D E E PA K S A M U E L

LECTURE ON C++ AND

ROOT FOR PHYSICISTS

I N D I A - B A S E D N E U T R I N O O B S E R V AT O R Y

Copyright 2012 Deepak Samuel

published by india-based neutrino observatory
[email protected]
Licensed under the Apache License, Version 2.0 (the License); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/
LICENSE-2.0. Unless required by applicable law or agreed to in writing, software distributed under the
License is distributed on an as is basis, without warranties or conditions of any kind, either
express or implied. See the License for the specific language governing permissions and limitations under
the License.
First printing, May 2012

Contents

Languages, Operating Systems and Compilers

Complier flags for creating objects and libraries
Data types and operators
Control Structures
More about functions

39

File I/O

45

Classes

51

More Features in C++

11

17

25

31

Storing collection of data: Arrays and Structures

Pointers

65

ROOT: Introduction and Installation

77

35

Programming with ROOT: Starting with Histograms

Histograms and Graphs

87

The TFile class for ROOT objects storage

The TTree class: The spreadsheet for ROOT

99
103

81

Introduction
This guide was prepared as an aid for the graduate students of the
India-based Neutrino Observatory programme at the Tata Institute of
Fundamental Research, Mumbai. I will begin by giving an overview
of what happens inside a computer when a program is run and then
making you acquainted with the tools like required to write and run
a program and then the ways to use the tools. Therefore, I might
introduce some concepts in the first few classes which you might not
understand. These concepts will become clearer as we progress but
they are necessary to explain how the tools work.
I feel that, swimming and programming languages should not be
taught using the black board alone. The audience will understand everything that is taught in that way but when pushed into real waters,
they are very likely to drown. This comes from my personal experience of learning C++. Therefore we will jump right away into the
real waters, holding the side walls of the pool we shall learn first how
to float and then how to swim. This will make the classes interesting
right from the start and less time consuming as well. We will learn
from our mistakes by deliberately introducing bugs in our program.
This way you will be ready for handling such things when you start
programming for your research work. No program is perfect-Even
the operating systems that we use give us security updates everyday. Therefore do not worry if your program crashes but strive to
make it better with every update. I would appreciate if you come
along with your laptop (i.e., the swimming pool) to my classes. You
will enjoy the classes as you start understanding the power of the
C++ language.

Languages, Operating Systems and Compilers

High Level Languages and Low Level Languages
Though most of this part is not completely right, I hope that it will
give a inside view of what happens when a program runs. Read this
only to aid your brains to get a picture. You will find a lot of terms
within quotes to suggest that they are not to be taken literally. You
need to have these concepts in your brains before we jump into C++.
At the core of the computer is its brain called the Processor. It is
made up of millions of logic gates in addition to other electronics.
Look at how the addition of two numbers is made using a gate.The
top one is an XOR gate and the bottom one is an NAND gate. This
circuit gives a SUM and a CARRY output as shown in the truth
table.
A
0
0
1
1

B
0
1
0
1

S
0
1
1
0

C
0
0
0
1

Remember that the Adder circuit above is a simple picture of

logical operation using gates. In a processor, the inputs of the NAND
gate are not tied directly to the inputs A and B as shown. Rather,
there are Digital Switches in between those connections. Why? If
they are connected directly, then this combination can be used ONLY
as an Adder. On the other hand if there are Switches, then one can
connect/disconnect to form a logic circuitry of any kind. In the figure
shown, if we have a switch in the inputs so that both the inputs of
the NAND gate are connected to B, then the NAND gate can also
operate as a logical inverter.
In a typical programming of a microprocessor this is what happens: Any processor comes with a set of instruction set (A set of binary numbers which will do switching and mathematical operations
plus other stuff in the processor), which is provided by the manufacturer. Have a look at some of the instructions in the instruction set of

Figure 1: A Half Adder Circuit.

8085 microprocessor:
Mnemonic
ADD
CALL
CC
CMP

Binary Code (in Hexadecimal)

87
CD
DC
BF

Operation
ADD
Call unconditional
Call on Carry
Compare with Memory

For example to add two numbers, you have give the hexadecimal
code 87 to the processor input which will suitably switch and connect the gates so that an adder is formed and then the address of
the locations of the two numbers to be added are given. So A and B
are connected to these memory locations. The output is also connected to a memory location but we need not get into details of that
for now. 1
Programming using the instruction set is usually called Assembly Level Programming. This is also called Low level language/programming. Programs written in low level languages are
faster in some sense. However, there are inherent disadvantages in it.
One writes a mnemonic code and then replaces them by the binary
codes and then checks if the codes are right and then checking if the
program runs properly. This is not an efficient way to develop a code.
Programmers nowadays use high level languages which remove (in
fact, hide) all the complexities involved in putting together the pieces
in the assembly language. High level languages give an advantage
to a developer to write a code by using natural and understandable
expressions.2
Examples of High Level Languages: C, C++, Java etc., High level
languages classified into two types3 :
Compiled Languages: Here there is an intermediate converter
called a compiler which converts the statements in your program
into binary instructions that the processor understands, apart from
checking your code for possible syntax errors. Examples are C,
C++ etc.,
Advantage: Speed of execution
Disadvantage: Portability-You need to recompile your code on
each machine it runs.
Interpreted Languages: Each statement in the code and then the
corresponding instruction is converted (using a java virtual machine (JVM) for Java) and executed one after the other. Examples
are Java, Python etc.,
Advantage: Portability-The same Java code will run in any machine and in any platform where JVM is installed.

In the instruction set shown you see

words like ADD, CALL etc., They are
called mnemonics and are there only
to aid programmers. What will be
fed into the processor is not ADD,
CALL but the binary number 87,
CD etc., here given in hexadecimal
for simplicity. Before a programmer
writes a program, he writes it using
the mnemonics on a piece of paper and
then converts them to the binary code
and feeds them.
1

Two types of languages exist in high

level programming: compiled languages and interpreted languages.
2

Compiled languages (C,C++ etc.,)

are faster but are not easily portable
across platforms. The contrary is true
for interpreted languages (Java etc.,).
3

Disadvantage: Speed-at least 10-50 time slower than a compiled

code.

Operating Systems
The operating system is the manager of your computer. It is also
another program written in C (well, at least Linux). It manages and
controls the use of various resources in your computer. Have a look
at the Linux kernel in the figure given below. The bottom block is
your hardware and most of the upper blocks are software pieces
which build the operating system.
Figure 2: The pieces of codes that make
a typical operating system. The most
famous OSs are Windows, Linux and
Apple. Linux is available for free (you
can download the source code of Linux
and see what is inside) and Ubuntu is a
flavor of Linux.

So what do you need to write a C++ program?

An Operating System: Although you can use any OS you wish
but I prefer to teach the examples in Ubuntu/Scientific Linux as I
myself am comfortable with that.
An Editor: To write your code (gedit, emacs etc.,); You should be
comfortable using it. I was never comfortable with these editors
and feel that you should start using professional tools like eclipse
or Qt Creator soon but only for the first few classes I recommend
you to use gedit or emacs.
A Compiler: As earlier mentioned, it is an intermediate program

10

called which converts the statements in your program into binary

instructions that the processor understands, apart from checking
your code for possible syntax errors and linking to other libraries.4

First Program: Hello World

So let us start writing our program. The steps that we will are:
Step 1: Writing the code using an editor
Open the text editor and type the following lines in it:5
1

#include <iostream>

using namespace std;

int main ()

cout << "Hello World!";

return 0;

For windows there is Visual Studio

and for Linux gcc is widely used.
You can check if gcc is installed in
your computer by typing gcc -v in the
terminal. If that is not installed, then do
it using your package manager.
4

Line 1 is the preprocessor directive

which will include the contents of the
file iostream to the current file. Line 3,
the main function, is the entry point for
all C/C++ exectuables. Line 5 will print
out the statement Hello World using
the function cout defined in iostream
(included in line 1). Line 6 makes the
program return from the main function.
5

}
main.cpp: Hello World

Step 2: Compile the code. This is a process which checks syntax

errors, links with libraries and creates executable/objects etc., 6
Typical command for compiling a simple program is: "g++ filename", where, filename is the name of the file in which the code is
saved. For example, g++ Hello.cpp. on successful compilation,
this command creates an output file named a.out
Step 3: Execute the output file:7
Run the code using: ./a.out
Do it yourself:
Check what happens if the preprocessor directive in the first line is
removed and the program is compiled.
Check what happens if you remove the line return 0 and compile.
How do you comment out a line?
How do you comment out a block of statements?

In a Linux machine, the compile

commands are typed in a so called
terminal.
6

a.out is called an executable. The

name of the executable can be changed
by slightly modifying the compile command to g++ Hello.cpp -o output,
where output is the name of the output
file.
7

Complier flags for creating objects and libraries

Functions
In this chapter, we will see how programs are organized and the role
of compilers in such a process. You will have to probably memorize
some of the commands that you will learn now. I will use the concept
of functions to demonstrate some of the compiling options which will
be useful later on. You will learn how to create objects and link to
them and how to create libraries using these objects.
Operations that are repeated often are usually written as functions.
This will not only reduce the size of the program but the program
will look more organized. Let us see how to add two numbers, first
without using functions and next using functions.

#include <iostream>

using namespace std;

int main ()

int a=99;

int b=100;

cout<<a+b<<endl;// returns the sum of a and b

return 0;

8
9

Simple addition. Compile using g++

filenameand run using ./a.out

}
main1.cpp: Adding two numbers

#include <iostream>

using namespace std;

int Add(int x, int y) // The Add function definition

return (x+y);

int main ()

int a=99;

Adding using a function. Compile

using g++ filename. Run it using
/a.out

12

10

int b=100;

11

cout<<Add(a,b)<<endl;// returns the sum of a and b

12

return 0;

13

}
main2.cpp: Adding using a function

The use of functions will not be obvious in simple cases like these but
you will see the use of functions in almost all the C/C++ programs.
Though both the programs yield the same answer, functions give you
more flexibility as they stand out as a separate entity.

Function prototyping and forward declarations

Now try compiling the program using functions after the modification as shown below: Move the function definition after the end of
the main function. You will find the compiler returns an error. This is
because, the compiler converts the program by reading it line by line.
At line 7, the compiler does not know what Add means as it has
not seen it before.
1

#include <iostream>

using namespace std;

int main ()

int a=99;

int b=100;

cout<<Add(a,b)<<endl;// returns the sum of a and b

return 0;

10

int Add(int x, int y) // The Add function definition

11

12

return (x+y);

13

To make sure that the compiler knows what Add means, the function should be ideally defined before the main function but this
method has its own disadvantage. Imagine two functions that call
each other-you will obviously fail in trying to decide which function to be placed first. The method of function prototyping is used in
such cases to declare a function so that the compiler knows what the
function looks like but does not know what the function does. For
example, the prototype of the Add function is done in a single line
like:

Moving the Add function definition

below the main function returns a
compiler error.

main3.cpp: Function prototyping and

forward declaration

13

int Add(int a, int b); \\ a function prototype declaration

Declaring a function before the main function in order that the compiler is aware of a function is called Forward Declaration. Using
forward declarations, you now have the freedom of writing your
function definitions anywhere you want. An example is shown below.
1

#include <iostream>

using namespace std;

int Add(int a, int b);\\forward declaration

int main ()

int a=99;

int b=100;

cout<<Add(a,b)<<endl;\\ returns the sum of a and b

return 0;

10

11

int Add(int x, int y) \\ The Add function definition

12

13

return (x+y);

14

Example of forward declaration. Compile using g++ filenameand run using

./a.out

In large software developments, the usual process is to move the

function declarations in a header file and the definitions in a source
file. Let us try it out with the Add function, you will make 3 files:
Add.h, Add.cpp and main.cpp
1

\\Add.h

int Add(int a, int b); \\function declaration

Add.h: Addition function header file

//Add.cpp

#include Add.h //include contents of Add.h

int Add(int a, int b) // function definition

return (a+b);

}
Add.cpp: Addition function source file

//main.cpp

#include <iostream>

The Add function definition is in

Add.cpp and is declared in Add.h. In
this case the forward declaration is
done by including the file Add.h in the
main program as shown in line 3.

14

#include Add.h //include contents of Add.h

using namespace std;

int main ()

int a=99;

int b=100;

cout<<Add(a,b)<<endl;// returns the sum of a and b

10

return 0;

11

}
main3.3.cpp: Compiling multiple files

The compile command is slightly different if your code includes files

other than your main source file. Compile this code using:
g++ main.cpp Add.cpp
Run the program using ./a.out.

Creating objects and linking

Imagine a function (like the Add function) that is used by most
of your programs. You obviously do not want to repeat writing the
Add function either in the main program or in a different source
file as described earlier. Doing so is a wastage of time, especially in
big programs and it also increases the size of the code.8 There is an
alternative to overcome such things. You can compile the functions
that are needed often and create so called objects. Objects are binary
versions of the functions which can be called by any program by a
process called linking. Creating objects is done by using a special
flag while compiling. For example to create an object for the Add
function saved in Add.cpp, the compile command is:

Remember compiling a huge program

also is a time consuming process. For
example compiling ROOT, a plotting
and data analysis program from CERN,
takes about 2 hours on a laptop.
8

g++ -fPIC -c Add.cpp

Now, if the compilation is successful, you will find a file name Add.o
in the directory. This is called an object file.9 Now to call the function
Add in your program, instead of doing:
g++ main.cpp Add.cpp
as we did earlier, we do:
g++ main.cpp Add.o
The difference between these two commands is that the first one
compiles main.cpp and then Add.cpp and then makes an executable.
The second one compiles main.cpp only and the object file Add.o is
inserted at the right place which reduces the compilation time.

