Guide for CS 1371

Introduction: The MATLAB Command Interface

MATLAB is basically just a really complicated program that functions as an extremely powerful
calculator. As such, it can perform tasks than most calculators cannot, usually in fractions of a second.
However, like all computer programs, MATLAB is incredibly stupid and is only ever as useful as your
coding capability allows it to be. Never underestimate how irritatingly picky MATLAB can be about small
sections of code. The programmers designed some functions to be flexible with input values, but for the
most part they didnt bother making it user-friendly.
It is not necessary to actually purchase the MATLAB software to complete this course, although
it is probably pretty useful; I do not personally own the software. Many, not all, computers in the library
do have it. Additionally, some computer geeks have set up a virtual desktop call Citrix that allows you to
access your library desktop from your laptop. Unfortunately, the remote desktop generally runs very
slowly and is prone to freezing at frequent intervals. It also likes to give you nonsensical login error
messages such as We are sorry. This desktop has been locked, and only you can use it.
Whenever you run MATLAB in the library or from your remote desktop, it will tell you its
initializing and then proceed to take an unreasonable amount of time to generate a pointless message
about customizing shortcuts. I have yet to find a method for preventing this. The MATLAB screen is
divided into a number of sections with specific names and titles. These are important to learn because
they make up a large part of the terminology used in the course.

The Command Window:

This is the large and obvious block in the middle of your screen labeled Command Window. It
can be used to run individual lines of code or display the progress of a script. Although youd imagine it
must play a very important role in writing codes, it is actually pretty much useless. If you use the
command window to run multiple lines of code sequentially and realize you made a mistake
somewhere, you have to redo the WHOLE process. The command window will not allow you to edit
anything previously entered, and once a variable is overwritten the original value cannot be restored.
To enter a command in the command window, simply type it in the little box underneath and
pres enter. It is possible to enter multiple lines of code such as loops and conditional statements using
shift+enter, but I wouldnt suggest it.
You can use the up-arrow to scroll through your most recent commands so you dont have to
retype them. If you attempt to do something that MATLAB doesnt like or if there is a typo in your
command, MATLAB will make an annoying beeping sound and display a red error message that tells you
what the problem is. Learn to fear the error message. At the very bottom-left part of the screen
MATLAB will display statuses such as initializing, busy, or ready. If you accidentally run an infinite
loop or an extremely long recursive function, you can cancel your last command by pressing ctrl+c. You
can enter clc to clear whatever is in the command window at the time.

The Workspace:
This is the little box off to the right labeled Workspace. Its purpose is to keep track of all the
variables and data types you are using. Anytime you save a value to a particular variable or enter any
assignment statement, the workspace is automatically updated with the new information. It displays
both the name of the variable and its data type (a brief description of data types will be given in section
one). If the workspace is becoming too cluttered with useless variables, you can reset it by entering clear
in the command window.

Command History:
This little box keeps track of all the commands you have entered since the beginning of the
MATLAB session. If you accidentally clear the command window and workspace but still need the
information from them, you can access it from command history. Otherwise, you should just pretend it
doesnt exist.

The Current Directory:

This box is labeled Current Folder and is slightly confusing to become accustomed to. MATLAB
can only access files and programs that reside in your current directory; if you attempt to run a function
or script from a different folder, MATLAB will give you an error message saying that it doesnt exist. It
would therefore behoove you to create an organized set of folders and sub-folders for CS 1371, with
each homework assignment having its own folder. If you need to access a function from another
homework, which will inevitably happen for some of the more difficult problems, simply copy/paste it
into the current directorys folder and submit it with your homework problems.
To change the current directory, locate the button on the primary toolbar at the top of the
screen and to the right of the current folder display. It should look like an ellipsis. Then just navigate to
the folder you wish to select and click OK. I suggest saving your entire CS 1371 folder on your GT Prism
drive so you can access it from any computer on campus.

Section One: Scripts and the Assignment Operator

I will be writing these sections with the primary focus of assisting in concepts used to complete
homework assignments, but all of the basic information is useful on the tests.

The Assignment Operator:

Any line of code that uses a single equals sign (the assignment operator) is called an assignment
statement. Assignment is used primarily to set values to variables and works the same way with any
data type. All variable names are case-sensitive, so a and A can be used as separate variables. Be
very careful when naming and accessing variables. Variable names are generally short enough to take up
little space but long enough to be easily distinguished. The last thing you want to do is have to fix your
code when you cant even figure out what it does.
There are three important restrictions on naming variables:
1. A variables name must begin with an uppercase or lowercase letter. The rest of the name may
contain any combination of letters and numbers.
2. A variables name cannot contain spaces. Obviously, MATLAB would then be unable to
distinguish between the variable and the surrounding code.
3. A variables name cannot contain any special characters except underscores.
4. A variable should never be given the same name as a built-in MATLAB function (min, max,
sort,etc.)
By these rules, the following would be legal variable names: A , a , a1 , a1a , a_a
The following are therefore illegal variable names: 1 , 1a , a: , : , a 1, a?
Assignment statements take the form A = B, where A is a variable or collection and B is an
expression that can be immediately evaluated. MATLAB first evaluates B then sets A equal to it.
Assignment statements do NOT establish relationships between variables; they can only be used to set a
variable equal to a known value. Once again, changing the values of other variables used in an
assignment statement has no effect on the previously outputted variable.
If no A is inputted, the value of B will automatically be stored in MATLABs temporary
variable ans. ans can be accessed at any time and is always placed in the workspace, but each new
non-specific assignment statement overwrites whatever is stored in it. It is generally a bad idea to rely
on ans to store any values you may be using later.
By default, the result of an assignment statement is immediately displayed on the command
window. However, this can become very irritating when you run long scripts or change large collections
of values; the command window will quickly become clogged with useless information that slows the
execution of your code. To prevent this, you can suppress the output of individual lines of code by
ending them with a semicolon. The result of a suppressed output is still stored in the workspace but will
not be displayed in the command window. With practice, you will learn to subconsciously suppress all
outputs as you write. MATLAB will automatically notify you of unsuppressed outputs by discoloring the
equals sign.

Whenever MATLAB encounters spaces, it automatically ignores them. You can therefore add
spaces in lines of code to make them more legible.
Using the Dot with Basic Arithmetic:
The MATLAB notation to add, subtract, multiply, divide, or raise to a power is exactly what you
would expect (+, -, *, /, ^). However, MATLAB must have a way of distinguishing between regular
arithmetic and special matrix operations. For instance, you can either multiply each element in one
matrix by the corresponding element in another or actually perform matrix multiplication using linear
algebra. Note: MATLAB uses decimals instead of fractions whenever it encounters a non-integer.
To differentiate between the two, MATLAB utilizes a dot (.) system with the symbols *, /, and ^.
To perform element by element multiplication, division, or exponentiation, simply add a single period
before the arithmetic symbol (.*, ./, or .^). Linear algebra functions are performed whenever the dot is
absent and will cause errors if the dimensions of the matrices do not match properly. Note that the dot
has no effect on scalar values (5*5 is the same as 5.*5).

Scripts:
As previously stated, using the command window to solve complex problems is inadvisable
because it does not allow you to edit commands you have already run. Consequently, most coding is
completed using scripts and the MATLAB function editor. A script is a blank space for inputting blocks of
code or functions. When you finish writing a script, you can run the entire thing at once; if an error
occurs, you can edit and rerun your code to your hearts content.
To open a new script, click the icon in the upper-left corner that looks like a blank sheet of paper
or press ctrl+n. When you finish typing out your code, you can run the script by using the green triangle
icon or by pressing F5. To save, go to filesave as or press ctrl+s. Scripts are automatically saved as .m
files, the default file type for MATLAB. Always leave the file type as .m if want to receive credit.
If MATLAB dislikes something youve written on a script, it will inform you by enclosing a
variable inside a discolored box or by underlining part of your code in red. This is a very quick way to
spot typos, but dont rely on it for everything.
If you plan to test your script frequently, you can dock the script editor inside the command
window. This will allow you to view the values of different variables from the workspace without having
to minimize the script. To dock the editor, click the small arrow on the left just under the minimize
button. The editor can be undocked in the same way by clicking the up-arrow above the minimize
button.
If you wish to run a script one line at a time rather than all at once, you can enter debug mode
by clicking the gray space to the right of the line numbers on the left side of the screen. A red circle will
appear on the debugged line. Any number of lines can be debugged. When run, the script will
immediately proceed all the way to the first debugged line, then stop. You can then continue running
each individual line by pressing the icon with the curvy blue arrow on top of a sheet of paper labeled
step. The step in icon allows you debug any internal functions that your script runs. Continue
allows the code to run normally until it reaches the next debugged line. While in debug mode, you can
display the values of any variables by hovering the mouse over them or by simply viewing their progress

in the workspace. If you cannot locate the source of an error, debugging your script is an excellent way
to track the problem.
You can add your own personal comments to any of your code to make it easier for you or other
people to understand. Simply precede any comments with a % percentage sign. MATLAB completely
ignores any green text following a percentage sign when it executes code. To see an example, just look
at a hw.m file. To comment a line of code, click on it and press ctrl+r. To uncomment something, press
ctrl+t. It is not necessary to actually highlight the code you sigh to comment or uncomment. You can
earn extra credit points on your homework assignments by commenting your code.

Built-In MATLAB Functions:

The designers of MATLAB included hundreds of built-in functions to assist users in writing code.
These functions allow you to perform otherwise complex tasks in a single line of code. To call a function,
type its name followed by any inputs in parentheses, separated by commas. I will list any useful
functions as well as brief descriptions of them in each section of this guide.
If you wish to perform a relatively common task and are wondering whether MATLAB has a
built-in function, it probably does. Go to Helpfunction browser (shift+F1) and search for the name of
the function you are looking for. Function names are generally incredibly obvious, exactly what you
would expect them to be. Function names are also case-sensitive.
sin()
cos()
sqrt()
abs()
exp()

Calculates the sine

Calculates the cosine
Calculate the square root
Calculates the absolute value
Calculate ex

Examples:
Input:

Output to Command Window:

5
x = 5
x = 5:

ans = 5
x = 5

Input:

Output to Command Window:

x = 5
y = x
x = 6
y

x
y
x
y

=
=
=
=

5
5
6
5

Note that changing the value of x had no effect on y.

y = X

Error: undefined variable X

Remember that variable names are case-sensitive.

x = x + 1

x = 7

Write a script that will calculate the radius and surface area of a sphere
given a volume of 15.

volume = 15;
radius = (volume.*3./4./pi).^(1./3);
surface_area = 4.*pi.*radius.^2;

(The constant is expressed as pi rather than with symbols)

Write a script that uses the linear relationship between the points (3,4) and
(-2,5) to find the y-value for x = 20.
x1 = 3;
y1 = 4;
x2 = -2;
y2 = 5;
slope = (y2 y1) ./ (x2 x1);
y_int = y1 slope .* x1;
y3 = 20 .* slope + y_int;

To Complete Homework Assignments:

1. Create a folder in your CS 1371 folder for the homework assignment. Make it easy to find.
2. Go to T-SquareAssignments and click on the homework assignment.
3. Save each of the files under Additional resources for this assignment by right-clicking and
selecting save target as.
4. Open MATLAB and change the directory to the specific homework folder.
5. Open the ABCs.m file and follow the directions. When you are finished, you can check your
answers by running the ABCs pretest file. Right click the pretest in the directory and select Run
File. The pretest will display which answers are correct or incorrect on the command window.
6. Open the hw.m file and write your personal information at the top. The descriptions for the drill
problems will include all inputs, outputs, and character classes. Always remember to check the
character class of your outputs. Dont forget to comment your code.
7. Create a new file in the script editor for each drill problem. When you finish, test them
thoroughly. If there are test cases on the hw.m file, you can uncomment them and run the hw
file as long as your scripts are saved in the same directory. Alternatively, you can copy the test
cases and run them from the command window.
8. After you complete the drill problems, update the information regarding collaboration on the
hw.m file. Navigate back to the same assignment page on T-square and submit all files listed in
the Files to submit list.
9. You can submit as many times as you want before the deadline and will frequently have
opportunities to resubmit after you receive your grade. Resubmissions are graded as the
average of the two submissions except on test weeks.
Homework assignments are graded automatically by a computer. It will test each of your scripts
or functions multiple times with various inputs determined by the TAs. They will use any means possible
to make your code run an error and cause you to get a zero on the problem. For this reason, you should
never rely solely on the test cases provided with the homework file. Make sure your functions can deal

with inputs such as negative numbers, fractions, and empty vectors or strings. If you can find one output
but not another, set the difficult output equal to something pointless like 6 so you can get partial credit.
Remember: any code that runs an error results in a zero.

Section Two: Functions

The primary drawback of scripts is their lack of variability; to run a script with different starting
values, you must manually change each of the predetermined variables. Additionally, many problems
require running the same block of code many times. Using functions allows you to complete such tasks
in a more concise, easily understandable, format.
A function is simply a script with variable inputs and outputs. You call the function with
whatever inputs you wish to use, it does something, and it returns the final result to whatever called it.
Once you have written a function, you dont even have to know exactly what it doesjust what it takes
in and what it regurgitates.

Writing Functions:
Except for a few lines of code, functions are absolutely identical to scripts. The main difference
and, coincidentally, the easiest way to spot a function is the first line, the function header. The header
contains all of the basic information concerning the function in the following form:
function [output1, output2, output3] = name_of_function (input1, input2)

or
function output1 = name_of_function (input1)

When you type the word function, MATLAB should recognize it and color it blue. Please
remember that spaces do not matter, but commas do. In situations with only one input/output, brackets
may be excluded but parentheses are always necessary. A function can have any number of inputs or
outputs, even zero. You can use the same variable as both an input and an output as long as you dont
need to keep track of its original value. Although you will probably never encounter a function with no
inputs, it is quite common for one to have no outputs if its purpose is simply to produce a plot or print
something out to the command window. If the function has no outputs, eliminate the equals sign and
outputs. Thus:
function name_of_function (input1, input2, input3)

Most functions are ended with the word end to let MATLAB know that the function is
complete. Doing so is only completely necessary if you plan on using helper functions, which I will now
explain. Although functions are normally called directly from the command window, they can actually
call other functions as well, as long as all the files needed are saved in the current directory or added to
the path.
If you wish to perform a small task multiple times in a single function, you can write a small
helper function to assist you. Just place the helper function after the end of the main function. Doing
so can save space as well as make your code more readable. Here is an example of a helper function.
function [out1, out2] = helperExample (a,b,c)
out1 = confangleNumber(a);
out1 = out1 - confangleNumber( a + (b - a).^3);
out2 = confangleNumber(b + 6.*a);
end

function num = confangleNumber(num)

num = abs(ceil(num + (num ./ 2).^2 round( (num +1)/(num 1).^2)))
end

As previously stated, it is not necessary to have any earthly idea about what a function actually
does. When you attempt to save your function, MATLAB will automatically suggest the same name as
the one given in the function header. Never name a function anything else.
Because function are so versatile and easy to grade, this course relies primarily on functionwriting for homework assignments and coding problems on tests. If you cant write a function properly,
youre probably going to fail the class.

Calling Functions:
Many people become very confused when they actually attempt to use a function theyve
written. The easiest and most obvious way to remember the format is that you call your own functions
in EXACTLY THE SAME WAY as you would call built-in MATLAB functions. Here is a comparison to the
built-in min() function.
[minimum, index] = min(vector);
[distance, speed] = helperExample(length, width, height);

Notice that the input and output variable names of the function call above do not match the
names given in the function itself. This is because all variables used in a given function are temporary.
When you enter the function call above, MATLAB first opens a new command window for the function
helperExample. It then sets the values of length, width, and height equal to a, b, and c
respectively (always in the same order given in the function header). When MATLAB reaches the end of
the helperExample function, it closes the new command window and sets whatever values are stored in
out1 and out2 equal to distance and speed. However, only those final values will be stored in
your original workspace. All other variables are deleted as soon as the function call is complete.
Let me reiterate: MATLAB will return only the output variables specified in the function header;
all other variables are considered temporary. Additionally, each function call opens its own workspace,
which does not have access to the variables in any other workspace. Therefore, any values used within a
function must be given as inputs. If you do not specify output variables in your function call, the first
results will automatically be stored in ans.
Function calls utilize a simple system known as the stack. Basically, the most recent command is
executed first in any given situation. In the helperExample function, each function call to
confangleNumber is completed (another command window is opened and closed) before the next line
of code is executed. Really, the stack concept is straightforward and easy to remember.
Thats pretty much it for functions. Theres really not that much to know.

Examples:
Lets just take the scripts from section one and turn them into functions. Remember to make your
functions easy to understand by naming variables properly.
function [radius, surface_area] = circleStuff (volume)
radius = (volume.*3./4./pi).^(1./3);
surface_area = 4.*pi.*radius.^2;
end
function y3 = extrapolate (x1, x2, y1, y2)
slope = (y2 y1) ./ (x2 x1);
y_int = y1 slope .* x1;
y3 = 20 .* slope + y_int;
end

Section Three: Vectors

So far, we have only considered situations that involve working with single values. However, we
may wish to operate on large sets of numbers at the same time. In these situations, we utilize vectors,
large collections of scalar quantities. As a matter of fact, even single numbers in MATLAB are interpreted
as vectors of length one. Vectors are easy to work with and easy to understand.

