MATLABr Handbook

L. Scuderi
English translation v.1.0 by A. Tibaldi

1 Starting commands
In Windows environments, Matlab can be started by selecting with the mouse the corresponding
icon and double-clicking on it. With MsDOS or Unix/Linux operating systems it is sufficient to
write on the shell matlab and press enter. The symbol >> that will appear indicates the Matlab
prompt. An expression or a command can be executed by writing it and pressing enter. The working
session can be terminated by executing the expressions exit or quit.
Table 1. Some useful commands during a Matlab working session.
Command Description
help visualize all the available help topics
help arg visualize information about topic arg
doc arg visualize detailed information about topic arg
clc clear the command window display, giving you a clean screen
; after an expressions, avoid to print on screen its result
... continue to write an expression in the successive row
who visualize the variables in the Matlab workspace (i.e., in memory)
whos visualize information about the variables in the Matlab workspace
clear clear all the variables from the workspace
clear var1 var2 clear the variables var1 and var2 from the workspace

2 Variables in MATLAB
The maximum number of characters allowed in a variable name is 32. Allowed characters are letters
(capital or not), numbers, and the character “_” (underscore). The name of a variable must begin
with a letter (a-z, A-Z). Matlab is case-sensitive (e.g., the names a1 and A1 are associated to
different variables). A variable is automatically created by assigning a value or the result of an
expression to it. The assignment is performed by using the symbol = as in the following example:
>> variable_name=expression
In order to create a string-type variable, the expression must be enclosed within a couple of
inverted commas. Table 2 reports some default scalar variables.
Matlab works with sixteen significant digits (double precision). In the command window, an
output integer variable is usually printed without any decimal digit, whereas a real (non-integer)
variable is plotted with only four decimal digits. A different output format can be adopted, by
using one of the commands from Table 3. For example, in order to print all the sixteen digits of the
Matlab working precision, it is possible to use the command format long e.
Table 4 reports the main operations between scalar variables. In addition to the basic operations,
Matlab features the function listed in Table 5.
In order to define a vector, its elements should be introduced between square parentheses; the
elements of a row vector should be separated by a space or a comma, whereas those of a column

1
 Table 2. Examples of predefined Matlab variables.
Variable Description
ans temporary variable containing the result from the last operation
1i,j imaginary unit
pi π, 3.14159265. . .
eps epsilon of the machine
realmax maximum positive machine number
realmin minimum positive machine number
Inf ∞, i.e., a number greater than realmax, or the result of 1/0
NaN Not a Number (for example, the result of 0/0)

Table 3. Some possible output formats in Matlab.

Format Description
format default format, equivalent to format short
format short fixed-point representation with 4 decimal digits
format long fixed-point representation with 14 decimal digits
format short e floating-point representation with 4 decimal digits
format long e floating-point representation with 15 decimal digits
format rat irreducible fraction representation

Table 4. Operations between scalar variables.

Operation Description
+ addition
- subtraction
* multiplication
/ division
^ exponentiation

vector by a semicolon or by pressing enter after the introduction of each element. In Matlab it is
not possible to use null or negative indexes to identify the components of a vector. The command
x(i) identifies the i-th component of the vector x, x(end) is the last element of x and length(x)
returns the length of x. Table 6 reports some commands aimed at generating and manipulating
vectors.
Table 7 reports few operations that can be applied on the vector x = (xi )i=1,...,n to compute
some quantities associated to it.
The function diag can be used also as diag(x,k), with k integer positive or negative; in this case
it generates a square matrix with dimension n + |k| with all-zero entries except those of the k-th
diagonal above (k > 0) or below (k < 0) the main diagonal, which are set to be the components of
the vector x.
The elements of a matrix should be inserted between square parentheses by rows, where each
row is terminated by a semicolon or by pressing enter. The command A(i,j) identifies the (i, j)
element of the matrix A, size(A) generates a row vector containing the number of rows and of
columns of A, and length(A), applied to a matrix, is equivalent to max(size(A)).
Table 8 reports some Matlab functions that generate particular square and/or rectangular
matrices. It is to be remarked that, in Table 8, when m ≡ n it is sufficient to insert only the
parameter n.
Table 9 reports some commands that, applied to the matrix A = (aij )i,j=1,...,n , return particular