You will soon encounter another

concept called objects in C++. C++
is called an object oriented language
where the term object refers to another
concept altogether but here we refer to
an object file which has nothing to do
with it. Do not get confused between
these two.
9

15

Creating libraries and linking

A collection of objects is called a library. For example, a math library
will contain functions for addition, subtraction, multiplication etc.,
Let us create a simple math library for addition and subtraction.
Since we already made an object for addition, we will now make the
object for subtraction and then create a simple math library. Create
the files called Sub.cpp and Sub.h and enter the following in those
files.
1

\\Sub.h

int Sub(int a, int b); \\ function declaration

Sub.h: Subtraction function header file

\\Sub.cpp

#include Sub.h \\ include contents of Sub.h

int Sub(int a, int b) \\ function definition

return (a-b);

Now we shall create an object (Sub.o) for this file using the command
described earlier:

Sub.cpp: Subtraction function source

file

g++ -fPIC -c Sub.cpp

Now we will create a math library called myMath which will contain the basic addition and subtraction functions in it. The command
to do it is:
g++ -shared -o libmyMath.so Add.o Sub.o
Of course, this command will be successful only if Add.o and Sub.o
already exist.

Linking to a library in your program

Now we have a library of functions, compiled and ready to be called
by any programs. You do not have to write separate files and compile
them and then link them. All you have to do is to include the requisite header files to forward declare the functions and then link to the
libraries in the compiler.
Using this library, let us see how to write a program which returns
the sum and the difference of these two numbers. First we create the
main file:

The prefix lib and the extension .so

is a standard convention that has to be
followed. The flag -shared is used to
create shared libraries.

16

\\main.cpp

#include <iostream>

#include Add.h \\include contents of Add.h

#include Sub.h \\include contents of Sub.h

using namespace std;

int main ()

int a=99;

int b=100;

10

cout<<Add(a,b)<<endl;\\sum of a and b

11

cout<<Sub(a,b)<<endl;\\diff of a and b

12

return 0;

13

}
main5.cpp: usage of libraries

The program above is compiled using the command:

g++ main.cpp -L. -lmyMath
The flag -l before myMath tells the compiler to link to the library
myMath. Notice that the prefix lib and the extension .so have
been dropped. The flag -L. tells that the library is existing in the
current directory. If the library is in another directory replace the .
by the directory name. Now run the code using ./a.out to see the
output.10
So far, you have seen ways of compiling a simple program. Now it
is time to jump into real programming. We shall use these concepts
while writing the programs that follow. You should now be knowing
how to compile a program and how to link to a library. The commands will not be explicitly mentioned as often it was in this section.
Do it yourself:
What are environment variables?
List the common environment variables in Linux.
How do you set an environment variable in Linux?

In some systems, the execution might

fail with a statement like error while
loading shared libraries in which case
you should add the directory where
the library exists, to the environment
variable LD_LIBRARY_PATH
using the command export
LD_LIBRARY_PATH=$LD_LIBRARY_PATH:dirname,
where dirname is the name of the
directory.
10

Data types and operators

In any programming language you will encounter the term called
variable. For a computer, a variable is nothing but a memory slot. In
the previous programs, you might have seen the following statement:
int a;
What internally happens is a memory location is allocated for storing
an integer.11

The main memory of the computer is

called as the RAM.
11

int a=99;
The above statement not only allocated memory but also set the bits
in that allocated memory space to the corresponding value. So when
the term memory comes, bring the picture shown here to your mind.
Though the value of the variable can be changed, the address cannot
be altered. You can only read the address of a variable using the &
operator. The following code prints the address of the variable a and
the date stored in that address i.e., the value of a.
1

#include <iostream>

using namespace std;

int main ()

int a=99;

int b=100;

cout<<&a<<endl;\\ print the address of a

cout<<a<<endl; \\print the value of a

return 0;

10

Figure 3: Representation of the physical

memory in the computer. For every
variable there is an address field and a
data field.

}
main6.cpp: Address of a variable

Fundamental data types

The fundamental data types in C/C+ are listed in Table 1. In addition to the fundamental types there is an option to use data types

18

Name
char
short
int
long
bool
float
double
long double

Description
character (1 byte)
short integer (2 bytes)
integer (4 bytes)
long integer (4 bytes)
boolean value-true/false (1 byte)
floating point (4 bytes)
double precision float (8 bytes)
long double precision float (8 bytes)

like int as an unsigned type where the numbers are treated as having a positive value. Therefore if the range of an int type is from
-2147483648 to 2147483647, the range for the unsigned int will be
from 0 to 4294967295. The statements for declaring a unsigned types
look like:
unsigned int a;
unsigned long b;
unsigned float c;
Do it yourself:
Try subtracting two unsigned integers a and b. First try a-b and
then b-a. What is the problem here?

Working with hexadecimal notation

You will have to work with the hexadecimal notation at some point
of time in programming, especially if your program is interfaced
with some hardware. A hexadecimal number is denoted with a prefix
0x. Thus the statement
unsigned int a=0xFF;
denotes the integer value of 255.

Escape characters
There are some special characters that are used in C/C++ which have
a special meaning especially in a cout statement. They are usually
not printed. They are listed in Table 2.
Lets write a simple program that takes an input from the user and
gives the mean as the output using the concepts that we learnt so far
and in a interactive manner. The cout command is used to print
out a statement while the cin command accepts a user input. These
commands are useful for an interactive program and are illustrated
in the example below.

Table 1: Fundamental data types in

C/C++.

19

Escape code
\n
\r
\t
\v
\b
\f
\a
\
\"
\?

Description
newline
carriage return
tab
vertical tab
backspace
form feed (page feed)
alert (beep)
single quote ()
double quote (")
question mark (?)

#include <iostream>

using namespace std;

float mean(float x, float y) \\ mean function

return (x+y)/2;

int main ()

float a;

10

float b;

11

cout<<Enter number 1<<endl;

12

cin>>a;

13

cout<<Enter number 2<<endl;

14

cin>>b

15

cout<<The mean is<<\t<<mean(a,b)<<endl;

16

return 0;

Table 2: Some of the escape codes in

C/C++.

main7.cpp: scope of a variable

Scope of a variable
In the above code, the variables a and b are seen only by the main
function and therefore they have a local scope. If instead for some
reason, you want the function mean to also see these variables,
you can declare outside main before the mean function. Then these
variables are said to have global scope.12
1

#include <iostream>

using namespace std;

float a;

float b;

float mean(float x, float y) \\ mean function

Local variables are can be used only

within the function in which they are
declared. Global variables can be used
by any function that can see it.
12

20

cout<<a;

return (x+y)/2;

10

int main ()

11

12

cout<<Enter number 1<<endl;

13

cin>>a;

14

cout<<Enter number 2<<endl;

15

cin>>b

16

cout<<The mean is<<\t<<mean(a,b)<<endl;

17

return 0;
main8.cpp: scope of a variable

Do it yourself:
Check what happens if the float function in the above example
takes int parameters instead of float.
What happens if in the above code, a and b are once again declared inside the main function along with those outside?

Assignment Operator
The most basic operator is the assignment operator =. This assigns
a particular value to the variable. For example, the expression:
int a=5;
assigns a value of 5 to the variable a.

Mathematical Operators
Mathematical operators are the most frequent ones that you will
use in a program, especially if you are a physicist. They are listed in
Table 3
Operator
+
?
/
%

Description
Addition
Subtraction
Multiplication
Division
Modulo

Compound assignment operators

Compound assignment operators operate on the variable and store
the output in the same variable. They are followed by an = sign
after a mathematical or a logical operator as shown below.

Table 3: Mathematical operators in

C/C++.

21

+=, -=, *=, /=, %=, >>=, <<=, &=, ^=, |=

For example, the expression:

a -= 5;
is similar to the expression:
a = a-5;
i.e., if a has a initial value of 0, then after this expression its value is
changed to -5.

Increment and decrement operators

To increase the value of a variable by 1 the increment operator ++
is used and to decrement by 1 the operator is used. For example,
after the following expressions:
1

int a=100;

a++;\\a=101

the value of a becomes 101 and after the following expressions:

1

int a=100;

a--;\\ a contains 99

the value of a becomes 99. The increment and decrement operators

can be used either as a suffix to a variable as shown in the previous
expressions or as a prefix to a variable. Consider the case:
1

int a;

int b=99;

a=b++;a contains 99, b contains 100

Here, after the last expression, the value of a is set to 99 and the
value of b is set to 100. That is, the value of b is incremented after the
assignment operation. If the increment operator is used as a prefix:
1

int a;

int b=99;

a=++b;\\ a contains 100, b contains 100

the value of a and b are set 100. That is, the value of b is incremented
before the assignment operation.

22

Relational operators
These operators compare the values of two expressions and return
a boolean value (true or false). The result of various relational operations for the variables a and b after the following expression is
tabulated in Table 4:

int a=100;

int b=99;

Relational Operator
==
!=
>
<
>=
<=

Description

Expression

Return

equal to
not equal to
greater than
lesser than
greater than or equal to
lesser than or equal to

(a==b)
(a!=b)
(a>b)
(a<b)
(a>=b)
(a<=b)

false
true
true
false
false
false

Table 4: Relational operators in C/C++.

Warning: The relational operator

== should not be confused with the
assignment operator =. This is a
frequent mistake made by beginners.

Logical operators
These operators perform logical operations on expressions that return
a boolean value. The result of logical operations for the variables a
and b after the previous expression is tabulated in Table 5:
Logical Operator
!
&&
||

Description

Expression

Return

logical inversion
logical AND
logical OR

!(a>b)
(a>b)&&(b<1000)
(a<b)||(b<50)

false
true
false

Conditional Operator
The conditional operator ? evaluates an expression and returns one
value if it is true and returns another value if it is false. The synatax
is:
condition ? result1 : result2
The use of this operator is illustrator is explained in the following
code.
1

#include <iostream>

using namespace std;

Table 5: Logical operators in C/C++.

23

int main ()

int a,b,c;

a=100;

b=99;

c = (a>b) ? a : b;

cout << c<<endl;

10

c = (a<b) ? a : b;

11

cout << c<<endl;

12

return 0;

13

}
main9.cpp: Conditional operator

The first cout statement will print out 100 since a is greater than b.
The second one will print out 99.

Bitwise operators
These operators modify the bit patterns of a variable. They are listed
in the following Table 6.
Bitwise Operator
&
|

Description

Expression

Return

AND
OR
XOR
NOT
Shift Right
shift Left

0x01&0x01
0x01|0x01
0x01|0x01
0xFFFFFFFF
0x011
0x011

0x01
0x01
0x00
0x00
0x00
0x02

Table 6: Bitwise operators in C/C++.

An alternative to cout is the printf

statement which can be useful for
printing hexadecimal numbers. To
print the value of the variable a in
hexadecimal format, use printf(%x,a);

Typecasting
There are some data types which can be converted into other types
using the typecasting operator. A typical example is given below:
float pi=3.14;
int i = (int)pi;
The value of i is now 3 and the numbers after the decimal point is
lost. This is a trivial example and as you can also check that the value
pi can be directly assigned to i without typecasting and it still
works the same way. This is because the compiler takes the freedom
of typecasting is some cases. Typecasting will become more obvious
when you start dealing with pointers.
Do it yourself:
What is the result of the expression int a= 100 + 2 * 5 ?

24

Is it 110 or 510?
Learn more about operator precedence.

Control Structures
Often in your program, you may have to taken decisions and route
your program to a particular segment or repeat a particular segment
of code. These are done by control structures which are discussed in
this chapter.

Checking if a condition is met: if and else

The if and else keywords are used for checking if a condition
is satisfied and then executing a block of statements which are put
inside the curly braces which follow. An example is given below:13
1

#include <iostream>

using namespace std;

int main ()

int a=99;

int b=100;

if (a<b)

{ \\ begin braces

Notice the begin and end braces

which enclose the group of statements
that have to be executed if the condition
is satisfied. These braces are not needed
if there is only one statement. Though
in the example there is only one statement, the braces are put for purpose of
illustration but it is a good practice to
put them during the learning phase. We
shall call the set of statements between
a begin brace and an end brace as a
block.
13

cout << "a is less than b";

9
10

} \\ end braces

11

else

12

{ \\ begin braces
cout << "a is greater than b";

13
14

} \\end braces

15

return 0;

16

}
main10.cpp: if else structure

The above code prints out a is less than b as the condition a<b
satisfied. But this code will fail if the values of a and b are the same.
First it checks if a<b and since it fails and therefore it goes into the
else structure to print out a is greater than b which is wrong. In
such cases, you can use a combination of if and else using the
else if keyword.

26

if (a<b)

{ \\ begin braces
cout << "a is less than b";

3
4

} \\ end braces

else if (a>b)

{ \\ begin braces
cout << "a is greater than b";

7
8

} \\end braces

else

10

{ \\ begin braces
cout << "a is equal to b";

11
12

} \\end braces
main11.cpp: else if structure

Repeating a set of instructions: Loops

Loops repeat a set of expressions until a condition is met. There are
different types loops each having a specific purpose in C++.

while loop
The while loop is repeated until the condition checked by the while
statement is satisfied as shown below:
1

#include <iostream>

using namespace std;

int main ()

int a=0;

while (a<100)

{ \\ begin

8
9

cout << a << ", ";

a++;

10

} \\end

11

cout << a exceeds 100;

12

return 0;
main12.cpp: while loop

In the above code the value of a is printed and then incremented

until condition checked by the while statement (a<100) is satisfied
(when a exceeds 100). There is a variant of the while loop called the
do while loop which has a slightly different functionality:

27

#include <iostream>

using namespace std;

int main ()

int a=0;

do

{ \\ begin

cout << a << ", ";

a++;

10

} \\end

11

while(a<100);

12

cout << a exceeds 100;

13

return 0;
main13.cpp: do while loop

Unlike the earlier version, the condition in the while statement is

checked at the end of the execution of the statements. There are no
rules to state which type to be used. You can choose the one which is
most appropriate for you code.

for loop
The for loop is also similar to the while loop but with extra functionalities. Unlike the simple while statement, there is an initialization
part, conditional part and a statement which is executed in the for
statement.
1

#include <iostream>

using namespace std;

int main ()

for (int a=0; a<101; a++)

{ \\begin
cout << a << ", ";

7
8

}\\end

cout << "a exceeds 100";

return 0;

10
11

}
main14.cpp: for loop

The code above does the same operation that was earlier done by
the while loop. The for statement has three parts separated by a
semicolon:
1. int a=0; This initializes a to zero

28

2. a<101; This checks if a is less than 101 and allows the block to be
executed only if it is true
3. a++; This increments the value of a before the end braces

Breaking and skipping in a loop structure

In a loop block you might want to break the loop if some condition
is met or you might want to skip executing a part of the block. These
are done using the break and continue keywords as shown in the
codes below.
1

#include <iostream>

using namespace std;

int main ()

{
for (int a=0; a<100; a++)

6
7

cout << a ;

if (a==50)

10

cout << "a is 50";

11

break; \\ come out of the loop

}

12

13

return 0;

14
15

}
main15.cpp: Breaking a loop

In the above code, the for loop is broken when a reaches 50 and the
loop is not more executed.
1

#include <iostream>

using namespace std;

int main ()

int a;

for (a=0; a<100; a++)

if (a==50) continue; \\skip the following lines

cout << a;

10
11

12

return 0;

29

main16.cpp: Skipping an iteration

In the above code, when a reaches 50, due to the continue keyword,
the following lines of code in the block (cout) are not executed but
the unlike the break statement the loop is not broken, the execution
continues with the next iteration.

switch-case structure
The switch case structure is similar to a multiple if else structure.
An example is given below:
1

#include <iostream>

using namespace std;

int main ()

int a =1;

switch (a) {

case 1:
cout << "a is 1";

break;

case 2:

10
11

cout << "a is 2";

12

break;

13

default:
cout << "value of a unknown";

14
15

16

return 0;

17

}
mian17.cpp: Switch Case

In the above code, the switch statement checks the value of a with
the possible cases. In case a is 1 then that segment is executed. It
might happen that a is neither 1 nor 2 in which case the default section is executed. Notice, the use of break after the end of each case
except the default case.

