MATLAB Basics - Help Sheet

David Lindsay

November 7, 2002

1
2 LIST OF FIGURES

Contents
1 Explanation of MATLAB screen layout 3

2 Essentials for creating basic programs 4

3 Using matrices in MATLAB 5

3.1 Basic matrix notation 5
3.2 MATLAB’s built in matrix functions 5
3.3 Manipulating matrices in MATLAB 7
3.3.1 Using the colon to create ranges 7
3.3.2 Accessing and manipulating elements of matrices 8
3.3.3 Concatenating matrices 8
3.3.4 Deleting rows and columns 9
3.3.5 Filtering content of matrices 9
3.3.6 Formatting output 9

4 Plotting graphs 10
4.1 How to use the plot command 10
4.1.1 Setting the range 10
4.1.2 Selecting colour 10
4.1.3 Selecting line style 11
4.1.4 Selecting marker types 11
4.2 How to set axis titles 12
4.3 Building up multiple plots using hold 12

5 Basic programming 13
5.1 The if statement 13
5.2 The switch statement 14
5.3 The while and for loops 14
5.4 The break and continue statements 14

6 The quadprog optimiser 15

List of Figures
1 The MATLAB desktop layout 3
2 Example of a multiple plot 13
3 Plot of the function f (w) = 21 w12 + w22 − w1 w2 − 2w1 − 6w2 17
 Explanation of MATLAB screen layout 3

Aim of this document

MATLAB is a very ‘weakly’ typed programming language, as compared to
languages such as C++ and java. It does not require formal type or
dimension declarations such as int, float, double, char etc. When MATLAB
encounters a new variable it automatically allocates the appropriate amount
of storage. If the variable already exists, MATLAB changes its contents and if
necessary allocates new storage. All of this can lead to some initial confusion,
as seemingly short code seems to magically perform wonderfully complex
operations. The aim of this handout is to introduce you to some of the basic
(and most important) features of MATLAB using simple examples.

1 Explanation of MATLAB screen layout

To start MATLAB type in matlab at the linux command prompt, you will
then be presented with the MATLAB desktop environment as seen in Figure 1.

Figure 1 The MATLAB desktop layout: On starting up MATLAB you

will be presented with a basic windows environment like so. Some of the most
important features are highlighted above.

An explanation of these tools is as follows:

Command window This is where all the magic happens; this is where you
will enter data, commands, run functions, see results etc.
4 LIST OF FIGURES

Command history Log of lines you enter in the Command window,

which allows you to copy/select/re-execute lines easily.
Launch pad Provides easy access to other MATLAB tools (eg. Optimisation
toolbox you will use for your coursework), as well as demos and
documentation.
Help browser This allows you to search and view all documentation on
MATLAB products.
Current directory browser This specifies which directory you are
currently working in. Make sure this is pointing to the directory where
your files are located.
Workspace browser This lists the names and sizes of arrays being used in
your current session. This can be used as a basic visual debugging
environment of your code.

2 Essentials for creating basic programs

You can either create your programs by typing in the commands directly into
the MATLAB command window (one line at a time), or preferably structure
the lines of code in a “.m” file, and then call the file within the Command
window.
You could structure your MATLAB program in a file using MATLAB’s text
editor by clicking on File→New→M-file. For example, try to create the file
helloworld.m containing the code shown below:
x = (-10: 0.1 :10);
y = x.^2;
plot(x,y,’r-’);
title(’My first MATLAB program’);
Then (making sure that the current working directory is pointing to where the
file is located) type helloworld in the command window to run the program.
To add comments to your code type the % character and whatever follows on
that line will be ignored by MATLAB
If you terminate a line statement with the ; character then MATLAB will not
display the results of executing that particular line of MATLAB code. For
example if a line of code specifying a matrix A does not end with a semi colon
then the contents of matrix A will be output in the Command window.
This can be useful for debugging your code, and also cutting down the output
of the program.
To find a formal specification about any of MATLAB’s built in functions type
at the Command window, replacing functionname with the desired
function:
help functionname
 Using matrices in MATLAB 5

3 Using matrices in MATLAB

These are probably the most important core components of MATLAB and
some basic understanding of matrix manipulations is required to use them
effectively. Pretty much every operation you perform will involve
manipulation of matrices and column/row vectors to derive solutions.

You can enter matrices with MATLAB in several ways:

1. Enter an explicit list of elements (using brackets []).

2. Load matrices from external data files (using the load command).

3. Generate matrices using MATLAB’s built in functions such as eye, diag

and ones etc.

4. Create matrices with your own functions in m-files.

3.1 Basic matrix notation

Matrices use the following format:
 Separate the elements of a row with a space or a comma.

 Use ; to indicate the end of a row of a matrix.

