MAMA5 CTP

Matlab
Notes for use with Matlab version 5.

1 Introduction
Matlab is a program that has been written to make certain numerical calculations
easier on a computer. It enables us to use the computer as a high-powered ‘pro-
grammable, graphical calculator’. One of the main strengths of Matlab (and the
feature it is named after) is its ability to manipulate matrices.
The command line is where we type our commands; it starts with the prompt >>.
Commands typed in at the command line are not executed before a press of the RE-
TURN key.
In what follows pieces of text that appear in this font show what we should see on
the computer screen as these notes are followed. Anything following the >> prompt
represents what is typed in at the Matlab command line.

1.1 Further reading

All the information you require for this course is contained within these notes (either
explicitly or via on-line help facilities built in to Matlab).
Further information, including some relatively advanced applications, can be found in
“Numerical methods using Matlab” by John Penny and George Lindfield.
This book is in the library under classmark 517.6018-PEN.

1.2 Getting help and getting out

Matlab has built-in facilities for giving help. If, for example, we require more infor-
mation about the command trace we could type

>> help trace

If the name of a particular command is not known, but we do know (for example) that
matrix inverses are of particular interest we can type

>> lookfor inverse

This will cause Matlab to search through all1 of its files (this can take a while!) and
report back giving the names of commands that might be what is required. In the
example above Matlab returns over 20 possibilities including things such as ‘inverse
hyperbolic cosine’. A quick look through the list tells us that inv is the command we
wanted. If more information is required then use the help feature.

One other fact worth knowing before we begin is that to exit Matlab type in the
command quit.

1
There is a technical issue here concerning ‘search paths’, but we can gloss over that point for the
purposes of this course

1
2 Vectors and matrices
Most work carried out in Matlab is done using vectors and matrices. We therefore
begin by seeing how these quantities can be defined and manipulated.

2.1 Basics
A row-vector can be created by specifying the first entry, an increment and the last
entry, for example
>> 1:3:10
ans =
1 4 7 10
(The final entry in the resulting vector is arranged to be the largest member of the
arithmetic progression which does not exceed the specified last entry, for example
>> 1:3:15
ans =
1 4 7 10 13
The next entry in the progression (the value 16) is larger than 15 and is therefore not
included.) Vectors can also be defined using square brackets (with entries separated
by spaces or commas)
>> [11 2 -3 1 0]
ans =
11 2 -3 1 0
>> [5,6,-9,12,3]
ans =
5 6 -9 12 3
Arithmetic calculations can be carried out on this result
>> 2*ans
ans =
10 12 -18 24 6
>> ans + 5
ans =
15 17 -13 29 11
>> sum(ans)
ans =
59
We can give names to the quantities we manipulate in Matlab for easier reference.
(If no name is given to the last result then Matlab will call it ans - short for answer.)
>> a = [1 3 4 7]
a =
1 3 4 7
>> b = [2 5 6 2];
The semicolon prevents Matlab from echoing the last result to the screen (this is a
useful feature to know about when we manipulate large vectors and matrices).

2
Matrices can be entered in a similar way. Either of the commands

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

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

will create a 3 × 3 matrix A. We can use either a semicolon or a press of the RETURN
key to separate rows. It is possible to alter parts of the matrix as follows

>> A(3,3) = 9
A =
1 2 3
4 5 6
7 8 9

More details of this sort of procedure can be found in §2.8.

The 2 × 3 matrix B defined below will appear in some examples to follow.

>> B = [1 2 5;3 5 -6]

B =
1 2 5
3 5 -6

2.2 Exercises
1. Define a matrix C by

>> C = magic(15)

Use Matlab to verify that C has equal row, column and diagonal sums.