Exiting and returning from a code

The exit and the return functions are used to exit from a running
program or to return from a function. The return function additionally takes an argument depending on the return type of the function.
Examples are:
return -1;
return 1.5;

30

Do it yourself:
Implement the switch case code above using if else structures.
Get an integer input from the user and calculate the factorial using
loops. The code should output an error if the user enters a negative
integer.

More about functions

So far, you have been using functions to for performing simple operations like adding two numbers. The purpose of introducing functions
then was to illustrate the use of libraries and related flags in a compiler. We will now venture a bit into it and come back.

The void type

Consider the following function:
1

int add(int x, int y){

return (x+y);

This function takes integers and returns an integer. Sometimes

you want to write a function that does not return anything in such
cases you can use the void return type as shown below:
1

void Error(int ErrNum){

cout<<An Error !!!<<\t<<ErrNum<<endl;

In fact the void type can be used as an input to functions but that
is an advanced concept which you can learn when you understand
pointers.

Passing reference to a variable

The add function defined earlier can be called in the main function
using the following syntax:
1

int add(int x, int y){

return (x+y);

32

int main ()

int a = 101;

int b = 100;

int c = add(a,b);

cout<<The value of c is <<c<<endl;

return 0;

10
11

When we called the function add with the inputs a and b what
happens actually is that the value of a and b are copied into the
local variables x and y and the addition is done using them. The
result of the addition is then stored in c. Notice that you cannot
change the value of a or b inside the add function or for that
matter any variable declared inside the main function. Of course, one
can make these global variables so that the add function sees them
but, there are disadvantages in doing so. The way to do it in C++ is
by passing the reference instead of the value itself. Have a look the
example below:

int add(int x, int y, int& z){

z = x+y;

return (x+y);

int main ()

int a = 101;

int b = 100;

int c=0;

10

cout<<c before addition is <<c<<endl;\\c=0

11

add(a,b,c);

12

cout<<c after addition is <<c<<endl;\\c=201

13

return 0;

14

main18.cpp: Passing values and references

Here the add function is slightly modified-now it takes three parameters instead of two. Also notice the & sign before the third parameter.
Using this syntax (&), the value of a parameter which is not seen
by the function can be changed. Here we are storing the result of the
sum of the numbers in the third parameter c which is not actually
seen by the function. We call this method as referencing a variable in
a function. This method is also useful when you want a function to
return more than one output. For example, you want a function to

main19.cpp: Passing values and references

33

return both the sum and the difference, you could do this way:

void add_sub(int x, int y, int& sum, int &diff){

sum = x+y;

diff = x-y;

int main ()

int a = 101;

int b = 100;

int s=0;

11

int d=0;
add_sub(a,b,s,d);

12

cout<<The sum is <<s<<endl; \\s=201

13

cout<<The difference is <<d<<endl;\\d=1

14

return 0;

10

15

What happens when a variable is

passed by referencing (i.e., using &) is
that instead of the value of the variable,
the address of the variable is given as
an input. The function manipulates
the value at that address. Remember,
the address of any variable is given
by the OS and you have no way of
changing it (although you can change
the value at that address). That is why
though the call add_sub(101,100,s,d)
is a perfect alternate, the call
add_sub(101,100,1000,2000) is wrong
and the compiler will complain as the
last two are not proper addresses as
they should have been.

}
main20.cpp: Passing values and references

Default parameters in functions

By giving default values to some parameters in a function, those
parameters can be skipped while calling the function in which case
the default value is taken and the operation is performed.

int add(int x, int y=0){

return(x+y);

int main ()

int a = 101;

int b = 100;

cout<<add(a); \\ prints 101

return 0;

10

}
main21.cpp: Default Parameters

In the above example, the second parameter is given a default value

of 0, therefore if the function is called without the second parameter,
then the default value for that parameter is used. However, not all
parameters can have default values; either all the parameters or the
last parameter(s) only can have default values. The legal statements
are illustrated in the example below:

34

function
operation(int x=0, int y=0, int z=0, int a=0, int b=0)
operation(int x, int y=0, int z=0, int a=0, int b=0)
operation(int x, int y, int z, int a=0, int b=0)
operation(int x, int y, int z, int a, int b=0)
operation(int x=0, int y, int z, int a, int b)
operation(int x=0, int y, int z, int a, int b=0)
operation(int x=0, int y=0, int z, int a, int b)
operation(int x=0, int y=0, int z, int a, int b=0)

legal
yes
yes
yes
yes
no
no
no
no

Function overloading
Sometimes you need to write functions which have almost similar
functionality but take different input types and return different return types. You can do this using the concept of function overloading
as illustrated below:
1

int add(int x, int y){

return(x+y);

float add(float x, float y){

return(x+y);

int main ()

int a = 101;

10

int b = 100;

11

float c = 101.2;

12

float d = 102.5;

13

cout<<add(a,b); \\ prints 201

14

cout<<add(c,d); \\ prints 203.7

15

return 0;

16

}
main22.cpp: Function overloading

In the above example, there are two functions defined with the same
name add. These two have almost similar functionality but one
operates on integers and the other operates on float. This is an example of function overloading. You could have as well given a different
name for the function that adds floats but the benefit of using function overloading is that the user need not worry which function to
use for adding integers, the function overloading takes care of it by
checking the input parameter types and uses the appropriate function.

Storing collection of data: Arrays and Structures

Arrays
In the previous sections we stored single values in variables using
instructions like:
int a;
We now move a step ahead to see how to store a collection of values.
For example, if you want to store the age of 5 people, you would do:
1

int age1=50;

int age2=45;

int age3=13;

int age4=15;

int age5=76;

There is a much more efficient way of doing it in C/C++. For such

cases where you want to store a collection of values belonging to a
data type, you can use arrays. The above example can be done in an
efficient way using array as illustrated:
1

int age[5]; \\ array declaration

age[0]=50;\\ array initialization 0

age[1]=45;\\ array initialization 1

age[2]=13;\\ array initialization 2

age[3]=15;\\ array initialization 3

age[4]=76;\\ array initialization 4

main23.cpp: 1d Array

The first line declares an one dimensional integer array of size 5.

14 The following lines initialize the values at the respective index.
Another way of initializing the same array is as follows:
1

int age[5]={50,45,13,15,76}; // array declaration

Notice the index of the array starts

from 0 and ends at 4. This has to be
remembered by beginners who are
usually tempted into starting from 1
and ending at 5.
14

36

To access the value at a particular index for example 2 is as simple as:

int

age2 = age[2];

Do it yourself:
Try to see what happens if you try to access an element outside the
array limit, say 6 using age[6]
What is the output you get when the arrays are not initialized?

Multidimensional arrays
Multidimensional arrays are another variation of one dimensional
arrays. For example, you want to store the age of students in 2 classes
class 1 and class 2, each having 5 students then you could do it with
a two dimensional array like:

int age[5][2]; \\ array declaration

\\ class 1

int age[0][0]=6;\\ array initialization 0

age[1][0]=7;\\ array initialization 1

age[2][0]=7;\\ array initialization 2

age[3][0]=6;\\ array initialization 3

age[4][0]=8;\\ array initialization 4

\\class 2

age[0][1]=7;\\ array initialization 0

10

age[1][1]=8;\\ array initialization 1

11

age[2][1]=6;\\ array initialization 2

12

age[3][1]=6;\\ array initialization 3

13

age[4][1]=7;\\ array initialization 4

main2.cpp: 2d Array

(5*2).15

The first line declares a two dimensional array of size 10

Of
course, we could have interchanged the indexing (2*5), and changed
the initializing part as well to obtained similar results. We could have
as well called an one dimensional array of size 10 and stored the
values but using a multidimensional array has made the code look
structured. For example, If we want to loop on the age of the class
2 students, it is now as simple as keeping the second index as 1 and
looping over the first index from 0 to 4.
Do it yourself:
Try to implement a 10 dimensional array with each index of size
100 and see if you can compile the code. If the compilation fails try to
reduce the dimension one by one and then run the code.

An array can have any number

of dimensions but it has practical
limitations imposed by the size of the
physical memory in the computer
15

37

Using array as input parameters to functions

Like any other data type, arrays can also be passed as a parameter to
function. However, the syntax is a bit different:

\\ function taking a one dimensional array as a parameter

int addArray(int arr[], int size){

int c=0;

for(int ii=0;ii<size;ii++{

c += arr[ii];

return c;

\\ function taking a two dimensional array as a parameter

10

int addArray(int arr[][2], int size){

11

int c=0;

12

for(int jj=0;jj<2;jj++{

13

for(int ii=0;ii<size;ii++{

14

c += arr[ii][jj];

15

16

17

return c;

18

}
main25.cpp: Arrays as parameters

The above example shows two functions one takes a one dimensional
array as its input. As you have noticed, the syntax is that the size
of the array is not given (i.e int arr[]). For a two dimensional array,
the size of the second dimension is given but not the size of the first
one.16

Structures
Arrays are useful for storing a collection of values which are of similar data type. Structures are a collection of a different data types. For
example you could use structures to store the age, height and weight
of a student using structures:

struct student {

int age;

float height;

float weight;

};

The general syntax for giving arrays

as input parameters to functions is that
the index all but the first dimension
should be given. The first dimension is
optional.
16

38

We have used multiple data types like int and float in the structure
student. The structure student is now a data type by itself and can
be used now the way int is used to denote a integer. An example
of declaring a variable of type student and initializing the values is
shown in the example below.
1

struct student {

int age;

float height;
float weight;

4
5

};

student a;

student b;

\\ initialize for student a

a.age=10;

10

a.height=100;

11

a.weight=30;

12

\\ initialize for student b

13

b.age=13;

14

a.height=127;

15

a.weight=36;
main26.cpp: Structures

The variables inside the structures are accessed by the .. a and b

are variables of type student or we call them objects of type student
and their respective values like age, height and weight are accessed
by suffixing a . and the variable name. Another way of declaring
structures is:
1

struct student {

int age;

float height;

float weight;

}a,b;

Structures used in conjunction with arrays is a neat way to organize a database. For example, to have the database of a class having
100 students you just have to call an one dimensional array of structures:
1

student class1[100];
main27.cpp: Structures and Arrays

Pointers
What is a pointer?
Earlier, we have seen how to visualize the physical memory in the
computer and how a variable is stored in it.
Any variable is stored in a memory location and every memory
location has an associated address to it. A pointer is a special variable which stores the address of a memory location. A pointer to an
integer is declared like:
int *p;

Figure 4: Representation of the physical

memory in the computer. For every
variable there is an address field and a
data field.

There is a star after the data type int to denote that this variable is
only allowed to store address of memory locations. There can be
pointers to all the data types we have studied so far (float, double
etc.,). Pointers are not so easy for beginners to understand but once
you get a pictorial representation, it is easier to follow. Let us assume
that we are declaring a variable:
int a;
a=100;
When the first statement is executed, a memory location is allotted
but that memory location is likely to contain a random value as this
variable is not yet initialized. The pictures below show the memory
location of a before initialization and after initialization to 100.
Figure 5: The memory location of
the variable a. The variable is not yet
initialized (int a) and therefore contains
a random value 56739.

40

Figure 6: The memory location of the

variable a. The variable is now initialized to 100 by the second statement
(a=100).

Now let us store the address of the variable a in a pointer variable.

int a;
int *p;
In the above statements, the first line declares a variable of type int
but the second line declares a pointer that points to an integer. At
this stage since we have not initialized the variables, the memory
locations will look like:
Figure 7: The memory location of the
variables a and p before initialization.

After the statements below:

int a = 100;
int *p = &a;

Figure 8: The memory location of the

variables a and p after initialization.

As shown above, the way to get the address of a is to put a & before it. Now pa contains the address of a as shown in the picture.
This is the correct way to initialize a pointer, that is only addresses
can be passed to a pointer. You cannot initialize a pointer to a random value by yourself:17
int a = 100;
int *p = 5000;
The compiler will complain if you try to do it.

An exception is that you can set a

pointer to 0 in which case it is called
as a Null pointer. A Null pointer
should not be used for any operations,
a concept which will become clearer
soon.
17

41

Setting the value of a variable using pointers

Now since we know the address of a variable, we can manipulate the
value at that address using the pointers itself. To set the value of a
to 99 using its pointer pa, we do:
*pa = 500;
We use the * to denote that the value at the memory location stored
in pa is 500. Obviously there is a confusion with the * here. Previously we used it to denote a pointer variable now it has different
definition. An easier way to understand is that if there is data type
like int, float etc., before a *, then the star denotes that the variable
next to it is a pointer. If not, as in the case above, then the star should
be read as the value at memory location pa is ...

main28.cpp: Pointers

The & is called as the reference operator

and * is called as the dereference
operator.