 Surround the entire list of elements with brackets [].

 Adding ; at the end of a statement triggers the ‘don’t display’ option.

An example of explicitly creating a matrix is as follows:

A = [16 3 2 13; 5 10 11 8; 9 6 7 12; 4 15 14 1]

This will create the matrix A:

 
16 3 2 13
 5 10 11 8 
A=
 
9 6 7 12 


4 15 14 1

Note: If we added the ; to the end it would not display the matrix in the
command window.

3.2 MATLAB’s built in matrix functions

As mentioned earlier, there are many built in MATLAB functions which are
very useful. A basic example and description of these functions will be
provided below. For further details about each function use MATLAB’s help
browser, or at the command window eg. help eye. You will find most of
these functions can be used on both matrices and individual numbers. The
best way to learn is to experiment with them.
6 LIST OF FIGURES

B=sum(A) row matrix containing the column sums of A

prod(A) (when A is a list) returns the product of the array elements in A

prod(A) (when A is a matrix) returns row matrix whose elements are products of
column elements.

A’ creates the transpose of matrix A (swap rows for

columns)

inv(A) create the inverse of matrix A (eg. A−1 A = I)

det(A) calculates the determinant of a square matrix A

A+B addition of matrices A and B

A+2.1 adds 2.1 element wise to matrix A

A-B subtraction of B from A

A*B multiplication of A and B

A\B same as inv(A)*B

A^3 power of A, same as A*A*A

B=sum(A’)’ stores rows of sums of A

b=diag(A) stores the diagonal elements of a matrix A in the row

vector b.

B=diag(a) creates a matrix B which has row vector a’s elements

as its leading diagonal (all other elements = 0).

sum(diag(A)) sum of the diagonal elements of matrix A

0 0 0 0
zeros(2,4) creates matrix
0 0 0 0

5 5
5*ones(3,2) creates matrix 5 5
5 5

1 0 0
eye(3) creates a 3 × 3 identity matrix 0 1 0
0 0 1

[n,m]=size(A) calculates the dimensions of the matrix A, the number

of rows =n, the number of columns = m
 Using matrices in MATLAB 7

max(A) returns the largest value(s) within each column of a

matrix

min(A) returns the smallest value(s) within each column of a

matrix

mean(A) returns the average(s) of each column of a matrix

std(A) returns the standard deviation(s) of each column of a

matrix

rand random number generator: creates uniformly between

0 and 1

randn random number generator: creates using normal dis-

tribution

fix rounds numbers downwards towards 0

rand(4,5) 4×5 matrix of random numbers

pi constant for defining π = 3.14159265 . . .

Inf constant for defining infinity

NaN constant for defining not a number

A.*3 performs element by element wise multiplication by 3

of matrix A

A.^2 performs element by element wise raising to the power

2 (squaring) of matrix A

sqrtm(A) matrix of square roots of elements of A

eig(A) gives the eigenvalues of A

3.3 Manipulating matrices in MATLAB

Along with basic functions there are some commonly used features of matrices
in MATLAB which can save a lot of time when writing programs.

3.3.1 Using the colon to create ranges

Often in MATLAB you may want to create a range that you can pass to the
plot command to specify which values to display the desired function for.
8 LIST OF FIGURES

This could be done explicitly by specifying a list of numbers, but this can
become tedious if you wish to plot over a large range. This can be easily
achieved using the : colon symbol like so:
A=1:10 creates a row vector A of integer values ranging from
1 to 10 like so

A=[1 2 3 4 5 6 7 8 9 10]

A=100:-7:50 creates a row vector A of integer values taking 7 away

each time, starting from 100, but greater than 50
A=[100 93 86 79 72 65 58 51]

A=0:pi/4:pi creates a row vector A containing a range of values

from 0 to π, using increments of π4
A=[0 0.7854 1.5708 2.3562 3.1416]

3.3.2 Accessing and manipulating elements of matrices

Another useful function of the colon symbol is to add/remove rows or

columns, and access individual elements of a matrix like so:
A(1,4)=39.7 This will set the element row 1, column 4 to be value
39.7

B=A(1:k,j) stores the first k elements of the jth column of A in B

B=A(:,end) stores all the elements of the last column of A in B

B=A(:,[permutation]) stores A in B interchanging columns according to the

given permutation

B=A(:,7) stores the 7th column of A in B

B=A(5,:) stores the 5th row of A in B

3.3.3 Concatenating matrices

To create even more complex matrices, it is often useful to concatenate two

(or more) simpler matrices. This is achieved by expressing the new matrix as
a list (using the [] brackets and ; semi colon) of matrices like so:
 Using matrices in MATLAB 9