2. The commands who and whos are useful. Type them in and see what happens.

3
2.3 Built-in functions
Matlab has many built-in functions, most with obvious names. Suppose that we want
to know the cosine of each entry (in radians) of B above

>> cos(B)
ans =
0.5403 -0.4161 0.2837
-0.9900 0.2837 0.9602

More decimal places can be shown by altering the format of the output

>> format long

>> ans
ans =
0.54030230586814 -0.41614683654714 0.28366218546323
-0.98999249660045 0.28366218546323 0.96017028665037

The command

>> format short

returns us to the default output style (which is assumed in the rest of these notes).
There are other styles of format available, the help command will list these. It should
be pointed out that when using the default format Matlab performs calculations to
its maximum accuracy - only the output is truncated to a small number of significant
figures.

2.4 Exercises
1. Create a vector x containing the values 0.1, 0.2, 0.3, . . . , 0.6 and hence find the
following functions of the entries

(a) The hyperbolic sine.

(b) The inverse tangent.
(c) The natural logarithm.
(d) The base-10 logarithm.
20
X
2. Evaluate sin(xi ) in which xi = iπ/20
i=1
P20
Answer: i=1 sin(xi ) =

3. The format command was used above to increase the number of decimal places
displayed by Matlab. It may also be used to give rational number output. Find
a rational number approximating cos(π/6).
Answer: cos(π/6) ≈

4
2.5 Operations
The following matrix operations are available in Matlab: + (addition), − (subtrac-
tion), * (multiplication), 0 (transpose),ˆ(power), \ (left division) and / (right division).
Multiplication and addition follow the usual rules for matrices. Using the 3 × 3 matrix
A and the 2 × 3 matrix B defined above consider the following Matlab session

>> A+B
??? Error using ===> +
Matrix dimensions must agree.
>> A*A
ans =
30 36 42
66 81 96
102 126 150
>> A^2 - A*A
ans =
0 0 0
0 0 0
0 0 0
>> B*A
ans =
44 52 60
-19 -17 -15
>> A*B
??? Error using ===> *
Inner matrix dimensions must agree
>> B^2
??? Error using ===> ^
Matrix must be square

The three error messages above were caused by us asking Matlab to perform mathe-
matically invalid operations.
If A is an invertible n × n matrix and b is a vector of the appropriate size then

>> x = A\b

gives the n-row solution x to A*x=b and

>> x = b/A

gives the n-column solution x to x*A=b.

The command x = inv(A)*b has the same effect as x=A\b. The command x = b*inv(A)
has the same effect as x = b/A. (In each case we again assume that b is of an appropri-
ate shape.) Calculating inverses can be very time-consuming for large matrices, left-
or right-division is more efficient.

5
The command ’ gives the transpose of a real matrix. Using the matrix A we defined
earlier,

>> A
A =
1 2 3
4 5 6
7 8 9
>> A’
ans =
1 4 7
2 5 8
3 6 9

as we would expect. What the ’ command actually does is give the complex-conjugate
transpose, so care is required when dealing with complex numbers. Consider

>> c = [1+i 2-2i 3-4i]

c =
1.0000+ 1.0000i 2.0000- 2.0000i 3.0000-4.0000i
>> c’
ans =
1.0000- 1.0000i
2.0000+ 2.0000i
3.0000+ 4.0000i

Use the following for non-conjugate transpose

>> c.’
ans =
1.0000+ 1.0000i
2.0000- 2.0000i
3.0000- 4.0000i

There are other examples of when a full-stop changes the effect of an operation as we
see after this exercise.

2.6 Exercise
Use Matlab to find the solution of the system of equations
    
1 2 3 4 w 7
 2 4 6 6  x   2 
    
  = .
 1 2 4 3  y   3 
3 4 6 2 z 0

Answers: w = , x= , y= , z= .

6
2.7 ‘Pointwise’ operations
If we precede the multiplication, power or either division operation with a full stop
then the operation is carried out separately on each of the positions in the matrix, for
example (using A again)

>> A.^2
ans =
1 4 9
16 25 36
49 64 81
>> [1,2,3].*[4,7,4]
ans =
4 14 12
>> [15 14 17]./[3 2 17]
ans =
5 7 1
>> [2 3 4].\[10 12 20]
ans =
5 4 5
>> [1 2 3]./[4 5]
??? Error using ==> ./
Matrix dimensions must agree

Example
1 + cos2 (x)
Consider the function f (x) = for x taking certain values between 0 and 1.
cosh(x)
We might define vectors x and f as follows.