2
 Table 5. Examples of predefined Matlab functions.
Function Description
sin sine
cos cosine
asin arcsine
acos arccosine
tan tangent
atan arctangent
exp exponential
log natural logarithm
log2 logarithm with base 2
log10 logarithm with base 10
sqrt square root
abs absolute value
real real part
imag imaginary part
sign sign function
factorial factorial
round round to the closest integer number
floor round to the closest lower integer number
ceil round to the closest upper integer number
chop(x,t) round x with t significant digits

Table 6. Some commands aimed at generating and manipulating vectors.

Command Description
x.’ generate the transposed vector of x
x=[] generate the empty vector x
sort(x) sort in ascending order the components of x
x=[a:h:b] generate the row vector x = (xi )i=1,...,m+1 where xi = a + (i − 1)h,
where m is the floor of (b − a)/h
x=linspace(a,b, n) generate the row vector x = (xi )i=1,...,n where xi = a + (i − 1)h,
where h = (b − a)/(n − 1)
x=logspace(a,b,n) generate the row vector x = (10xi )i=1,...,n where xi = a + (i − 1)h,
where h = (b − a)/(n − 1)
x(r) return the components of x which indexes are specified in the vector r
x(r)=z assign to the components of x (which indexes are specified in the vector r)
the respective values of z
x(r)=[] remove from the vector x some components (which indexes are specified in r)
x([i j])=x([j i]) switch the components i and j of the vector x

quantities associated to it. The functions described in Table 9 can be applied to rectangular matrices
as well. Moreover, the functions sum, max and min can be applied also as sum(A,k), max(A,[],k)
and min(A,[],k), with k = 1, 2. For k = P 1 they act as in Table 9. For k = 2 they generate a
n
column vector x = (xi )i=1,...,n , with xi = j=1 aij , xi = maxj aij e xi = minj aij , respectively.
The function diag can be used also as diag(A,k), with k positive or negative integer; in this case
it generates a column vector containing the elements of the k-th diagonal above (k > 0) or below
(k < 0) the main diagonal. Also the function tril (triu) can be used as tril(A,k) (triu(A,k)),

3
 Table 7. Some predefinite Matlab functions acting on a vector x.
Command Description
Pn
a=sum(x) generate the scalar a = Q i=1 xi
n
a=prod(x) generate the scalar a = i=1 xi
a=max(x) generate the scalar a = maxi xi
a=min(x) generate the scalar a = mini xi
a=norm(x) generate the scalar a = ||x||2
a=norm(x,1) generate the scalar a = ||x||1
a=norm(x,inf) generate the scalar a = ||x||∞
A=diag(x) generate the diagonal matrix A = (aij )i,j=1,...,n , con aii = xi

Table 8. Some commands aimed at generating and manipulating matrices.

Command Description
A=[] generate the empty matrix A
A.’ generate the transposed of the matrix A
A=eye(n) generate the identity matrix A = (aij )i,j=1,...,n , with aij = δij
A=zeros(n,m) generate the matrix A = (aij )i=1,...,n,j=1,...,m , with aij = 0
A=ones(n,m) generate the matrix A = (aij )i=1,...,n,j=1,...,m , with aij = 1
A=rand(n,m) generate the matrix A = (aij )i=1,...,n,j=1,...,m , with pseudo-casual 0 < aij < 1
A=hilb(n) generate the Hilbert matrix A = (aij )i,j=1,...,n , with aij = 1/(i + j − 1)
A=vander(x) generate the Vandermonde matrix A = (aij )i,j=1,...,n , with aij = xin−j
A(r,c) return the elements of A belonging to the intersections of the rows and
columns specified in r and c respectively
A(r,c)=C assigns to the elements of A (whose row and column indexes are specified
in r e in c) the respective values defined in C.
A(r,c)=[] removes from the matrix A some elements (whose row and column indexes
are specified in r and c)
A([i j],c)=A([j i],c) switch the elements of the rows i and j of A belonging to the columns
specified in c
A(r,[i j])=A(r,[j i]) switch the elements of the columns i e j di A belonging to the rows
specified in r