Creating Vectors:
To create a vector, simply enclose a list of numbers or variables separated by spaces or commas
in straight brackets. Vectors can be of any length, even one or zero. Vectors of length zero, or empty
vectors, are usually used to initialize variables that will be filled later. If you want to create a new
variable of a non-predetermined length, first initialize it as an empty vector.
You can also create column vectors in much the same way. Simply separate all elements with
semicolons to indicate that they should be placed on top of each other. MATLAB indexes and slices
column vectors in the same way.
To convert row vectors from row to column or column to row, transpose them by following
them with an apostrophe. The transpose function is not limited to transposing vectors; it actually swaps
the rows and columns of any one or two-dimensional collection.
vec = [1 2 3 4 5];
vec2 = [6, 7, 8, 9];
emptyVec = [];
columnVec = [1; 2; 3; 4; 5]

The Colon Operator:

Many vectors are too large to create manually. To generate sets of evenly incremented
numbers, MATLAB uses a colon operator, not to be confused the semicolon used to suppress outputs.
The colon operator takes in a start value, a step value, and a final value; it generates a vector of
numbers ranging from the start value to the stop value and incremented by the step size. Expressions
using the colon operator do not have to be enclosed in brackets. All vectors are of type double.
colonVec = start:step:stop
If a step size is not specified, MATLAB will use the default step of one. Step sizes can be positive
or negative. Any of the three inputs can be substituted for a variable stored in the current workspace.
Putting stupid expressions into the colon operator almost never produces an error. Here is a list of
references for absurd situations.
1. If the step size is equal to zero, regardless of the other two inputs, produces an empty vector.
2. If the start and stop values are equal, returns a vector containing only that number unless the
step value is zero.

3. If the start value is less than the stop value but the step size is negative, produces an empty
vector.
4. If the start value is greater than the stop value but the step size is positive, produces an empty
vector.
Although the start value is always included in the resulting vector, the stop value will only be
included if it falls within the exact step size of the value before it. Thus, 0:4:7 produces the vector [0 4],
which does not include 7. You must always be careful to set the step value properly if you want the stop
value to be included.
Linspace ():
If you absolutely must include both the start and stop values or are interested in generating a
vector of a particular length, you can use the linspace() function. Linspace() takes in a start value, a stop
value, and the desired length of the output vector (NOT a step size) and produces a vector of linearly
spaced numbers between the start and stop values. The start and stop value are guaranteed to be
included, and the space between any two adjacent numbers is always equal. However, you will probably
end up with some weird fraction as your step size. If you do not input a desired length, MATLAB will use
the default value of 100.
linspaceVec = linspace(start, stop, length);

Basic Arithmetic with Vectors:

Any basic arithmetic involving a vector and a constant changes the value of all elements in the
vector. Remember to use the dot (.) with multiplication and division.
vec = [1 3 5 7 9];
vec = vec + 2
vec = vec.*2

vec = [3 5 7 9 11]
vec = [2 6 10 14 18]

Indexing Vectors:
To return a particular range of values from a given vector, you can index into it using
parentheses. The item in parentheses, or the range you wish to return from the vector, can be either a
double or collection of doubles (another vector) representing the positions in the original vector that
you wish to return. By using another vector as the index range, you can return the elements in any order
or even the same element multiple times. The length of the vector returned will always be equal to the
length of the index.
Because any index that returns multiple values is by definition a vector, indices easily be
generated using the colon operator. Doing so allows you to perform many nifty indexing tricks such as
reversing a vector or returning only even-numbered positions. If you wish to index up to the last
element in a vector, use the keyword end. Because MATLAB interprets end as a number equal to the
length of the vector, you can use subtraction or division to index positions based on the end value. Any
attempt to index with a number less than one, a fraction, or a number greater than the length of the

vector youre indexing into will produce an error. You will probably lose a majority of the points on
homework assignments from the message Error: Index exceeds matrix dimensions.
vec = [1 3 5 7 9];
vec(3)
vec([1 4 2 2 3]
vec(1:2:5)
vec(6)
vec(1:3:end)
vec([1:3, end, end-2])
vec(round(end/2):end)

ans = 5
ans = [1
ans = [1
Error
ans =[1
ans = [1
ans = [5

7 3 3 5]
5 9]
7]
3 5 9 7]
7 9]

Slicing Vectors:
The word slicing seems to be standard terminology for this course even though it makes no
sense to me. Slicing involves replacing some of the elements in one vector with some of the elements in
another vector. The most important thing to remember when slicing is to be careful about what you
choose to overwrite. To delete elements from a vector, set them equal to empty brackets. Here are the
basic forms of slicing:
1. A(index) = B(index)
This is the most common form. MATLAB first evaluates B(index), then replaces A(index) with it.
The rest of A remains unchanged. Produces an error if B(index) and A(index) are not the same
length.
2. A(index) = B
Replaces the elements in A(index) with whatever is stored in B. Once again, A(index) and B must
have identical lengths.
3. A = B(index)
This is the mistake many students make when slicing. This overwrites the variable A and sets it
equal to B(index). All elements in A are replaced, including the ones outside the range of B, so
the lengths can be unequal without producing an error.
4. A(index) = []
Setting anything equal to empty brackets deletes it. This can be used to shorten the length of A
or delete all of the elements entirely, leaving A as an empty vector.
Contrary to indexing, you can slice elements that are out of bounds of the original vector. If you
assign a value to an element that did not originally exist, MATLAB will extend the vector by filling all
intermediate positions with zeros. To slice outside the bounds of a vector of unknown length, use
addition or multiplication with end.
vec1 = [1 3 5 7 9];
vec2 = [0 2 4 6 8 10];
vec1([1 2 4]) = vec1(3:end)
vec1(1:3) = vec2(end:-1:4)
vec2(2:end) = vec1
vec1 = vec2(1:2:end)
vec2([1 4 5]) = []
vec1(end+2) = 6

The possibilities are endless

vec1
vec1
vec2
vec1
vec2
vec1

=
=
=
=
=
=

[5 7 5 9 9]
[10 8 6 7 9]
[0 1 3 5 7 9]
[0 4 8]
[2 4 10]
[1 3 5 7 9 0 6]

Concatenating Vectors:
MATLAB uses square brackets for just about everything. You can also use them to concatenate
vectors, or combine multiple small vectors into one large one. Simply create a vector in the standard
manner using vectors as the elements rather than scalar values. To concatenate vectors column-wise,
separate the different elements with semicolons.
vec1 =
vec2 =
bigVec
colVec

[1 5 9];
[6 8];
= [vec2, vec1]
= [vec1; vec2]

bigVec = [6 8 1 5 9]
colVec = [1; 5; 9; 6; 8]

Logical Indexing:
So far, we have only dealt with data of type double, representing real, measurable numerical
values. However, another data type frequently used in MATLAB is type logical. Logical data can only take
two forms: true or false. In most cases, the number one represents true, while the number zero
indicates false; however, MATLAB actually interprets anything not equal to zero as true, including
negative numbers and strings.
When using any of the expressions below, keep in mind that the two objects you wish to
compare must be dimensionally IDENTICAL. Attempting to compare two vectors of unequal length will
produce an error.
A logical expression will generate a collection of logical indicating where the statement is true
and where it is false (one for true and zero for false). Logical expressions generally involve comparisons
and can take numerous forms:

==, equal to
>, greater than
<, less than
>=, greater than or equal to
<=, less than or equal to
~=, not equal to

The tilde (~) is always used to refer to the word not. When preceding a logical collection, it
changes every true to false and every false to true. Nifty!
vec1 = [1 3 5 7 9];
vec1 == 3
vec1 > 3

ans = [0 1 0 0 0] (logical)
ans = [0 0 1 1 1]

Masking
Indexing with logicals is very different from indexing with doubles. The position numbers of
values being indexed are represented by the values of double indices and the position numbers of logical
indices. Thus, a true at position three in a logical collection will index the element in position three of

the collection being indexed. The index may be shorter than the vector being indexed, but not longer.
For some reason, logicals can even be used to index other logicals. For some reason, this is known as
masking.
vec1 = [1 3 5 3 9];
vec1(vec1==3)
vec1(vec1 <= 5)
vec1(true false true)

ans = [3 3]
ans = [5 9]
ans = [1 5]

Masking can be used with even more precision using and (&) and or (|). These can be used to
combine collections of logicals to form a final index representing all of them. Just remember these
seemingly random rules:
1. An and statement is only true if all of the elements are true.
2. An or statement is only false if all of the elements are false.
When multiple logical collections are combined using & or |, each element in the collections is
compared individually to the corresponding elements in the other collections. This strategy can be used
to make a single index that matches many different parameters.
vec1 = [1 3 5 7 9];
vec1 >1 & vec1 <= 7
[true false true] | [false false true]
(true | false) & (false & true)

ans = [3 5 7]
ans = [1 0 1]
ans = 0

Useful Functions:
Vector Stuff:

min(x) returns the smallest element in the collection (Note: if the minimum value occurs
multiple times, min() and max() will return only the first instance)
[value, index] = min(x) returns the smallest element along with its position number
max(x) returns the largest element in the collection
[value, index] = max(x) returns the largest element along with its position number
sort(x) sorts the elements in ascending order
[newX, index] = sort(x) returns x in ascending order along with a vector containing the position
numbers of the original x to which each value corresponds
fliplr(x) flips x left-to-right
x transposes x (row vector column vector)
mod(x,num) returns the remainder if x is divided by num
round(x) rounds x up or down
ceil(x) rounds x up to the next-highest integer if x is fractional
floor(x) rounds x down to the next-lowest integer if x is fractional
ones(x,y) creates an array of ones of dimension x,y (Use 1 for x or y to generate vectors)
zeros(x,y) creates an array of zeros of dimension x,y
true(x,y) creates an array of logical trues of dimension x,y
false(x,y) creates an array of logical falses of dimension x,y
sum(x) computes the sum of x

prod(x) computes the product of x

rand(x,y) generates an array of dimension x,y of linearly spaced random numbers between 0
and 1
randn(x,y) identical to rand() except all values are truly random
find(logical expression) generates a vector of all indices where the logical expression is true

Logical Stuff:

all(x) true if everything in x is true

any(x) true if anything in x is true
not(x) same as ~x
or(x,y) same as x | y
and(x,y) same as x & y

Section Four: Arrays

While vectors are merely one-dimensional collections of data stored as rows or columns, arrays
are two-dimensional collections stored as grids containing both rows and columns. They operate in
exactly the same way as vectors except for a few minor differences. Although arrays are supposed to be
different from matrices, theyre actually not. Arrays can contain doubles, logicals, and various other data
types.
While MATLAB is capable of storing data in more dimensions than are readily understandable,
this course never involves data collections of more than two dimensions, excluding images.

Creating Arrays:
Arrays are created using square brackets in much the same way as vectors. Simply enter the
individual rows of the collection separated by semicolons. Yes, the semicolon also has many diverse
uses. Arrays must always be rectangular; if at any point MATLAB attempts to create an array that is not
rectangular, it will produce an error.
arr = [1 2 3; 4 5 6; 7 8 9]

ans =

arr2 = [1 2; 3 4; 5 6]

ans =

arr3 = [1 2 3; 4 5; 6 7 8]

Error

Indexing Arrays:
The items in arrays are organized based on their location as determined by rows and columns.
Both rows and columns begin at one and increase as you move down and to the right, respectively.
Elements are indexed by their row numbers followed by their column numbers, with a comma in
between. To index all rows or all columns, replace the row or column numbers in the index with a colon
symbol. You can also use the keyword end to access the final element in a particular row or column, or
both if it makes you happy.
arr = [1 2 3; 4 5 6; 7 8 9]

ans =

arr(2,3)
arr(2,:)

ans = 6
ans = [4 5 6]

arr(:,2)
arr(end,end)

ans = [2;5;8]
ans = 9

Splitting Arrays into Halves and Quarters:

Dont bother memorizing these. You can easily reproduce them with some practice.
arr(1:round(end/2),:)
arr(round(end/2):end,:)
arr(:,1:round(end/2))
arr(:,round(end/2):end)
arr(round(end/2):end,round(end/2):end)

Top half
Bottom half
Left half
Right half
Bottom-right quarter

MATLABs Secret Trick:

Although arrays are displayed in the command window as two-dimensional matrices of values,
MATLAB actually stores the information as a very long column vector. It numbers the elements by
reading down the columns one at a time, meaning that arrays can actually be indexed as if they were
vectors. This can be useful to remember because it allows you to reshape arrays into vectors very easily.
arr = [1 2 3; 4 5 6; 7 8 9]

ans =

arr(:)
arr(4)
arr(8)

ans = [1;4;7;2;5;8;3;6;9]
ans = 2
ans = 6

Slicing Arrays:
By slicing arrays in creative ways, you can do all sorts of interesting things such as swapping
rows or columns, mirroring, deleting elements, and flipping things upside down. Make sure you know
and understand the first two, as they are extremely useful.
arr(:,:) = arr(:,end:-1:1)
arr(:,:) = arr(end:-1:1,:)
arr(2:2:end,:) = arr(end:-2:2,:)
arr(1:3:end,2:2:end) = []

Mirrors left to right

Flips upside down
Reverses the even-numbered rows
Deletes every third row and second column

Concatenating Arrays:
Arrays can be concatenated horizontally or vertically just like vectors. Once again, any command
that would produce a non-rectangular array results in an error.
[arr1, arr2]
[arr1; arr2]

Concatenates horizontally
Concatenates vertically

Logical Indexing and Masking:

Seriously, its the same thing.
arr = [1 2; 3 4];

ans =

arr > 3
arr(mod(arr,2))==0)=7

ans = [0 0; 0 1]
ans = [1 7; 3 7]

When you need to determine which elements meet multiple overlapping conditions, you can
use and (&) and or (|) to combine multiple sets of logicals.
mod(arr,2)~=0 & arr >= 5
arr > 10 | arr==7

All odd elements greater than or equal to five

All elements greater than 10 or equal to 7

Basic Arithmetic:
All arithmetic calculations involving arrays are performed element-by-element. Remember to
use the dot with multiplication and division.
arr = [1 2; 3 4]

ans =

arr2 = [5 6; 7 8];

ans =

arr + arr2
arr .* arr2
arr 1

ans = [6 8; 10 12]
ans = [5 12; 21 32]
ans = [0 1; 2 3]

Transposing and Rotating Arrays:

Just as the apostrophe turns row vectors into column vectors, it also transposes the rows and
columns in arrays. The first column becomes the first row, the second column becomes the second row,
etc. This is not to be mistaken with rotating an array clockwise or counterclockwise, requires both
transposition and indexing.

arr
arr(end:-1:1,:)
arr(:,end:-1:1)
arr(end:-1:1,end:-1:1)

ans = [1 4 7; 2 5 8; 3 6 9]
Rotates clockwise
Rotates counterclockwise
Rotates 180 degrees

Reshape()
This function allows you to change the number of rows and columns in an array without altering
its overall size. MATLAB fills the spaces in the new array in the same order as the original, namely
reading down the columns one at a time. Thus arr(4) will be equal to new_arr(4).
You probably wont be using reshape() more than once or twice.
new(arr) = reshape(arr, new_rows, new_columns)
arr2 = [1 2; 3 4; 5 6]

ans =

new_arr = reshape(arr,2,3)

ans =

Useful Functions:

ones(x,y) creates an array of ones of dimension x,y (Use 1 for x or y to generate vectors)
zeros(x,y) creates an array of zeros of dimension x,y
true(x,y) creates an array of logical trues of dimension x,y
false(x,y) creates an array of logical falses of dimension x,y
magic(x)creates a magic square of dimension x
[row,col] = size(arr)returns the number of rows and columns in the input array

The following functions work differently with arrays than with vectors. Built-in functions generally treat
arrays as individual column vectors and express their outputs as row vectors.

[value,index] = min(arr)returns a row vector of the minimum values in each column. The
second output contains the row number of the minimum value in each column
[value,index] = max(arr)returns a row vector of the maximum values in each column. The
second output contains the row number of the maximum value in each column
[sorted_arr,index] = sort(arr)sorts each column of the array in ascending order from top to
bottom. The second output consists of the indices used to sort each column concatenated
together as an array
sum(arr)returns a row vector of the sum of each column in the array
mean(arr)returns a row vector of the mean of each column in the array

To determine the absolute minimum, maximum, sum, or mean of an array, simply call the respective
functions twice.

[value, index] = max(max(arr))returns the absolute maximum of the array along with its
column number
[value, index] = min(min(arr))returns the absolute minimum of the array along with its column
number
sum(sum(arr))returns the overall sum of the elements in the array
mean(mean(arr))returns the overall mean of the elements in the array

I know these explanations can be somewhat confusing, so do yourself a favor and play around
with the min(), max(), mean(), and sort() functions for a few minutes. All four frequently appear on
homeworks and test problems.