>> x = 0:.2:1
x =
0 0.2000 0.4000 0.6000 0.8000 1.0000
>> f = (1 + cos(x).^2)./cosh(x)
f =
2.0000 1.9220 1.7097 1.4182 1.1106 0.8372

The use of two pointwise operations in the definition of f enables us to easily evaluate
the function over the given range of x values.
Pointwise operations are an exceptionally useful feature when using Matlab graphics
as we will see later on.

7
2.8 Referencing parts of matrices
Let us define a vector containing the integers −5 to 5

>> d = -5:5
d =
-5 -4 -3 -2 -1 0 1 2 3 4 5

We can pick out parts of this vector as follows

>> d(1)
ans =
-5
>> d(2:4)
ans =
-4 -3 -2
>> [d(1) d(10) d(11)]
ans =
-5 4 5
>> d([1 10 11])
ans =
-5 4 5
>> d([1:2:9])
ans =
-5 -3 -1 1 3

Returning to our matrix

>> A
A =
1 2 3
4 5 6
7 8 9

We can isolate elements, rows or columns as shown next.

>> A(2,3)
ans =
6
>> A(:,2)
ans =
2
5
8

A colon in this context can be thought of as ‘every’ so that the command above could
be read as ‘matrix A, every row, second column’.

8
We can reference rows in a similar way, the following command could be read as ‘matrix
A, third row, every column’
>> A(3,:)
ans =
7 8 9
The next command gives entries on the third row in the first and third columns.
>> A(3,[1 3])
ans =
7 9
As a final example of this point we now select entries on the first two rows in all
columns
>> A([1 2],:)
ans =
1 2 3
4 5 6
For further illustration suppose that we want to perform row operations on
>> A
A =
1 2 3
4 5 6
7 8 9
to find an upper triangular matrix that is row-equivalent to A. The row operations
r2 → r2 − 4r1 and r3 → r3 − 7r1 will create zeros below the diagonal in the first column
and may be carried out in Matlab with the two commands
>> A(2,:) = A(2,:) - 4*A(1,:)
A =
1 2 3
0 -3 -6
7 8 9
>> A(3,:) = A(3,:) - 7*A(1,:)
A =
1 2 3
0 -3 -6
0 -6 -12
Finally we eliminate the non-zero below -3 with the row operation r3 → r3 − 2r2
implemented by the command
>> A(3,:) = A(3,:) - 2*A(2,:)
A =
1 2 3
0 -3 -6
0 0 0
which is an upper-triangular matrix.

9
Reference to matrices can also be carried out with only one input (rather than one
each for the row and column), the thing to remember is that Matlab counts down
the columns so that if we define

>> B = [7 -8 3;4 6 9;1 2 5]

B =
7 -8 3
4 6 9
1 2 5

then Matlab gives the following responses to the inputs shown

>> B(3)
ans =
1
>> B(4)
ans =
-8
>> B([5:8])
ans =
6 2 3 9
>> B(1:9)
ans =
7 4 1 -8 6 2 3 9 5

It is possible to get Matlab to list the entries of a matrix in the order that it has
them stored, the command

>> B(:)

gives, as output, the transpose of the previous output. In this context the colon means
‘list all the entries of B in a column’.

2.9 Exercises
1. Using the matrix C = magic(10) create a column vector z which is equal to the
3rd column of C plus the transpose of the 6th row. What are the minimum and
maximum entries in z?
Answers: minimum = , maximum = .

