Introduction to Matrices for Engineers

C.T.J. Dodson, School of Mathematics, Manchester University

What is a Matrix?

A matrix is a rectangular array of elements, usually numbers, e.g. 0 2 -1 117

1 -2 4
1 2

3 8 0 0

The above matrix is a (4 3)matrix, i.e. it has three columns and four rows.

1.1

Why use Matrices?

We use matrices in mathematics and engineering because often we need to deal with several variables at onceeg the coordinates of a point in the plane are written (x, y) or in space as (x, y, z) and these are often written as column matrices in the form: x y and x y z

It turns out that many operations that are needed to be performed on coordinates of points are linear operations and so can be organized in terms of rectangular arrays of numbers, matrices. Then we nd that matrices themselves can under certain conditions be added, subtracted and multiplied so that there arises a whole new set of algebraic rules for their manipulation. In general, an (n m)matrix A looks like:

A= an1,1 an,1

a11 a21 a31 ...

a12 a22 a32 ... an1,2 an,2

a13 a23 a33 ... an1,3 an,3

... ... ... ... ... ...

a1,m1 a2,m1 a3,m1 ... an1,m1 an,m1

a1,m a2,m a3,m ... an1,m an,m

Here, the entries are denoted aij ; e.g. in Example 4.1.1 we have a11 = 1, a23 = 2, etc. Capital letters are usually used for the matrix itself.

1.2

Dimension

In the above matrix A, the numbers n and m are called the dimensions of A.
Based

on original lecture notes on matrices by Lewis Pirnie

Introduction to Matrices

1.3

Addition

It is possible to add two matrices together, but only if they have the same dimensions. In which case we simply add the corresponding entries: 1 -2 4
1 2

3 8 0 0

4 0 2 0 + -1 5 1 117 2

0 -8 1 1

1 5 -3 -2 = -2 9 1 -50

3 0 1 1

1 -1 -3 67

If two matrices dont have the same size (dimensions) then they cant be added, or we say the sum is not dened.

1.4

Example
1 2 0 -2 + 3 7 1 2 0 -2 + 3 7 6 1 0 4 3 1 4 7 2 = 1 0 3 0 -2
7 4

0 7

is undened

2
2.1

Multiplying Matrices
Why should we want to?

We can motivate the process by looking at rotations of points in the plane. Consider the point P on the x-axis at x = x, y = 0, so it has coordinates x 0 Now, let P be the point obtained by rotating P round the origin in an anticlockwise direction through angle , keeping the distance from the origin (which is actually the square root of the sum of squares of the coordinates) constant. What are the coordinates of the point P ? It is not dicult trigonometry to work out that these are: x cos x sin Next consider the point Q on the y-axis with coordinates 0 y and rotate this point round the origin in an anticlockwise direction through angle , keeping the distance from the origin constant. We nd that the new coordinates are: y sin y cos Now the payo, it is actually a 2 2 matrix that does the rotating here because the operation is linear on the coordinate components. We can express the rotation of any point with coordinates (x, y) as the following matrix equation: cos sin sin cos x y = x cos y sin x sin + y cos

Can you guess what will be the matrix for rotation in a clockwise direction?

C.T.J. Dodson

2.2

Exercise on Trigonometry

1. Three common right angled triangles can be used to obtain the following values for the trigonometric functions: 30 = 6 sin = 6 cos = 6 tan = 6 45 = 4 sin = 4 60 = 3 sin = 3 cos = 3 tan = 3

1 2

3 2 1 3

cos = 4 tan = 1 4

1 2 1 2

3 2 1 2

2. Obtain the rotation matrices explicitly for rotations of = 30 , 45 , 60 , 90 , 180 .

2.3

Rules

When multiplying matrices, keep the following in mind: lay the rst row of the rst matrix on top of the rst column of the second matrix; only if they are both of the same size can you proceed. The rule for multiplying is: go across the rst matrix, and down the second matrix, multiplying the corresponding entries, and adding the products. This new number goes in the new matrix in position of the row of the rst matrix, and the column of the second matrix. For example: 1 3 2 4 2 1 0 5 3 7 = 1.2 + 2.1 3.2 + 4.1 10 5 17 37 1.0 + 2.5 3.0 + 4.5 1.3 + 2.7 3.3 + 4.7

4 10

Symbolically, if we have the matrices A and B as follows: a11 a21 a31 ... a12 a22 a32 ... an1,2 an,2 b12 b22 b32 ... bp1,2 bp,2 a13 a23 a33 ... an1,3 an,3 b13 b23 b33 ... bp1,3 bp,3 ... ... ... ... ... ... ... ... ... ... bp1,q1 ... a1,m a2,m a3,m ... an1,m an,m b1,q b2,q b3,q ... bp1,q bp,q ,

A= an1,1 an,1 b11 b21 b31 ...