Section Five: Strings

Vectors and arrays are only useful for organizing sets of doubles and logicals. Now, we will be
dealing with a new data type called char (short for character). Using characters allows you to set
variables and outputs equal to meaningful words and phrases rather than just numbers.
A string is a vector of numbers that represent the characters and symbols on your keyboard.
MATLAB identifies these number-character combinations based on a predetermined set of conversions
called an Ascii Table.

The columns labeled Dec contain the ascii codes for the corresponding red characters in the
columns labeled Char. For example, the ascii code for a capital Z is 90. Note that the code for a
lowercase letter is exactly 32 plus its uppercase equivalent. The ascii code for a space is 32.
The most important thing to remember about strings is that they function exactly like vectors in
most scenarios. To better understand the functionality of strings, well be looking at a lot of examples.

Creating Strings:
Strings are created by enclosing letters, numbers, and symbols in single quotations marks ().
Unlike with creating vectors, you neednt separate the elements with spaces or commas (remember that
spaces and commas now count as elements). You can also create empty strings that function just like
empty vectors.
If you have a string saved as a variable in your workspace, it will have a cute little ab symbol
next to it. The class() of a string is char. In fact, the class() function always returns an answer of type
char.

str = Happy Birthday;

str2 = 770-253-5796;
emptyString = ;
class(str)
class(true)
class(logical)

ans = char
ans = logical
ans = char

Char() and Double():

MATLAB contains two built-in functions for converting between different data types. double()
changes a character string into a vector of each characters representative ascii code; char() converts a
vector of doubles into the characters they represent. You can therefore create a vector of ascii codes
and use char() to change it into a string.
str = Ishmael;
double(str)
vec = [77 111 115 101 115];
char(vec)

ans = [73 115 104 109 97 101 108]

ans = Moses

Basic Arithmetic with Strings:

You can use basic arithmetic to change characters in strings to other characters. From our
previous example, we know that adding 32 to a string of uppercase letters will convert them to
lowercase. Just remember the golden rule of strings: whenever you add or subtract elements in strings,
MATLAB always changes them back to doubles. You will therefore have to use char() to keep your
original string format.
str = SARAI;
str + 32
char(str + 32)
char(50) + char(50)

ans = [115
97
ans = sarai
ans = 100

114

97

105]

Indexing Strings:
You can index the elements in a string the same way you would with a vector. Just keep in mind
that MATLAB now interprets EVERYTHING in the string as a separate element, including spaces and
commas. Indexing strings returns the actual characters, not doubles.
str = Adam and Eve;
str(2)
str(5)
str(3:7)
length(str)

ans
ans
ans
ans

=
=
=
=

d

am an
12

Slicing Strings:
If you can slice vectors, you can slice strings. If you slice beyond the bounds of the original string,
MATLAB will fill the intermediate spaces with ascii zeros, not 32s.
str1 = Isaac;
str2 = Rebecca;
str1(3:end) = str2(2:4)
str2(1:5) = str1
str1 = str2(1:end-1)
str1([1 3]) = []

str1
str2
str1
str1

str3 = Abram;
str3(8) = O
double(str3(7))

ans = Abram
ans = 0

=
=
=
=

Isebe
Isaacca
Rebecc
sac

Concatenating Strings:
You only need to worry about concatenating strings horizontally. Technically, they can be
concatenated vertically to make freaky string arrays, but doing so is virtually useless. Keep in mind that
strings are concatenated just like vectors, so MATLAB will not add spaces to separate words.
You can also concatenate strings with doubles since theyre virtually the same thing. MATLAB
always retains the string format after concatenation.
tstr2 = Ham;
str3 = Japheth;
[str1 str2 str3]
[str1 65]
[65 str1]

ans = ShemHamJapheth
ans = ShemA
ans = AShem

Logical Indexing with Strings:

Strings can be used in logical expressions much like vectors. The only difference is that you can
now make comparisons between characters and doubles. Remember that comparing vectors of
different lengths produces an error. It is impossible with strings of unlike lengths as well; theres a
special function for that.
Logical expressions are very useful for determining capital and lowercase letters and for
differentiating between letters and symbols. Once again, you can locate elements that fulfill multiple
logical conditions using and (&) or or(|).
str1 = Jacob;
str2 = Rachel;
str1==c
str2==82
str1==str2

ans = [0 0 1 0 0]
ans = [1 0 0 0]
Error

str>=65 & str<=90

Capital letters
str>=97 & str<=122
Lowercase letters
(str>=65 & str<=90)|(str>=97 & str<=122) All letters

Strcmp() and Strcmpi():

The strcmp() function allows you to compare strings of unequal length. This is important
because attempting to use logical expressions to do so produces an error. Strcmp() outputs a logical 1 or
0 representing whether the two input strings are exactly equivalent.
Strcmp() is case-sensitive. There is another built-in function, strcmpi(), that performs the same
function while ignoring case
Strcmp() and strcmpi() can also be used to compare a string with a cell array of strings, in which
case they output a vector of logicals.
str1
str2
str3
str4

=
=
=
=

Ephraim;
Manasseh;
Ephraim;
manasseh;

strcmp(str1, str2)
strcmp(str1, str3)
strcmp(str2,str4)
strcmpi(str2,str4)

ans
ans
ans
ans

=
=
=
=

0
1
0
1

Sprintf():
The sprintf() function allows you to create strings containing variables. Consider the following
situation: You need to create a string stating how old someone is, but the persons name and age are
stored in variables in your workspace. You can either concatenate the string manually or use the
sprintf() function.
sprintf() takes in a string containing variable markers followed by the parameters you wish to
use. The number of inputs will always be equal to one plus the number of variables in the output string.
Variable markers are placed using a percent (%) sign followed by a letter designating the type of variable
(double or string, usually). Type %d for a double variable and %s for a string. Remember to list the
variable names in the same order you used them; you can use the same variable multiple times in the
formatted string, but you still have to list it once for each time it is used.
name = Methuselah;
age = 969;
[name is age years old]
sprintf(%s is %d years old, name, age)
sprinf(%d %d %d %d, age, age, age, age)

ans = Methuselah is 969 years old

ans = Methuselah is 969 years old
ans = 969 969 969 969

Strtok():
The strtok() function is used to split (tokenize) an input string into two parts based on the
location of a specific character called the delimiter. Basically, you choose what character to search for
(the delimiter) and MATLAB will divide the input string into two output strings according to the following
rules:
1. MATLAB locates the first non-delimiter character and deletes all delimiters preceding it.
2. MATLAB locates the next delimiter after the character in (1). The first output is every character
from the remaining string up to the character before the delimiter. The first output can NEVER
contain the delimiter.
3. The second output is everything else, including the delimiter found in (2).
4. If the input is an empty string, both outputs will be empty strings.
5. If the string does not contain any delimiters, the first output is the entire string, and the second
output is an empty string.
6. If the string contains only the delimiter, both outputs will be empty strings.
Strtok() is one of the most important functions to know for MATLAB. It is used heavily in both
string manipulation and file input/output.
str = eeeeAbel Cain;
[A,B] = strtok(str,e)

A = Ab
B = el Cain

[C,D] = strtok(B, )
[E,F] = strtok(D, )

C
D
E
F

=
=
=
=

el
Cain
Cain

Lower() and Upper():

The lower() and upper() functions take in a string and convert all letters to lowercase or
uppercase letters, respectively. Because they ignore special characters in strings, they are generally very
useful.

Solving Madlib Problems:

For some reason, Madlib problems are very popular in string-related drill problems. The
difficulty arises from the concept of replacing part of a string with another string of unequal length.
Consider the following Madlib sentence:
str

= I can do all things through @ who strengthens me.

A typical problem may ask you to replace the @ symbol with the word Christ. Unfortunately,
simply slicing letters into the original string will overwrite the end of the sentence rather than insert the
word in place (just think of the same problem using vectors instead of strings). You must therefore use
either strtok() or manual concatenation to accomplish the insertion.
Remember that the find() function returns a vector of all true positions, not merely the first one.
index
index
str =
OR
[A,B]
str =

= find(str==@);
= index(1);
[str(1:index-1), Christ, str(index+1:end)];
= strtok(str, @);
[A, Christ, B(2:end)];

Section Six: Conditionals

Conditional statements allow you to determine whether or not to run a certain block of code
based on some non-predetermined information. Basically, its what you wish you could have been using
for all the previous homework assignments. Conditionals come in two formsif and switch statements,
but knowing switch statements is more useful than absolutely necessary.

If Statements:
All if statements are coded in the following basic format:
if <logical expression>
<code block>
end

MATLAB will automatically turn the words if and end blue to designate them as markers
pertaining to the code block in between them. MATLAB first evaluates the logical expression to the right
of if. If the statement is true, the code block between if and end is run as if the conditional
statement did not exist. Otherwise, MATLAB jumps from if to end and ignores the code in between.
More complicated conditionals can be created by adding the keywords elseif and else. Elseif
statements function like extra if expressions (logical1 was false, but maybe logical2 is true). You can
have as many elseif statements as you want but each must contain its own corresponding code block. If
you find that multiple statements contain the same code block, you should consider combining them
into one logical statement using or (|).
An else statement, if you use one, must be the final statement in your overall conditional. It
functions like an elseif statement that is automatically evaluated as true. However, an else statement
will by definition only be run if all preceding conditional statements are evaluated as false.
The order of conditional statements is extremely important because MATLAB will run only the
code block corresponding to the FIRST true logical expression. Thus, it will not evaluate any expressions
following the first true one and will skip to end instead.
Finally, EVERY conditional statement (if or switch) must have a corresponding end. Failing to
include enough ends will cause an error. MATLAB automatically aligns any end you type with the
nearest designated conditional statement.
If <logical expression1>
<code block1>
elseif <logical expression2>
<code block2>

elseif <logical expression3>

<code block3>
else
<code block4>
end

Nested Conditionals:
Believe it or not, you can actually place conditionals INSIDE other conditionals . Sometimes,
nesting conditionals is the only way to solve a particular problem. More frequently, however, you can
use it to make your code more legible, pretty, and/or cool. Consider the following example:

Write a function that takes in a vector and checks to see if all its elements
are perfect squares between 6 and 23 that are not divisible by four. The
function should output a logical true or false.

function out= Swordfish(vec)

out = false;
if all(mod(sqrt(vec),1))==0
if all(vec>6)
if all(vec<23)
if all(mod(vec,4)~=0)
out = true;
end
end
end
end
end

As you can see, this code is much easier to understand than its one-line equivalent. Notice that a
common strategy is to arbitrarily initialize the output variable as one possibility and change it depending
on the result of the if statements; doing so ensure that your output variable will always be defined,
thereby reducing the chance of your code producing an error.

Switch Statements:
Switch statements are if statements that determine the output based on the value stored in a
defined variable. The terminology is slightly different for no necessary reason. case now replaces
elseif, and otherwise replaces else. Once again, you can have as many cases as you want. The first
line always reads switch followed by the name of the variable you are switching over.

switch <variable>
case <possible value>
<code block>
case <possible value>
<code block>
otherwise
<code block>
end

For instance, if you have a string stored in the variable str, you can use a switch statement to output
whether it is four, five, or six.
switch num
case four
out = 4;
case five
out = 5;
case six
out = 6;
otherwise
out = Pharaoh;
end

Seriously, folks, conditionals are the easiest thing in the world to write. If you encounter a
difficult problem involving conditionals, the actual difficulty will come from the other elements and not
from forming the conditionals themselves.

Section Seven: Iteration

Iteration is a fancy MATLAB word for loops, which is a fancy computing word for running a single
code block multiple times. Loops are reasonably difficult to understand because, unlike many other
coding tools, they are not intuitive. Whereas you would tend to focus on the whole problem to find all
solutions at once, MATLAB can look at every possible solution individually.
Consider a problem asking you to solve a fourth-order polynomial equation for all real zeros.
One approach would be to algebraically or graphically find the solutions (all at once); another approach
would be to try plug numbers into the equation until you find all four solutions. Loops allow you to
implement the latter method, typically called a brute force approach.
There are two types of loops, for loops and while loops. Technically, for loops are unnecessary
because all for loops can be rewritten as while loops. Paradoxically, students almost never resort to
using while loops because they are slightly more difficult to understand.

For Loops:
All for loops are written in the following basic format:
for i = <vector>
<code block>
end

When MATLAB reaches the loop, it will automatically set i equal to the first value in the vector,
henceforth called the iteration paramater. When it reaches the end of the loop, namely the word end,
MATLAB returns to the first line and changes the value of i to the next value in the vector. This process
continues until there are no values remaining for i to assume, at which point MATLAB terminates the
loop and reads the rest of your code.
Determining the first line is generally the most complicated part of writing a loop. The
parameter you set determines how many times the loop will run. Thus, the primary drawback of using
for loops is that you must be able to predetermine how many times to iterate. You will generally use the
variable i as an index that changes for each iteration. There are two basic strategies:
1. Set i equal to a range of values using the colon operator.
for i = 1:10
for i = 1:length(variable)
2. Set i equal to a predetermined or inputted vector. People usually forget that the vector can
be ANYTHING, not just a range of values with step size 1. MATLAB will still iterate a number
of times equal to the length of the vector, but i will automatically assume values from the

vector itself rather than indices.

for i = vec
I always use the variable i when implementing for loops because thats what my TA used.
However, use whatever variable system makes sense to you. They best way to learn loops is by looking
at examples.

Write a function sumVec that takes in a vector and outputs the sum of its elements using iteration.
function out = sumVec (vec)
out = 0;
for i = 1:length(vec)
out = out + vec(i)
end
end

However, this solution has an unnecessary complication. I deliberately ignored the more sensible
approach of simply setting i equal to the vector itself.
function out = sumVec (vec)
out = 0;
for i = vec
out = out + i
end
end

Write a function compareVec that takes two vectors and outputs a vector of all elements that appear in
both vectors in the same position with the same value.
Any problem that asks you to build an output vector inside a loop will have a solution in the following
format:
out = [];
for i = <vector>
if <logical expression>
out(end+1) = <whatever your output should contain>
end
end

The idea is to initialize the output as an empty vector OUTSIDE the loop and build it element-byelement INSIDE the loop. By indexing the vector at position end+1, we continually increase the size of
the output vector by adding elements onto the end. A simple assignment statement (out = i) would
overwrite the current output variable every time the loop is run. Returning to the compareVec problem:

function out = compareVec (vec1, vec2)

out = [];
L = length(vec1);
if length(vec2) < length(vec1)
L = length(vec2);
end
% We must first determine the length of the shorter vector so we know how
%many times to iterate.
%Further iteration will result in indexing out-of-bounds.
for i = 1:L
if vec1(i)==vec2(i)
out(end+1) = vec1(i);
%We could also tack on values to the end by concatening the current vector
%with the new value.
%out = [out, vec1(i)];
end
end
end

Loops placed inside other loops are called nested loops. Nesting loops allows you to iterate over
multiple parameters simultaneously. As a final example, we will consider a case that cannot be solved
with only a single loop.

Write a function compStr that takes in two strings of equal length and
returns the largest set of consecutive letters that appear in both strings at
the same position, disregarding case.
compStr(henceforth, heraldrth) = rth

function out = compStr(str1,str2)

out = ;
%Once again were building an output element-by-element, so we have to
initialize it OUTSIDE the loop.
for i = 1:length(str1)
for j = i:length(str2)
%If j is less than i, MATLAB will waste time comparing nonexistent ranges
like str(4:3). Since str1 and str2
%must be of equal length, we could have used the length of str1 instead.
if strcmpi(str1(i:j), str2(i:j)) && j-i+1 > length(out)
%We compare each range of values from str1 with the equivalent range from
str2. Unless the match is
%longer than the ones weve already found, it is discounted.
out = str1(i:j)
end
end

end
end

While Loops:
All while loops look like this:
while <logical expression>
<code block>
end

MATLAB first evaluates the logical expression and runs the loop until the expression becomes
false. Thus, you must have something inside the loop that changes the outcome of the logical
expression. Otherewise, MATLAB will enter an infinite loop that can only be terminated using control+c.
Although there is no iteration variable associated with while loops, all for loops can be rewritten
as while loops using a counter variable inside the loop. The idea is to increase the counters value by one
each time the loop is run. Lets rewrite the sumVec function as a while loop.
function out = sumVec (vec)
num = 1;
out = 0;
while num <= length(vec)
%Dont forget to include the equals sign!
out = out + vec(num);
num = num + 1;
%This line moves the while loop toward the terminating condition.
end
end

However, while loops are usually used for loops in which the number of iterations is impossible
to evaluate beforehand.

Write a function Gideon that takes in a number. If the number is even, divide it by two; if it is odd,
multiply it by three and subtract one. The function should repeat this process until the number equals
one (this always happens) and outputs the number of iterations required.
function
out =
while
if

out = Gideon (num)

1;
num ~= 1
mod(num,2)==0
num = num/2;
else

num = num*3 1;
end
out = out + 1;
end
end

In more colloquial terms, loops are dead useful. You can solve almost any problem with a wellimplemented for loop. Make sure you understand loops, because almost every problem from now on
will involve them.