2. Use the help command to tell you about the Matlab function eig. Hence find
the eigenvalues and eigenvectors of the matrix
 
1 2 8 4
 2 4 6 7 
 
A= .
 1 2 5 3 
2 4 5 5

10
    
   
   
   
   
   
   
   
   
Answers: λ1 = , x1 =  , λ2 = , x2 =  
   
   
   
   
   
   
   

   
   
   
   
   
   
   
   
   
λ3 = , x3 =  , λ4 = , x4 =  
   
   
   
   
   
   
   

Use Matlab to verify that Axk = λk xk (k = 1, 2, 3, 4).

Is (2, −1, 0, 0)T an eigenvector of A? Answer:
Is (1, 1, 0, −1)T an eigenvector of A? Answer:

3. What is the rank of the matrix A that appeared in the previous question?
Answer: rank(A) = .

4. Use the help feature to familiarise yourself each of the following commands.
(Space has been left for you to make additional notes.)

• eye

• zeros

• ones

• size

• length

• diag

• sort

11
 • diff

• rem

• disp

5. Using some of the commands above create a 20 × 20 matrix K with the following
properties (
i + j if i = j
(K)ij =
1 otherwise

Hence find the largest eigenvalue of K.

Answer:

12
3 Graphics
We now describe how Matlab can be used to create graphical output.

3.1 Two-dimensional graphics

Let us start with an example, we create a vector of linearly spaced points between 0
and 2π

>> x = linspace(0,2*pi);
>> plot(x,cos(x))
>> plot(x,sin(x))

The first graph was replaced by the second, we can hold on to the current graph and
cause subsequent plots to appear overlayed as follows

>> hold on
>> y = sin(x).^2;
>> plot(x,y,’r-.’)
>> title(’Plots of sine and sine squared’)

(Notice that we used .ˆ for squaring sin(x) to ensure that this operation was per-
formed entrywise.)
Use the help plot command to see what other options are available. There is a
selection of line styles and colours to choose from.

3.2 Exercise
2)
Create a plot of the function y(x) = e(−x for x in the range −0.5 to 1.5. Your plot
should have the following features

• The curve should be a black broken line.

• The y-axis should be labelled ‘This is the y axis’.

• There should be a grid of dotted lines in the background corresponding to the

x- and y-axis ticks.

• The range of x-values in the graph should be −0.5 to 1.5.

• The range of y-values should be 0 to 1.

• The aspect ratio should be set so that equal increments on the x and y axes
appear equal on your graph.

Hint: you may have to resort to the lookfor and help commands to achieve all of
this.

13
3.3 Functions of two variables
Consider the function
f (x, y) = x2 − ( y4 )2 + ( y5 )4
for −1 ≤ x ≤ 1 and −4 ≤ y ≤ 4. (It can be shown that this function has a saddle point
at (x, y) = (0, 0).) We can begin by defining vectors of appropriate x and y values
>> x = linspace(-1,1,25);
>> y = linspace(-4,4,25);
(These commands will define x and y as vectors with 25 entries, larger numbers of
points will produce smoother graphics, but using too many points can cause printing
to be very slow.) To obtain a plot of f we need to evaluate the function at every
x,y combination. This is best done by defining matrices X and Y with the meshgrid
command as follows
>> [X,Y] = meshgrid(x,y);
We can now evaluate the function f
>> f = X.^2 - (Y/4).^2 + (Y/5).^4;

3.4 Exercise
Use the built-in help command to explain more about meshgrid. Examine the re-
sponse to the command
>> [a,b] = meshgrid([1 2 3],[8 9])

3.5 Three dimensional graphics

There are many commands that can be used to generate and manipulate plots of
f (x, y). Try each of the following and note its effect - don’t forget the help command
for extra information.

>> surf(X,Y,f)
>> colorbar
>> shading interp
>> view(-40,30)
>> view(-45,30)
>> close

>> surfl(X,Y,f)
>> shading interp
>> colormap(copper)
>> colormap(pink)
>> close

>> contour(X,Y,f)
>> colorbar
>> C = contour(X,Y,f);
>> clabel(C)

14
3.6 Exercises
1. Create a surface plot of the function

3y 4 − 4y 3 − 12y 2 + 18
f (x, y) = (−1 ≤ x ≤ 1, − 1.5 ≤ y ≤ 2.5)
12(1 + 4x2 )

adjusting the viewing angle to best effect. Which values of az and el in the view
command did you decide gave the best view?
Answers: az = , el = .

