MATLAB Tutorial

Everything starts in the MATLAB command line.

Contents

 Declaring a Variable

 Functions

 Writing Functions and Scripts

 Loops and Vectorization

 Plotting Results

Declaring a Variable

The first thing we'll do is make a variable. It's pretty easy actually; just use the

equals sign.
x=4

x =

Note how it outputs to the window. That can be annoying. See?

x = 5
y = 6
z = 7

x =

y =

6
z =

How do we stop it? Simple: put a semicolon!

x = 5;
y = 6;
z = 7;

MATLAB's real power is in dealing with vectors and matrices. To utilize that

power, first we must make a vector and a matrix. Let's make a vector first.
x = 1:11

x =

1 2 3 4 5 6 7 8 9 10 11

Note that we have made a vector that goes from 1 to 6 with stepsize 1. Pretty

cool, eh? What if we wanted a different stepsize, like 2?

x = 1:2:11

x =

1 3 5 7 9 11

That's how. If we want to make an arbitrary vector, we use brackets:

x = [1 4 3 2 5 6]

x =

1 4 3 2 5 6

That gives us a row vector, but what about a column vector? Well, just transpose

the row vector. To transpose a MATLAB vector, use the single quote.
x = x'
x =

1
4
3
2
5
6

Matrices are vectors put together. We go back to the brackets ...

x = [1 2 3 4; 5 6 7 8]

x =

1 2 3 4
5 6 7 8

Note how I used the semicolon in the brackets to put the rows on top of each

other. Suppose I wanted to put two columns together instead; then, I use a

comma or a space
y = [1 2 3 4]'
z = [5 6 7 8]'
x = [y, z]
x = [y z]

y =

1
2
3
4

z =

5
6
7
8
x =

1 5
2 6
3 7
4 8

x =

1 5
2 6
3 7
4 8

All this "concatenation" can be boiled down to

 Put elements together by column with spaces or commas.

 Put elements together by row with semicolons.

Think of everything in MATLAB as a matrix. So numbers are 1x1 matrices, rows

are 1xn matrices and columns are nx1 matrices. That will make the manipulation

a lot easier. Addition and subtraction are easy ...

x = y - z

x =

-4
-4
-4
-4

... as long as the elements are the same. One issue is multiplication and division.

In MATLAB, the multiplication operator is MATRIX multiplication and not

element‐wise multiplication. What does that mean? It means

x = [1 2; 3 4];
y = [2 2; 2 2];
z = x*y

z =

6 6
14 14

is not just element‐wise; it's real matrix multiplication. But what if we just want

to multiply each element together? We can do that! For multiplication, division

and exponentiation, we put a period before it to indicate element‐wise

application of the operation. See?

z = x.*y
z = x.^y

z =

2 4
6 8

z =

1 4
9 16

We also have some special vectors that can be made via functions:
x = zeros(3)
y = ones(3,1)
z = rand(4,2)

x =

0 0 0
0 0 0
0 0 0

y =
 1
1
1

z =

0.8147 0.6324
0.9058 0.0975
0.1270 0.2785
0.9134 0.5469

The last one is very important, as it allows us to make a matrix of values that are

random (from a uniform distribution to be exact). To draw from a Gaussian

distribution, use
z = randn(4,2)

z =

3.5784 0.7254
2.7694 -0.0631
-1.3499 0.7147
3.0349 -0.2050

Functions

Also, the last example shows us that calling functions are easy. Given a function

name you can figure out how the function works by the help command

help interp1
interp1 - 1-D data interpolation (table lookup)

This MATLAB function returns interpolated values of a 1-D function at

specific
query points using linear interpolation.

vq = interp1(x,v,xq)
 vq = interp1(x,v,xq,method)
vq = interp1(x,v,xq,method,extrapolation)
vq = interp1(v,xq)
vq = interp1(v,xq,method)
vq = interp1(v,xq,method,extrapolation)
pp = interp1(x,v,method,'pp')

See also griddedInterpolant, interp2, interp3, intern

doc interp1

This tells us how the interp1 function works. Pretty cool function, eh? If you

want to see all the description and examples of each syntax, one click on doc

interp1.

If you want to see all the variables, use

who

Your variables are:

ans fs m t x y z

and to see the details (how big the variables are, etc.) use
whos

Name Size Bytes Class Attributes

ans 1x27 54 char

fs 1x1 8 double
m 1x1 8 double
t 1x21 168 double
x 3x3 72 double
y 3x1 24 double
z 4x2 64 double

which shows the current variables in your "workspace". We could also click into

your "workspace" to see the details

 and
clear x

which clears the specified variables. To clear all, you just

clear all

Writing Functions and Scripts

At some point you guys will be making your own functions. To do that, you make

a new M‐file and name it by the function name. Let's say our function is called

syed; then the new file will be syed.m ... let's let it do something simple: function

x = syed(y) x = sum(y.^2) end

Doing a help will tell you that "sum" sums up all the elements of a vector, and

for a matrix it sums up each row. Since the .^2 squares the elements, we get a

sum‐of‐squares! Note first the keyword function. If that keyword is not there,

then we have a script. What's the difference? A script is just a list of commands

that are run. A function has inputs and outputs. So if you want a script (generally

the "main" file for your work will be a script) then make sure the first word is

not function. When you make a script, it is also an M‐file. For example, this

tutorial was created in an M‐file called tutorial.m and I called it in the command

line just by typing the M‐file name "tutorial" ... make sure the M‐file is in your

path. You can see your path at the top of MATLAB or by the command
pwd

ans =

C:\Users\ROG\Desktop\worksapce\matlab

Your current directory is the directory you're working in, so generally you start
by changing your directory using the command cd
cd C:\Users\ROG\Desktop\worksapce\matlab

and to see all the files in your directory, use

ls
. .. tutorial.m

Loops and Vectorization

The next step to complicate things is loops. To make a for loop use the for

command:
for m = 1:10
x(m) = m^2;
end

Note the syntax: after the for loop you give the iterating variable (don't use i or

j because in MATLAB, they are the square root of ‐1). Then you set it equal to a

vector. Note I said vector and not just 1:10. See?

