property of determinants, if we can convert it to a triangular form, the determinant
can be found by multiplying diagonal elements.

Example:
1 2 3 4 1 2 3 4
−4𝑅1 + 𝑅3 → 𝑅3
|𝐴| = |4 3 2 1
| ~𝑅2 → 𝑅3 ~(−1) |
0 −1 2 3
|~
0 −1 2 3 4 3 2 1 −𝑅1 + 𝑅4 → 𝑅4
1 6 4 −2 1 6 4 −2
1 2 3 4 1 2 3 4
0 −1 2 3 −5𝑅 2 + 𝑅3 → 𝑅3 0 −1 2 3
~(−1) | |~ ~(−1) | |
0 −5 −10 −15 4𝑅2 + 𝑅4 → 𝑅4 0 0 −20 −30
1 4 1 −6 0 0 9 6
1 2 3 4 1 2 3 4
0 −1 2 3 0 −1 2 3
= (−1)(−10)(3) | | ~−𝑅4 + 𝑅3 → 𝑅3 ~(30) | | ~−3𝑅3 + 𝑅4 → 𝑅4 ~
0 0 2 3 0 0 −1 1
0 0 3 2 0 0 3 2
1 2 3 4
0 −1 2 3
~(30) | | = (30)(1. −1. −1.5) = 150
0 0 −1 1
0 0 0 5

3- Zeroing - Reducing elements of a row or column except one

Reducing the elements of a row or column except one and then applying Laplace expansion.

Example:
1 2 3 4
|𝐴| = |4 3 2 1
|
0 −1 2 3
1 6 4 −2

Will try to make these elements ‘0’

1 2 3 4 1 2 7 10
2𝐶2 + 𝐶3 → 𝐶3 4 3
|𝐴| = |4 3 2 1
|~ ~|
8 10
| : 3𝑟𝑑 𝑟𝑜𝑤 𝑒𝑥𝑝𝑎𝑛𝑠𝑖𝑜𝑛: 𝑎32 . 𝐶32
0 −1 2 3 3𝐶2 + 𝐶4 → 𝐶4 0 −1 0 0
1 6 4 −2 1 6 16 16

5/ 10
= (−1)(−1)3+2 𝑀32 =
1 7 10 −4𝑅 + 𝑅 → 𝑅 1 7 10
1 2 2
(−1)(−1)3+2 |4 8 10| ~ ~ |0 −20 −30| : 1𝑠𝑡 𝑟𝑜𝑤 𝑒𝑥𝑝𝑎𝑛𝑠𝑖𝑜𝑛: 𝑎11 . 𝐶11 :
−𝑅1 + 𝑅3 → 𝑅3
1 16 16 0 9 6
−20 −30
= (1)(−1)1+1 | | = (−20.6) − (−30.9) = −120 + 270 = 150
9 6

NOTE: There is no checking method for determinants other than applying another method.

Inverse of a Matrix

Let 𝐴 be an nxn (square) matrix, 𝐴−1 is the inverse of 𝐴 if and only if

𝐴−1 𝐴 = 𝐴𝐴−1 = 𝐼
|𝐴−1 𝐴| = |𝐴||𝐴−1 | = |𝐼| = 1
1
So, |𝐴−1 | = |𝐴| , where |𝐴| ≠ 0 and |𝐴−1 | ≠ 0

Hence, if a matrix has determinant |𝐴| = 0 , it is noninvertible (singular) matrix.

If matrix 𝐴 has determinant |𝐴| ≠ 0, it is called nonsingular or invertible.

If matrix 𝐴 has an inverse, it is unique.

For identity matrix, 𝐼𝑛 = 𝐼𝑛−1 , So, 𝐼𝑛 is nonsingular.

If 𝐴 is nonsingular, (𝐴−1 )−1 = 𝐴

If 𝐴, 𝐵 are two nonsingular matrices ; (𝐴𝐵)−1 = 𝐵−1 𝐴−1

If 𝐴 is nonsingular and 𝐴𝑇 is nonsingular, then (𝐴𝑇 )−1 = (𝐴−1 )𝑇

If 𝐴 and 𝐵 are nxn singular matrices, then, 𝐴𝐵 and 𝐵𝐴 are both singular

If 𝐴𝑛𝑥𝑛 is nonsingular (𝐴−1 𝑒𝑥𝑖𝑠𝑡𝑠), |𝐴| ≠ 0 , then rankA=n

Hence, if r[A]=n : A is nonsingular

if r[A]<n : A is singular

Methods of finding 𝑨−𝟏

1-Gauss-Jordan Elimination Method:

In this method, we use the matrix (𝐴) and the identity matrix, 𝐼, to form the augmented matrix
[𝐴|𝐼].

Then we apply Gauss-Jordan Elimination to have 𝐼 on the left site, right site will be 𝐴−1 .