2. Create a contour plot of the above function arranging for there to be 20 contour
lines.

3. (a) The function f has one saddle point, where is this? (You need not find the
position of the saddle point analytically, try to ‘spot’ the location of this
point from the graphs you have produced.)
Answer: saddle point is at (x, y) = .
(b) What value does f take at the saddle point?
Answer: value of f at the saddle point = .
(c) Produce a contour plot with only one contour line, this corresponding to
the value f takes at its saddle point.

15
4 Logical operations
Matlab can deal with logical expressions, that is expressions that are either true or
false. This is achieved by attaching a numerical value to the words ‘true’ and ‘false’;
a value is ‘true’ if and only if it is non-zero. By way of illustration let us define two
vectors
>> u = [1 0 2 4 0 2];
>> v = [5 6 1 0 0 7];

• The following command “u and v” gives, as response, the value 1 (true) if and
only if the corresponding entries in u and v are both non-zero.

>> u & v
ans =
1 0 1 0 0 1

• This next command is “u or v”

>> u | v
ans =
1 1 1 1 0 1

• “not u”

>> ~u
ans =
0 1 0 0 1 0

• This command asks if the entries of u are equal to 2 (note the double equal sign)

>> u == 2
ans =
0 0 1 0 0 1

• This next example demonstrates the ‘exclusive or’ function which is true if one
argument is zero and the other argument is non-zero

>> xor(u,v)
ans =
0 1 0 1 0 0

• Are any of the entries of v non-zero?

>> any(v)
ans =
1

• Are all of the entries of v non-zero?

>> all(v)
ans =
0

16
4.1 The find command
The find function identifies places where a logical expression is true, for example
>> find(u == 2)
ans =
3 6
which tells us that the 3rd and 6th entries in u are equal to 2. We can use this feature
to create index arrays as we show next, consider
>> A = magic(4)
A =
16 2 3 13
5 11 10 8
9 7 6 12
4 14 15 1
The following command identifies which entries are greater than 13
>> A>13
ans =
1 0 0 0
0 0 0 0
0 0 0 0
0 1 1 0
The array index defined below identifies these positions
>> index = find( A>13 )
index =
1
8
12
(remember that Matlab indexes matrices by counting down the columns). We can
now use this index array to alter A
>> A(index) = 999
A =
999 2 3 13
5 11 10 8
9 7 6 12
4 999 999 1

4.2 if, else and elseif

if evaluates a logical expression and executes a group of statements based on the value
obtained. In its simplest form the syntax is typified by the following example in which
a is an integer
>> if rem(a,2) == 0
disp(’the value is even’)
end
Further details and a more complicated example are available via help.

17
4.3 Examples
These two examples should help you with the exercises that follow.

1. The following command produces a row vector found by summing down the
columns of the matrix A ignoring any negative values.

>> sum( A.*(A>0) )

2. The command

>> all(diff(sort(a)))

in which a is a vector, returns the value 0 if and only if a contains a repeated

entry.

4.4 Exercises
In the spaces provided write down commands which do the following tasks.

1. Devise a command which returns 0 (false) if a matrix contains a repeated entry

and 1 (true) if a matrix contains distinct values.
Command:

2. Find a command which returns 1 if and only if each column of the matrix contains
distinct entries.
Command:

3. Use a one-line command to sum down the diagonal of a square matrix where
the sum is taken only over values larger than 1 in magnitude (i.e., values in
(−∞, −1) ∪ (1, ∞).)
Command:

4. Devise a one-line command which overwrites the diagonal entries of a square

matrix so that the resulting matrix has every row sum equal to 1.
Command:

Test each of your commands on these two matrices

 
  1 8 2 1
1 4 7  
   3 2 −3 6 
 2 −3 6  ,  .
 6 7 1 4 
7 1 4
7 2 3 8

18
5 M-files
M-files are files containing lines of Matlab commands. There are two different kinds
of m-file.