Section Eight: Cell Arrays

Cell arrays are another data type MATLAB utilizes to store information. Despite being
intimidating for most people, cell arrays are easy to work with as long as you UNDERSTAND what they
are. Seriously, you dont want to end up guessing when cell arrays appear on exams.
Cell arrays are a unique data type in that they are capable of storing data of multiple other
types. Each element in a cell array is a cell, and a cell can contain anything (doubles, logicals, strings,
structures, or even other cells). My shorthand for a cell array is ca.

Creating Cell Arrays:

Cell arrays are collections enclosed in curly brackets. The only difference is that elements can
now be of any data type. This is the easy part.
ca = { [1 2 3], Judah, [true true false] };

Indexing Cell Arrays:

The complicated thing concerning cell arrays is that they can be indexed in true different ways,
using parentheses () or curly brackets {}. Indexing with parentheses returns the actual element (in a cell
array, a cell), whereas indexing with curly brackets returns the CONTENTS of the element. In other
words, curly brackets open a cell to display what it contains; only cells can be indexed with curly
brackets.
To return specific components of the contents of a cell, use curly brackets to open the cell
followed by parentheses to index its contents. When cells are nested inside each other, you must use
multiple sets of curly brackets to reach whatever is ultimately inside. The rule, which is well worth
memorizing, is that you can only index with parentheses once, and the parentheses must come last in a
set of indices; this makes sense if you pause to think about it.
ca = {[true false], Reuben, {[1 2 3], Gad}};
class(ca)
ans = cell
class(ca(1))
ans = cell
class(ca{1})
ans = logical
ca(3)
ans = {1x2 cell}
ca{2}
ans = Reuben
ca{2}(2)
ans = e
ca{3}{1}(3)
ans = 3

Thats pretty much it for cell arrays. Remember that you still use square brackets, not curly
brackets, when concatenating and deleting cell arrays. Most of my examples will involve intricate
combinations of cell arrays and loops.
When using cell arrays and especially iteration and conditionals, it is generally necessary to keep
track of data types. The functions isdouble(), islogical(), ischar(), and iscell() return a logical true or false
of whether the input is of the specified data type. The function isempty() returns true if the input is an
empty string, vector, cell array, or structure array.
Write a function Manasseh that takes in a cell array containing vectors of
equal length and meshes them together into one long output vector. The first
element in the output vector should be the first element of the first vector,
followed by the first element of the second vector, and so on.
Manasseh({[1 2 3] [4 5 6] [7 8 9]}) ans = [1 4 7 2 5 8 3 6 9]

function out = Manasseh (ca)

%Perhaps the most intuitive way to solve this problem would be to create a
simple test case and focus %on the identity of the output in relation to the
inputted cell array. For the cell array {[1 2] [3 4] [5 6]},
%the output is [ca{1}(1), ca{2}(1), ca{3}(1), ca{1}(2), ca{2}(2), ca{3}(2)].
It now becomes apparent that the
%first index starts at one and approaches the number of vectors, and the
second index starts at one and
%approaches the length of the vectors.
out = [];
%The solution is deceptively simple but requires two simultaneous for loops.
for i = 1:length(ca{1})
%The outer loop runs once for every element in the vectors (the vectors are
all of equal length).
for j = 1:length(ca)
%The inner loop runs once for each vector in the cell array.
out(end+1) = ca{j}(i);
end
end
end

Write a function Mephibosheth that takes in a string and divides it into

individual words. The function should output a cell array containing one word
per cell.
Mephibosheth(Jacob I have loved)

function out = Mephibosheth (str)

out = {};

ans = {Jacob I have loved}

while ~isempty(str)
%Iterate until there is no string remaining.
[word, str] = strtok(str, )
%Reassigning the value of str moves us toward the condition that terminates
the loop.
out{end+1} = word;
end
%Youll be using this code to break down text from notepad files in a few
weeks.
end

Section Nine: High-Level File Input/Output

File input/output, hereafter referred to as file IO, is a process by which MATLAB can read
documents from other from other computer programs, manipulate them, and save them as new
documents in your current directory. High-level file IO specifically involves dealing with highly organized
data in notepad and Microsoft Excel. We will be working with three types of files: csv files, dlm files, and
excel spreadsheets.
The primary difficulty with file IO involves conversions between different data types. If you find
yourself using loops to accomplish simple conversions or comparisons, youre probably doing something
wrong. There are a few easy tricks that will make these problems much, much easier.

Reading and Writing CSV Files:

CSV stand for comma-separated values. Basically, a CSV file is a Notepad .txt document
containing only an array of numbers separated by commas. Any spaces and empty lines in the file are
completely ignored. If the document contains anything other than doubles (negative numbers and
fractions are okay) and commas, it is not a CSV file. There are two functions you need to know.
arr = csvread(filename)
The function csvread() takes in the filename (always a string) of a .txt CSV file in your current
directory, removes the commas, and outputs the resulting array of doubles. If the array is not
rectangular, it fills any unused space with zeros.
For example, csvread will turn the following file into the resultant array:
1,2,3
4,5,6
7

csvwrite(filename, array)

The other significant function is csvwrite(), which does exactly the opposite. It takes in an array
of doubles and saves the corresponding CSV document as a file in your current directory. Please note
that csvwrite(), and any writing function for that matter, does not have any output; you should not
suppress it with a semicolon or set it equal to anything.

Reading and Writing DLM Files:

A dlm file is exactly the same as a csv file except that the doubles can be separated by any
delimiter rather than just commas. Delimiters can be characters such as exclamation points, letters, or
even numbers. Remember that the delimiter is always considered a string even if it is a number. Once
again, there are two function you need to know:
arr

= dlmread(filename, delimiter)

Unlike with csv files, MATLAB does not automatically know what character is being used as the
delimiter in a text file; you will therefore need to specify the delimiter in string format as a second input.
If you dont feel like including the second input, MATLAB will attempt to guess the delimiter for you; it
likes to choose things like commas and spaces. You can also write delimited files just like csv files; the
only difference is that you must now specify the delimiter you want MATLAB to use.
dlmwrite(filename, array, delimiter)

Reading and Writing Excel Spreadsheets:

Peforming tasks on excel spreadsheets is easily the most useful application of high level file IO,
which is why most homework and test problems focus on this topic specifically.
[num, text, raw] = xlsread(filename)

The function to read in excel documents has three outputs instead of one. Although the names
you assign them do not matter, the standard convention in CS 1371 is to use the names num, text, and
raw. Obviously, MATLAB will always produce the same outputs in the same order, so knowing the order
of the outputs is very important.
The first output, num, finds all cells in the spreadsheet that contain numbers and turns them
into an array of doubles, much like csvread(). MATLAB finds the smallest rectangle of cells that includes
all numbers in the spreadsheet and forms them into an array. If any cells in the rectangle contain nonnumbers or are empty, MATLAB pads the empty space with NaN, which stands for Not a Number.
The second output, text, finds all the words (non-numbers) in the spreadsheet and formats
them into a cell array of strings in the same way, creating the smallest rectangle possible. Empty cells
and cells containing numbers are padded with empty strings.
The third output, raw, takes the entire spreadsheet and turns it into a cell array. Empty spaces
are padded with cells containing NaN.
The most important things to remember are the respective data types of the three outputs.
Num is an array of doubles, whereas text and raw are both cell arrays. When performing numerical
calculations, it is often easier to use num; the drawback is that num is not necessarily the same size as
the spreadsheet itself. Raw is by definition the same size but carries the drawback of being a cell array.
When solving problems, choose whichever output seems easiest for you.

Given the spreadsheet people.xls

[num, text, raw] = xlsread(people.xls)

num =

text = {

raw = {

You can also write arrays and cell arrays into excel spreadsheets. The function name is pretty
obvious.
xlswrite(filename, arr)

And thats basically it for high level file IO. Ill quickly recap the six functions before moving onto
example problems.
arr = csvread(filename)
csvwrite(filename, arr)
arr = dlmread(filename, delimiter)
dlmwrite(filename, arr, delimiter)
[num, text, raw] = xlsread(filename)
xlswrite(filename, arr)

Write a function Haggai that takes in the name of an excel document containing information about a
footrace in the following format. The document may have any number of rows but will always contain

three columns with the following column headers:

The function should read in the spreadsheet and add an additional column called Result as the fourth
column. This column contains the final position of each competitor based on his total time. The function
should also add an additional row to the end of the spreadsheet. The first column in this row will be the
string Fastest Lap, and the second column will contain the overall fastest lap in the race. The function
should write the resultant cell array to an excel file. The filename should be the input filename with
_edited appended to the end. The final output will look like:

function Haggai(fn)
[num, ~, raw] = xlsread(fn);
% Reading in the file is usually the first step.
totals = num(:,end);
%For the first step, were only interested in the total time of each
%competitor. This is a race, so lower
% total times will receive lower position numbers. To achieve this, well use
%the second output of the
%sort() function, which is the indices used to sort the input.
[~, rank] = sort(totals);
%Unfortunately, a simple sorting method doesnt quite work with ranking
%people, so we have to call the
%sort function AGAIN on the indices. Ill overwrite both variables from the
%previous function since we %wont be using them anymore.

[~, rank] = sort(rank);

%Now we have a column vector called index that contains the final rank of
%each competitor. To
%concatenate this onto the raw spreadsheet data, we have to convert it to
%type cell.
%index = {index} will put the entire vector inside a SINGLE cell, which is
%unfortunately not what we want.
%We need to create a vector of cells, each containing one number from our
%index vector. Luckily,
%theres a MATLAB function that does just that.
rank = num2cell(rank);
%Now we need to concatenate rank onto the string Result. Since were
concatenating cells, both
%elements must be of type cell. We therefore enclose the string in curly
brackets before concatenating.
column = [{'Result'}; rank];
%Were finished with the last column, so we can add it onto the raw
%spreadsheet data.
raw = [raw, column];
%Now we just have to determine the fastest lap and perform the same
%algorithm. Remember that were
%adding a row instead of a column, so horizontal and vertical concatenations
%will be reversed.
fastest_lap = min(num(:,1));
%We cant concatenate this time because the dimensions are not consistent, so
%well use MATLABs
%indexing out of bounds trick. MATLAB will pad the remaining spaces with
%empty cells.
raw(end+1,1) = {'Fastest Lap'};
raw(end,2) = {fastest_lap};
%Now all we have to worry about is writing the array to the correct filename.
%Simply appending
%_edited onto the end will produce something like race.xls_edited, which
%is not a valid filename.
%The .xls part must come last. There are probably a dozen ways to do this.
%Heres an elegant one.
fn2 = sprintf('%s_edited.xls', strtok(fn, '.'));
%You can also use find() to locate the period.
xlswrite(fn2, raw)
end

Useful Functions:

num2cell(arr)converts an array of doubles into a cell array where each cell contains one
double
cell2mat(ca)converts an array of cells containing doubles into an array of doubles
[sort, index] = sort(vec)always important for sorting

Section 10: Low-Level File Input/Output

Obviously, not all data are organized into csv files, dlm files, and excel spreadsheets. In
particular, MATLAB is useful for manipulating text documents in notepad, hereafter called low-level file
IO. Because low-level file IO involves looping through documents one line at a time, it is much more
confusing than high-level file IO. I will outline a basic procedure for any coding problems you may
encounter.
1. Use fopen() to read in the .txt file. You may need to use fopen() to create an output file as well.
2. Use fgetl() or fgets() to read in the first line of the document.
3. Create a while loop that runs as long as the line is of type char.
a. You may wish to use strtok() and another while loop to pull out each individual word.
b. Use fgetl() or fgets() to read in the next line of the document.
c. Somewhere in the while loops, use conditionals to implement whatever changes you need.
d. Depending on your loop, you may need to use a loop-and-a-half.
4. Once you have completed your output, fclose() all documents.
Just like most MATLAB concepts, understanding the functions and concepts is easy; the difficult
part is using the correct loops and conditionals in the correct places. For this reason, tracing problems
can also be fairly annoying.

Fopen()
The first step to any problem is to actually open the file youre interested in. You can assume
that all files are .txt documents written in notepad.
fh = fopen(filename, permission)

Once again, the filename is always a string with .txt at the end. The output for fopen() is called a
file handle (please take note of my abbreviation fh). A good pictorial representation of a file handle is
an arrow that points to a particular line in the document; the file handle is initialized at the first line of
the document and moves down one line at a time until it reaches the end. The file handle is always a
double. Although you cant directly do anything with the file handle itself, it becomes the input for other
file IO functions that are useful.
The second input for fopen() is a permission statement, a string representing what youre
allowing MATLAB to do with the file. For instance, MATLAB cannot alter previously existing documents
unless you give it a write permission. The three most common permissions are r, w, and a (read,
write, and append).
The read permission allows you to read in a file but not change it, basically for extracting data to
perform other functions on. Write allows you to overwrite an existing document (useless) or create an

entirely new blank document. If you specify the write permission and the filename does not exist in your
current directory, MATLAB will autotmatically create a blank notepad file along with its respective file
handle. The append permission allows you to add more information onto the end of an existing file.
Fortunately, youll probably never be appending anything in this class, so dont worry about it.
There is one interesting error you may encounter when using fopen(). If a call to the file handle
later in your script produces an error like Error using fgetl: invalid file identifier, fopen() probably failed
to open the input file. This tends to happen only when using the write permission to create a new file
and is caused by some sort of MATLAB permission error. MATLAB will continue reading your code but
will set the file handle equal to -1 to indicate a failure in opening the file. If you encounter this error,
change your current directory to something else and then change it back; this will reset MATLABs
permissions.
fh = fopen(fn, r)
fh = fopen(fn, w)
fh = fopen(fn, a)

Fgetl() and Fgets()

As I stated earlier, the file handle is actually relatively useless; you need a different function to
move through the document and take out each line. The fgetl() and fgets() function return the line the
file handle is currently pointing to, although in slightly different ways. Please note my shortand line to
represent a line of text.
line = fgetl(fh)
line = fgets(fh)

These two functions output the current line (whichever one the file handle is pointing to) as a
string. Each time you call either of these functions, the file handle moves down to the next line in the
document, so you can actually move through the entire document simply by running the same line of
code over and over.
Because of this fact, a simple while loop could easily pull out every line from a particular
document. When the file handle has moved down far enough that there are no new lines remaining in
the document, fgetl() and fgets() output a -1 instead of a line. Because this -1 is a double rather than a
string, you can set your while loop to run as long as the current line is of type char.
while ischar(line)
<code block>
end

There is only one noticeable difference between fgetl() and fgets(). Each line of text in a
document ends with a new line character: \n. This symbol tells computer programs to skip to the next

line on the page before continuing. Although the new line character is usually invisible (its not really
text), it is inherently there.
Fgetl() removes the new line character from the output and returns only the relevant text as a
character string. Fgets() returns the entire line including the new line character. Fgetl() is more useful if
you are only concerned with interpreting data in a notepad file; fgets() is more useful when you are
copying data to a different file because it automatically copies new line characters as well.

Fprintf()
The fprintf() function allows you to print text to the command screen or a notepad file. The
word print may be confusing if you are unused to it. To see an example of printed text, type a line of
code without suppressing the output; the result of your code will be printed in the command window.
Fprintf() does not have an output.
fprintf(filename, str, var1, var2,)

Like sprintf(), fprintf() can be used to print formatted strings, or strings containing variables.
Simply replace the doubles and strings in the formatted string with %d and %s, respectively. If you dont
specify the file handle of the document you wish to print to, MATLAB will print to the command
window. To utilize tabs and new lines, print \t and \n, respectively.
In the following examples, were printing to a notepad file called Rahab.txt.
fh = fopen(Rahab.txt, w)
fprintf(fh, %s, Mordecai)
fprintf(fh, %s\n, Mordecai)
fprintf(fh, \n)

prints Mordecai
prints Mordecai and new line character
prints new line character

Strtok Loops
File IO coding problems frequently entail analyzing every individual word of a document. This
task is most easily accomplished using strtok() and a while loop, with a space as the delimiter. The loop
should continue until the remainder of the string, the second output of strtok(), no longer contains
anything.
line = Jeroboam
while ~isempty(line)
[word, line] = strtok(line, )
<code block>
end

Initializing line is perhaps the easier method to understand. The other option is to actually run
strtok() or fgetl() before entering the respective while loop. Note also that overwriting line instead of
using a different variable name allows us to reduce our number of strtok() calls by one.

Loop-and-a-Half:
Depending on how you structure your while loops, it may be necessary to implement a loopand-a-half at some point. Certain problems can occur if the terminating condition has been met while
you still have data left to use. In, in our above example with strtok(), we decided to have the code block
ABOVE the strtok call, MATLAB would pull out the final word of the line (making rest empty) and fail to
reenter the loop. Adding a simple conditional statement fixes this problem.
line = Jeroboam
while ~isempty(line)
<code block>
[word, line] = strtok(line)
if isempty(line)
<code block>
end
end

Fclose()
The fclose() function closes documents youve opened. Having a bunch of extraneous
documents open will tend to slow down MATLAB and is considered sloppy (that means you lose points).
Fclose() does not have an output.
fclose(fh)
Fclose() is easy to use and even to forget about. Remember, fclose() is always called on the file
handle, not the filename; this is a very common mistake. Lets work an example problem to see how all
these elements combine.

Write a function Elisha that takes in the filename of a .txt document.