with k positive or negative integer; in this case it returns the upper (lower) triangular part starting
from the k-th diagonal above (k > 0) or below (k < 0) the main diagonal.
The elementary operations acting on scalars can be naturally extended (when well-defined) to
vectors and matrices, exception made for division and power. The operation * returns the row-
column product. For square matrices the command A^k with integer positive k returns the row-
column product of the matrix A with itself, for k times; A^(-1) generates the inverse of A, if it is
nonsingular.
In addition to the classic sum and product operations between vectors or in general matrices,
Matlab supports element-wise operations, which act directly element by element. These operations
can be performed by writing a point before the symbol defining the operation.
Given the row (column) vectors x = (xi )i=1,...,n , y = (yi )i=1,...,n , the matrices A = (aij )i,j=1,...,n ,
B = (bij )i,j=1,...,n and the positive real number e, Table 10 reports a list of element-wise operations.

4
 Table 9. Some predefined Matlab functions acting on a matrix A.
Command Description
a=norm(A) generate the scalar a = ||A||2
a=norm(A,1) generate the scalar a = ||A||1
a=norm(A,inf) generate the scalar a = ||A||∞ Pn
x=sum(A) generate the scalar x = (xj )j=1,...,n , with xj = i=1 aij
x=max(A) generate the row vector x = (xj )j=1,...,n , with xj = maxi aij
x=min(A) generate the row vector x = (xj )j=1,...,n , with xj = mini aij
x=diag(A) generate the column vector x = (xi )i=1,...,n , with xi = aii
B=abs(A) generate the matrix B = (bij )i,j=1,...,n , with bij = |aij |
B=tril(A) generate the lower triangular matrix B = (bij )i,j=1,...,n ,
with bij = aij , i = 1, . . . , n, 1 ≤ j ≤ i
B=triu(A) generate the upper triangular matrix B = (bij )i,j=1,...,n ,
with bij = aij , i = 1, . . . , n, i ≤ j ≤ n

Table 10. Element-wise operations in Matlab.

Operation Description
z=x.*y generate the row (column) vector z = {zi }i=1,...,n , with zi = xi ∗ yi
z=x./y generate the row (column) vector z = {zi }i=1,...,n , with zi = xi /yi
z=x.^y generate the row (column) vector z = {zi }i=1,...,n , with zi = xyi i
z=x.^e generate the row (column) vector z = {zi }i=1,...,n , with zi = xei
C=A.*B generate the matrix C = (cij )i,j=1,...,n , with cij = aij ∗ bij
C=A./B generate the matrix C = (cij )i,j=1,...,n , with cij = aij /bij
b
C=A.^B generate the matrix C = (cij )i,j=1,...,n , with cij = aijij
C=A.^e generate the matrix C = (cij )i,j=1,...,n , with cij = aeij

3 Graphics in MATLAB
In order to plot a function f with variable x it is possible to use the function fplot

fplot(’f’,[xmin xmax])

where f is the expression to be plotted and [xmin xmax] is a vector of two variables, containing
the boundaries of the x axis for the plot. It is possible to specify also the boundaries of the
representation for the y axis, by using the vector [xmin xmax ymin ymax] as second argument of
the function fplot.
Alternatively, it is possible to use the command plot according to the syntax

plot(x,y)

where x is a vector containing a set of coordinates in the x axis, and y is obtained by evaluating the
function f in the points specified in x.
Both plot and fplot commands allow to customize the color and type (solid, dashed, dash-
dotted...) of the line. Moreover, plot allows to introduce markers correspondingly to the points
(xi , yi ) contained in the vectors x and y. Tables 11 and 12 list some of the possible options and
existing commands aimed at commenting a plot.
In order to use a logarithmic scale on the x, y or both axis, instead of plot it is possible to
use the commands semilogx, semilogy or loglog, respectively. The syntax and options of these
commands are the same as those of plot.