Incrementing and decrementing pointers

Like other data types, pointers can also be incremented and decremented (of course, other mathematical operations are permissible but
the increment and decrement operations are the most often used).
But there is a difference in the way pointers are incremented. For
example, if you increment an integer using the increment operator
++, the value is increased by 1 but a pointer of an integer is incremented not by 1 but by the size of an integer. This is a point that
should be understood clearly. Memory locations have a fixed width
(8 bits usually). We know that the size of a integer is 32 bits which
obviously cannot be written in a memory location. Therefore the
integer is split into 4 parts and stored sequentially in the memory
location. Thus when a pointer to an integer is incremented, it jumps 4
locations.
Figure 9: Incrementing the pointer pa
(having the address of the integer a)
does not increment it by one but by the
size of the integer divided by the width
of the memory field. In the example
shown here the width of the memory
field is 8 bits and the size of the integer
is 32 bits.

main29.cpp: Pointer arithmetic

42

Pointers and Arrays

There is a close connection between pointers and arrays. An array
when declared, is allocated a contiguous region of memory, therefore
if the address of the first element is stored in a pointer then the memory location of the next element can be pointed by just incrementing
the pointer. The address of the first element can be done by:
int a[5];
int *pa=a;
Now pa contains the address of the first element (i.e., a[0]). To get the
address of, for example, the third element in the array we do:
int a[5];
int *pa=&a[2];
main30.cpp: Pointers and Arrays

Do it yourself:
In the above example, what happens when you make a statement
like *pa++? Is the pointer incremented or the value pointed by the
pointer incremented? Check it out by printing out the values of pa
and *pa after the statement

Pointer to pointers
As mentioned earlier, pointers contain the address of a variable.
In the similar way, the address of the pointer can also be stored in
another pointer. This is illustrated in the example below:

1
2
3
4

int a;
int *pa=&a; // pa contains the address of a
int **ppa=&pa;//ppa contains the address of pa
int ***pppa=&ppa;//pppa contains the address of ppa

You may add as many stars as you want but you are most likely to
use only pointers and not anything more than that and therefore we
shall not dwell much into it.

Other types of pointers

So far, in the illustrations we have seen only pointers to int and float
but there are types of pointers apart from double, long etc., which
you will often encounter. The first type is the void pointer. The void
pointer has no type as is declared as:

43

void *p;// a void pointer

A void pointer cannot be incremented or decremented for obvious

reasons. The computer does not know how many memory locations
to skip, as it does not know the type. But a void pointer has the flexibility that any type of pointers can be stored in it (int, float, double
etc.,).
Do it yourself:
How do you increment a void pointer? Learn more about the
function sizeof() and try to use it to increment a pointer.
There can be pointers to structures as well but due to the similarities with pointers to classes, we shall do it later on. Pointers to
functions are also allowed in C++ but I hope that you will not need it
at least in the near future and therefore we shall skip it.

main31.cpp: void pointer

File I/O
In the next chapter, we will jumping into the concept of C++ but
before that I will give a short glimpse of file I/O. We will learn how
to do various operations like opening, reading, writing and closing a
file. This chapter will be brief so that you will just know how to do
simple things using files. There are other ways to manipulate files but
I will just cover only one.
The first thing to note is that we will include a header file called
fstream whenever file related operation are done18 . The general
sequence to do any file related operation is:
Open the file
Read or Write from/to the file
Close the file

File objects
Prior to doing any file operations, we should first define a file object
and that is done like:
fstream outFile;
fstream is a class in C++ which is used for basic file reading and
writing. Since we will get into classes in the next chapter, we will just
treat it like we treat int, float etc., We call outFile is an object of type
fstream.

Opening a file
Opening a file is done using the function open as shown below:
outFile.open("out.txt", ios::out|ios::app );
The first argument is the file name and the second argument is a
combination of some flags as described below:

There are other headers like ifstream and ofstream specifically

used for reading and writing respectively. fstream can be used for both
reading and writing.
18

46

flag
ios::out
ios::in
ios::app
ios::trunc

description
open file for writing
open file for reading
append file for appending
delete file (if it already exists) and open

These flags can be combined using the Bitwise OR operator as

shown in the previous example which would open a file for writing
and appends the data written in it. However not all combinations are
possible. For example, you cannot open a file both for reading and
writing (ios::out|ios::in) neither can you open a file with a option like
(ios::app|ios::trunc) as these are conflicting each other.

Writing to a file
Once you have opened a file for writing, the file object can be used
the way cout was used to write on the screen as shown in the example below:

#include <iostream>

#include <fstream> // required for file related operations

using namespace std;

int main () {

fstream outFile;

outFile.open("test.txt",ios::out|ios::trunc);

outFile<<"I am here"<<endl; // write this into the file

cout<<Data written...<<endl;

outFile.close(); //close the file

return 0;

10
11

main32.cpp: Writing to a file

Reading from a file

For reading out data from a file, the flag ios::in should be used first
to open the file. The reading syntax is similar to cin which takes
inputs from a user and stores it in a variable. You should therefore
know what the data type is before reading the data.

#include <iostream>

#include <fstream>

using namespace std;

int main () {

fstream inFile;

main33.cpp: Reading from a file

47

inFile.open("test.txt",ios::in);

int a;

inFile>>a;

cout<<a<<endl;

10

inFile>>a;

11

cout<<a<<endl;

12

inFile>>a;

13

cout<<a<<endl;

14

inFile>>a;

15

cout<<a<<endl;

16

inFile.close();

17

return 0;

18

Reading out line by line using string

For reading out lines of data from a file, you can use the getline
function. You need to use the data type string for that and declare
the string header file as shown below:

#include <iostream>

#include <fstream>

#include <string> //header file for strings

using namespace std;

int main () {

fstream in;

in.open("test1.txt",ios::in);

8
9

string s;

10

cout<<"Line 1:"<<endl;

11

getline(in,s); // read first line

12

cout<<s<<endl;

13
14

cout<<"Line 2:"<<endl;

15

getline(in,s);// read second line

16

cout<<s<<endl;

17

in.close();

18

return 0;

main34.cpp: Reading lines of data from

a file

48

Closing a file
Closing a file is done using the function close() as illustrated in the
examples above. If the file is not closed, the memory might not be
freed for other uses or the file might get corrupted.

Checking the end of file (EOF)

You might want to read a file which contains a long list of integers.
So in that case you implement a loop to read the data but the problem is most of the times you do not know how many numbers (or
data elements) are stored in it. To overcome this, you can use the
eof function which will return true if the end of the file has been
reached. The use of this function is shown in the example below:

#include <iostream>

#include <fstream>

using namespace std;

int main () {

fstream inFile;

inFile.open("test.txt",ios::in);

int a;

while(!inFile.eof()){ // loop until end of file

inFile>>a;

cout<<a<<endl;

10

11

return 0;

12
13

main35.cpp: Checking end of files

The function eof will return false until the end of the file is reached
and therefore the reading will continue until then.
Do it yourself:
Check what happens if you continue reading even after the end of
file in the above example.

Reading binary files

Binary files are special files which contain only 0s and 1s.19 . The
reading and writing of binary files is somewhat different from the
writing shown in the previous examples. You also need to know the
manipulation of pointers. You might very rarely be asked to use a
binary file and therefore we need not get into it but one thing you
should remember is to use the ios::binary while opening it.

Of course all files are basically a series

of 0s and 1s but they are encoded in
some format (ASCII for example) but
binary files have no encoding they are
in some sense just a series of 0s and 1s
and therefore very compact. This was
useful in the olden days when memory
was very costly
19

49

As far as possible you try to avoid binary files. You will be using
ROOT files for storing most of the data in particle physics which you
will learn soon. If at all you need to read or write simple data, follow
the simple examples given in this chapter.
Do it yourself:
TIFR wants you to write a program for that stores the details like
name, age, department and salary of an employee in a database.
Implement it with the use of structures and make the program interactive. The program should asked the user to input the values and
should allow the user to decide whether to exit the program.

Classes
From C to C++
The concepts and features we have learned so far are more or less
common to C and C++. We shall now learn about classes, the main
ingredient of C++. C++ can be called as a higher version of C in that
whatever we could do with C++ can be done with C but not in a
straight forward way. Data abstraction and object oriented programming are the main features of C++. For a beginner it might seem that
the use of C++ is a bit of work around and that most of the things
could be achieved with C itself. This is true but the question here
is not the achievement but the way it is achieved and the time it
takes to achieve. It will take sometime to convince yourself about the
power of C++ and probably you might not be convinced at all until
you write a program that deserves the power of C++. You should be
aware that a language is used only to express an idea and how powerful a language is depends on how easily we can express not only
simple ideas but also abstract ideas. That is to say that just because
the expressions are easier in a particular language does not necessarily imply that the program will perform better in that language.
It depends on the algorithm and other things also. A program written in C++ can be slower than a program written Java if it is not well
written. We shall move on with these things in mind.

Built-in and user-defined types

We have so far used the built-in data types like int, float etc., By
built-in, we mean that when we do not have to give a description
about them anywhere. We also saw how to group data of similar
data types using arrays and finally try to mix multiple data types
using structures. In structures, we had to describe our data type
explicitly, i.e., the structure student was described using built-in data
types int and float. Now the structure student is an abstract data type
which internally contains the attributes like age, height, weight of a
student. Here the structure student is an user defined data type. The

52

user has defined it according to his convenience (and therefore omits

some attributes that he might deem unnecessary in his case).

struct student {

int age;

float height;

float weight;

};

We could have easily done away with structures by just having the
age, height and weight variables as simple variables and still would
have got the program to run successfully.

int age;

float height;

float weight;

What then is the advantage of structures? The answer is that the program is more structured and organized when the details belonging
to an abstract data type are a sort of "tied together" so that the access and manipulation becomes simpler. Expressing the attributes
becomes more elegant by using structures. When you want to store
details of students of a class room without the use of structures, you
would declare some variables like
1

int age1, age2, age3, age4, age5 ....;

float height1, height2, height3, height4, height5 ....;

float weight1, weight2, weight3, weight4, weight5 ....;

for the entire class which is not an elegant way.20 . So far, I have just
tried to convince you that using structures makes your program
look more concise by tying the elements of an abstract data type like
student. Let us look at the other features that structures offer.

Member functions in structures

We shall now use a structure to define a three vector and see how
functions can be implemented in structures:

struct Vector{

int x,y,z; // member variables

void Initialize(int a, int b, int c); //member function

Of course, you can put them in arrays

to make it look elegant but there is
a reason for arranging things in a
particular manner. For example, putting
apples in one basket and oranges in
another one is a good strategy when
you want to pack them for selling but
when you want to test the quality of
fruits coming from a particular country,
the best strategy is to make baskets for
each country and then fill them with
the fruits from that country and do the
testing. (Each countries basket may
have smaller baskets for each fruit.)
20

53

};

void Vector::Initialize(int a, int b, int c){ // definition

x=a;

y=b;
z=c;

8
9

10

int main () {

11

Vector v;

12

v.Initialize(1,2,3);

13

cout<<v.x<<endl;

14

cout<<v.y<<endl;

15

cout<<v.z<<endl;
return 0;

16
17

In the above example, "Initialize" is a member function of the structure Vector which sets the values of the three components of a vector.
The function is declared inside the structured but defined outside of
it. Notice the use of scope operator ::. This operator tells that this
function belongs to the structure Vector. This is needed because there
could be other structures which might have an Initializefunction.
The scope operator tells us to which structure the function belongs
to. The above example is a simple revelation of how the data abstraction, that is, to organize data types that build the abstract structure
and implement useful methods to manipulate them and to extract
useful information using the the fundamental data types in the structure, brings in simplicity into code development.

structureFunction.cpp: Function in a
structure

From structures to classes

Structure are a subset of so called classes. In the last example we see
that the member variables of the structure vector could be accessed
using directly by using v.x etc., . However for some reasons you
want these variables to be hidden from the user or to be manipulated
directly by user. Such a thing cannot be done using structures as
all the members are public i.e visible to everyone. This can be
implemented using classes21 Lets start our first example of classes by
implementing the vector structure using a class:

1
2
3
4

class Vector{
int x,y,z; // member variables
public:
void Initialize(int a, int b, int c); //member function

For example, you want to avoid the

user from accidently forgetting to
initializing one of the values.
and this is not the only thing that can
be done using classes.
21

54

};

void Vector::Initialize(int a, int b, int c){ // definition

x=a;

y=b;
z=c;

9
10

11

int main () {

12

Vector v;

13

v.Initialize(1,2,3);

14

cout<<v.x<<endl; // error: x is private

15

cout<<v.y<<endl; // error: y is private

16

cout<<v.z<<endl; // error: z is private

return 0;

17
18

As you see, there is not much of difference in implementation of

classes and structures. The keyword struct has been replaced by
class. There is however another keyword called public in the class
definition. This keyword tells that the function Initialize is open to
all. There are another couple of keywords which define the access
level of the members of a class- private and protected. The access
definitions are given below:
keyword
private
protected
public

2
3
4

Friends and derived classes will be

discussed in the next chapter.

accessible by
members of the same class or friends
members of the same class, friends and derived classes
all

By default, if any access specifier is not given, the member is given

only private access. In the example above, the members x,y,z are
given the default access level (private) and therefore are not accessible by the usual method (i.e., using v.x etc.,) and hence the compilation will fail. However they could still be accessed by the function
Initialize because it is a member of the same class.
Do it yourself:
Implement some extra methods pertaining to a vector object.
Now let us implement some useful methods for this class vector
like a method to get the length of the vector and the dot product
between two vector objects. This can be implemented in the following
way:
1

firstClass.cpp: Example of a simple

vector class

class Vector{
int x,y,z;
public:
void Initialize(int a, int b, int c);

You might have to include the math.h

header file for some math operations

55

float Length();

float dotProduct(Vector a);

};

void Vector::Initialize(int a, int b, int c){

x=a;

10

y=b;
z=c;

11
12

13

float Vector::Length()

14

15

return (sqrt((x*x)+(y*y)+(z*z)));

16

17

float Vector::dotProduct(Vector a)

18

19
20

return(x*a.x + y*a.y + z*a.z);

}

The implementation of the Length method is straight forward. For

taking the dot productwe define a method which takes another
vector object and operates on components of that vector. In that
method notice how the variable x belonging to the calling object and
that belonging to input object are being accessed. In the first case the
variables are used directly but in the latter case they are called using
the dot operator. This distinction should be clearly understood.