The function should output the number of words in the document
(anything separated by spaces counts as a word). The input file
contains a space after the last word of each line. Elisha should
create a new file with these extraneous spaces removed. The filename
of this new file should be the origninal filename followed by
_edited.
function num = Elisha(fn)

fh = fopen(fn, 'r');
fn2 = [strtok(fn, '.') '_edited.txt'];
fh2 = fopen(fn2, 'w');
num = 0;
%We now have both files open and two respective file handles. We have also
%initialized our output.
line = fgetl(fh);
while ischar(line)
while ~all(line==32)
%Since we're also eliminating spaces, we use a slightly different terminating
%condition.
[word, line] = strtok(line);
num = num + 1;
if all(line==32)
fprintf(fh2, word);
else
fprintf(fh2, [word 32]);
end
end
line = fgetl(fh);
if ischar(line)
fprintf(fh2, '\n');
end
end
end

Section 11: Structure Arrays

For most people, structures constitute the most difficult MATLAB concept. Structures and
structure arrays are merely another useful method for organizing information. Much like cell arrays,
structure arrays can be of any length and are capable of storing every data type, including other
structures.
A structure is basically a list of information stored in different headings, termed fields. Each field
is a string capable of storing one piece of information (double, logical, string, cell, structure, etc.) per
structure. Fieldnames are case-sensitive, just like everything else.

Creating Structures Manually:

There are two methods for creating structures. The first is to assign each datum manually using
the dot (.) operator. Structure assignments require three inputs: the name of the structure, a fieldname,
and something to assign to it, organized in the following fashion using the dot.
structure.field = something

If the structure does not already exist, MATLAB automatically creates one and stores it in your
current workspace. If the field does not already exist, MATLAB automatically adds it to the respective
structure below any fields already present. Otherwise, the structure is updated to include whatever new
information youre storing in it.
My shorthand for structures is sa, standing for structure array.
sa.name = 'Kevin';
sa.age = 19;
sa=
name:Kevin
age:19

Individual structures can be concatenated in rows and columns just like arrays of doubles
(homework and test problems never seem to move beyond structure vectors, fortunately). To create a
second structure under the same variable name, simply add its position number in parentheses before
the dot.
sa(x).field = something

Just like with vectors, you use the dot assignment to index out of bounds. MATLAB will
automatically fill any in-between space with unassigned fields and structures with empty vectors.

Similarly, adding an additional field to one structure will also add it to every other one, with an
empty vector for any unassigned fields.
sa(2).name = Micah;
sa(1) =
Name:Kevin
Age:19

sa(2) =
Name:Micah
Age:[]

You will frequently encounter situations where the fieldname you wish to create or access is not
given but stored as a string in one of your variables. In that case, you can enclose the variable referring
to the fieldname in parentheses; MATLAB interprets items in parentheses as variable names rather than
direct fieldnames.
Joseph = Age;
sa.Joseph = 19
sa =
Joseph: 19
sa2.(Joseph) = 19
sa2 =
Age: 19

Creating structures manually is easy to remember and implement, so there obviously must be a
more complicated method used to torment students on tests. That method is called the struct()
function.

Struct():
Entire structure arrays can be created in a single line of code via the struct() function. It takes in
any number of input fieldnames and cell arrays of data and creates the corresponding structure.
sa = struct(field1, {}, field2, {}, )

The number of structures in the resultant array is always equal to the length of the input cell
arrays. The first structure will contain all of the fieldnames (the odd-numbered inputs) and the first
element of each cell array (the even-numbered inputs). The second structure will contain all fieldnames
and the second elements of each cell, and so forth.
sa = struct(Name, {Kevin, Micah}, Age, {19, 20})
sa(1) =
sa(2) =
Name:Kevin
Name:Micah
Age:19
Age:20

If one of the even-numbered inputs is replaced with a double, a string, a structure, or a cell
containing only one value, MATLAB will place that element in every structure of the resultant array. The
number of structures in the array is still determined by the length of the other cell arrays in the function
call.

sa = struct(Name, {Kevin, Micah, Joash, Abel}, Age, 19)

sa(1) =
sa(2) =
sa(3) =
sa(4) =
Name:Kevin
Name:Micah
Name:Joash
Name:Abel
Age:19
Age:19
Age:19
Age:19
sa = struct(Name, {Kevin}, Age, {19, 20})
sa(1) =
sa(2) =
Name:Kevin
Age:19
Name:Kevin
Age:20

The struct() function is like the Forceyou can put faith in your blasters and try to survive
without it, but eventually some Jedi will show up and hack you to pieces.

Indexing Structure Arrays:

The code to pull individual elements out of structure arrays is simply the reverse of the dot (.)
assignment used to create them.
something = sa(x).field

This code is only useful for pulling out individual elements from a particular structure.
Fortunately, there are two other ways to index structure arrays. You can index an entire structure by
leaving out the filedname.
something = sa(x)

In this case, the output of the index is another structure array. Obviously, you could also index a
range of structure arrays using the colon operator. Perhaps the most useful method for indexing
structure arrays is to access every element in a particular field; once again, you can use parentheses for
fieldnames stored as variables.
something = sa.field
something = sa.(field)

Unfortunately, indexing an entire field produces multiple outputs that MATLAB does not
automatically store. It saves each value in the temporary variable ans, effectively overwriting all but
the final element. You will therefore need to enclose the variable youre creating in square or curly
brackets to create vectors and cell arrays representing a fieldname.
sa = struct(Name, {Matthew, Mark, Luke, John}, Age, {19, 20, 21,
22});
Names = sa.Name
Names = John
{Names} = sa.Name
Names = {Matthew, Mark, Luke,
John}
[Ages] = sa.Age
Ages = [19, 20, 21, 22]

And now well take a look at the bunch o functions associated with structure arrays.

Setfield()
Manipulating elements in structure arrays is much more complicated than in vectors or cell
arrays. Fieldnames serve as rather fixed entities, and there are no little tricks involving the colon
operator or empty brackets available.
Setfield() takes in a structure array, the name of a field in the structure array, and a new
fieldname (fieldnames are ALWAYS strings) and changes the name of the pre-existing field to the new
string. Youll probably use this function once and then forget it exists.
sa = setfield(sa, field, new_field)

Rmfield()
Alternatively, you may actually want to delete a field entirely from a structure array. Once again,
the only way to accomplish this is to use the built-in function. Rmfield() takes in a structure array and a
field and deletes the field from the structure array.
A common mistake is to forget to set rmfield() and setfield() equal to anything; if you dont,
MATLAB will store the updated structure array in the temporary ans variable, which is not what you
want. Always remember to overwrite the structure array by putting it on BOTH sides of the equals sign.
sa = rmfield(sa, field)

Getfield()
Getfield() does the same thing as indexing a structure array by fieldnameit takes in the
structure and the field and returns the corresponding element in the structure. Getfield() is useless.
something = getfield(sa(x), field)

Fieldnames()
Fieldnames() takes in a structure array and outputs a cell array containing the names of all fields
in the structures. A common homework problem ploy is to not preallocate the fieldnames in the input
structure array, in which case you would use fieldnames() along with indexing indirectly (with
parentheses) to acess the contents.
ca = fieldnames(sa)

Isfield()
Is a certain string one of the fieldnames in a structure? Isfield() returns a logical true or false.
logical = isfield(sa, field)

Istruct()
Works just like ischar(), is numeric(), and iscell(). Can you guess what it does?
logical = isstruct(something)

Thats pretty much it for structure arrays, believe it or not. Just memorize all the little rules and
functions because structure problems love to pop up on the final exam. I just have one theoretical
discussion to add that may be useful.
Whenever you have a collection of data such as a vector, array, cell array, or structure array and
you wish to pull out on element at a time (probably to perform some kind of logical or conditional
statement) you have to use a series of nested for loops. The number of for loops necessary will always
be equal to the number of dimensions in the data collection (one for vectors, two for array, etc.). The
important thing to remember about structure arrays is that the set of fieldnames adds an extra
dimension, so a one-dimensional structure array acts like a two-dimensional object.
Given a one-dimensional structure array with unknown fieldnames, how would you index each
element individually?
fields = fieldnames(sa);
for i = 1:length(sa)
for j = 1:length(ca)
x = sa(i).(fields{j});
end
end

Second, remember that you can use strcmp() to compare a string to a cell array of strings to find
matches. Fieldnames() outputs a cell array of strings.

Section 12: Recursion

A function can actually call itself in the same way it would call another function. When a
function contains a call to itself, it is called a recursive function (the act of a function calling itself is
called recursion). A recursive call causes MATLAB to open another command window called a stack
frame on top of the one your function is running. Each successive call opens yet another stack frame,
and the process continues until no more recursive calls are made. MATLAB then evaluates the each
individual stack frame starting with the MOST RECENT ONE.
Think of a stack of papers on a table. If you set your calculus, physics, and chemistry homework
on a desk and then pick them up one at a time, youll finish your chemistry homework first because it
ends up on top. This will all seem rather ethereal until we look at an example.
Unfortunately, having your function call itself imposes a few very specific limitations to avoid
infinite loops. Here are three pillars of recursion.
1. The function calls itself.
2. The function contains a terminating condition.
3. Successive calls move the function toward the terminating condition.
And here is the basic template youll be using with recursion.
if <terminating condition>
<code block>
elseif <maybe another terminating condition, ad infinitum>
<code block>
else
<recursive call>

So what on earth is a terminating condition? The most severe limitation of recursion is that, if
left unchecked, the function will call itself infinitely, opening an infinite number of stack frames and
eventually crashing your computer. MATLAB has a built-in safeguard against this possibility; it will give
you a warning and ask for your permision before opening too many stack frames at once. Never give it
permission to do so.
To prevent this tragedy, you simply change the inputs of each recursive call (make them
variable) in such a way that they approach a known value. Eventually, the function will reach this
terminating condition and immediately resolve all waiting stack frames. Lets look at the famous
fibonacci example to see how this works.
A fibonacci sequence is a series of numbers where each element is equal to the sum of the
previous two elements. The first two numbers are arbitrarily chosen as zero and one. 0,1,1,2,3,5,8,13,21

and so forth. We want to write a recursive function that takes in a number and returns the fibonacci
value at that position.
function out = fib (num)
if num==1
out = 0;
elseif num==2
out = 1;
%These are our two terminating conditions. Obviously, the sequence cannot
%contain any elements before position one.
else
out = fib(num-1) + fib(num-2);
%This line constitutes the recursive call. MATLAB evaluates each recursive
%call completely before continuing finishing the code. For higher numbers,
%each call will open many stack frames that quickly become compounded like
%branches of a tree. Each successive call reduces the value of the input
%until it is equal to one or two (the terminating condition).
end
end

If you attempt to use this function to evaluate a moderately high number like 50, you will find
that MATLAB takes all eternity to return the answerit would almost be faster to evaluate the
sequence by hand. Why? The enormous branches of recursive calls actually end up evaluating the same
numbers over and over again. Since MATLAB doesnt inherently save the value for fib(20), it has to
reevaluate it each time it is prompted to do so.
Thus, the other major limitation of recursion is that it becomes pointlessly redundant for all but
the simplest tasks. Why, then, would you use recursion for a problem such as this? You wouldnt.
The only reason you learn recursion in this class is so you can be tested on it. On rare occasions,
recursion is the only way to efficiently solve a problem, but for something like this you will always be
better of using iteration instead.
As a final note and warning, consider what would happen if anything but a positive integer were
inputted into the above function. It would run forever.
The fibonacci function involved approaching the terminating condition by reducing the value of
the input. Lets look at an example involving a different type of terminating condition.
Write a function recursiveSum that takes in a vector and returns the sum of its elements using recursion.
function out = recursiveSum(vec)
if length(vec)==1
out = vec;
else
out = vec(1) + recursiveSum(vec(2:end));
end
end

Once again, we only used recursion because the direction told us to. Any sane person would
have simply used the sum() function. This time, we approached the terminating condition by reducing
the length of our input but leaving the values unchanged.
Finally, some problems may ask you to keep track of the number of recursive used to obtain the
solution.
Write a function recursiveLength that takes in a vector and returns its length using recursion.
function out = recursiveLength(vec)
if length(vec)==0
out = 0;
elseif length(vec)==1
out = 1;
else
out = 1 + recursiveLength(vec(1:end-1));
end
end

The fact that using the length() function was necessary is a testament to the uselessness of
recursion.
Perhaps the only task for which recursion is absolutely necessary involves the variability of cells
and structures. The variable could potentially contain any number of cells and structures contained
within other cells and structures. Thus, we would use recursion to unwrap the contents of the
variable. Lets take cells as an example.
Write a function unwrapCell that takes in a one-dimensional cell array and locates every non-cell
element. The function should print the value of all non-cell elements to the command window,
separated by spaces. The cells will not contain structures but may contain other cells.
function unwrapCell(ca)
for i = 1:length(ca)
if iscell(ca(i))
unwrapCell(ca(i))
elseif ~isempty(ca(i))
fprintf(ca(i))
end
end
end

Finally, the TAs may for some reason decide to assign a drill problem that is unsolvable the way
it is written. Remember that the ONLY way to approach a terminating condition during recursive calls is
to vary the inputs of the function. If you need to change a variable that is not an input to the function in
order to approach the terminating condition, the problem is unsolvable. Simply write a helper function
for your function that uses more inputs than the original.

Section 13: Plotting

Compared to the concepts weve been dealing with recently, plotting is both fun and easy.
While MATLAB has the capability to produce almost any conceivable kind of graph or chart, CS 1371
seems to be concerned only with basic two-dimensional graphs and three-dimensional surfaces. The
course schedule sometimes presents these as two separate sections, but for clarity we will keep it as
one guide.
Basic plotting presents no difficult or theoretical concepts. All you have to do is memorize a new
slew of functions and then use the knowledge you already have to determine the appropriate inputs to
produce the desired graph.
Recall from the discussion on functions that some functions can be run without specifying any
output. Plotting problems frequently make use of this nuance by requiring you to produce a plot rather
than return an output. Therefore, you need to remember this function header template:
function name_of_function (input1, input2, input3)

Plot()
Strangely enough, MATLABs plotting function is actually called plot(). Plot() takes in a vector of
x-data, a vector of y-data, and an optional character string; it produces (not outputs) a two-dimensional
xy plot of the input data. Each point is automatically connected to the previous one with a straight line.
MATLABs default plot utilizes a solid blue line. However, you can change this by adding a
character string as a third input to the plot function. The first character specifies a color, the next
character specifies a plot symbol, and the remainder of the string specifies a line type (solid line, dashed
line, etc.). Here is a chart of relevant symbols for you to peruse:
b
g
r
c
m
y
k
w

blue
green
red
cyan
magenta
yellow
black
white

plot(x,y,'b*')
plot(x,y,'k-.')
plot(x,y,'wd--')

.
o
x
+
*
s
d

point
circle
x-mark
plus
star
square
diamond

:
-.
--

solid
dotted
dashdot
dashed

When creating closed shapes, remember to repeat the first set of coordinates; otherwise, MATLAB will
not know to draw a final line back to it. For example,
plot([-1, -1, 1, 1], [1, -1, -1, 1])

plots only three sides of the unit squarethe top is not connected. We would need to add a fifth
coordinate identical to the first.
plot([-1, -1, 1, 1, -1], [1, -1, -1, 1, 1])
2

1.5

1.5

0.5

0.5

-0.5

-0.5

-1

-1

-1.5

-1.5

-2
-2

-1.5

-1

-0.5

0.5

1.5

-2
-2

-1.5

-1

-0.5

0.5

1.5

Polar Coordinates:
Plotting with polar coordinates is a technique used to draw circle-like shapes. Drawing a circle by
plotting individual x-y coordinates would be difficult at best, so we turn x and y into functions of r
and theta (radius and angle) using the following conversions:
x = r.*cos(theta)
y = r.*sin(theta)

By simply adjusting the radius and angle from the origin, we can sweep out any circle-like area
on the xy plane. Heres a picture from Wikipedia.

So, to plot an ordinary circle of radius three, we would set r equal to three and allow theta to
range from 0 to 2. Linspace() is particularly useful here for generating a bunch of points, namely 100.

r = 3;
theta = linspace(0,2*pi);
plot(r.*cos(theta), r.*sin(theta))

-1

-2

-3
-3

-2

-1

By setting r equal to a range of values as well, we can make interesting two-dimensional spiral
shapes. Here well let theta repeat three times to ensure multiple revolutions.
r = linspace(0,5,300);
theta = linspace(0, 2*pi);
theta = [theta theta theta];
plot(r.*cos(theta), r.*sin(theta))
4
3
2
1
0
-1
-2
-3
-4
-5
-5

-4

-3

-2

-1

When plotting two-dimensional shapes, just be aware of the geometry involved and use polar
coordinates when necessary.

Figure Functions:
There are numerous functions MATLAB uses to adjust axes and label figures. Just know them,
and please remember that many of the inputs MUST BE STRINGS.
Plot labeling
title(string)
xlabel(string)
ylabel(string)
zlabel(string)
legend(string)

%puts a title on the plot

%labels the x-axis
%labels the y-axis
%yes, we will be using the z-axis as well
%you probably wont ever use this

Axis adjusting
axis
axis
axis
they
grid

square %makes the current axis box square in size

tight %makes the smallest possible axes to display the plot
equal %makes the axes of equal length. There are other axis options, but
dont matter.
on %turns on the grid on your current plot

CLF:
In the same way that clc clears the command window, clf clears the current figure.