[𝐴|𝐼] → [𝐼|𝐴−1 ] ,if 𝐴−1 exists, check whether |𝐴| ≠ 0 or not.

6/ 10
Example:
1 2 3 1 0 0
[𝐴] = [1 4 3 ] , [𝐼] = [0 1 0]
1 3 4 0 0 1
1 2 3 1 0 0 −𝑅 + 𝑅 → 𝑅 1 2 3 1 0 0
1 2 2
[𝐴|𝐼] = [1 4 3 |0 1 ] ~
0 −𝑅 + 𝑅 → 𝑅 ~ [ 0 2 0 |−1 1 0] ~1/2𝑅2 → 𝑅2 ~
1 3 3
1 3 4 0 0 1 0 1 1 −1 0 1
1 2 3 1 0 0 −𝑅2 + 𝑅3 → 𝑅3 1 0 3 2 −1 0
~ [0 1 0 |−1/2 1/2 0] ~
−2𝑅2 + 𝑅1 → 𝑅1
~ [0 1 0 |−1/2 1/2 0]
0 1 1 −1 0 1 0 0 1 −1/2 −1/2 1
1 0 0 7/2 1/2 −3
~ − 3𝑅3 + 𝑅1 → 𝑅1 ~ [0 1 0 |−1/2 1/2 0]
0 0 1 −1/2 −1/2 1
𝐼𝑛 𝐴−1
Check 𝐴𝐴−1 = 𝐼

If 𝐼𝑛 cannot be obtained, then |𝐴| = 0 ,inverse does not exist.

2-Adjoint (Cofactor) Method:

𝑇
1 𝑇 ((−1𝑖+𝑗 )𝑀𝑖𝑗 )
|𝐴−1 | = (𝐶𝑖𝑗 ) =
|𝐴| |𝐴|
Example:
1 2 3
[𝐴] = [1 4 3], Find |𝐴−1 |
1 3 4
1 2 3 −𝑅 + 𝑅 → 𝑅 1 2 3
|𝐴|=|1 4 3| ~−𝑅1 + 𝑅2 → 𝑅2 ~ |0 2 0|:1st column expansion:
1 3 3
1 3 4 0 1 1

|𝐴| = (1)(−11+1 ) |2 0| = 2 , in adjoint method we first check if |𝐴| ≠ 0

1 1
4 3
𝐶11 = (−11+1 ) | |=7
3 4
1 3
𝐶12 = (−11+2 ) | | = −1
1 4
1 4
𝐶13 = (−11+3 ) | | = −1
1 3
2 3
𝐶21 = (−12+1 ) | |=1
3 4
1 3
𝐶22 = (−12+2 ) | |=1
1 4

7/ 10
 1 2
𝐶23 = (−12+3 ) | | = −1
1 3
2 3
𝐶31 = (−13+1 ) | | = −6
4 3
1 3
𝐶32 = (−13+2 ) | |=0
1 3
1 2
𝐶33 = (−13+3 ) | |=2
1 4
𝐶11 𝐶12 𝐶13 7 −1 −1
𝐶𝑖𝑗 = [𝐶21 𝐶22 𝐶23 ] = [ 1 1 −1]
𝐶31 𝐶32 𝐶33 −6 0 2

𝑇
7 1 −6
(𝐶𝑖𝑗 ) = [−1 1 0]
−1 −1 2
7 1 −6
[−1 1 0] 7/2 1/2 −3
−1
1 𝑇 −1 −1 2 = [−1/2 1/2
|𝐴 | = (𝐶𝑖𝑗 ) = 0]
|𝐴| 2
−1/2 −1/2 1
Inverse of a Diagonal Matrix

𝐴 = [𝑎𝑖𝑗 ], 𝑎𝑖𝑗 = 0 only for 𝑖 ≠ 𝑗 → diagonal matrix

𝑎11 0 0 1/𝑎11 0 0
−1
𝐴=[ 0 𝑎22 0 ] , then 𝐴 = [ 0 1/𝑎22 0 ]
0 0 𝑎33 0 0 1/𝑎33
Inverse exist if nd only if none of the diagonal elements are zero. If any of the diagonal elements is
zero, then it would have |𝐴| = 0 →singular =noninvertible

Solutions of Linear Systems of Equations Using Matrix Inverse

HLS

⃗⃗
𝐴𝑥⃗ = 0

Multiply both sides with 𝐴−1

⃗⃗ → 𝐼𝑥⃗ = 0
𝐴−1 𝐴𝑥⃗ = 𝐴−1 0 ⃗⃗

So, if a homogeneous system has a coefficient matrix which has an inverse (A is nonsingular) it has
only trivial (zero) solution.