B= bp1,1 bp,1 then the product AB is given by:

a1i bi1 a2i bi1 ... m i=1 ani bi2

m i=1 m i=1

a1i bi2 a2i bi2 ... m i=1 ani bi2

m i=1 m i=1

... ... ... ...

a1i biq a2i biq ... m i=1 ani biq

m i=1 m i=1

where i=1 a1i bi1 stands for a11 b11 + a12 b21 + a13 b31 + . . . + a1n bn1 , etc. Note that we must have m = p, i.e. the number os columns in the rst matrix must equal the number of rows of the second; otherwise, we say the product is undened.

Introduction to Matrices

2.4

Example
matrices: 3 -1 = 5

We multiply the following 1 -1 2 0 0 (i) 4 1 -2 7 4 7 2 0 -8 1 (ii) 2 3 -1 2 1 (iii) 4 2 1 2 1 2 -1

1.1 + 2.0 + 0.7 4.1 + 1.0 + 2.7

1.3 + 2. 1 + 0.5 4.3 + 1. 1 + 2.5

-1 -10

-5 1

is undened 4.2 + 1 . 1 2 2.2 + 1. 1 4.2 + 1 .4 2 2.2 + 1.4

2 4

6 8

-1 1 0 -1 2 1 (iv) 0 2 3 7 0 -1 1 4 1.0 + 1.3 1. 1 + 1.7 1.2 + 1.0 0. 1 + 2.7 0.2 + 2.0 = 0.0 + 2.3 1.0 + 4.3 1. 1 + 4.7 1.2 + 4.0 (v) (vi) = = 1 3 1 -2 0 1 3 8 0 2 2 5 = 4 0 5 1.2 + 0.5 3.2 + 1.5 0 -8 1 1 -3 -2 =

1.1 + 1. 1 3 0.1 + 2. 1 = 6 1.1 + 4. 1 12 2 11

8 14 27

-2 0 2

-2 -2 -3

6 0 1

1.4 + 3.0 + 0.5 2.4 + 8.0 + 2.5

1.0 + 3. 8 + 0.1 2.0 + 8. 8 + 2.1

1.1 + 3. 3 + 0. 2 2.1 + 8. 3 + 2. 2

1.6 + 3.0 + 0.1 2.6 + 8.0 + 2.1

4 -24 -8 6 2 -62 -30 -10 We see from the examples: (i) (iii) (iv) (v) (vi) Product Product Product Product Product of of of of of (2 3)matrix (2 2)matrix (3 2)matrix (2 2)matrix (2 3)matrix with with with with with a a a a a (3 2)matrix (2 2)matrix (2 4)matrix (2 1)matrix (3 4)matrix is is is is is a a a a a (2 2)matrix. (2 2)matrix. (3 4)matrix. (2 1)matrix. (2 4)matrix.

Note that the two middle numbers must be the same if the product is dened; and then the dimensions of the answer is just the two outer numbers. Thus, the product of an (n m)matrix with a (m q)matrix is an (n q)matrix.

Scalar Multiplication

There is another type of multiplication involving matrices called scalar multiplication. This means just multiplying each entry of the matrix by a number. For example: 3 -1 4 2 1 0 -2 = -3 12 6 3 0 -6

3.1

Rules

There are some rules which matrix addition and multiplication obey: Associative Commutative (A + B) + C = A + (B + C) A+B =B+A

C.T.J. Dodson

Distributive Distributive Associative NonCommutative Moving Constants

A(B + C) = AB + AC (A + B)C = AC + BC (AB)C = A(BC) AB = BA (usually) A(B) = (AB)

(Assuming that the sums and products are dened in all cases.)

3.2

Example
Let A = 1 3 0 2 , B= -1 1 4 -2 , C= 2 3

Consider the following: AB = 1 3 -1 1 0 2 4 -2 -1 1 1 3 4 -2 0 2 = 1. 1 + 0.1 3. 1 + 2.1 1.1 + 4.3 1.1 + 2.3 1.4 + 0. 2 3.4 + 2. 2 1.0 + 4.2 1.0 + 2.2 = -1 -1 11 -5 4 8 8 -4

BA =

Now, AB = BA, and we see that two matrices are not the same if they are multiplied the other way around. Also consider: AC = 1 3 -1 1 0 2 4 -2 2 12 0 2 0 4 + 4 0 + -1 1 2 3 2 3 10 -4 4 -2 2 3 = 2 12 10 -4 = 12 8 0 4 12 8 4 0

BC =

AB + BC = 1 3

A+B =

(A + B)C =

and we notice that (A + B)C = AC + BC as required.