Hold on
Yes, this is actually a function. Unfortunately, MATLAB automatically overwrites your current
plot whenever you try to plot something else. To prevent this, you must toggle a command called hold.
Hold on allows you to add more plots to the same figure, while hold off resets MATLABs default
behavior. This is EASY to forget on tests.

Subplot()
If you want to place multiple individual plots in the same figure, you can use subplot() to
organize them into a grid. The inputs are the number of rows in the resulting figure, the number of
columns in the resulting figure, and the individual plot you wish to modify.
Unlike when indexing arrays, MATLAB reads individual plots from left to right, like a book. As an
example, well make six plots in one 2x3 figure. Subplot() must called once for every new plot.
x = 0:5;
subplot(2,3,1)
plot(x,log(x))
subplot(2,3,2)
plot(x,x)
subplot(2,3,3)
plot(x,x.^2)
subplot(2,3,4)
plot(x,sin(x))
subplot(2,3,5)
plot(x, 1./x)
subplot(2,3,6)
plot(x,x./4)

2
1.5

25

20

15

10

1
0.5
0

0.5

0.8

0.6

-0.5

0.4

-1

0.2

Notice the order of the subplots:

subplot(rows, columns, current_plot)

1.5

0.5

Plot3()
Plot3() works exactly the same way as plot(), except that it takes in three vectors and produces a
three-dimensional plot. The only difference here is that we add a dimension z that is a function of x
and y. If you havent worked with this before, dont worryits actually pretty simple. Just note that
plot3() produces line plots, not surfaces (there are other functions for that). Lets draw some pointless
three-dimensional curve:
120
100

x = 0:10;
y = 5:15;
z = x.^2 + y;
plot3(x,y,z)
grid on
xlabel(x-axis)
ylabel(y-axis)
zlabel(z-axis)

z-axis

80
60
40
20
0
15

10

y-axis

10

x-axis

Three-dimensional plotting will also work with polar coordinates. Just specify a z value to
correspond to the other coordinates. You can plot three-dimensional spirals by using x and y values
for a circle and letting z range slowly upward. Well allow theta to repeat again to obtain a coolerlooking spiral.

z-axis

r = 5;
theta = linspace(0,2*pi);
theta = [theta theta theta];
x = r.*cos(theta);
y = r.*sin(theta);
z = linspace(0,10,300);
plot3(x,y,z)
grid on
xlabel(x-axis)
ylabel(y-axis)
zlabel(z-axis)

10
8
6
4
2
0
5
5
0

y-axis

0
-5

-5

x-axis

Surf() and Mesh()

One of the most powerful assets of MATLAB is its ability to produce surface plots from matrix
operations. A surface plot resembles a three-dimensional line plot in that it uses z as a function of x
and y. The difference is that x and y are not fixed to any particular range of values; for every
combination of an x value and a y value, there is a corresponding z coordinate.
Obviously, to use plot3() with vector inputs would be nauseatingly complicated if not
impossible. Luckily, MATLAB has a few built-in functions that produce surface plots using arrays as
inputs; the two important functions are surf() and mesh(), which are very similar.
Here we introduce a new notation for function inputs: when we use matrices rather than
vectors for plotting inputs, x becomes xx, y becomes yy, and z becomes zz. This notation can be
useful for distinguishing between vectors and arrays produced by meshgrid().
Surf() and mesh() take in three arrays (xx, yy, and zz) of EQUAL size and match each set of
corresponding elements to produce a point. For instance, the very first element in each array becomes a
single point with a unique xyz value, as does each other position in the arrays. MATLAB plots the
resulting surface and adds color based on the z value of each point. It also interpolates to produce a
reasonably smooth plot regardless of the inputs.
Surf() produces a pretty-looking plot with all interior points colored in, while mesh() uses only
the original points and produces a sort of wire-frame of what the surface should actually look like. There
are many other variations on surf() and mesh() but were not really concerned with them for the
purposes of this class.
surf

yy =
18

18

16

16

14

14

z-axis

zz = xx.^2 + yy;
subplot(1,2,1)
surf(xx,yy,zz)
title(surf)
subplot(1,2,2)
title(mesh)
mesh(xx,yy,zz)
xlabel(x-axis)
ylabel(y-axis)
zlabel(z-axis)

z-axis

xx =

mesh

12

12

10

10

6
8

6
8
3

7
2

6
y-axis

5 1

x-axis

7
2

6
y-axis

5 1

x-axis

Meshgrid()
Next to such infamous functions as strtok() and fprintf(), meshgrid() is one of the most confusing
functions in CS 1371. The obvious question in the previous example is how to generate arrays like xx
and yy from x and y vectors. This is certainly possible but would require some complicated loops and
a lot of unnecessary work. Meshgrid() takes in two vectors and outputs those arrays.
The vectors I used to create the arrays in the previous example were x = 1:3 and y = 5:8. We can
immediately see that the resulting arrays (both of the SAME size) were of dimension 4x3. The four
comes from the length of the second input, while the three is the length of the firstvery tricky.
Mesgrid() has two outputs. The first array consists of the first vector repeated along the rows
once for each element in the second vector. The second array consists of the second vector repeated
along the columns once for each element in the first vector. This is just something youll have to
memorize.
x = 1:3;
y = 5:8;
[xx,yy] = meshgrid(x,y)

xx =

yy =

Obviously, we dont need to use meshgrid() to create a zz array since z will always be a
function of x and y.

Shading:
We can also adjust the manner in which MATLAB shades surace plots to make them look
prettier. Shading, like hold on, does not require any parentheses.
shading flat %shades all pieces as one solid color determined by the enpoint
values
shading interp %uses linear interpolation to estimate the values of interior
points, producing a smooth shading
shading faceted %flat shading with black mesh lines to divide regions. This
is the default shading.

[xx, yy] = meshgrid(0:5,4:8);

zz = xx.^2 - yy.^2;
subplot(1,3,1)
surf(xx,yy,zz)
shading flat
title('shading flat')
subplot(1,3,2)
surf(xx,yy,zz)
shading interp
title('shading interp')
subplot(1,3,3)
surf(xx,yy,zz)
shading faceted
title('shading faceted')

shading flat

shading interp

shading faceted

20

20

20

-20

-20

-20

-40

-40

-40

-60

-60

-60

-80

-80

-80

5
6

5
6

4 0

5
6

4 0

4 0

Colormap:
Finally, we can adjust the colors MATLAB uses for shading surfaces. I will list the options below,
but there is no particular need to remember them.
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap
colormap

hsv
hot
gray
bone
copper
pink
white
flag
lines
colorcube
vga
jet
prism
cool
autumn
spring
winter
summer

Sorting Points:
One of CS 1371s favorite test questions involves plotting points that arent necessary in
ascending order. Remember that MATLAB only plots points in the order they are inputted.
Unfortunately, calling the sort() function on the x and y values separately will mix up the points to be
plotted because y is a function of x (not necessarily ascending). The secret is to use the second output
of the sort() function to sort each vector in the same manner.
[x, index] = sort(x);
y = y(index);

Section 14: Bodies of Rotation

People despise bodies of rotation because, unlike regular plotting, it is not intuitive unless you
REALLY understand what youre doing. We can rotate two-dimensional plots either around the origin by
a fixed angle theta or around an axis, creating an interesting-looking surface. The latter is actually the
same bodies of rotation example as is used in calculus I to study integration.

Rotation Around the Origin:

To rotate a two-dimensional plot around the origin by a fixed angle, simply matrix multiply the
coordinates by the two-dimensional rotation matrix (the rotation matrix MUST be on the left). The
coordinates themselves must be organized in a 2xn matrix, with a row of x data on top of a row of y
data.
[

][

Note that we should NOT use the dot (.) to multiply the matrices, which would only result in an
error anyway. After the multiplication step, simply plot the x and y date by indexing the rows of the
resulting matrix. I like to call the rotation matrix R. Note that theta must be in radians.

theta = pi;
x = -3:3;
y = x.^2;
subplot(1,2,1)
plot(x,y)
grid on
R = [cos(theta) -sin(theta);...
sin(theta) cos(theta)];
mat = R *[x;y];
subplot(1,2,2)
newx = mat(1,:);
newy = mat(2,:);
plot(newx, newy)
grid on

-1

-2

-3

-4

-5

-6

-7

-8

0
-4

-2

-9
-4

-2

This method only works for a counterclockwise rotation given in radians. To convert from
degrees to radians, multiply by /180. To rotate clockwise, multiply by -1.

Rotation Around an Axis:

This is much more complicated than rotation about the origin because it turns a twodimensional plot into a three-dimensional body, requiring us to specify another dimension.
Here I must introduce more bizarre CS 1371 notation. Rotation examples are usually done with z
vs. x data, which actually makes more sense for creating three-dimensional bodies. The zx plane is
generally depicted as directly in front of you and perpendicular to your line of sight (your computer
screen), with the y axis stretching ahead of you in space (behind your computer screen). Rotating
around the y axis would now be rotation about the origin, so we only need to worry about rotation
around the x and z axes.
25

20

z-axis

15

10

0.5

1.5

2.5
x-axis

3.5

4.5

From my experience, problems are usually given with u vs. v data rather than x vs. z data to
avoid confusion. Theyre the same thing. To create a three-dimensional rotation body, simply follow (or
memorize) this list of steps.
1.
2.
3.
4.
5.

Make theta using linspace().

Use meshgrid() to convert u and theta to uu and ttheta.
Use meshgrid() to convert v and theta to vv and ttheta.
Decide which variable (uu or vv) is always equal to the radius and set it to rr.
Set the variable (xx, yy, or zz) of the rotation axis equal to uu or vv.

6. Set the other two variables equal to rr.*cos(ttheta) and rr.*sin(ttheta).

7. Use surf() or mesh() to plot the resulting body.
That list probably made absolutely no sense, so lets work through an example. We want to plot
the rotation of the above xz data around the x-axis.
u = 0:5;
v = u.^2;
%These will be given in the problem statement.
theta = linspace(0,2*pi);
%This ensures a complete rotation.
[uu, ttheta] = meshgrid(u,theta);
%Step one.
[vv, ttheta] = meshgrid(v,theta);
%Step two. ttheta is exactly the same, so don't worry about overwriting it.
rr = vv;
%Visualize the final figure. If we take a circular slice parallel to the
%y-axis, the radius of that circle is always equal to the value of 'v' at
%that point.
xx = uu;
%The 'x' value doesn't change, since we're rotating around the x-axis.
yy = rr.*cos(ttheta);
zz = rr.*sin(ttheta);
%These are arbitrary. Just pick one.
surf(xx,yy,zz)
shading interp
xlabel('x-axis')
30
ylabel('y-axis')
zlabel('z-axis')
20

z-axis

10
0
-10
-20
-30
40
20

5
4

3
2

-20
y-axis

-40

1
0

x-axis

If we view the figure in the xz plane, we see that the original graph has been mirrored along
the x-axis, exactly as expected.

view(0,0)

25
20
15
10

z-axis

5
0
-5
-10
-15
-20
-25

0.5

1.5

2.5
x-axis

3.5

4.5

Now well rotate the same plot around the z-axis.

u = 0:5;
v = u.^2;
theta = linspace(0,2*pi);
[uu, ttheta] = meshgrid(u,theta);
[vv, ttheta] = meshgrid(v,theta);
rr = uu;
%The imaginary circular slice now
%radius(distance from the z-axis)
zz = vv;
%Once again, the axis of rotation
%along the z-axis(up and down) is
xx = rr.*cos(ttheta);
yy = rr.*sin(ttheta);
%Very repetitive.
surf(xx,yy,zz)
shading interp
xlabel('x-axis')
ylabel('y-axis')
zlabel('z-axis')

lies in the 'xy' plane. The

is equal to 'u'.
become fixed. This time, the distance
equal to the original 'v'.

25

z-axis

20
15
10
5
0
5
5
0

0
-5

y-axis

-5

x-axis

view(0,0)
25

20

z-axis

15

10

0
-5

-4

-3

-2

-1

0
x-axis

Just for practice, we can rotate some xy data.

u =[0, 5, 2, 0];
v = [0, 5, 5, 30];
plot(u,v)
xlabel(x-axis)
ylabel(y-axis)

30

25

y-axis

20

15

10

0.5

1.5

2.5
x-axis

3.5

4.5

Around x-axis:

z-axis

u =[0, 5, 2, 0];
v = [0, 5, 5, 30];
theta = linspace(0,2*pi);
[uu, ttheta] = meshgrid(u,theta);
[vv, ttheta] = meshgrid(v,theta);
rr = vv;
xx = uu;
zz = rr.*cos(ttheta);
yy = rr.*sin(ttheta);
mesh(xx,yy,zz)
shading interp
xlabel('x-axis')
ylabel('y-axis')
zlabel('z-axis')

30
20
10
0
-10
-20
-30
40
20

5
4

3
2

-20
y-axis

-40

1
0

x-axis

Around y-axis

z-axis

u =[0, 5, 2, 0];
v = [0, 5, 5, 30];
5
theta = linspace(0,2*pi);
[uu, ttheta] = meshgrid(u,theta);
[vv, ttheta] = meshgrid(v,theta);
rr = uu;
yy = vv;
xx = rr.*cos(ttheta);
0
zz = rr.*sin(ttheta);
mesh(xx,yy,zz)
shading interp
xlabel('x-axis')
ylabel('y-axis')
zlabel('z-axis')
-5
30

4
2
0
-2
20

10

-4
0

x-axis

y-axis

Consider these examples carefully. Rotation will seem incredibly complicated until you
understand the thought process.

Three-Dimensional Rotation Matrices:

Finally, you can rotate a plot by a fixed angle theta around the x, y, or z axes by multiplying
the appropriate three-dimensional rotation matrix. This method works just like two-dimensional
rotation about the origin. There is probably no need to remember these since they rarely appear on
homework or test problems.

Section 15: Images

MATLAB is capable of reading in and manipulating images much like Notepad or Excel files from
file IO, although I cant understand why anyone would want to use MATLAB to play with images.
Fortunately, there are only a few important functions and concepts to learn, so this is one of the easiest
topics in the course.
Image arrays are HUGE. Never forget to suppress your outputs.

Background on Images:
Computers store and interpret images as enormous arrays of elements called pixels. Each pixel
has three values associated with it: red, green, and blue. The computer interprets different
combinations of these three numbers as different colors; each individual pixel becomes one solid color,
and large collections of these colors become images.
The pixel values are based on mixing the three primary colors of light: red, green, and blue. All
colors can be made from different combinations of those three colors of light. A pixel with all zeros
wouldd represent the absence of light, the color black, while a pixel with the highest possible values for
all three would be pure white.
MATLAB stores images in MxNx3 three-dimensional arrays, with the third dimension
representing the three primary colors. The image itself is an array of red values on top of an array of
green values on top of an array of blue values.

Uint8():
Because the addition of light must be bounded by a finite limit (the color white), images are
stored in a special new data type called uint8, which stands for unsigned 8-bit integer. A thorough
discussion of this definition would be pointless here, so I will cover the important points. Uint8 numbers
function like regular doubles except for three differences.
1. All numbers lower than the minimum value of 0 are automatically set to 0.
2. All numbers above the maximum value of 255 are automatically set to 255.
3. All decimals are automatically rounded up or down to the nearest whole number.
Basically, uint8 is a data type that only allows for whole numbers ranging from 0 to 255. This
makes sense in terms of images because fractional pixel values would be meaningless, and black (0) and
white (255) must have finite values.
Luckily, the function that converts other data types to uint8 is very easy to remember. Its called
uint8().
arr = uint8(arr)

In most image problems, we will first convert the image array to doubles because they can be
more easily manipulated. Usually, the final step will be to convert all the data back to type uint8.
uint8(300)
uint8(-5)
uint8(3.5)
class(uint8(5))

ans
ans
ans
ans

=
=
=
=

255;
0;
4;
uint8

Imread():
MATLAB has a built-in function to read images called imread(). Imread() takes in a string
representing an image file in the current directory and outputs an image array of type uint8. Image
arrays are just like regular arrays with a third dimension. The number of rows and columns will vary
between images, but there will always be three layers: red, green, and blue, in that order. My shorthand
for an image array is im.
im = imread(filename)

In the same way that two-dimensional arrays are indexed by rows followed by columns, threedimensional arrays are indexed by rows, then columns, then layers. You can create three-dimensional
arrays either by manually creating each layer or by using the cat() function, explained below.
A = magic(3);
im(:,:,1) = A;
im(:,:,2) = A;
im(:,:,3) = A;

Alternatively, it is easy to divide images into layers using the same method.
im = imread(filename)
red = im(:,:,1);
green = im(:,:,2);
blue = im(:,:,3);

Cat():
There is also a magical MATLAB function that concatenates equivalently sized arrays along the
third dimension, much like using square brackets. The inputs should be the number 3 (for threedimensional concatenation) followed by all the arrays you wish to concatenate.
A = magic(3);
im = cat(3,A,A,A);

Imwrite():
Imwrite() takes in a three-dimensional uint8 array and saves the corresponding image to the
current directory. You must also specify the format in which you wish to save the image (jpg, bmp, etc.)
as a string.
imwrite(arr,filename,format)
imwrite(A,hello.jpg,jpg)
imwrite(A,greetings.bmp,bmp)

If the input array has only one layer instead of three, imwrite() will create a grayscaled image
with the same array as all three layers.