Constructors and Destructors

In the previous example we had an explicit method called Initialize
to set the values of the vectors. Calling the other methods without
calling the Initialize method would give us junk values. Therefore
it is good that all classes have such a method which initializes the
variables properly and it is better if such crucial methods are not
avoided accidentally. This feature is built in C++ using constructors.
A constructor can be called as an initialization method of a class.
The constructor method has the same name of the class and takes
in variables as parameters just like any other function. However a
constructor can be called only at the time of construction of an object
but not anywhere else. Have a look at the constructor in the example
below. The constructor has replaced the Initalize function.

1
2
3

class Vector{
int x,y,z;
public:

classVectorMethods1.cpp: The vector

class with additional methods

56

Vector(int a, int b, int c); // ctor 1

Vector(); // ctor 2
Vector(); // ctor 2

6
7

~Vector(); // dtor

float Length();

float dotProduct(Vector a);

10

};

11

Vector::Vector(int a, int b, int c){

12

x=a;

13

y=b;
z=c;

14
15

16

Vector::Vector(){
x=0;

17

y=0;

z=0;

18

19

Vector::~Vector()

20

21

cout<<"Deleting object..."<<endl;

22

23

float Vector::Length()

24

26

return (sqrt((x*x)+(y*y)+(z*z)));
}

27

float Vector::dotProduct(Vector a)

28

25

30

return(x*a.x + y*a.y + z*a.z);

}

31

int main () {

29

32

Vector v1(1,2,3);// using ctor 1

33

Vector v2;// using ctor 2 (no brackets!!!)

34

cout<<"Length of V1: "<<v1.Length()<<endl;

35

cout<<"Length of V2: "<<v2.Length()<<endl;

return 0;

36
37

The main difference is, unlike the Initialize function which can be
called anywhere and any number of times, the constructor is called
only when the objects are created.

Constructor overloading
A class can have many constructors each taking different arguments,
a concept similar to overloading of functions which we saw earlier.
In the above example we see the explicit declaration of two construc-

constDestr.cpp: Constructors and

Destructors

57

tors one that takes three parameters and another one that take no
parameters. To create an object of this class you can use either one of
these but notice the way how the object is created using the first constructor and the second constructor. There is no bracket used when
created using the second constructor though that constructor definition has them. This is an important point to remember that if the
constructor has no parameters you need not use the brackets when
you create object.

You will have to use brackets if you

create a pointer to the object instead but
that will be treated later on.

Default Constructor
What happens if the constructor for a class is not explicitly declared?.
The compiler takes over and writes a dummy constructor- one that
does nothing. This is called as the default constructor. A default
constructor will be made only if there are no existing constructors
defined.

Destructor
A destructor is the opposite of the constructor. It is called when an
object is deleted. The syntax for defining a constructor is a (tilde) followed by the class name as shown in the above example. As you will
notice, the destructor is called before the end of the program automatically even without an explicit call to it22 . Destructor overloading
is not allowed in C++.

Pointers to classes
Just like we had pointers to other data types, we can also have pointers to classes. Before that let us see how a pointer to a structure is
created and how we can access the variables of a structure using
pointers.

struct Vector{

int x,y,z;

void Initialize(int a, int b, int c);

};

5
6

void Vector::Initialize(int a, int b, int c){

x=a;

y=b;
z=c;

9
10

11

int main () {

explicit calls to the destructor can be

made using the delete method which
will be introduced when we deal with
pointers to objects
22

58

Vector v;

12

14

Vector *pv = &v; // pointer init

pv->Initialize(1,2,3);

15

cout<<"Values of x,y,z"<<endl;

16

cout<<v.x<<endl;

17

cout<<v.y<<endl;

18

cout<<v.z<<endl;

19

return 0;

13

20

As demonstrated in the example above, we create a pointer to a structure pv as usual using the *. This pointer is initialised with the
address of v using the address operator &. We call v as a object
and pv as the pointer to the object. To access a member function
using the object we use the . (dot) operator. For example to call
the initialize function using the object, we would have called
v.Initialize(1,2,3). In the above example we call this function using the pointer instead. Notice the arrow operator in place of the
dot. This is the operator to be used to access a member using a
pointer.
The same procedure applies for classes as well. However creation
of pointer is made in a different way for classes as shown below.

structurePointer.cpp: Pointer to a
structure

Vector *v1 = new Vector(1,2,3);

Vector *v2 = new Vector();
The new keyword returns a pointer to the class. Notice how the
constructors are used here. The second constructor which does not
take input parameters has brackets here (unlike when just the object
was created). To access a member of the class using the pointer we
use the arrow operator instead of the dot operator:
v1Length();
An object can and should be deleted so that the memory is freed for
other purposes by calling the destructor using the delete keyword:
delete v1;
When a pointer to a class is created instead of an object, the destructor should be called explicitly failing so will result in what we call
as memory leaks. The delete function should be called when you
think that the object will be of no use anymore.

1
2

class Vector{
int x,y,z;

Pointer to arrays can also be created

using the new keyword. An example:
int *parr = new int [5] ;

59

public:

Vector(int a, int b, int c); // ctor 1

Vector(); // ctor 2

~Vector(); // dtor 1

float Length();

float dotProduct(Vector a);

9
10

};
Vector::Vector(int a, int b, int c){

11

x=a;

12

y=b;
z=c;

13
14

15

Vector::Vector(){
x=0;

16

y=0;

z=0;

17

18

Vector::~Vector()

19

{
cout<<"Deleting object..."<<endl;

20
21

22

float Vector::Length()

23

{
return (sqrt((x*x)+(y*y)+(z*z)));

24
25

26
27
28

float Vector::dotProduct(Vector a)

29

{
return(x*a.x + y*a.y + z*a.z);

30
31

32

int main () {

33

Vector v1(1,2,3);// using ctor 1

34

Vector v2;// using ctor 2

35

cout<<"Length of V1: "<<v1.Length()<<endl;

36

cout<<"Length of V2: "<<v2.Length()<<endl;

37

// using pointers
Vector *v3 = new Vector(1,2,3); // using ctor 1
Vector *v4 = new Vector();// using ctor 2

38
39
40

cout<<"Length of V3: "<<v3->Length()<<endl;

41

cout<<"Length of V4: "<<v4->Length()<<endl;

42

delete v3; // try commenting this out

43

delete v4; // try commenting this out

return 0;

44
45

}
classPointer.cpp: Pointer to a class

60

The example above demonstrates the use of pointers to classes. The

main things to remember when you work with pointers are:
1. new: to create a pointer
2. : to access members
3. delete: to delete and freed the memory used by the object
Do it yourself:
Create a class for a simple two by two matrix and implement all
methods pertaining to it like determinant, addition, subtraction etc.,

Operator Overloading
Suppose you make a class for a 2 x 2 matrix, you will include all
relevant methods in the class like the determinant. Now, if you want
to add two matrices, you cannot just use the +operator and get the
answer. The + operator works directly for integers, floats etc., but
when using with classes, it is not so because it is not directly relevant
which is to be added to what. However, you make it relevant by the
mechanism of function overloading.
1

class twoBytwoMatrix{

public:

float a11,a12,a21,a22;

twoBytwoMatrix(float a, float b, float c, float d);

twoBytwoMatrix(); // ctor 2

~twoBytwoMatrix(); // dtor 1

float determinant();

twoBytwoMatrix operator - (twoBytwoMatrix);

twoBytwoMatrix operator + (twoBytwoMatrix);

10

};

11
12

twoBytwoMatrix twoBytwoMatrix::operator+ (twoBytwoMatrix m)

13

14

twoBytwoMatrix out;

15

out.a11 = a11 + m.a11;

16

out.a12 = a12 + m.a12;

17

out.a21 = a21 + m.a21;

18

out.a22 = a22 + m.a22;

return out;

19
20

21

twoBytwoMatrix twoBytwoMatrix::operator- (twoBytwoMatrix m)

22

23

twoBytwoMatrix out;

61

24

out.a11 = a11 - m.a11;

25

out.a12 = a12 - m.a12;

26

out.a21 = a21 - m.a21;

27

out.a22 = a22 - m.a22;

return out;

28
29

30

twoBytwoMatrix::twoBytwoMatrix(float a, float b, float c,

float d){

31
32

a11=a;

33

a12=b;

34

a21=c;

35

a22=d;

36

37

twoBytwoMatrix::twoBytwoMatrix()

38

{
a11=0;

39

a12=0;

a21=0; a22=0;

40

41

twoBytwoMatrix::~twoBytwoMatrix()

42

{
// cout<<"Deleting object..."<<endl;

43
44

45

float twoBytwoMatrix::determinant()

46

{
return ((a11*a22)-(a21*a12));

47
48

49

int main () {

50

twoBytwoMatrix m1(1,2,3,4);

51

twoBytwoMatrix m2(2,3,4,5);

52

twoBytwoMatrix m3;

53

m3=m1+m2;

54

cout<<"Determinant: "<<m3.determinant()<<endl;

55

return 0;

56

In the above example, we have overloaded the addition and subtraction operators. The syntax is as follows:
return type operator + (input parameter type)
The + operator can be replaced by most of the operators we have
learnt so far like -,*,/ etc.,. When the operator is a binary operator
(an operator which has operands before and after it) the left hand
operand is an object of the class. The type of the right hand operand
is given in the input parameter type and the output type is given in
the return type.

operatorOverloading.cpp: Overloading
operators

62

In the above example we saw that the + operator has been overloaded to perform the addition of two matrices. However, note that
the same thing can be done by replacing it by the - operator. That
is you can overload the subtraction operator to perform an addition.
This is completely legal and valid but such things are never done.
It is always better to overload the operators to perform what their
sign signifies rather than confusing yourself and other end users
eventually.

Static members
In the above example, we created three matrix objects m1, m2 and
m3. When each of these objects are created, a memory location is
allocated for their members. Now m1 has its set of member data and
functions (a11, a12 ... etc.,) and so do m2 and m3. There might be
cases where you want to share a variable across all the objects of a
class. That is, a variable that has an unique value for all the objects.
Such variables are called static variables and they are somewhat like
global variables to a class. The way they are declared and initialized
is different from the way it is done for other variables. Lets see an
example using static variables to keep track of the number of objects
created.
1
2

class twoBytwoMatrix{

public:

static int N; // declare inside class

float a11,a12,a21,a22;

twoBytwoMatrix(float a, float b, float c,

7
8

};

twoBytwoMatrix::twoBytwoMatrix(float a, float b,
float c, float d){

10
11

a11=a;

12

a12=b;

13

a21=c;

14

a22=d;

15

N++; // incrementing N

16
17

18

int twoBytwoMatrix::N=0; // init outside the class

19

int main () {

20
21

twoBytwoMatrix m1(1,2,3,4);

63

22

cout<<"Creating Matrix Number "<<m1.N<<endl;

23

twoBytwoMatrix m2(2,3,4,5);

24

cout<<"Creating Matrix Number "<<m2.N<<endl;

25

cout<<"The matrix number accessed directly:"<<

twoBytwoMatrix::N<<endl;

26

return 0;

27
28

29

staticVarcpp: Static Variables

In the above example a static integer is created using the keyword

static insided the class. A static variable should be initialized only
outside the class as shown in the above example. Though the variable
was declared as an integer already inside the class, the type (i.e.,
int)is used while initializing also:
int twoBytwoMatrix::N=0; // init outside class
The variable N is incremented in the constructor and therefore keeping track of the number of objects created. Though there might be
many objects created, the value of N will be same for all of them.
Static variables can be accessed just like other member variables or
using the scope operator :: as shown in the above example.

The this keyword

Sometimes you need to access the pointer of the object inside the
class functions. In such cases you can use the this keyword. It
returns the pointer of the object that calls that function.
1

twoBytwoMatrix::twoBytwoMatrix(float a, float b,
float c, float d){

2
3

a11=a;

a12=b;

a21=c;
a22=d;

cout<<"This pointer value of this object:"<<this<<endl;

7
8

int main () {

10
11

twoBytwoMatrix m1(1,2,3,4);

12

cout<<"Address of m1: "<<&m1<<endl;

13

return 0;

14

}
this.cpp: The this keyword

More Features in C++

In this section we will review the main features of C++. Although
you might not use these features in your research, it is important to
have an idea of how these features are used in order to understand
codes written by others who would have used these features. The
main features are:
Friendship
Inheritance
Polymorphism
Templates

Friendship
In the last class we saw the use of access specifiers: public, private
and protected. The private and protected members can be accessed
by friend functions and friend classes. Let us see the use of friend
function using examples:

class rect{
float length, breadth;// private mem

2
3

public:
rect(float a, float b){

4
5

length = a;

breadth = b;

float GetArea();

friend float GetLength(rect r);

10

};

11

float rect::GetArea()

12

{
return length*breadth;

13
14

66

15

// function outside class

16

float GetLength(rect r)

17

{
return r.length;

18
19

20

int main() {

21

rect first(2,3);

22

cout<<"The area of rect:"<<first.GetArea()<<endl;

23

cout<<"The length of rect:"<<GetLength(first)<<endl;

24

return 0;

25

}
friendFunction.cpp: Friend function

In the above example, we have made a class for the shape called
rectangle which has its length and breadth as its private members. It
also has a public method to compute the area of the rectangle. Now
the user has implemented a function called Getlength() which returns
the length of the rectangle. Since length is a private member it is not
accessible by ordinary functions. However, if the function is declared
as a friend inside the class, then it can access the private members
also. The syntax to declare a function as a friend is as shown in the
example above:
friend float GetLength(rect r);
Like friend functions, there can be friend classes too. If a class is
declared as a friend, the friend class can access the members of the
class that declared as a friend (not vice-versa):
1

class square{
float length;// private

2
3

public:
square(float a){//ctor

length = a;

5
6

// give access to rect

friend class rect;

9
10
11
12
13

};
class rect{
float length, breadth;// private
public:
rect(float a, float b){//ctor

14

length = a;

15

breadth =b;

16

17

float GetArea();

18

float GetArea(square);

67

19

};

20

float rect::GetArea()

21

{
return length*breadth;

22
23

24

float rect::GetArea(square s)

25

26

length=s.length;

27

breadth=s.length;

28

return GetArea();

29

30

int main() {

31

square s(2);

32

rect r(3,4);

33

cout<<"The area of rect is "<<r.GetArea()<<endl;

34

cout<<"The area of square is "<<r.GetArea(s)<<endl;

35

}
friendClass.cpp: Friend class

In the above example, we have implemented a method to calculate

the area of a square. This method takes a square object has its input,
and calculates the area by setting the length and breadth variables to
the length of the square. The GetArea(square s) is able to access the
member s.length only because the rect class is declared as friend in
square class. If we remove the friend declaration in the square class,
the compiler will give out errors complaining that the member length
is private to square.