5
 Table 11. List of options for commands plot and fplot.
Color Description Symbol Description Line Description
w white . dot - solid line
y yellow o circle : dotted line
r red x cross -. dash-dotted line
g green + plus -- dashed line
b blue * star
k black s square

Table 12. List of commands for commenting a plot.

Command Description
title introduce a title to the plot
xlabel introduce a label for the x axis
ylabel introduce a label for the y axis
grid introduce a grid for the x and y axis
legend introduce a legend to describe the different curves in the plot
text introduce a text string in a specific location
gtext introduce a text string in a location selected with the mouse

At every call of plot, Matlab erases by default the previous plot and traces the new one. In
order to draw more graphs in the same figure, it is possible to use the command hold on before
calling the command plot to trace the second curve. Alternatively, it is possible to use the command
plot with the following syntax:

plot(x_1,y_1,’-’,x_2,y_2,’:’)

It is possible to draw several plots in the same graphical window, in different sub-plots, by using
the command subplot

subplot(rows,columns,subfigure)

where rows and columns indicate the number of rows and columns of the matrix of subfigures of the
window, and subfigure indicates the subfigure to be considered for plotting a graph. Subfigures
are enumerated from left to right and from top to bottom. It is possible to open several graphical
windows (or “figures”). The figure n can be activated by means of the command figure(n); to
close this figure, it is possible to use the command close(n). In order to close all the figures, it is
possible to use the command close all.

4 MATLAB programs
A file containing several Matlab instructions is generally referred to as m-file. This name derives
from its extension “.m”. A m-file name can contain letters and/or numbers, and the character _
(underscore). In a m-file, commands should be introduced on different lines or on the same line
(separated by comma or semicolon). Comments can be introduced after the special character %.
There are two categories of m-files: scripts or functions.

• Scripts are given by a sequence of Matlab commands. A script m-file can be executed by
selecting the directory where it is saved, and inserting its name (without extension) in the
Matlab prompt. Script files do not allow input or output passing parameters. Moreover, all

6
 the variables defined in a script are kept in memory, in the workspace, just like those defined
directly in the command prompt.
• The function-type m-files must start with the syntax

function [y_1,y_2,...,y_n]=function_name(x_1,x_2,...,x_m)

where y_1,y_2,...,y_n are the output parameters, and x_1,x_2,...,x_m are the input
parameters. The string function_name is the name of the function; this must coincide with
the m-file name (appending, of course, the file extension .m) where the function is saved. A
function can be executed in the command prompt, o from a script, or by another function, by
writing

[y_1,y_2,...,y_n]=function_name(x_1,x_2,...,x_m)

or

function_name(x_1,x_2,...,x_m)

In the latter case the function returns only the first output parameter, which is then saved in
the variable ans.
It is to be remarked that a value must be assigned for each output parameter; otherwise,
Matlab returns the following error message

??? One or more output arguments not assigned during call to ’function_name’

Unlike scripts, function m-files require input and output parameters, and the variables defined
during their execution are local, and erased from the Matlab memory workspace after the
execution of the function.

4.1 MATLAB coding: syntax constructs

The Matlab programming language supports classical syntax constructs. Table 13 lists the main
Matlab relational operators.

Table 13. Relational operators in Matlab.