Transpose

Another operation on matrices is the transpose. This just reverses the rows and columns, or equivalently, reects the matrix along the leading diagonal. The transpose of A is normally written At thus a11 a12 a13 ... a1,m1 a1,m a21 a22 a23 ... a2,m1 a2,m a31 a32 a33 ... a3,m1 a3,m , A= ... ... ... ... ... ... an1,1 an1,2 an1,3 . . . an1,m1 an1,m an,1 an,2 an,3 ... an,m1 an,m

6 a11 a12 a13 ... a21 a22 a23 ... a2,n1 a2,n a31 a32 a33 ... a3,n1 a3,n ... ... ... ... ... ... am1,1 am1,2 am1,3 ... am1,n1 am1,n

Introduction to Matrices

At = a1,n1 a1,n

am,1 am,2 am,3 ... am,n1 am,n

Note that the transpose of a (n m)matrix is a (m n)matrix.

4.1

Example

As an example of the transpose: 2 0 4 3 -1 5 2 , At = 4 -1 0 3 5

A=

Square Matrices

Given a number n, (n n)matrices have very special properties. Note that if we have two (n n)matrices, the product is dened and will also be an (n n)matrix.

5.1

The Identity Matrix

There also exists a special matrix, known as the identity, Inn : 1 = 0 0 0 1 0 1 0 0 0 , I44 = 0 1 0 0 1 0 0 0 0 1 0 0 0 , etc. 0 1

I22 =

1 0

0 1

, I33

which has 1s on the main diagonal, and 0s everywhere else. This matrix has the property that, given any (n n)matrix A: AInn = A and Inn A = A i.e. multiplying by the identity on either side doesnt change the matrix. This is similar to the property of 1 when multiplying numbers. We usually abbreviate Inn to just I when its obvious what n is.

5.2

Example
Let A = 1 3 7 2 , I = I22 = 1 0 0 1

Now, we check the properties of the identity: AI = 1 3 1 0 7 2 0 1 1 0 1 3 0 1 7 2 = 1.1 + 7.0 3.1 + 2.0 1.1 + 0.3 0.1 + 1.3 1.0 + 7.1 3.0 + 2.1 1.7 + 0.2 0.7 + 1.2 = 1 3 1 3 7 2 7 2

IA = as required.

C.T.J. Dodson

Determinants

One of the most important properties of square matrices is the determinant. This is a number obtained from the entries.

6.1

Determinant of a (2 2)Matrix
a c b d . Then, the determinant of A, denoted det A or |A| is given by ad bc.

Let A =

6.2

Example
det det -1 3 2 1 2 -6 3 5 = 2.5 3.1 = 10 3 = 7 = (1)(6) 2.3 = 6 6 = 0

Before we go on to larger matrices, we need to dene minors.

6.3

Minors

Let A be the (n n)matrix a11 a21 a31 ... a12 a22 a32 ... an1,2 an,2 a13 a23 a33 ... an1,3 an,3 ... ... ... ... ... ... a1,m1 a2,m1 a3,m1 ... an1,m1 an,m1 a1,m a2,m a3,m ... an1,m an,m

A= an1,1 an,1

Then, the minor mij , for each i, j, is the determinant of the (n 1 n 1)matrix obtained by deleting the ith row and the j th column. For example, in the above notation: a22 a23 ... a2,m1 a2,m a32 a33 ... a3,m1 a3,m ... ... ... ... m11 = det . . . an1,2 an1,3 . . . an1,m1 an1,m an,2 an,3 ... an,m1 an,m m21 a12 a32 ... a13 a33 ... an1,3 an,3 ... ... ... ... ... a1,m1 a3,m1 ... an1,m1 an,m1 a1,m a3,m ... an1,m an,m

= det an1,2 an,2

6.4

Example
1 4 0 -1 3 -2 0 -5 2 -5 3 -2 -1 -2 = 15 = 9 m13 = m23 = 0 -5 2 -5 4 0 1 0 = 20 =5

2 We compute all the minors of A = 0 -5 m11 = m21 = 4 0 1 0 3 -2 -1 -2 = 8 = 2 m12 = m22 =

8 1 4 -1 3 2 0 -1 3

Introduction to Matrices

m31 =

=7

m32 =

=6

m33 =

2 0

1 4

=8

6.5

Minors and Cofactors

The numbers called cofactors are almost the same as minors, except some have a minus sign in accordance with the following pattern: + + ... + + ... + + ... + + ... ... ... ... ... ...