A=[1 2 3] this will create a row vector 1 2 3

4 5 6
B=[4 5 6; 7 8 9] this will create a 2×3 matrix
7 8 9

C=[A; B] this will create a 3×3 matrix C, with A as the first

1 2 3
row and B as the second and third rows 4 5 6
7 8 9

3.3.4 Deleting rows and columns

To delete rows or columns from a matrix use the following techniques:

A(:,2) = [] deletes the 2nd column of matrix A

A(2:3:12)=[] combining the range technique earlier, this will remove

columns 2, 5, 8 and 11 from matrix A

3.3.5 Filtering content of matrices

Often you will need to be able to search/extract a matrix for specific values.
You may also wish to find the position index of these values in that matrix.
This can be achieved like so:
A=[2.1 1.7 1.6 1.5 1.9 1.8 1.5 5.1 1.8 1.4 2.2 1.6 1.8]
explicitly creates the matrix A

A=A(A<2)
this searches and removes all elements in A which are greater than or
equal to 2
A=[1.7 1.6 1.5 1.9 1.8 1.5 5.1 1.8 1.4 1.6 1.8]

A=find(isprime(10*A))
this searches through matrix A (each element multiplied by 10) for all
elements which are prime and returns their index, eg. 1.9×10=19 which
is prime and has the 4th position in matrix A
A=[1 4]

3.3.6 Formatting output

Sometimes it may be useful to convert numbers into a more appropriate

number format. There are many different types available, some of which are
demonstrated below:
10 LIST OF FIGURES

A=[4/3 0.0000123456 log(2)] explicitly creates the matrix A

A=[1.3333 1.2346e-005 0.69315]

format short this rounds numbers to 4 decimal places

A=[1.3333 0.000 0.6931]

format long this rounds numbers to 14 decimal places

A=[1.33333333333333 0.00001234560000 0.69314718055995]

format bank this rounds numbers to 2 decimal places

A=[1.33 0.00 0.69]

format rat approximates numbers as rational fractions

A=[4/3 2/162001 1588/2291]

4 Plotting graphs

4.1 How to use the plot command

The plot command has the following simple layout:
plot(x,y,’ colour style marker ’)
The x and y is the range of values to be plotted, these are specified as row
vectors. The x values are plotted horizontally, and the y values are plotted
vertically.
These (x, y) coordinate pairs are then plotted on a 2 dimensional plot.
Further settings such as connecting the points with a line, the type of markers
used for each point and the colour of these lines and markers can be specified
as mentioned below.

4.1.1 Setting the range

It is very important that the range of values is specified correctly. You must
make sure that the row vectors x and y are the same dimension, so that it can
properly group each pair of coordinates.

4.1.2 Selecting colour

The colour of the line and markers can both be set using the following codes:

 c = cyan

 m = magenta
 Plotting graphs 11

 y = yellow
 r = red
 g = green
 b = blue
 w = white
 k = black

4.1.3 Selecting line style

If no line is specified to connect the points then the resulting plot will look
like a scatter plot of individual points. The type of the line used to connect
the points can be specified using the following codes:

 - = solid
 -- = dashed
 : = dotted
 -. = dash-dot
 none = no line

4.1.4 Selecting marker types

The type of marker that is used to draw each point can be specified using the
following codes:

 + = plus sign
 o = circle
 * = star
 x = cross
 s = square
 d = diamond
 ^ = up-triangle
 v = down-triangle
 > = right-triangle
 < = left-triangle
 p = pentagram
 h = hexagram
12 LIST OF FIGURES

4.2 How to set axis titles

To make the graphs that you produce easier to interpret you can introduce
titles to your plot. Remember that you can put latex into these strings, to
introduce mathematical notation. Some of the most useful title commands are
displayed below.
title(’Nice graph of y=x^{2}+\alpha’,’FontSize’,12,’FontWeight’,’bold’)
this will create the plot title to be “Nice graph of y = x2 + α”, with font size
= 12pt and font wieght = bold

xlabel(’w_{1}’,’FontSize’,11)
this will place the text w1 on the horizontal axis

ylabel(’w_{2}’,’FontSize’,11)
this will place the text w2 on the vertical axis

4.3 Building up multiple plots using hold

In many applications you will probably want to superimpose many plots on
top of each other to accurately compare results. If you were to type multiple
plot commands one after another you would normally just be left with the last
plotted graph. To superimpose you must therefore place the text hold on
before all of your plot commands, and then type hold off at the end.
This following example demonstrates all of the previous sections in
constructing the simple graph shown in Figure 4.3.
x=-4:0.5:4;
sets the range for the x values from -4 to 4 using increments of 0.5
y1=(x.^2)-8;
specifies the first function y = x2 − 8
y2=x;
specifies the second function y = x
hold on;
suppress any plot commands
plot(x,y1,’r-+’);
plot the first function
plot(x,y2,’b:s’);
plot the second function
title(’Graph comparing y=x and y=x^{2}-8’,’FontSize’,12,’FontWeight’,’bold’)
xlabel(’x’,’FontSize’,11)
ylabel(’y’,’FontSize’,11)
set appropriate titles
hold off;
allow graphs to be plotted
 Basic programming 13

Figure 2 Example of a multiple plot : This is a very basic plot comparing

two functions.