Operator Description
< lesser
> greater
<= lesser or equal
>= greater or equal
== equal
∼= not equal

Relational operators allow to perform comparison between expressions. In the last Matlab versions,
the result of these comparisons is assigned to logical-type variables. Specifically, 0 is used to
represent “false” and 1 represents “true”.
Relational operators can be combined through the logical operators listed in Table 14. Table 15
lists the results of logical operators when applied to two conditions a e b.
In the following we summarize some programming structures available in Matlab.

7
 Table 14. Matlab logical operators.
Operator Description
& and
| or
∼ not
xor exclusive or

Table 15. Logical operators applied to two conditions a and b.

a b a & b a | b ∼a xor(a,b)
0 0 0 0 1 0
1 0 0 1 0 1
0 1 0 1 1 1
1 1 1 1 0 0

• unconditional loop controlled by a counter

for index = expression

instruction block
end

where index assumes the values defined by expression.

• conditional loop

while condition
instruction block
end

where condition is an expression that Matlab evaluates and interprets as true if different
from 0, and false if equal to 0.
• conditional structures

if condition_1
instruction block 1
elseif condition_2
instruction block 2

..
.

else
instruction block n
end

where instruction block 1 will be executed only if condition_1 is true, instruction block 2
only if condition_1 is false and condition_2 is true, and so on. The instruction block n
is executed only if no one of the previous conditions is true. The instructions elseif ande
else can be omitted if unnecessary.

8
The command return allows to terminate the execution of a program before it reaches the last
instruction.
Moreover, Matlab supports the command break, which allows to terminate a cycle before
it reaches its natural end. When this command is executed, Matlab jumps directly to the end
instruction, terminating the loop.
In order to estimate the efficiency of a program, it is possible to measure its execution time by
means of the commands tic and toc. These commands return the number of seconds required to
execute an instruction block, by means of the syntax

tic
instruction block;
elapsed_time=toc

tic activates the timer, toc stops it and saves the time needed to execute the instructions between
tic and toc in the variable ’“elapsed_time”.

5 Main MATLAB functions for numerical analysis

Table 16. Linear algebra.

Function Description
lu(A) generate the PA=LU (Gauss) factorization with partial pivoting
chol(A) generate the Choleski factorization of the matrix A
qr(A) generate the QR factorization of the matrix A
x=A\b solve the linear system Ax = b
cond(A) compute the spectral conditioning number (norm 2) of A
cond(A,1) compute the norm 1 conditioning number of A
cond(A,inf) compute the norm ∞ conditioning number of A
rcond(A) compute the reciprocal of the norm 1 conditioning number of A
rank(A) compute the rank of A
det(A) compute the determinant of A
inv(A) compute the inverse of A
eig(A) compute the eigenvalues and eigenvectors of A

Table 17. Polynomials, functions and approximation.

Function Description
polyval evaluate a polynomial
f=inline(’expression’,’x_1’,...,’x_n’) define the function f (x1 , . . . , xn ) = expression
y=f(x_1,...,x_n) evaluate the function y = f (x1 , . . . , xn ) defined
by inline
y=feval(f,x_1,...,x_n) evaluate the function y = f (x1 , . . . , xn ) defined
by inline or by a function m-file
polyfit compute the coefficients of the polynomial interpolating
or approximating a set of data
according to the least squares criterion
spline evaluate a cubic interpolating spline

9
 Table 18. Equations and systems of nonlinear equations.
Function Description
fzero evaluate the zeros of a nonlinear function
roots evaluate the zeros of a polynomial
fsolve solve a system of nonlinear equations

Table 19. Integrals.

Function Description
quad adaptive Simpson formula
quadl adaptive Gauss-Lobatto formula

Table 20. Equations and systems of ordinary differential equations (ODE).

Function Method
ode45 Order 4 and 5 explicit Runge-Kutta
ode113 Variable-order Adams-Moulton
ode23 Order 4 and 5 explicit Runge-Kutta
ode23t Trapezoidal method
ode15s Variable-order implicit linear multi-step
ode23s Order 2 implicit Runge-Kutta

10