Inheritance
Now we come to an important concept in object oriented programming called inheritance. Inheritance is a method by which classes
which share some of the features with other classes can inherit methods and variables common to both of them thus reducing the code
size.
1

class shape{

protected:

3
4
5

float length, breadth;

public:
shape(){

cout<<"In Shape ctor 0"<<endl;

length=0;

breadth =0;

68

shape(float l, float b){

10
11

cout<<"In Shape ctor 1"<<endl;

12

length=l;

13

breadth =b;
}

14
15

};

16

class square:public shape{

17

public:
square(float a){//ctor

18
19

cout<<"In square ctor"<<endl;

20

length = a;

21

breadth=a;

22

23

float area(){
return

24

length*breadth;

25
26

};

27

class rect:public shape{

28

public:
rect(float a, float b):shape(a,b){//ctor

29

cout<<"In rect ctor"<<endl;

30
31

32

float area(){
return length*breadth;

33

34
35

};

36

int main() {

37

square s(2);

38

rect r(3,4);

39

cout<<"The area of square is "<<s.area()<<endl;

cout<<"The area of rect is "<<r.area()<<endl;

40
41

}
inherit.cpp: Inheritance in classes

In the above example, there is a class called shape which has two
variables length and breadth. The shape and rect classes too have
these variables and therefore can make use of the shape class using
the feature of inheritance. The syntax is as shown in the example:
class rect:public shape
class square:public shape
The first class in the above syntax is called as the derived class
and the second class, the base class. The derived class can access the
members of the base class as if they if their own member variables.
The access specifier public can be replaced by other access specifiers
like protected and private.23 Note that the square and rect classes

If the access level in the base class is

less than the one specified while inheriting, then the members are inherited
with the inheritance specification. For
example, if base class access is public
and inheritance specification is private,
the members are inherited with private
access. If it is the other way round the
access specification in the base class in
retained.
23

69

have no mention of the length and breadth variables in their classes

but can access the variables of the shape class. Also note that the
members in the shape class have protected access. Private members
cannot be accessed by derived classes.

Initializing constructor of a base class using inherited class

How do we call the constructor of the base class using the derived
class? In the square class, its constructor is simple has no mention
about the constructor of the shape class in which case, the default
constructor (ctor 0) of shape is called. In the rect class however,
we see that the constructor has a different syntax: rect(float a, float
b):shape(a,b)
This constructor takes in two values and calls the constructor of
shape (ctor 2) giving these two values as its input.

Multiple inheritance
Classes can inherit multiple classes too. In the example below there
is a simple class for printing out the values of an input float variable.
The square inherits from the shape and the Print class. As with single
inheritance, the square class can access the members of the Print class
as if it were is its own member.
1

class Print{

public:

Print(){

void PrintValue(float a){

cout<<"Printing Value:"<<a<<endl;

};

class shape{

10
11
12
13

protected:
float length, breadth;
public:
shape(){

14

cout<<"In Shape ctor 0"<<endl;

15

length=0;

16

breadth =0;

17

18

shape(float l, float b){

19

cout<<"In Shape ctor 1"<<endl;

20

length=l;

21

breadth =b;

70

22
23

};

24

class square:public shape,public Print{

25

public:
square(float a){//ctor

26
27

cout<<"In square ctor"<<endl;

28

length = a;

29

breadth=a;

30

31

float area(){
return

32

length*breadth;

33
34

};

35

int main() {

36

square s(2);

37

s.PrintValue(s.area());

38

}
multInherit.cpp: Inheritance in classes

Polymorphism
The next feature in C++ is polymorphism. We have seen what a base
class is and what a derived class in previous examples. The shape
class was a base class and square was a class derived from shape.
Let us create a pointer for the derived class square: square *s=new
square(2);
Now let us create a pointer for the base class shape: shape *sh=new
shape();
Let us proceed to the next logical step. Now since square is a
derived class from shape, its pointer is type compatible with shape
and therefore we can as well do:
shape *sh=s;
This type compatibility of the pointer of the base class with the
derived class is called polymorphism. That is the pointer of the base
class can point to the pointer of the derived class. Note however the
vice-versa is not true:
square *s=sh;
The above (reverse polymorphism) is not allowed in C++. The
following is an example of polymorphism:

class shape{

protected:

float length, breadth;

71

public:
shape(){

5
6

cout<<"In Shape ctor 0"<<endl;

length=0;
breadth =0;

shape(float l, float b){

10
11

cout<<"In Shape ctor 1"<<endl;

12

length=l;

13

breadth =b;

14

15

void PrintValues(){
cout<<"Length: "<<length<<" Breadth: "<<breadth<<endl;

16

17
18

};

19

class square:public shape{

20

public:
square(float a){//ctor

21
22

cout<<"In square ctor"<<endl;

23

length = a;
breadth=a;

24
25

26

float area(){
return

27

length*breadth;

28
29

};

30

int main() {
square *s=new square(2);
shape *sh1= new shape();

31
32

shape *sh2 = s; // type compatible

// square *s2=sh2;// error

33
34
35

sh1->PrintValues();

36

sh2->PrintValues();

37

// sh1->area();// not allowed

38

}
poly.cpp: Polymorphism in classes

Virtual Members
In the above example, thougth there was type compatibility of the
base class with the derived class, there was one pitfall in it. The base
class pointer thus obtained cannot still access the members of the
derived class. square *s=new square(2);
shape *sh2= s;

72

In the above statements, the base class pointer sh2 points to the
derived class pointer s. Though the pointer s can access its member
variable area directly:
Allowed
s->area();
the pointer sh2 however cannot access it:
Not allowed
sh2->area();
There is a workaround in such cases: move the area method into
the base class. But remember that we will lose the flexibility in doing
so for if we implement a another class called triangle where the area
is calculated differently, this we lead to a complex code structure.
This is where one can use virtual members. If a particular method
is common to all the derived classes but have different implementations, a virtual method should be defined in the base class as shown
in the example below:

using namespace std;

class shape{

protected:
float length, breadth;

4
5

public:
shape(){

6
7

cout<<"In Shape ctor 0"<<endl;

length=0;

breadth =0;

10

11

shape(float l, float b){

12

cout<<"In Shape ctor 1"<<endl;

13

length=l;
breadth =b;

14
15

16

void PrintValues(){
cout<<"Length: "<<length<<" Breadth: "<<breadth<<endl;

17
18

19

virtual float area(){

20

return -1;
}

21
22

};

23

class square:public shape{

24

public:

25

square(float a){//ctor

26

cout<<"In square ctor"<<endl;

27

length = a;

73

breadth=a;

28
29

30

float area(){
return

31

length*breadth;

32
33

};

34

class tri:public shape{

35

public:
tri(float a, float b){//ctor

36
37

cout<<"In triangle ctor"<<endl;

38

length =

39

breadth=

a;
b;

40

41

float area(){
return

42

0.5*(length*breadth);

43
44

};

45

int main() {
square *s=new square(2);
tri *t=new tri(2,3);

46
47

shape *sh1 =s;

shape *sh2 =t;

48
49
50

cout<<"square area: "<<sh1->area()<<endl;

51

cout<<"triangle area: "<<sh2->area()<<endl;

52
53

}
virtual.cpp: Virtual methods in classes

In the above code, the two classes square and tri have a common
method called area which are implemented differently. For the base
class to access these members, a virtual method for area has been
defined in the base class.

virtual int area ()

{ return (0); }

The above virtual function is a sort of dummy function which has

a simple implementation which returns 0. We can also remove this
implementation using a statement like:
virtual int area () =0;
A virtual function which has no implementation is called a pure
virtual function and a class that contains a pure virtual function is
called a abstract base class.

74

Templates
Function templates are special functions than can adapt themselves
to different data types.
1

template <class T>

T Subtract (T a, T b) {

T diff= a-b;

return diff;

template <class T, class U>

U Subtract2 (T a, U b) {
U diff= a-b;

return diff;

9
10

11
12

int main () {

13

float a,b;

14

int c,d;

15

a=2.2;

16

b=5.6;

17

c=3;

18

d=4;

19

float d1=Subtract<float>(a,b);

20

int d2= Subtract<int>(c,d);

21

float d3=Subtract2<int,float>(c,a);

22

cout<<"Diff of floats: "<<d1<<endl;

23

cout<<"Diff of int: "<<d2<<endl;

24

cout<<"Diff of int and float "<<d3<<endl;

return 0;

25
26

In the above code we have declared a template function called Subtract which takes in two variables of type T and returns the same
type T, the difference between the inputs. T is just a place holder
and any other letter can be used in its place. The statement: template
<class T>
tells that the function following takes a template variable T. The
Subtract function can take any of the data types now by replacing the
template variable: float d1=Subtract<float>(a,b);
takes float as its inputs and
int d2= Subtract<int>(c,d);
takes int as its inputs. A template function can takes different
types as its input also by declaring a statement like:

functionTemplate.cpp: Function Templates

75

template <class T, class U>

The above statement tells that the following function will use two
template variables for its implementation. The Subtract2 function
can take an int and a float as an input and return a float as in the
following statement:
float d3=Subtract2<int,float>(c,a);
Just like we had function templates, we can also have class templates which uses template parameters.
1

template <class T>

class square{
T side;

public:

square(T a){//ctor

side=a;

6
7

T area(){
return a*a;

10
11

};

12

int main () {

13

square<int> s1(2);

14

square<float> s2(3.3);

15

cout<<"area of s1 "<<s1.area()<<endl;

16

cout<<"area of s2 "<<s2.area()<<endl;
return 0;

17
18

}
classTemplate.cpp: Class Template

In the above code, a template variable T is defined and the class

square makes use of this template variable. The template variable is
replaced by the user provided data type in the following statements:
square<int> s1(2);
square<float> s2(3.3);
The first statement replaces the template variable T by int for all
instances of its object and the second statement replaces the template
variable T by float.

ROOT: Introduction and Installation

This is the beginning of the second part of this lecture notes where
we will discuss about ROOT. ROOT is a famous open source data
analysis and graphing library developed by CERN. If you are a high
energy physicists you will probably be spending a good amount time
in writing codes using ROOT. ROOT has almost all the features that
you will require in your research work. Examples are histograms,
graphs, profile plots, 2d plots, fitting, image manipulation, random
number generation, FFT, signal analysis, PCA etc.,.24 ROOT is not an
application as such like MS Excel where you just open your data files
and start plotting. ROOT offers only the libraries that is to say the
header and source files required for plotting and data analysis. It is
your job to compile the source code of ROOT and then link it to your
program. I shall not be able to cover all the features of ROOT owing
to the limited time of this course. Once you get acquainted with the
way ROOT works, using others features will not be a big problem.

Installing ROOT
The installation of ROOT (in Linux) consists of three steps which will
be explained briefly in the following subsections. The sections will
not describe advanced features as it is mainly aimed at beginners
trying to learn ROOT. If you need more details about it you can refer
to the ROOT website.

Downloading ROOT
The source code of ROOT is available for free download at:
http://root.cern.ch/drupal/content/downloading-root
ROOT regularly releases new updates and patches for bugs found
in their released source code. In the link above, you will find a huge
list of ROOT versions available for download. It is usually recommended to use the latest version to make sure that you up to date
with the features. However, it might be the case that the codes written for your experiment depend on a specific version of ROOT in

For a complete list, please refer to the

ROOT website.
24

If you have forgotten what a library is

and how they are done, please go to the
second chapter and review it.

78

which case you should download the appropriate version.25 Once

you have clicked on the appropriate version, you will be taken to a
page that gives you options to either download the complete source
tree or the binaries. To have a stable version on ROOT and to prevent
any library conflicts, it is always better to download the source tree.
The source tree contains all the header and source file of ROOT in a
compressed format (.tar.gz)

The developers will remove/add

new features depending on the reviews
and requests they get on their forum.
Should you find something that does
not work as expected in ROOT you can
contact the developers by writing in
their forum.
25

Compiling ROOT
Once you have downloaded the ROOT source tree, follow the steps
here:
1. Un zip/tar in a convenient location and the contents will be stored
in a folder called root, by default.
2. cd to the root folder
3. type ./configure
4. make -j 4
As described in Chapter 2, one can create objects of functions (with
declarations in the header file and definitions in the source file) so
that another program can easily link to these objects without having to re-compile the sources and headers26 . The rules for compiling
ROOT will depend on the Operating System and the libraries that
are installed in your system. These rules are generated automatically by the ./configure command which will generate a makefile.
The makefile contains the commands for compiling the sources and
headers for your system. The number 4 in the make command is the
number of cores that will be used for the compiling process. You can
change it to the number of cores in your machine. Once you type the
make command, the compilation process will start and you see lines
of compile commands starting with g++. This is similar to what you
did for compiling the codes in the examples. ROOT contains thousands of files and each has to compiled and therefore it takes sometime to complete. Sometimes, the compilation process might stop
with an error. In that case, it might be that you have not installed
some libraries on which the compilation of ROOT depends. Check:
what the error is from the compilation output and try to check where
the error first started by scrolling up. That line will show you some
information like for example -lGl not found in which case you
should install the GL library. Once you have installed it, you can
restart the compilation by typing the make -j 4 command. The
compilation will restart from the place where it last stopped.