Image() and Imshow():

Image() and imshow() can be used to display image arrays in the current figure window. The
main difference is that image() causes MATLAB to do strange things to the image before displaying, such
as adding axis marks to indicate pixel numbers, while imshow() just displays the image. If the image is
too big to fit on the screen, imshow() will automatically scale it down to a more manageable size. When
in doubt, use imshow().
image(im)
imshow(im)

Salem:
Image problems involve manipulating image arrays using various algorithms to produce the
desired results. None of these concepts should be any different from what we could already perform
with arrays of doubles. For all of the following examples, I will be using an image of my cat to
demonstrate the results. His name is Salem, just like the talking cat from Sabrina, the Teenage Witch.

im = imread(salem.jpg);

Size():
We return now to a brief recapitulation of the size() function used to find the number of rows
and columns in arrays. When used regularly on a three-dimensional array, sizez() will rather stupidly
decide to make up for the extra dimension by tripling the number of columns, which will probably throw
you off if youre not watching for it. ALWAYS call size() with three outputs when dealing with images,
even though you already know there are three layers.
[row, col, layer] = size(im);

If it makes you feel better, just replace layer with a tilde so it wont be saved in your workspace
as a variable.
[row, col, ~] = size(im);

Resizing Images:
It is often desirable to make an image larger or smaller without significantly changing its
aggregate appearance. Unfortunately, adding or deleting random elements from the image array would
either add or delete information from the image itself.
The solution is to use indexing tricks to either repeat or delete evenly spaced rows and columns.
To make an image twice as large, we copy every row and column an extra time; to halve the size, we
delete every other row and column. The easiest way to accomplish this is by using the linspace()
function to create a range of the desired number of indices and then rounding the indices to the nearest
whole number. If we index the original image array at this new array of position values, MATLAB will
automatically repeat or delete informationthe use of linspace() ensures that this information is evenly
spaced and thus difficult to notice.
Here we simply set x and y equal to the desired effect on rows and columns, respectively. To
halve the rows and double the columns, use x=0.5 and y=2.
[row, col, layer] = size(im);
row_index = round(linspace(1, row, row.*x));
col_index = round(linspace(1, col, col.*y));
im = im(row_index, col_index, :);
imshow(im)

Once again, remember to index rows, columns, AND, all three layers or imshow() will produce a
grayscale image based only on a single layer.

Grayscaling Images:
If a certain pixel has the same intensity value for red, green, and blue, MATLAB interprets the
color as gray; smaller numbers generate darker shades of gray, as we would expect. To convert a
colorful image to grayscale (shades of gray varying according to the intensity of the individual pixels),
just take the average value of the three layers of the image array. That average value is then repeated
three times to create identical red, green, and blue layers.
Remember that taking averages does not work very well with data type uint8() because it
automatically rounds fractions and caps numbers at 255. Some course administrators will tell you that
you can avoid this issue by dividing through by 3 first when taking averages, but this method could also

potentially fail if MATLAB decides to round a number in an unexpected direction. Take my advice and
convert to double and back to uint8it will probably save you some headache in the long run.

im = double(im);
gray = (im(:,:,1) + im(:,:,2) + im(:,:,3))/3;
im = cat(3, gray, gray, gray);
im = uint8(im);
imshow(im)

Taking the Negative of Images:

Taking the negative is a fancy phrase for reversing the intensity of every pixelwhite becomes
black, and black becomes white. Just subtract the entire image array from 255.
negative = 255 im;

Swapping Layers:
Creative indexing also gives us the ability to change the order of the color layers in a threedimensional array. Remember waaay back from vectors that an range of indices is simply a vector of
position numbers. Thus, instead of using a colon (:) to access all layers of an image, we can use a vector
to swap them instead. Red=1, Green=2, and Blue=3. Lets swap red and green.
im = im(:, :, [2 1 3]);
imshow(im)

As you can see, the table now looks purple.

Dividing Images:
Dividing images into halves or quadrants is easy but tedious. This just requires indexing and
heavy use of the round(), floor(), or ceil() functions to avoid decimals. I always use round().
im(1:round(end/2),:,:);
im(round(end/2)+1:end,:,:);
im(:,1:round(end/2),:);
im(:,round(end/2)+1:end,:);
im(1:round(end/2),1:round(end/2),:);
im(1:round(end/2),round(end/2)+1:end,:);
im(round(end/2)+1:end,1:round(end/2),:);
im(round(end/2)+1:end,round(end/2)+1:end,:);

Top half
Bottom half
Left half
Right half
Top right corner
Top right corner
Bottom left corner
Bottom right corner

Transposing Images and Permute():

For some unknown reason, the designers of MATLAB decided that using the apostrophe () to
transpose three-dimensional arrays should cause an error. You can either divide the image into three
layers and transpose them individually or use the permute() function.
Permute() takes in an array and a vector indicating the order you wish to rearrange the
dimensions. The natural order for a three-dimensional array is [1 2 3], or rows, columns, layers. To swap
rows and columns, use [2 1 3] instead.
Once again, remember that transposing is fundamentally different from rotating. Notice that the
lizard is now in front of Salems RIGHT paw.
red = im(:,:,1);
green = im(:,:,2);
blue = im(:,:,3);
trans_im = cat(3,red',green',blue');
imshow(trans_im)
OR
trans_im = permute(im,[2 1 3]);

Rotating Images:
We can also use the same indexing tricks from the section on arrays to rotate images 90 degrees
clockwise or counterclockwise.
Remember that, in the first two cases, we tranpose the array and then reverse either the rows
or columns. To rotate 180 degrees, transposition is unnecessary.
im = permute(im,[2 1 3]);
im(end:-1:1,:,:)

Rotates counterclockwise

im = permute(im,[2 1 3]);
im(:,end:-1:1,:)

Rotates clockwise

im(end:-1:1,end:-1:1,:)

Rotates 180 degrees

To convert larger or smaller angles to their 90 degree equivalents, either add or subtract 360
inside a while loop or take the absolute value of the modulus from 360. The latter method returns 0, 90,
180, or 270. Thanks go to Carmie Cuda for this idea.
angle = abs(mod(angle,360));

For all other concepts such as concatenation, simultaneously changing pixel values (masking),
and basic manipulation ideas, refer to the section on arrays. Remember the golden rule of images: they
work just like regular arrays.

Section 16: Numerical Methods

Numerical methods is a fancy term for various problems that would normally be cumbersome
but that MATLAB can solve easily using brute force methods. Although the actually scope of numerical
methods encompasses very fascinating topics such as root finding, linear regression, and solutions to
partial differential equations, none of these are covered in CS 1371.
Instead, this section focuses on a few basic topics like solving linear systems, curve fitting,
numerical integration, and linear interpolation. Most problems can be solved easily using the
appropriate built-in MATLAB function.

Solving Linear Systems:

A system of many linear equations (no exponents greater than one) can be solved using basic
matrix division. Here I will not attempt to explain the finer points of linear algebra; the only important
equation is Ax = b, where A is a matrix of the coefficients for the linear equations, x is a column vector
of x-values (the solutions to the linear equations), and b is a column vector of the right-hand sides of all
the equations. Thus, the system
X1 + 2x2 = 3
4x1 7x=2 = -3
can be transformed into the equivalent linear algebra equation
[

][ ]

Al though this particular system would be easy to solve be hand, MATLAB can solve much more complex
systems using its amazing capability to invert matrices. There are three ways to solve any system of the
form Ax = b.
1.

Multiply the inverse of A times b using the inv() function. Remember that you must ALWAYS
multiply on the left-hand side.
2. Multiply the inverse of A times b using the MATLAB power operator.
3. Use matrix division to employ Gaussian elimination and back-substitution without finding the
inverse itself. This is by far the most efficient method. MATLAB uses the backslash \ to indicate
matrix division.
A
b
x
x
x

=
=
=
=
=

[1 2; 4 -7];
[3; -3];
inv(A)*b;
A^-1*b;
A\b;

Problems statements will often give the equations in some strange manner in an attempt to
confuse you. Make sure your two matrices are set up exactly as in the example above.

Curve Fitting:
MATLAB has one important function for fitting a polynomial curve to a collection of data points.
Polyfit() takes in a vector of x-values, a vector of corresponding y-values, and the degree of the
polynomial you wish to form. It outputs the coefficients of said polynomial.
The third input may require some discussion. The number you input will be equal to the highestorder exponent in the resulting polynomial (2 for quadratic, 3 for cubic, etc.). However, any polynomial
can have a constant coefficient as well as a coefficient for every power of the independent variable, so
inputting a three will actually give you a vector of length four. [A B C D] represents Ax3 + Bx2 + Cx + D.
Well use some contrived data as an example:
x = 0:5;
y = x.^3 - x.^2 + x - 1;
coeff = polyfit(x,y,3)

coeff = [1 -1 1 -1]

MATLAB can usually make reasonable fits for a variety of different highest order coefficient
choices. Mathematically, the highest order possible to produce a unique fit is one less than the length of
the dataa higher order fit would not have a sufficient number of points to go through. As an example,
consider a simple system of two points. You could feasibly plot an infinite number of parabolas through
those two points, but only one line.

Polyval()
Obviously, a vector of coefficients is not very useful if you want to evaluate the function at other
points. Fortunately, we have the polyval() function to fix that problem.
Polyval() takes in a vector of coefficients representing a polynomial equation and a vector of xvalues at which to evaluate the function. It outputs the associated y-values based on the input function.
Using polyfit() and polyval() can be used in conjunction to convert a collection of disconnected
points into a function and evaluate the function at more points to produce a smooth plot.
x = 1:5;
y = [1.3 2.6 3.2 4.7 5.9];
plot(x,y,'r*')
coeff = polyfit(x,y,4);
xi = linspace(0,5);
yi = polyval(coeff,xi);
hold on

plot(xi,yi)
title('Regression using polyfit() and polyval()')
Regression using polyfit() and polyval()
6

-2

-4

-6

0.5

1.5

2.5

3.5

4.5

Interp1():
Dont forget the numerical 1 at the end of this function name. Interp() is a completely different
function that is never used in CS 1371. Interpolation allows you to estimate the values at points in
between individual data points by connecting them with straight lines. Linear interpolation does NOT
produce smooth curves based on data trends; it merely connects each set of two points with a straight
line.
The inputs for interp1 are a vector of x-values, a vector of y-values, and a vector of new x-values
at which you wish to interpolate. Interp1() outputs the estimated y-values for those new x-values.
x = 1:5;
6
y = [1.3 2.6 3.2 4.7 5.9];
plot(x,y,'r*')
5.5
hold on
5
xi = linspace(0,5);
yi = interp1(x,y,xi);
4.5
plot(xi,yi)
title('Interpolation using interp1()') 4

Interpolation using interp1()

3.5
3
2.5
2
1.5
1

0.5

1.5

2.5

3.5

4.5

Interp1() can also be used for extrapolation. Extrapolation is the process of guessing the y-value
of a data point outside the initial data range. As such, extrapolation is often very untrustworthy because
there is no way of accurately estimating trends outside of a given data range. MATLAB appears to have
changed this particular function in its newest addition, so extrapolation now requires five inputs.
yi = interp1(x,y,xi,'linear','extrap');

Dont use extrapolation.

Spline():
Spline() is an odd function that is confusing for most people, but it is fortunately relatively
unimportant. Spline() is used to achieve nonlinear interpolation based on piecewise cubic functions.
Basically, it takes every set of three points and uses some sort of special MATLAB trick to connect them
with a cubic function. Spline interpolations always go through every point.
x = 1:5;
y = [1.3 2.6 3.2 4.7 5.9];
plot(x,y,'r*')
hold on
xi = linspace(0,5);
yi = spline(x,y,xi);
plot(xi,yi)
title('Interpolation using spline()')

Interpolation using spline()

6
5
4
3
2
1
0
-1
-2
-3

0.5

1.5

2.5

3.5

4.5

It looks a lot like the polyval() plot.

Differentiation and Integration of Polynomials:

Because the derivatives and integrals of polynomials are based only on a simple, easy formula,
we can use MATLAB to produce the coefficients very quickly. The secret is to create a vector of exponent
values, like [3 2 1 0] for a cubic function and use it to multiply or divide the original coefficients.
Remember the basic formulas:


So for differentiation we multiply each coefficient by its corresponding exponent and delete the
last one (remember that the derivative of a constant is zero). We integrate by dividing each coefficient
by its corresponding exponent plus one and adding an arbitrary constant (usually zero) to the end.
Consider the function f(x) = 5x3 + 3x2 2x + 1.
coeff = [5 3 -2 1];
der = coeff.*(length(coeff)-1:-1:0);
der(end) = [];
int = coeff ./ (length(coeff):-1:1);
int(end+1) = 0;

Diff():
The diff() function takes in a vector and outputs the differences between each set of two
elements in the vector. The length of the output vector will always be one less than the length of the
original vector. Diff() is not very useful but can be used with xy data to estimate the derivate dy/dx at
multiple points.
x = [1 2 5 8 13];
y = [-1 5 2 -10 5];
diff(x)
diff(y)./ diff(x)

ans = [1 3 3 5]
ans = [6 -1 -4 3]

Cumsum():
Another function with relatively little practical use, cumsum() takes in a vector and, for each
element, outputs the sum of that element and all previous elements. The output vector is of the same
length as the input vector.
x = [1 2 5 8 13];
cumsum(x)

ans = [1 3 8 16 29]

Trapz() and Cumtrapz():

Trapz() provides an estimate of the numerical integral of a set of points (the area beneath the
curve) by using the trapezoidal approximation from calculus. If you dont remember the details about
the trapezoidal approximation, dont worryits not very important. Trapz() takes in ONLY ONE vector,
the set of y-values, and seems to assume even spacing between them. The output is a single number,
the estimation of the definite integral.
Cumtrapz() works much like cumsum() in that it outputs the sum of the current trapezoid along
with all previous ones for each element in the vector. The output will also be of the same length as the
input vector. The first element of the output is always zero, and the last element is always equals the
total approximation given by the trapz() function.
y = [1 2 5 8 13];
trapz(y)
cumtrapz(y)

ans = 22
ans = [0 1.5 5 11.5 22]

Thats pretty much it for numerical methods; most of these functions have very little practical
value and will probably be ignored on homework and test problems. Most numerical methods problems
are relatively straightforward. Always remember how to use the linspace() function.

Useful Functions:

polyfit(x,y,order)outputs the coefficients of the best-fit polynomial with highest exponent

equal to order
polyval(coeff,xi)outputs the y-values of the function given by the coefficients in coeff at all
points in xi
interp1(x,y,xi)uses linear interpolation to approximate the y-values at xi
spline(x,y,xi)uses spline (strange cubic) interpolation to approximate the y-values at xi
diff(vec)calculates the difference between every pair of elements in vec
cumsum(vec)outputs the sum of each element plus all previous ones
trapz(y)estimates the numerical integral of the y-values using trapezoidal approximation
cumtrapz(y)outputs the sum of each trapezoid plus all previous ones, using trapezoidal
approximation

Section 17: Sound

Sound is the last homework topic and is usually somewhat difficult for people to grasp
conceptually, although the actual MATLAB coding is fairly straightforward. MATLAB can be used to read
in and manipulate sound, much like images.
Almost all coding with sound involves executing a specific series of commands to perform a task.
Many of these algorithms are difficult to understand, by which I mean that I dont understand them. You
should be fine if you just memorize them.

Wavread():
Wavread() reads in sound files from your current directory. The input is the filename, a string.
The first output is the sound itself stored in a COLUMN vector. This vector contains the amplitude values
(samples) for the sound in type doublehigher numbers represent louder sounds. The second output is
the sampling frequency, the rate at which the amplitude vector is read by MATLAB in samples per
second. If you prefer musical terminology, think of sampling frequency as the tempo of the sound.
Please notice my sound notation before it confuses you later, and never forget to suppress this function.
[data, fs] = wavread(fn);

Basic Algorithms:
I will give a brief explanation for each of these; if you forget the code, you can probably figure it
out logically.