5.1 Script files

These are simply lists of Matlab commands gathered together in a file. If there is
a sequence of commands that will be executed several times then it is often sensi-
ble to write these commands together in a script file and then access these commands
via the file name. For example, create a file called ma3vex.m with the following contents

x = linspace(0,pi);
plot(x,exp(-0.5*x).*cos(5*x),’g’)
hold on
plot(x,exp(-0.5*x).*sin(5*x),’r’)
legend(’First solution’,’Second solution’)
axis equal
axis([0 pi -1 1])

This plots two solutions of a particular differential equation. To make Matlab run
through these commands we type the file name (omitting the .m part) at the prompt,
in this case

>> ma3vex

Any sequence of Matlab commands can be collected together in a file and run in this
way. Such m-files are called scripts, or script files. Script-files behave as if we had
typed in each line at the >> prompt. One important thing to note is that script-files
‘know about’ all of the variables in the workspace (that is, the list of variables shown
in response to the command who), this is in contrast to function m-files as we see next.

5.2 Functions
Functions require one or more input arguments, they are self-contained and do not
use Matlab variables in the workspace. The only variables that a function will know
about are those passed through the argument list or those defined as global. We will
not discuss global variables in this course.
The following is a simple example of a function m-file which calculates the sum of the
squares of the eigenvalues of a square matrix. This should be saved in the file name
sum2eig.m in your file space.

19
 function s = sum2eig(A)
% SUM2EIG Sum of squares of eigenvalues
% sum2eig(A) returns the sum of the eigenvalues squared of A

% An error message is returned (by eig) if a non-square

% matrix is entered

s = sum( eig(A).^2 );

Note that Matlab ignores any statement on a line after the % character. We use this
feature to put comments into m-files.
This function has some elements that are common to all Matlab functions

• A function definition line: function s = ....

This line defines the function name and the input/output variable(s).

• A H1 line: % SUM2EIG Sum of ...

The ‘help 1’ line is the line printed out by the lookfor command.

• Help text: % sum2eig(A) returns ...

If we type help sum2eig at the command line then the output you get will be
the first block of comment lines in the file sum2eig.m. In the example here we
have only one line of help text after the H1 line, but there can be as many as
desired. The comment lines beginning % An error ... are separated from the
help text by the blank line.

The rest of the file contains the function text which leads up the to calculation of the
output variable which in this case is called s. We access the function in much the same
way as we access built in functions, for example

>> sum2eig([1 0;0 3])

ans =
10
>> C = [1 3 4;3 4 5;6 7 8];
>> k = sum2eig(C)
k =
217.0000

5.3 Exercise
Write a function called facsup which takes a positive value x as its argument and, as
output, gives the smallest natural number n such that n! > x.
Add to your m-file so that the user is given a warning if an invalid argument is passed
to the function.

5.4 More input/ouput arguments

As hinted above, functions can have several input and output variables. Consider the
following function which decomposes a square matrix into the sum of a symmetric

20
matrix and skew-symmetric matrix.

function [X,Y] = ssdecom(A)

% SSDECOM Decomposes matrix into symmetric + skew-symmetric
% [X,Y] = ssdecom(A)
% Given a square matrix A this function returns two square
% matrices X and Y such that A=X+Y, X=X.’ and Y=-Y.’

if nargout==1
warning(’ Only one output argument from SSDECOM’)
end

X = 0.5*( A + A.’ );
Y = 0.5*( A - A.’ );

We can access this function as follows

>> C = [1 7 2;-1 0 3;2 1 4];

>> [P,Q] = ssdecom(C)
P =
1 3 2
3 0 2
2 2 4
Q =
0 4 0
-4 0 1
0 -1 0

In the m-file above we have used the variable nargout (number of arguments to be
output). In the example immediately above we typed [P,Q] = ... thus requesting
the two outputs P and Q. If we type

>> a = ssdecom(C)

or simply

>> ssdecom(C)

then only one output is being requested and nargout will be equal to 1. This feature
has been used to warn the user that they may have made an error.
Recall the function eig which we first saw at the end of §2. If this function is called
with two output arguments then the function calculates eigenvalues and eigenvectors.
If the function is called with just one output argument then only the eigenvalues are
returned. This is an example of how the nargout variable can be used to affect output.