Compilation of ROOT may take more

than 30 minutes
26

79

Setting the environment variables

Once the compilation is successfully done, you have to set the environment variables for proper operation of ROOT. Your ROOT
programs will use these environment variables to locate the proper
location of ROOT libraries and other configuration files. The main
environment variables is:
ROOTSYS
which should point to the folder where the root folder is present.
For example, if you have extracted the source tree in the folder /
home/myprograms/root then you should set the ROOTSYS variable
like this:
export ROOTSYS=/home/myprograms/root/ 27
in a terminal.The system environment variables like:
LD_LIBRARY_PATH and
PATH
have to be modified by the appending the library folder location
and the bin folder location of ROOT respectively. If you have set your
ROOTSYS variable using the method above, then proceed to do the
following in the same terminal where you set the ROOTSYS variable:
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$ROOTSYS/lib
export PATH=$PATH:$ROOTSYS/bin
There is however a shortcut to do all the environment settings of
ROOT. The following file
$ROOTSYS/bin/hisroot.sh (csh)28
(where $ROOTSYS is the folder where ROOT was installed) contains all the commands required to set the environment variables for
ROOT. You should execute this file using the commands:
export ROOTSYS=/home/myprograms/root/
source $ROOTSYS/bin/thisroot.sh
in a terminal. You should remember that this should be done
every time you open a new terminal. In case you want it to be set
permanently so that you do not need to set the variables every time
you open a new terminal, you can do so by opening your .bashrc file
adding the above command in it. Follow the commands to do that:
cd
gedit .bashrc
The .bashrc file will be opened and in it you should add the following lines:
export ROOTSYS=/home/myprograms/root/
source $ROOTSYS/bin/thisroot.sh
and save the file.
In case you have multiple versions of ROOT in your system then
you should avoid using this method and try to invoke the right script

In some machines, you would have

to use the command setenv instead of
export and the syntax for setenv is a bit
different from export. This depends on
the shell you use. If you use the Bourne
shell then you should use export
and if you use the C shell then you
should use the setenv command. More
about setting environment variables in
different shells can be found here
27

If your shell is a Bourne shell use

the .sh file and if yours is a C shell,
use the .csh file. These are script files
which contain commands for various
operations for proper operation of
ROOT
28

You can ignore the export command

and set the ROOTSYS variable in the
source command to the actual location
of your ROOT installation.

80

yourself in the terminal using the previous method. Now you are all
set to run ROOT. In the terminal, type root and you will enter into
the ROOT terminal. Type: TCanvas t and press Enter and you will
see a canvas popping up. Enter .q in the terminal and you will be out
of the ROOT terminal.

Working with ROOT

There are two ways to work with ROOT,
1. Using CINT interface which will not be covered in this lecture
2. Using the traditional C++ way which we shall discuss in detail
Using the CINT interface can be useful while developing the code
where you will trying out all options until your code works properly.
You need not compile the program after it has been changed. However CINT interface is an interpreter module which is way slower
that a compiled program and therefore do not use it for running
huge programs. I will present one example here to show how it
works:
1

#include <iostream.h>

int rootTest()

cout<<"Hi welcome to ROOT CINT interface"<<endl;

return 0;

Let us see how the code above can be run using the ROOT CINT
interface. In a terminal, cd to the location where the above file is
present and start the ROOT CINT interface by typing root and then
type:
.x rootTest.C
The program will be executed and you should see the output. You
can exit the root terminal by typing:
.q
You can use any of the ROOT classes in these files which will be
executed by the ROOT CINT interface. More ways to ron a ROOT
program can be found in this wiki link.

rootTest.C: A macro for testing ROOT

CINT interface

Programming with ROOT: Starting with Histograms

Lets start our first example of ROOT by opening a canvas. A canvas
is a sort of drawing board for ROOT plots on which all the graphs,
histograms etc., will be drawn. The TCanvas class is responsible for
all canvas related operations like opening, dividing it into subpads
etc.,.
1

#include <iostream>

#include <TApplication.h>

#include <TCanvas.h>

using namespace::std;

int main()

{
TApplication *t = new TApplication("myapp",0,0);
TCanvas *C = new TCanvas;

7
8
9
10

t->Run();

11

delete C;
return 0;

12
13

If you try to compile the above program with the usual compile
command:
g++ canvas.cpp
The compiler will complain that it could not find the header
files (TApplication, TCanvas etc.,). These file reside in the include
folder present in the location where ROOT is installed ($ROOTSYS/include). You will have to inform the compiler about this location so that it can search for these file there. The flag used for the
search directory for include files is -I. The compile command now
looks like:
g++ canvas.cpp -I$ROOTSYS/include
Executing the above command gives errors like undefined reference to TApplication::TApplication etc.,. This is because though
the header files where the functions were declared are given, the pro-

canvas.cpp: A simple program to

instantiate a ROOT canvas

82

gram cannot find the definitions for those functions. We have already
compiled ROOT and have made the objects files for all the files so we
need to link to those object files (or libraries) using the flag -l. The
flag used for the search directory for library files is -L. The compile
command looks like:
g++ rootTest.C -I$ROOTSYS/include -L$ROOTSYS/lib -lCore
-lGpad
where $ROOTSYS/lib denotes the location where the ROOT libraries are located and libCore and libGpad are the libraries where
the definitions for the TApplication and TCanvas classes are found.If
you use other classes, you should link to the appropriate libaries
Now you will see the program compiles successfully and you can run
it using:
./a.out

When the -l flag is used the lib prefix

of the libraries is dropped as explained
in Chapter 2.

Getting the correct compiler flags for program using ROOT libraries
As you saw earlier, you need to know the correct flags for the include
directory and for linking to the correct libraries. There is easier and
faster way of doing this. ROOT provides a program called rootconfig to output these flags for your system. This program is found
in the folder:
$ROOTSYS/bin
If you just execute root-config on a terminal, it will shown the
options for using the program. FOr example using the option
glibs will give you the libraries required for linking and cflags
will give you other options requires for compiling like the include
folder path etc.,.Type:
root-config glibs
and you will see an output containing the libraries to be linked to
for a root program:
-L/home/myprogram/root/lib -lGui -lCore -lCint -lRIO -lNet
-lHist -lGraf -lGraf3d -lGpad -lTree -lRint -lPostscript -lMatrix lPhysics -lMathCore -lThread -pthread -lm -ldl -rdynamic
You can use execute the root-config program directly in your compile command and automatically append these flags in this manner:
g++ canvas.cpp root-config --glibs --cflags

Plotting a histogram
Let us move to a real world example where we will plot a distribution of a variable using a histogram. Let us see how to generate a
gaussian distribution and poisson distribution of random number
and plot it in a canvas using ROOT.

The symbol is not the symbol of single

quote but of the accent found close the
1 key (at least in my computer) and
there is no space in between the two
dashes.

83

#include <iostream>

#include <TApplication.h>

#include <TCanvas.h>

#include <TH1F.h>

#include <TRandom.h>

using namespace std;

int main()

TApplication app("my_app",0,0);

10
11
12

TCanvas *c = new TCanvas("myCanvas","My Canvas");

TRandom *ran = new TRandom(450);
TH1F *h1 = new TH1F("my_hist1","Gauss",50,-500,500);

14

TH1F *h2 = new TH1F("my_hist2","Poisson",50,-100,100);

h1->SetFillColor(4);

15

h2->SetFillColor(6);

16

for(int ii=0;ii<10000;ii++){

17

h1->Fill(ran->Gaus(0,100));

18

h2->Fill(ran->Poisson(40));

19

20

c->Divide(2,1);

21

c->cd(1);

22

h1->Draw();

23

c->cd(2);

24

h2->Draw();

25

gPad->Update();

26

app.Run();

27

delete ran;

28

delete c;

29

return 0;

30

13

For the above job, we will use the following three classes in ROOT:
1. TApplication: creates the ROOT application environment
2. TCanvas: Opens a canvas
3. TRandom: Generates random numbers
4. TH1F: Histogram class
In the above code, we create an object of TApplication which should
be done only once in your program. Without the TApplication class,
your canvases will not be shown. The Run method should be called

histo1.cpp: Histograms of a gaussian

and a possion distributed random
numbers.

84

before you return from your code. Next we create a pointer to a

TCanvas object using its constructor which takes a alias name as the
first parameter and the canvas title as the second parameter. Next
we create a pointer to the random number generator class TRandom.
This class gives you useful methods to generate different types of
distributions. The constructor takes an input value as its seed. Next
we create two pointers for creating histograms, one for the gaussian
and another for the poisson. The constructors syntax is:
alias name, title, number of bins, lower limit and upper limit
Of course there are other constructors as well about which you
can learn from the ROOT documentation. Next, we fill the histgrams
with 10000 entries in a loop. In each loop the following functions:
h1->Fill(ran->Gaus(0,100));
h2->Fill(ran->Poisson(40));
return a random number obeying the appropriate functional distribution. The first one returns a random number from a gaussian
distribution with mean 0 and sigma 100. The second one returns a
random number for a poission distribution with mean 40. The next
thing is to plot these two histograms. Before that we need to subdivide the canvas which we do using:
c->Divide(2,1);
The canvas is now split into sub-pads with 2 columns and 1 row.
We will plot the gaussian in the first pad and therefore we need to
do:
c->cd(1);
which makes the first pad active and then:
h1->Draw();
The poission function will be plotted in the next sub-pad so we
need to set the second sub-pad as active by doing:
c->cd(2);
and then we draw the second histogram:
h2->Draw();
The fill color of the histograms are set by the functions:
h1->SetFillColor(4);
h2->SetFillColor(6);
The Draw function of the histogram can take arguments which
will change the plotting style.

Saving a canvas
Once you have plotted anything on a canvas, you can save them in a
familiar format like pdf, png, jpg etc.,.You can use the Print function
in the TCanvas class like this:
c->Print("canvas.pdf");

All root classes derive from the base

class called TObject and most of ROOT
objects will have an alias name which
can be used as an handle to access the
object.

85

to generate a pdf file. The canvas file menu also has a Print or Save
option which also helps you to do it.

Histograms and Graphs

Plotting histograms
In this chapter we shall see what we options exists for plotting histograms and graphs. These things are the most often used objects
and therefore it is necessary to have an idea of what can be done
with them.

Weighing your histogram

When you fill a histograms with values, the bin in which that value
falls in is incremented by one. Sometimes, you may want to give
more importance to some values and less importance to some values.
This can done by giving weights to the values.

int main()

//App instance

TApplication app("my_app",0,0);

// create canvas and divide

TCanvas *c = new TCanvas("myCanvas","My Canvas");

c->Divide(2,2);

// create histograms

9
10
11

TH1F *h1 = new TH1F("my_hist1","Gauss",50,-500,500);

TH1F *h2 = new TH1F("my_hist2","Poisson",50,-100,100);
TH1F *h3 = new TH1F("my_hist3","Weighted Gauss",50,-500,500);

13

TH1F *h4 = new TH1F("my_hist4","Weighted Poisson",50,-100,100);

// set styles

14

h1->SetFillColor(4);

15

h2->SetFillColor(6);

16

h3->SetFillColor(5);

17

h4->SetFillColor(7);

18

// create random number and fill histogram

12

19
20

TRandom *ran = new TRandom(450);

for(int ii=0;ii<10000;ii++){

88

21

h1->Fill(ran->Gaus(0,100));

22

h2->Fill(ran->Poisson(40));

23

h3->Fill(ran->Gaus(0,100),2);// weighted by 2

24

h4->Fill(ran->Poisson(40),2);// weighted by 2

25

26

// draw histograms

27

c->cd(1);

28

h1->Draw();

29

c->cd(2);

30

h2->Draw();

31

c->cd(3);

32

h3->Draw();

33

c->cd(4);

34

h4->Draw();

35

// setting the axis title for h1

36

h1->GetXaxis()->SetTitle("Gauss Values");

37

h1->GetYaxis()->SetTitle("Counts");

38

gPad->Update();

39

app.Run();

40

delete ran;

41

delete c;

42

return 0;

The above example illustrates the use of weights. As you can see, the
second argument of the Fill function gives the weight. By default, it is
set to one. Histograms h1 and h2 are filled with no weights and in h3
and h4 all values are filled with a weight of 2. The number of counts
(not the entries, see the y axis) in the histograms h3 and h4 (Refer
Fig.10) are therefore double the counts in the histograms h1 and h2.

Adding histograms
Just like you add ordinary numbers, you can also add histograms
too. In histograms, the number of counts in a particular bin in the
first histogram is added to the number of counts in the same bin in
the second histogram. One has to be careful here while setting the
bin size and the number of bins in the histograms. They should be
similar for both the histograms or you will end up getting wrong
results.
1

int main()

// app instance

histo2.cpp: Adding weights in a histogram.

The axis title of an histogram can be

set using the functions GetXaxis or
GetYaxis and then SetTitle(title), as
shown in the example.

89

Figure 10: Example of weighted histograms.

TApplication app("my_app",0,0);

//canvas creation

TCanvas *c = new TCanvas("myCanvas","My Canvas");

c->Divide(1,3);

// histogram creation

9
10
11

TH1F *h1 = new TH1F("my_hist1","Gauss1",50,-500,500);

TH1F *h2 = new TH1F("my_hist2","Gauss2",50,-500,500);
TH1F *h3 = new TH1F("my_hist3","Gauss1+Gauss2",50,-500,500);

12

// styles

13

h1->SetFillColor(4);

14

h2->SetFillColor(6);

15

h3->SetFillColor(5);

16

//fill random values

17
18

TRandom *ran = new TRandom(450);

for(int ii=0;ii<10000;ii++){

19

h1->Fill(ran->Gaus(0,100));

20

h2->Fill(ran->Gaus(0,100));

21

22

c->cd(1);

23

h1->Draw();

24

c->cd(2);

25

h2->Draw();

26

// add histograms

27

h3->Add(h1,h2);

28

c->cd(3);

90

29

h3->Draw();

30

gPad->Update();

31

app.Run();

32

delete ran;

33

delete c;

34

return 0;

35

}
histo3.cpp: Adding two histograms.
Figure 11: Adding two histograms

In figure 11, the top one is the first histogram and the middle
one is the second histogram, having the same bin size and range as
the first one. The bottom one is the result of these two histograms.
Notice that the counts in the third for each bin is equal to the counts
in the first plus the counts in the second.