Number of samplesevery element in the amplitude vector represents one sample.

num = length(data);
t, the time per sampleif the sampling frequency gives us samples per second, its inverse
must be seconds per sample.
dt = 1/fs;
The duration of the soundnow that we have dt, we can multiply seconds per sample times
the total number of samples to calculate the total duration in seconds.
dur = dt*num;
The maximum possible sampling frequency for the amplitude vectortheres some weird
concept called the Nyquist Theorem that I probably understood at one point. It says that the
maximum frequency is equal to half of the given sampling frequency.
f_nyquist = fs/2;

The sound played in reversethis is just basic indexing of vectors.

reverse = data(end:-1:1);
The fast fourier transform of the soundthis is another idea that I dont understand. Luckily,
the TAs usually post a guide for the fft() function is MATLAB, which you can read if you have the
time before finals. Remember to take the absolute value, multiply by two, divide by the length
of the data, and delete the second half of the result.
FFT = 2*abs(fft(data))/length(data);
FFT = FFT(1:round(end/2));

Changing the Pitch of a Sound:

There are two main methods to make a sound higher or lower. Remember that this is
fundamentally different from changing the volume, which we would accomplish be simply adding to or
subtracting from the amplitude vector.
First, we can change the sampling frequency. Doubling the frequency raises the pitch by one
octave, while halving the frequency decreases the pitch by one octave. Unfortunately, doing so will
affect the total duration of the sound, which is not always desirable.
fs = fs*2*n; %where n is the number of octaves by which the pitch is raised
or lowered

Second, we can use the same method we used with images to index the sound vector at certain
intervals to artificially affect the rate at which samples are chosen. This, too, will by definition increase
or decrease the duration of the sound.
data = data(round(linspace(1,num,num/2^(n/12)))); %where n is the number of
half-steps by which the pitch is raised or lowered.

Because these methods affect the total duration of the sound, problems will generally ask you
to truncate or zero-fill the amplitude vector to decrease or increase the final duration, respectively.
Thus, to maintain the same number of samples after transposition:
function data = transpose_sound(fn,n)
[data, fs] = wavread(fn);
num = length(data);
dur = num/fs;
data = data(round(linspace(1,num,num/2^(n/12))));
new_dur = length(data)/fs;

if new_dur > dur

data = data(1:num);
elseif new_dur < dur
data(num) = 0;
end
end

Basic Musical Jargon:

Pitchthe frequency of the sound. Higher pitches sound higher.

Notea sound produced by a single pitch. The seven natural notes are A,B,C,D,E,F, and G.
Octavea series of eight notes, starting and ending with the same one. Octaves on the piano
are labeled with numbers, with C4 being the C closest to the center of the piano. A C5 is one
octave higher than a C4, so an example of an octave would be C4 to C5.
Whole stepthe difference in pitch between two natural notes, such as D and E, is defined as a
whole step.
Half-stepsome notes fall in between whole steps. These are called accidentals, and are
denoted by sharps (#) and flats (b). The difference in pitch between a natural note and the
nearest sharp or flat is defined as a half-step. Thus, there are two half-steps in a whole step. A
C# is one half-step higher than a C, while a Cb is one half-step lower than a C. Musically, the halfstep between any two natural notes can be denoted by either the flat or the sharpthey are
the same note. Thus, Eb and D# are termed isotonic, and are exactly the same pitch. The only
exceptions to this rule are B#, Cb, E#, and Fb; a B# is isotonic with a C, while a Cb is isotonic with
a B. Basically, there is no half-step between C and D or between E and F. Heres another great
picture from Wikipedia to illustrate this concept.

Chorda group of notes played simultaneously that blend to produce a more interesting pitch.

Generating Chords:
Supposedly, you can generate chords in MATLAB by simply adding together the amplitude
vectors for the different notes. Unfortunately, the examples given in this course usually involve a series
of irritating beeping noises that the Tas term a chord. Real chords are actually pleasing to the ears.
Plotting FFT vs. Normalized Frequency:
Once again, I dont know what FFT is. For this problem, we use the FFT algorithm described
above to generate the ydata for the plot. The xdata will be a series a values ranging from 0 to the
nyquist frequency, also described above. We use linspace() to generate a vector of xdata equal in length
to the ydata.
Finally, problems will often ask you to identify the principal frequency of a sound, which is
simply the frequency with the highest FFT value.
function Plot(data)
y = 2*abs(fft(data))/length(data);
y = y(round((1:end/2)));
f_nyquist = fs/2;
x = linspace(0,f_nyquist,length(y));
plot(x,y)

hold on
[Max,index] = max(y);
f_principal = x(index);
plot(f_principal, Max, 'r*')
end

Wavwrite():
The opposite of wavread. Wavwrite() takes in an amplitude vector, a sampling frequency, and a
filename and creates the appropriate .wav file. Wavwrite() is the only writing function in which the
filename comes LAST, not first. Try to remember that for the final, as it is very tricky.
wavwrite(data,fs,fn)

There isnt really too much to know for sound. Keep in mind that the sound plotting problem
with the fft() function is a VERY popular coding problem on the final exam. I have seen it pop up for two
consecutive semesters now.

Section 18: Sorting

Sorting constitutes one or two days of lecture at the end of the semester, one pointless day of
recitation, and a single multiple choice question on the final exam. While sorting appears to take place
instantly in MATLAB with the sort() function, it is actually a complex iterative process that can be
accomplished in many different ways. Sorting in CS 1371 involves understanding and learning to
recognize four of these sorting methodsyou will NOT have to code any of these methods.
Insertion Sort:
The simplest and most obvious sorting method, insertion sort is also one of the least efficient.
Insertion sort pulls out one number at a time from the original vector and builds the sorted vector
element-by-element. The actual algorithm requires MATLAB to iterate through the vector being built in
order to determine the proper location of the current element.
1.
2.
3.
4.
5.

Create an empty output vector.

Remove an element from the unsorted vector.
Iteratively determine the elements location in the output vector.
Place the element in the output vector.
Repeat steps 2 through 4.

vec = [3 1 7 2]
vec = [1 7 2]
vec = [7 2]
vec = [2]
vec = []

out = []
out = [3]
out = [1 3]
out = [1 3 7]
out = [1 2 3 7]

function out = insertion_sort(vec)

if isempty(vec) || length(vec)==1
out = vec;
return
end
out(1) = vec(1);
vec(1) = [];
while ~isempty(vec)
num = vec(1);
vec(1) = [];
for i = 1:length(out)
if num <= out(i)
out = [out(1:i-1),num,out(i:end)];
break
end
end
if i==length(out)
out = [out num];
end
end
end

Bubble Sort:
Bubble sort is similar to insertion sort in that it requires step-by-step iteration through the entire
starting vector. The difference is that bubble sort produces the output by repeatedly swapping side-byside elements in the starting vector instead of creating a new one.
1.
2.
3.
4.

Take the first element in the unsorted vector.

If it is greater than the element on its right, swap it with the element on its right.
Repeat step 2 until the element is no longer greater than the element on its right.
Repeat steps 1 through 3 until the vector is sorted.
This one is tough to visualize, which is once again why we have Wikipedia:

First Pass:
(51428)
(15428)
(14528)
(14258)
swap them.
Second Pass:
(14258)
(14258)
(12458)
(12458)

( 1 5 4 2 8 ), Here, algorithm compares the first two elements, and swaps them.
( 1 4 5 2 8 ), Swap since 5 > 4
( 1 4 2 5 8 ), Swap since 5 > 2
( 1 4 2 5 8 ), Now, since these elements are already in order (8 > 5), algorithm does not

(14258)
( 1 2 4 5 8 ), Swap since 4 > 2
(12458)
(12458)

function vec = bubble_sort(vec)

if isempty(vec) || length(vec)==1
return
end
for i = 1:length(vec)-1
for j = 2:length(vec)
if vec(j-1) > vec(j)
vec = [vec(1:j-2),vec(j),vec(j-1),vec(j+1:end)];
end
end
end
end

Merge Sort:
Merge sort is a recursive method based on the principle that combining two smaller, presorted
vectors is faster than sorting the entire thing manually. Merge sort arbitrarily breaks the unsorted vector
into smaller pieces, sorts the pieces, and then merges them back together step by step.
1.
2.
3.
4.

Divide the vector into halves.

Continue dividing the halves into smaller halves until all pieces have length 1.
Simultaneously combine and sort each pair of elements.
Repeat step 3 with each resulting vector until the entire list is rebuilt.

The principle of merging vectors is that, if the two vectors are already sorted, you only have to
compare the first element in each vector, thus saving considerable time by eliminating excessive
iteration.

function out = merge_sort(vec)

if isempty(vec) || length(vec)==1
out = vec;
else
left = merge_sort(vec(1:round(end/2)));
right = merge_sort(vec(round(end/2)+1:end));
out = merge(left,right);
end
end
function out = merge(left,right)
out = [];
while ~isempty(left) || ~isempty(right)
if ~isempty(left) && ~isempty(right)
if left(1) <= right(1)
out = [out left(1)];
left(1) = [];
else
out = [out right(1)];
right(1) = [];
end
elseif isempty(left)
out = [out right];
right = [];
else
out = [out left];
left = [];
end
end
end

Quicksort:
Quicksort is another recursive algorithm based on the idea of combining presorted vectors. It is
somewhat more efficient than merge sort, however, in that it sorts the elements while breaking them
apart rather than afterward. The quicksort algorithm selects a random element from the unsorted
vector, called the pivot, and separates the remaining elements according to whether they are greater
than or less than the pivot. The result of this separation is that the pivot automatically ends up in the
correct position, so recursively creating more pivots eventually sorts the vector.

function out = quicksort(vec)

if isempty(vec) || length(vec)==1
out = vec;
return
end
less = [];
greater = [];
pivot = vec(1);
for i = length(vec):-1:2
if vec(i) < pivot
less = [less vec(i)];
else
greater = [greater vec(i)];
end
vec(i) = [];
end
out = [quicksort(less) pivot quicksort(greater)];
end

Big O Values:
Each sorting method is assigned a big O value based on its overall efficiency. While they are all
capable of sorting small lists of numbers virtually instantaneously, constantly iterating through
databanks of millions of numbers incurs significant drawbacks for less efficient sorting methods. The
first two methods, insertion sort and bubble sort, are considered
, meaning that the time
necessary to sort the vector increases quadratically with the length of the vector.
The other two algorithms, quicksort and merge sort, have values of
, meaning that
the increase in time is logarithmically related to the length of the vector. Thus, quicksort and merge sort
are significantly faster when sorting longer lists, for the reasons explained above.

Identifying the Sorting Method Used:

Most sorting problems involve analyzing a piece of code like the ones above and identifying
which of the four sorting methods is being used. Just note the differences and follow this easy flowchart.

A problem might try to trick you by using the merge sort algorithm but naming one of the
variables pivot, or other such tricks. Just follow the general flowchart and you should be fine.
I abstained from commenting any of the above sorting codes in hopes that they will appear
clearer as a whole. Most of the ideas involved are fairly straightforward, and if you need some fun
coding practice in iteration or recursion before the final, these algorithms are highly instructive.

Section 19: Conclusion

Thank you for reading my guide for CS 1371. It has been a great pleasure to work with so many
fascinating students while completing this project, and I have garnered numerous friendships in the
process. I also appreciate the many thanks I have received both personally and through email.
Regarding payment, please dont offer me anything in return for this guideconsider it my free
gift to you. I only ask that you consider how much potential you have to make a difference in the lives of
others; helping people is as simple as being perceptive enough to notice a problem and caring enough to
correct it.
There are most likely countless small coding mistakes and other such errors in this document. If
you notice any specific ones, please let me know. You may also inform me of any additions, revisions, or
clarifications that may be beneficial for future readers of this study guide. My email address is
kmcarr@ymail.com.
Feel free to distribute these guides to whichever students may find them useful. I have tried my
best to generate my own practice problems and avoid giving answers to real homework or test
problems, so this guide should be considered an honest source of material for the course. No one has
my permission to sell this document or to remove my name from it.
I am planning to edit and revise the study guide next semester, most likely by adding a separate
example section for each topic as well as three practice exams and a practice final. If you would like the
final revised copy, let me know at the end of next semester.

The Final Exam:

The CS 1371 final is comprehensive and is generally worth 35% of your overall grade. There will
be twenty to thirty multiple choice tracing problems and three or four coding problems, if I remember
correctly. You should find the final significantly easier than any of the individual tests because the skills
you accumulate throughout the course build on each other.
For studying I suggest reworking all three of your previous exams as well as any homework
problems that were relatively straightforward but gave you trouble conceptually. Topics to focus on are
file IO, structures, recursion, images, and sound.

Aggregate Useful Functions List

Data Type Conversion:

double(x)converts x to type double

char(x)converts x to type char
uint8(x)converts x to type uint8
class(x)returns the data type of x as a string

Data Type Determination:

isnumeric(x)determines if x is of type double

ischar(x)determines if x is of type char
iscell(x)determines if x is of type cell
isstruct(x)determines if x is a structure
islogical(x)determines if x is of type logical
isempty(x)determines if x is an empty vector, string, etc.

Vectors and Arrays:

min(x) returns the smallest element in the collection (Note: if the minimum value occurs
multiple times, min() and max() will return only the first instance)
[value, index] = min(x) returns the smallest element along with its position number
max(x) returns the largest element in the collection
[value, index] = max(x) returns the largest element along with its position number
sort(x) sorts the elements in ascending order
[newX, index] = sort(x) returns x in ascending order along with a vector containing the position
numbers of the original x to which each value corresponds
fliplr(x) flips x left-to-right
flipud(x)flips x top-to-bottom
x transposes x (row vector column vector)
mod(x,num) returns the remainder if x is divided by num
round(x) rounds x up or down
ceil(x) rounds x up to the next-highest integer if x is fractional
floor(x) rounds x down to the next-lowest integer if x is fractional
ones(x,y) creates an array of ones of dimension x,y (Use 1 for x or y to generate vectors)
zeros(x,y) creates an array of zeros of dimension x,y
true(x,y) creates an array of logical trues of dimension x,y
false(x,y) creates an array of logical falses of dimension x,y
sum(x) computes the sum of x
prod(x) computes the product of x
find(logical expression) generates a vector of all indices where the logical expression is true
linspace(a,b,num)creates an evenly spaced vector of length num ranging from a to b

Logical Expressions:

all(x) true if everything in x is true

any(x) true if anything in x is true
not(x) same as ~x
or(x,y) same as x | y
and(x,y) same as x & y

Strings:

num2str(x)converts a double to its numerical equivalent as a string

str2num(x)converts a string to its numerical equivalent as a double
strcmp(x,y)determines if two strings are exactly identical
strcmpi(x,y)determines if two string are identical, ignoring case
sprintf(str,var1,var2,)creates a string using variable strings or doubles
strtok(x,delimiter)breaks up string x based on the location of the delimiter
lower(x)converts x to lowercase letters
upper(x)converts x to uppercase letters

Cell Arrays:

num2cell(x)places each element in x in an individual cell in a cell array

cell2mat(x)converts cell array x to its equivalent double or string array

File IO:

csvread(fn)stores csv file x as an array

csvwrite(fn,x)saves x as a csv file
dlmread(fn,delimiter)stores dlm file x as an array
dlmwrite(fn,x,delimiter)saves x as a dlm file using the input delimiter
[num,txt,raw] = xlsread(fn)reads in an Excel spreadsheet
xlswrite(fn,x)stores x as an Excel spreadsheet
fopen(fn,permission)opens the file and returns the file handle
fgetl(fh)returns the next line in the file
fgets(fh)returns the next line in the file, including the next line character (\n)
fprintf(fh,str,var1,var2,)prints the formatted string to the file or the command window
fclose(fh)closes the designated file

Structure Arrays:

struct(field1,{},field2,{},)creates the input structure

setfield(sa,field,new_field)changes a fieldname in a structure
rmfield(sa,field)removes a field from a structure
getfield(sa,field)returns the value of the field in the structure
fieldnames(sa)returns a cell array of the fieldnames in the structure
isfield(sa,field)determines if the fieldname exists in the structure

Plotting:

plot(x,y,parameters)plots two-dimensional data

axisspecifies various type of axes (tight, equal, square, etc.)
title(str)creates the plot title
x/y/zlabel(string)labels the x,y,or z axis
clfclears the current figure
hold onprevents the current plot from being overwritten
subplot(row,col,plot)creates grids of different plots
plot3(x,y,z)plots three-dimensional data
surf/mesh(xx,yy,zz)creates surface plots
[xx,yy] = meshgrid(x,y)meshes two vectors into two arrays
shadingsets the shading for the current plot (flat, faceted, interp)
colormapsets the color scheme used for the current plot

Images:

imread(fn)stores an image as a uint8 array

cat(dimension,var1,var2,)concatenates the inputs along the specified dimension
imwrite(fn,im,type)stores the uint8 array as an image
image/imshow(im)displays the image
salemmy cat

Numerical Methods:

polyfit(x,y,order)outputs the coefficients of the best-fit polynomial with highest exponent

equal to order
polyval(coeff,xi)outputs the y-values of the function given by the coefficients in coeff at all
points in xi
interp1(x,y,xi)uses linear interpolation to approximate the y-values at xi
spline(x,y,xi)uses spline (strange cubic) interpolation to approximate the y-values at xi
diff(vec)calculates the difference between every pair of elements in vec
cumsum(vec)outputs the sum of each element plus all previous ones
trapz(y)estimates the numerical integral of the y-values using trapezoidal approximation
cumtrapz(y)outputs the sum of each trapezoid plus all previous ones, using trapezoidal
approximation

Sound:

[data,fs] = wavread(fn)returns the amplitude vector and sampling frequency of the sound
wavwrite(data,fs,fn)saves the data as a sound file with the specified sampling frequency

FIN