Otherwise, if A is singular, it has nontrivial (parametric) solution. 𝑟[𝐴] < 𝑛 → |𝐴| = 0

NHLS

𝐴𝑥⃗ = 𝑏⃗⃗

𝐴−1 𝐴𝑥⃗ = 𝐴−1 𝑏⃗⃗ → 𝐼𝑥⃗ = 𝐴−1 𝑏⃗⃗

If 𝐴 is nonsingular 𝐴−1 exists, then

𝑥⃗ = 𝐴−1 𝑏⃗⃗ gives unique solution.

8/ 10
Example:

x1+2x2+3x3=4

x1+4x2+3x3=-8

x1+3x2+4x3=-1

For the given NHLS 𝐴−1 was found as:

7/2 1/2 −3
|𝐴−1 | = [−1/2 1/2 0]
−1/2 −1/2 1
7/2 1/2 −3 4 13
𝑥⃗ = 𝐴−1 𝑏⃗⃗ = [−1/2 1/2 0 ] [−8] = [−6]
−1/2 −1/2 1 −1 1

CRAMER’S RULE

For solving linear systems of n equations in n unknown we can use determinants.

𝑎11 𝑥1 + 𝑎12 𝑥2 = 𝑏1
𝑎21 𝑥1 + 𝑎22 𝑥2 = 𝑏2
𝑎 𝑎12
|𝐴| = |𝑎11 𝑎22 | = 𝑎11 𝑎22 − 𝑎12 𝑎21
21

We remove coefficient of that unknown and put the equality set into its place
𝑏 𝑎12 𝑎 𝑏1
| 1 | | 11 |
𝑏2 𝑎22 𝑎21 𝑏2
𝑥1 = |𝐴|
and 𝑥2 = |𝐴|
, where |𝐴| ≠ 0

If, |𝐴| = 0 → NHS: no solution, parametric solution

If, |𝐴| = 0 → HS: nontrivial solution

So, for a given linear system of equations with the same number of unknowns and equations exist:
|𝐴1 | |𝐴2 | |𝐴𝑛 |
𝑥1 = |𝐴|
, 𝑥2 = |𝐴|
, … , 𝑥𝑛 = |𝐴|
, where 𝐴𝑛 is the determinant obtained by replacing the nth column
with the equality 𝑏𝑛 .

Example:

x1+3x2+4x3=2

x1+2x2+x3=4

x1+4x2-2x3=1
1 3 4
|𝐴| = |4 2 1 | = 75
1 4 −2
2 3 4 1 2 4 1 3 2
|𝐴1 | = |4 2 1 | = 67, |𝐴2 | = |4 4 1 | = 9, |𝐴3 | = |4 2 4| = 14
1 4 −2 1 1 −2 1 4 1

9/ 10
 |𝐴1 | 67 9 14
𝑥1 = |𝐴|
= , 𝑥2 = , 𝑥3 =
75 75 75

Gauss Elimination

In solution of linear systems of equations, we apply elementary row operations. Each operation
transforms one system into an equivalent system. Two linear systems are equivalent if and only if
every solution of one system is also the solution of other.

For an efficient solution we use a scheme in which we aim to construct an upper triangular
transformation by applying row operations. For this, we eliminate x1 (first unknown) from all other
equations.

Given linear system equations in augmented form:

𝑎11 𝑎11 … 𝑎1𝑛 𝑏1 𝑎11 𝑎11 … 𝑎1𝑛
𝑎21 𝑎22 … 𝑎2𝑛 𝑏2 0 𝑎22 … 𝑎2𝑛
[𝐴|𝑏] = [ ⋮ | ] → [
⋮ ⋱ ⋮ ⋮ 0 0 ⋱ ⋮ ]
𝑎𝑚1 𝑎𝑚2 … 𝑎𝑚𝑛 𝑏𝑚 0 0 0 𝑎𝑚𝑛

For this row operation multiply the first row by (−𝑎21 /𝑎11 ) and add to 2nd row multiplying the first-
row equation similarly by (−𝑎31 /𝑎11 ) and add to 3rd row by continuing up to last row.

In the second step, try eliminating for x2 (2nd unknown) similarly by utilizing 𝑎22 as the operator:

(−𝑎32 /𝑎22 )𝑅2 + 𝑅3 → 𝑅3

(−𝑎42 /𝑎22 )𝑅2 + 𝑅4 → 𝑅4
We continue this process till the elimination of other unknowns.
𝑎11 𝑎11 … 𝑎1𝑛
0 𝑎22 … 𝑎2𝑛
[ 0 0 ⋱ ⋮ ]
0 0 0 𝑎𝑚𝑛
Then, by back substitution, we can solve the system. However, if we extend the procedure for the
upper rows also, we obtain RREF.
𝑎11 0 0 0
0 𝑎22 0 0
[ ]
0 0 ⋱ 0
0 0 0 𝑎𝑚𝑛

10/ 10