5 Basic programming
For completeness MATLAB offers many of the basic programming statements
and routines commonly used in languages such as C++ and java. A brief
example of each of these is given below.

5.1 The if statement

Much like if statements in many programming languages MATLAB requires
the following format:
if logical expression
statements
elseif logical expression
statements
else
statements
end
The following comparison operators can be used:
== = equals
~= = not equals
> = greater than
>= = greater than or equal to
<= = less than or equal to
< = less than
~< = not less than
To compare the equality of two matrices A and B use:
14 LIST OF FIGURES

if isequal(A,B)

5.2 The switch statement

To avoid messy nested if statements use the switch statement like so:
switch expression
case value1
statements
case value2
statements
.
.
otherwise
statements
end

5.3 The while and for loops

The for loop must be provided a range like so:
for i = 3:2:77
statements
end
The while loop requires a logical expression like so:
while logical expression
statements
end

5.4 The break and continue statements

The continue statement passes control to the next iteration of a for or
while loop in which it appears.
The break statement lets you exit early from a for or while loop.
 The quadprog optimiser 15

6 The quadprog optimiser

The quadprog MATLAB routine solves quadratic optimisation
  problems. The
w1
 w2 
 
function try’s to solve for a vector of n variables w =  . 

, that minimises
 .. 
wn
the function:
f (w) = 12 w0 Qw + u0 w
subject to the constraints
A0 w ≤ a
B0w = b
c≤w≤d

Note that in general solving n variables, with m constraints:

Q is an n × n matrix, u is a n × 1 matrix used to specify the main term in the
optimisation problem
A is an m × n matrix, a is a m × 1 matrix used to specify the ≤ side
constraints (note that these can be simply reversed to ≥ side constraints by
multiplying by -1).
B is an m × n matrix, b is a m × 1 matrix used to specify the = side
constraints.
c is an n × 1 matrix, d is a n × 1 matrix used to specify the lower and upper
bounds of the side constraints.
This can then be solved using the following MATLAB code:
w=quadprog(Q,u,A,a,B,b,c,d)
This will solve the optimisation problem and store the solution in the n × 1
matrix w.
Please note that you do not have to use all of these inputs to quadprog. For
example if you do not require the additional constraints such as B 0 w = b and
c ≤ w ≤ d then you could either enter them as matrices or don’t include them
at all eg. quadprog(Q,u,A,a) or quadprog(Q,u,A,a,[],[],[],[]).
16 LIST OF FIGURES

Let us consider a practical example of solving the following quadratic

optimisation problem:

f (w) = 12 w12 + w22 − w1 w2 − 2w1 − 6w2

subject to the constraints
w1 + w2 ≤ 2
−w1 + 2w2 ≤ 2
2w1 + w2 ≤ 3
w1 ≥ 0
w2 ≥ 0

Using the same matrix notation as used earlier for the quadprog function we
can formulate the problem using the following matrices:
   
! ! 1 1 2 !
1 −1 −2 0
Q= , u= , A =  −1 2  , a =  2  , c =
   
−1 2 −6 0
2 1 3

This can therefore be solved using the following MALTAB code:

Q=[1 -1; -1 2] % defines the quadratic part of the main optimisation term
u=[-2; -6] % defines the linear part of the main optimisation term
A=[1 1; -1 2; 2 1] % defines the less than or equal constraints
a=[2; 2; 3]
c=[0; 0] % defines the greater than or equal constraints
w=quadprog(Q,u,A,a,[],[],c,[]) % solves for the optimal w vector!
0.6667
The corresponding output of this code gives the solution for w = .
1.3333
If we look at a plot of this function as a 3-dimensional surface shown in Figure
3 we can see that these coordinates do indeed correspond to a minima. This
function was plotted in MATLAB using the code:
ezmeshc(’(0.5*x.^2)+(y.^2)-(x*y)-(2*x)-(6*y)’,[-3,3,-3,3])
 The quadprog optimiser 17

Figure 3 Plot of the function f (w) = 12 w12 + w22 − w1 w2 − 2w1 − 6w2 : This
plot clearly demonstrates a local minima in the surface of the function f (w).
Notice that the solution given by quadprog is located in the low blue area of
this plot.