The best way to remember this is as an alternating or chessboard pattern. The cofactors from the previous example are: c11 = m11 = 8 c21 = m21 = 2 c31 = m31 = 7 c12 = m12 = 15 c22 = m22 = 9 c32 = m32 = 6 c13 = m13 = 20 c23 = m23 = 5 c33 = m33 = 8

Determinant of a (3 3)Matrix

In order to calculate the determinant of a (33)matrix, choose any row or column. Then, multiply each entry by its corresponding cofactor, and add the three products. This gives the determinant.

7.1

Example
1 4 0 -1 3 as before, we compute the determinant using the top row: -2

2 Letting A = 0 -5

det A = a11 c11 + a12 c12 + a13 c13 = 2.(8) + 1.(15) + (1).20 = 16 15 20 = 51 Suppose, we use the second column instead: det A = a12 c12 + a22 c22 + a32 c32 = 1.(15) + 4.(9) + 0.(6) = 15 36 0 = 51 It doesnt matter which row or column is used, but the top row is normal. Note that it is not necessary to work out all the minors (or cofactors), just three.

7.2

Example
0 1 2 4 0 . We compute the determinant of B: 1 0 1 2 4 0 =1 1 1 2 0 1 -2 3 0 1 -2 3 1 2

1 Let B = -2 3

1 det -2 3

+4

= 1(1 0) 0(2 0) + 4(4 3) = 1 28 = 27

C.T.J. Dodson

Determinant of an (n n)Matrix

The procedure for larger matrices is exactly the same as for a (3 3)matrix: choose a row or column, multiply the entry by the corresponding cofactor, and add them up. But of course each minor is itself the determinant of an (n1n1)matrix, so for example, in a (44) determinant, it is rst necessary to do four (3 3) determinants quite a lot of work!

Inverses

Let A be an n n matrix, and let I be the n n identity matrix. Sometimes, there exists a matrix A1 (called the inverse of A) with the property: A A1 = I = A1 A In this section, we demonstrate a method for nding inverses.

9.1

Inverse of a 2 2 Matrix
a c b d . Then, the inverse of A, A1 is given by: A1 = 1 ad bc d c b a

Let A =

To check, we multiply: A1 A = 1 ad bc = d c b a a c ad bc 0 b d = 1 ad bc = da bc ca + ac 1 0 0 1 db bd cb + ad

1 ad bc

0 ad bc

In a similar fashion we could show that A A1 = I. Of course, the inverse could also be written A1 = 1 det A d c b a

Note, that if det A = 0, then we have a division by zero, which we cant do. In this situation there is no inverse of A.

9.2

Inverse of 3 3 (and higher) Matrices

Recall the denition of a minor from Section 4.3.3: given an (n n)matrix A, the minor mij is the determinant of the (n 1 n 1)matrix obtained by omitting the ith row and the j th column.

9.3

Example
0 1 2 =1 = 8 = 4 4 0 . We calculate the minors: 1 m12 = m22 = m32 = -2 0 = 2 3 1 1 4 = 11 3 1 1 4 =8 -2 0 m13 = m23 = m33 = -2 1 = 7 3 2 1 0 =2 3 2 1 0 =1 -2 1

1 Let A = -2 3 m11 = m21 = m31 = 1 2 0 2 0 1 0 1 4 1 4 0

10

Introduction to Matrices

Recall also the pattern of + and signs from + +

which we obtain the cofactors: + + +

Now, we put the minors into a matrix and change their signs according to the pattern to get the matrix of cofactors: 1 2 7 8 11 2 4 8 1 The next stage is take the transpose: 1 2 7 8 11 2 4 8 1

and nally we must divide by the determinant, which is 27, from Example 4.3.8: 8 1 4 27 27 1 8 4 27 1 2 11 8 2 11 8 = 27 A1 = 27 27 27 7 2 1 7 2 1 27 27 27 This shows how to calculate the inverse of a (3 3)matrix. We check the result: 1 8 4 1 0 4 1 16 12 0+88 1 2 11 8 -2 1 0 = 2 + 22 24 0 11 16 A1 A = 27 7 2 1 3 2 1 7 + 4 + 3 02+2 -27 0 0 1 0 0 1 -27 0 = 0 1 0 = 0 27 0 0 -27 0 0 1 as required. The same procedure works for (n n)matrices. (I) (II) (III) (IV) Work out the minors. Put in the signs to form the cofactors. Take the transpose. Divide by the determinant.