21
5.5 Summary
To summarise some of the above we now compare script m-files and function m-files.

Script m-files Function m-files

Do not accept input arguments or Can accept input arguments and
return output arguments return output arguments
Operate on data in the workspace Internal variables are local to the
function by default
Useful for automating a series of Useful for extending the
steps you need to perform many Matlab language for your
times application

22
6 Loops and when not to use them
Loops are used to repeat a statement (or group of statements) a certain number of
times.

6.1 ‘for’ loops

These loops repeat statements a fixed number of times, the syntax is
for i = start:increment:end
statement(s)
end
(When using loops in an m-file it is good practice to indent statements that are within
a loop, this makes the file easier for the user to read.) The statement(s) within the
loop are executed once for each value of i defined by the ‘for’ line. If the increment is
omitted Matlab assumes an increment of 1. For example this loop executes 3 times
as shown.

>> for i = 1:3

x(i) = i+1
end
x =
2
x =
2 3
x =
2 3 4

Each output of x was caused by Matlab passing through the line x(i)=i+1. It is
times such as this that use of a semicolon to suppress output is desirable.
An example of how nested for loops can be implemented is available via help.

6.2 ‘while’ loops

This structure is unlike ‘for’ loops in that we may not know how in advance how many
loops will be necessary. The syntax is
while logical expression is true
statement(s)
end
For example the commands

>> i = 1
>> while i < 4
x(i) = i+1;
i = i+1;
end

23
perform three loops and do the same job as the ‘for’ loop in the preceding subsection.
However, the commands
>> x = 0;
>> change = 1;
>> while abs(change) > 0.00001
newx = cos(x);
change = newx-x;
x = newx;
end
(which approximate the solution to x = cos x by the iteration xn+1 = cos xn with
x0 = 0) will run until two successive iterates differ by less than 0.00001. It is not
obvious in advance how many iterations (or loops) this will take.

6.3 When not to use loops

If it is possible to perform a task using built-in Matlab commands instead of loops
then do so, it is almost certainly more efficient.
Consider the two scripts below. One uses a loop to create a vector with the square
roots of 1 to 1000, the other uses vectorised Matlab commands. (Don’t forget to use
the help feature to tell you more about any unfamiliar commands you encounter.)

Script 1

clear
tic
for i = 1:1000
b(i) = sqrt(i);
end
t=toc;
disp([’Time taken for loop method is ’,num2str(t)]);

Script 2

clear
tic
a = 1:1000;
b = sqrt(a);
t = toc;
disp([’Time taken for vector method is ’,num2str(t)]);

Run both of these and compare the time taken in each case. (Do this a few times to
get an average time.)
One way to increase the speed of the first script (although it will still be slower than
the second) is to set aside memory for b. If the line
b = 1:1000;

24
is inserted immediately before the ‘for’ loop then Matlab will be able to set aside the
1000 ‘lumps’ of memory all in one go, instead of a ‘lump’ at a time.
Summarise the results of your experiments by completing the following table.

Typical run-time

Script 1

Script 2

Script 1 (improved)

6.4 Exercise
It is given that the iteration

1 + x2n (1 − tanh2 (xn ))

xn+1 = (x0 = 1.5)
tanh(xn ) + xn (1 − tanh2 (xn ))

converges to the unique positive solution of

x tanh(x) = 1. (1)

Use this fact to write an m-file which obtains an approximation to the solution of
(1) accurate to 6 decimal places. (This can be assumed to have happened when two
successive iterates differ by less than 12 × 10−6 .)
Answer: x ≈ .

25