for m = [2 4 3 7]
m^2
end

ans =

ans =

16

ans =

ans =
 49

So for loops are cool right? WRONG. At all costs, do not use for loops. They are

super‐slow and basically only for the most necessary situations. Take the

previous for loop. Instead of doing that, we could just do

x = (1:10).^2

x =

1 4 9 16 25 36 49 64 81 100

and get what we want much faster. Always try to vectorize your code. One way

to help you out is this tool, which takes any matrix and turns it into a vector:
x = rand(3,3)
x(:)

x =

0.4218 0.9595 0.8491

0.9157 0.6557 0.9340
0.7922 0.0357 0.6787

ans =

0.4218
0.9157
0.7922
0.9595
0.6557
0.0357
0.8491
0.9340
0.6787

If we want to compare every # of a vector with a variable, just do

alltheplayers = rand(3,5);
threshold = rand(1,1);
selectplayers = (alltheplayers(1,:)>threshold);

alltheplayers =

0.1816 0.2914 0.3182 0.0278 0.8027

0.5757 0.4617 0.4599 0.6585 0.4086
0.1859 0.3470 0.2359 0.1588 0.3274

threshold =

0.7460

selectplayers =

0 0 0 0 1

we compare the first row of matrix alltheplayers with a single variable threshold,

and express the result in logical.

If you have to use a for loop, try to initialize the variable that's being updated.

So in our for loop, we should do

x = zeros(10,1);
for m = 1:10
x(m) = m.^2;
end

to prevent major time issues due to having to constantly add size to the vector.

For conditioning (that is, if‐else ladders) use this construct

m = 7;
if m == 5
disp('Hi.');
elseif m == 7
disp('What''s up?');
else
disp('Hello.');
end

What's up?

Note that the "end" ends the if loop. disp displays a string; note how you use

consecutive single quotes (not double quotes) to get a single quote.

Plotting Results

So let's make some time series:

fs = 20;
t = 0:1/fs:1; % Time segment of one second with sampling rate 20
x = t.^2;

Whoa, wait. What was that percent sign and green thing? You probably guessed

that it was a comment. The percent sign is MATLAB's version of the // for C. Use

it often, and use it well. Please. For our sanity. And yours. We want comments

that are good and frequent. They help; I promise. Ok, back to plotting. The plot

command does the trick

figure, plot(t,x);
figure creates a new figure upon which we plot the data. Note that the xaxis

goes first, followed by the y‐axis. This plot is pretty lame. If x were filter

coefficents, we'd like for you to use plot's cousin, stem

stem(t,x);
But for signals, let's go back to plot.
plot(t,x);
Note my habit of using semicolons? Very useful. Ok how do we edit the plot?

Well, I'm a command‐line guy, so I use it a lot. Let's start with annotation:
title('INSERT TITLE HERE');
xlabel('Descriptive name of x variable');
ylabel('Descriptive name of y variable');

Please add labels to all your plots, k? K. K? K. To change the color or the dots,

use help plot to get a list of things to add as the third argument to the plot

function. So I can do
plot(t, x, 'rp-.');
to make the plot red with pentagrams at the points and a dot‐dashed line. As

you guys get better you guys can mess around with the Figure window and make

really good plots. We find it very important that you make good plots because

good plots mean good work :) Suppose you wanted to plot two signals on the

same axis. How do we do that?

x = rand(100,1);
y = randn(100,1); % Constant vector of 1.1
plot(x)
hold on, plot(y, 'r'), hold off;
hold on means every plot afterwards goes on the same axis, until you give a

hold off command. If you would rather have the plots side by side on the same

figure, use the subplot command

subplot(2,1,1), plot(x);
subplot(2,1,2), plot(y);
Subplot takes a # of rows and # of columns to make a "matrix of axes" within

the figure. So we have 2 rows and 1 column, which means one plot on top of

the other. Then the numbering is done row‐wise first, so that 1 means the first

axis. 2 means the 2nd axis on that row, etc. We can change

the axis of the second plot by doing

axis([0 100 -0.5 0.5]);
and we can turn the grid on (to see where the curve hits points better) using
grid on;
So we've gone over

 setting and initializing variables,

 vectors and matrices,

 concatenation,

 element‐wise operations vs matrix operations,

 functions and scripts,

 for loops and the emphasis on vectorization, and

 plotting results.

This should get you a decent head start into MATLAB. Here's a list of other useful

functions; use help to determine how they work:

 sum/max/min

 dlmread/dlmwrite

 size/length

 fprintf

 logical

 xor

 ceil/floor/round

 polyfit

 polyval

 interp1

Published with MATLAB® R2014a