4+04 8+08 28 + 0 + 1

Furthermore, an n n matrix has an inverse if and only if the determinant is not zero. So, its a good idea to calculate the determinant rst, just to check whether the rest of the procedure is necessary.

10

Linear Systems

We discuss one very important application of nding inverses of matrices.

10.1

Simultaneous Equations

Often, when solving problems in mathematics, we need to solve simultaneous equations, e.g. something like: 2x 5x + + y 3y = = 3 7

C.T.J. Dodson

11

from which we would obtain x = 2 and y = 1. The process we have used up until now is a little messy: we combine the equations to try and eliminate one of the unknown variables. There is a more systematic way using matrices. We can write the equations in a slightly dierent way: 2x + y 5x + 3y = 3 7

Now we can check that the rst matrix is equal to the product: 2x + y 5x + 3y and so altogether we have a matrix equation: 2 5 1 3 x y = 3 7 = 2 5 1 3 x y

The next stage is to use the inverse of the (2 2)matrix, so lets calculate that now. Let A = 2 5 1 3 , then A1 = 1 2.3 1.5 3 -5 -1 2 = 3 -5 -1 2 .

Now, we take the matrix equation above, and multiply by A1 2 5 3 -5 -1 2 2 5 1 3 1 3 x y x y = 3 7 3 -5 -1 2 3 7

Then, doing the multiplication: 1 0 0 1 x y x y = 3.2 + 1.7 5.3 + 2.7 2 1 2 1

and so x = 2 and y = 1, as required. So, provided we can work out the inverse of the matrix of coecients, we can solve simultaneous equations.

11

Larger Systems

The same thing works with 3 equations in x, y and z. Suppose we have x 2x + 2y 3y y + + 2z 2z 8z = = = 1 2 7

Then, the matrix form is 1 0 2 2 3 -1 2 x -1 -2 y = 2 8 z 7

Now, we denote the 3 3 matrix by A, and calculate the inverse of A. The minors are as follows:

12 3 -2 = 22 -1 8 2 2 = 18 -1 8 2 2 = 10 3 -2 0 2 1 2 1 0 -2 =4 8 2 =4 8 2 = 2 -2

Introduction to Matrices

m11 = m21 = m31 =

m12 = m22 = m32 =

m13 = m23 = m33 =

0 2 1 2 1 0

3 = 6 -1 2 = 5 -1 2 =3 3

Recall the chessboard pattern: + + So we have the following matrix of cofactors: 22 18 10 + + +

4 4 2

6 5 3

We can calculate the determinant by taking any row or column, and multiplying the original matrix entry by its corresponding cofactor, and then adding lets choose the top row: det A = 1.22 + 2. 4 + 2. 6 = 22 8 12 = 2 and so the determinant is 2, and so we will be able to nd the inverse. From the matrix of cofactors, we take the transpose, and then divide by the determinant to get A1 : 11 9 5 22 18 10 1 2 1 4 2 = 2 A1 = 4 2 5 3 6 5 3 3 2 2 Now, we return to solving the simultaneous equations, where x 1 2 2 0 3 -2 y = z 2 -1 8 we had: -1 2 7

Multiplying both sides on the left by A1 , we have: 11 9 5 11 1 2 2 x 2 2 1 0 3 -2 y = 2 5 3 z 2 -1 8 3 3 2 2

9 2
5 2

5 -1 1 2 3 7 2

and we know that A1 A = I, so 11. 1 + 9.2 + 5.7 1 0 0 x 0 1 0 y = 2. 1 + 2.2 + 1.7 5 0 0 1 z 3. 1 + 2 .2 + 3 .7 2 64 x y = 13 37 z 2 and so we get x = 64, y = 13 and z =
37 2 .

Its a very good idea to check calculations like this! = 1 =2 =7

x + 2y + 2z = 64 + 2(13) + 2( 37 ) = 64 + 26 + 37 2 3y 2z = 3(13) 2( 37 ) = 39 37 2 2x y + 8z = 2(64) 13 + 8( 37 ) = 128 13 + 148 2 as required.

C.T.J. Dodson

13

12
12.1

Appendix
A Scientic Wordprocessing with L TEX

A This pdf document with its hyperlinks was created using L TEX which is the standard (free) mathematical wordprocessing package; more information can be found via the webpage: http://www.ma.umist.ac.uk/kd/latextut/pdfbyex.htm

12.2

Computer Algebra Methods

The computer algebra package Mathematica can be used to manipulate and invert matrices. See: http://www.ma.umist.ac.uk/kd/mmaprogs/AREADMEFILE for beginning Mathematica. Similarly, Maple and Matlab also can be used for working with matrices.