Plotting multiple histograms on a single canvas

Sometimes you want to plot multiple histograms one on top of each
other. This can be achieved in two ways: one is using the SAME
option and second is using the THStack Class. Examples are given
below:
1
2

TH1F *h1 = new TH1F("my_hist1","Gauss",50,-500,500);

TH1F *h2 = new TH1F("my_hist2","Gauss",50,-500,500);

// styles

h1->SetFillColor(4);

91

h2->SetFillColor(6);

// fill histograms with random numbers

TRandom *ran = new TRandom(450);
for(int ii=0;ii<10000;ii++){

7
8

h1->Fill(ran->Gaus(0,50));

h2->Fill(ran->Gaus(0,100));

10
11

12

//draw

13

h1->Draw();

14

// draw over the last

h2->Draw("SAME");

15
16

first

The option SAMEis given as an argument to the Draw function of

the histogram. The first histogram drawn on a pad should not use
this option. The same option should be used carefully and only if a
histogram is already plotted on the canvas otherwise you are likely
to end up in a crash.
The SAMEoption is not always useful

histo4.cpp: Plotting multiple histograms

using the SAME options.

Figure 12: The "SAME" option for

drawing mulitple histograms.

especially in the management of the axis. The axis scale is managed

by the one that was plotted first and therefore any histograms plotted
thereafter will have to be within that range. To overcome this, you
can use the THStack class.
1
2
3

TH1F *h1 = new TH1F("my_hist1","Gauss",50,-500,500);

TH1F *h2 = new TH1F("my_hist2","Gauss",50,-500,500);
// styles

92

h1->SetFillColor(4);

h2->SetFillColor(6);

// fill histograms with random numbers

TRandom *ran = new TRandom(450);
for(int ii=0;ii<10000;ii++){

7
8

h1->Fill(ran->Gaus(0,50));

h2->Fill(ran->Gaus(0,100));

10

11
12
13

THStack hs("Stack","Stack of my histograms");

14

hs.Add(h1);

15

hs.Add(h2);
hs.Draw(); //try "nostack" option

16
17

For using the THStack class, you should have your histograms created and call the Add function giving the pointers as the argument as
shown in the above example.

histo5.cpp: Plotting multiple histograms

using the THStack class.

Two dimensional histograms

The histogram class TH1F is for one dimensional binning. To plot
a 2D histogram, you should use the TH2F class. It is almost similar to the 1D class except for the fact that the Fill function take two
arguments x and y. Of course, the weight can be given as the third
argument. The constructor here takes the binning information for the
yaxis also.

TH2F *h1 = new TH2F("my_hist1","Gauss",50,-100,100,50,-100,100);

// fill histograms with random numbers

1
2

TRandom *ran = new TRandom(450);

for(int ii=0;ii<50000;ii++){

3
4

h1->Fill(ran->Gaus(0,50),ran->Gaus(0,50));

5
6

// draw histogram

h1->Draw("SURF2");

histo6.cpp: Plotting a 2D histogram.

The Draw function takes the drawing options which are listed in the
ROOT documentation.

93

Figure 13: A 2D histogram using the

TH2F class.

Profile histograms
Profile histograms are a replacement for 2d histograms and are plotted using the TProfile class. The x axis contains the x bins and the y
axis shows, the mean of y and the corresponding error as against the
y bins in a 2d histogram.
TProfile *hprof = new TProfile("hprof","Profile of pz versus px",100,-4,4,0,20);
// fill histograms with random numbers

1
2

TRandom *ran = new TRandom(450);

float px,py,pz;

for(int ii=0;ii<50000;ii++){

//Return 2 numbers distributed following a gaussian with mean=0 and sigma=1.

ran->Rannor(px,py);
pz = px*px + py*py;
hprof->Fill(px,pz,1);

8
9
10

11

// draw histogram

12

hprof->Draw();

13

}
histo7.cpp: Plotting a profile histogram.

Filling is similar to a 2D histogram but the constructor is slightly

different from the TH2 class in that it does not take the binning information for the y axis as it is not required. The drawing options are
listed in the source code.

94

Figure 14: A profile histogram.

Fitting a histogram
The next part of histogram is the fitting. The fit panel can be accessed
by right clicking on the histogram data points in the canvas. To do it
in C++, you should first create the function to fit using the TF1 class.
There are some pre-defined functions like gauss, pol0, pol1, pol2,
expo, landau etc., which do not require any initialization. However
if you user your own function for fitting, you should initialize the
parameters.

1
2

TH1F *h1 = new TH1F("my_hist1","Gauss",50,-500,500);

// styles

h1->SetFillColor(4);

// fill histograms with random numbers

5
6

TRandom *ran = new TRandom(450);

for(int ii=0;ii<10000;ii++){
h1->Fill(ran->Gaus(0,50));

7
8

//draw

first

10

h1->Draw();

11

// change style to show fit parameters

12

gStyle->SetOptFit(1011);

13

// Fitting using pre-defined function

14

//h1->Fit("gaus");

15

// Fitting using user-defined functions

16

TF1 *func = new TF1("myGauss","[0]*exp(-0.5*((x-[1])/[2])**2)",-100,100);

95

17

// parameter names

18

func->SetParNames("Factor","Mean","Sigma");

19

//Set Parameters:

20

func->SetParameter(0,1);

21

func->SetParameter(1,0);

22

func->SetParameter(2,50);

23

h1->Fit(func);// use option "R" to restrict to a certain region for fitting

24

// fixing parameters: FixParamter, SetParamterLimits

//To get the paramter values programmatically:

25
26

cout<<"Parameter Factor: "<<func->GetParameter(0)<<endl;

27

cout<<"Parameter Mean: "<<func->GetParameter(1)<<endl;

28

cout<<"Parameter Sigma: "<<func->GetParameter(2)<<endl;

// update

29
30

}
histo9.cpp: Fitting a histogram.

In the above example, the user has defined his own gaussian function
using the TF1 class and initializes the parameters using the SetParameters function. After this, the Fit function of the histogram class
is called with the TF1 pointer as its argument. The option R can be
used to restrict the fitting range to the one set in the TF1 pointer. The
fitting results can be accessed by the GetParameter function.
Figure 15: Fitting a histogram.

Graphs
Graphs are plotted using the TGraph class. Unlike histograms, there
is no binning here. There are manys to fill a graph one of them is

96

to fill the x and y arrays and give them as a input to the TGraph
constructor.
1

int n =5000;

double x[n], y[n];

// create random number and fill graph

4
5

TRandom *ran = new TRandom(450);

for(int ii=0;ii<n;ii++){

x[ii]= ran->Gaus(0,50);

y[ii]=x[ii]*x[ii];

9
10
11
12
13
14

TGraph *graph = new TGraph(n,x,y);

// another way of filling individually
//graph->SetPoint(ii,xval,yval);
// draw histogram
graph->Draw("A*"); //options L,A,F,C,*,P,B
histo9.cpp:Plotting a graph.

Like the histogram classes, the TGraph class also takes drawing options.
Figure 16: Plotting a graph.

Plotting multiple graphs on a single canvas

Just like we plotted multiple histograms using the SAMEoption, we
can plot multiple graphs too using it. When doing so, the option a
should be added along with a plotting style (a*, al etc.,) to the

97

first graph that is drawn and the SAME option should be added
for the following graphs (lsame,*same etc.,). The a option
should be used only for the first graph.
Similar to the THStack, there is the TMultiGraph class which takes
care of all the axis management for plotting multiple graphs.
1

int n =5000;

double x[n], y[n];

double x1[n], y1[n];

// create random number and fill graph

TRandom *ran = new TRandom(450);
for(int ii=0;ii<n;ii++){

5
6

x[ii]= ran->Gaus(0,50);

y[ii]=x[ii]*x[ii];
x1[ii]= ran->Gaus(0,150);

8
9

y1[ii]=x1[ii];

10

11

TGraph *graph1
TGraph *graph2

12
13

= new TGraph(n,x,y);
= new TGraph(n,x1,y1);

14

graph1->SetLineColor(kBlue);

15

graph2->SetLineColor(kRed);

16

18

TMultiGraph *m = new TMultiGraph();

m->Add(graph1,"lp");

19

m->Add(graph2,"lp");

20

m->Draw("a");

17

The graphs are added to the TMultigraph object with their appropriate drawing options, however without the a option. The a option
has to be given only while drawing the multigraph.

Plotting a 2D graph
If you want to plot a surface or a 2d graph, then the TGraph2D class
in used. The filling can be done using an array like the TGraph class
or by using the SetPoint function:
1
2
3
4
5
6
7

//

TGraph2D *dt = new TGraph2D();

// create random number and fill values
TRandom *ran = new TRandom(450);
for(int ii=0;ii<n;ii++){
x = 2*6*(ran->Rndm(ii))-6;
y = 2*6*(ran->Rndm(ii))-6;
z = (sin(x)/x)*(sin(y)/y)+0.2;

98

dt->SetPoint(ii,x,y,z);

8
9

10
11
12

//

gStyle->SetPalette(1);
dt->Draw("surf1");
histo10.cpp:Plotting a 2D graph.
Figure 17: Plotting a 2D graph using
the TGraph2D class.

Fitting a graph
Fitting can be done following the same procedure illustrated for
histograms.

The TFile class for ROOT objects storage

Like we saved ASCII data in text file, there is a way to save ROOT
objects like graphs, histograms and trees. The class used for storing is called TFile. Let us begin with a small example of saving two
histograms into a TFile:
1
2
3
4
5

TCanvas *c = new TCanvas("myCanvas","My Canvas");

// create file
TFile *f = new TFile("rootfile.root","RECREATE","My root file");
// create histograms
TH1F *h1 = new TH1F("my_hist1","Gauss",50,-500,500);

TH1F *h2 = new TH1F("my_hist2","Gauss",50,-500,500);

// set style

h1->SetFillColor(4);

h2->SetFillColor(6);

10
11
12

// fill histograms
TRandom *ran = new TRandom(450);
for(int ii=0;ii<10000;ii++){

13

h1->Fill(ran->Gaus(0,50));

14

h2->Fill(ran->Gaus(0,100));

15

16

h1->Draw();

17

h2->Draw("SAME");

18

gPad->Update();

19

cout<<"Press anything to quit..."<<endl;

20

cin>>a;

21

// close file

22

f->Write();

23

f->Close();

As usual we instantiate the TApplication class and then create the

TCanvas for plotting our histograms. Now, before we create the histograms, we should create the TFile object. Whatever ROOT objects
created after the creation of TFile object will belong to that file. The
first parameter is the file name and the second parameter is an option

filewrite.cpp: A simple program to

write histograms to a root file.

100

which describes the access options and the third parameter is the title
of the file. The access option can be one of the following:
NEW or CREATE: create a new file and open for writing. If the file
already exists it is not opened.
RECREATE: Same as NEW or CREATE but if the file already exists
it is overwritten.
UPDATE: Open an existing file for writing. If no file exists, it is
created.
READ: Open file for reading (default).
Other options exist and can be seen in the ROOT documentation.
As shown in the example, after the file is created, the histogram
pointers are created and therefore these histograms belong to that
file. The histograms are filled with values. The objects are written to
the file using the Write() function and then closed using the Close()
function. Let us see how to read the contents of a TFile.

1
2
3
4
5

TCanvas *c = new TCanvas("myCanvas","My Canvas");

// create file
TFile *f = new TFile("rootfile.root","READ","My root file");
// read objects by type casting
TH1F *h1 = (TH1F *)f->Get("my_hist1");

h1->Draw();

gPad->Update();

cout<<"Press anything to quit..."<<endl;

cin>>a;

10

// close file

11

f->Close();

To read, you should open the file using the READ option. You
should also know the object name and the object type that you want
to read from. In the above example, we are reading from the file
created using the last file write example. The Get function in the
TFile takes the object name as the argument and then returns a TObject pointer. This TObject pointer should then be type-casted to the
type of the object you are reading like in the example above. The file
should be closed at the end of all operations.

Handling multiple files

Working with multiple files is a tricky thing to do in ROOT. To have
a simple picture, imagine a TFile to be like a directory. If you have
multiple files, you need to use the cd() function in the TFile class to
go to the appropriate file, just like you use the cd bash command.

fileread.cpp: A simple program to read

a histograms from a root file.

101

1
2

TFile *f1 = new TFile("rootfile1.root","RECREATE","My root file1");

TFile *f2 = new TFile("rootfile2.root","RECREATE","My root file2");

// create histograms

//f1->cd();

5
6
7

TH1F *h1 = new TH1F("my_hist1","Gauss",50,-500,500);

//f2->cd();
TH1F *h2 = new TH1F("my_hist2","Gauss",50,-500,500);

// set style

h1->SetFillColor(4);

10

h2->SetFillColor(6);

11

// fill histograms

12
13

TRandom *ran = new TRandom(450);

for(int ii=0;ii<10000;ii++){

14

h1->Fill(ran->Gaus(0,50));

15

h2->Fill(ran->Gaus(0,100));

16

17

h1->Draw();

18

h2->Draw("SAME");

19

gPad->Update();

20

cout<<"Press anything to quit..."<<endl;

21

cin>>a;

22

// close file

23

f1->Write();

24

f2->Write();

25

f1->Close();

26

f2->Close();

In the above example we have created two files f1 and f2. Let us see
what happens for the following cases in the above example:
1. f1->cd() and f2->cd() both commented: The histograms are written
in f2 and nothing is written in f1. This is because the file that is
created last will be the active one and all the objects will be written
into it. (Just like a directory).
2. f1->cd() is used and f2->cd() is commented: The histograms will
be written in f1 and nothing in f2. The command cd changes the
active file to f1.
3. f1->cd() is commented f2->cd() is used: The histograms will be
written to f2 and nothing in f1.
4. Both f1->cd() and f2->cd() are used: The first histogram will be
saved in f1 since f1->cd() is called before its creation and then
since f2->cd() is called, the second histogram is saved in f2.

multifilewrite.cpp: Program to explain

handling multiple files.

102

Viewing TFile contents in a ROOT terminal

Though you can open a file for reading using C++, there is a simpler
way to have an overview of the contents. Open the file using the
following command in a terminal:
root filename.root
where filename.root is the file you want view. The ROOT terminal
is automatically opened wherein you should type:
TBrowser th
which will open a browser showing a list of all the files, with a
special icon next to the file that you opened. Clicking on the file
will show you the contents of the file and clicking on objects like
histogram, graphs and trees (about which you will learn in the next
chapter) will plot them in a canvas.
Figure 18: The TBrowser window of
ROOT.

The TTree class: The spreadsheet for ROOT

Writing to a TTree
The last thing that is most important and often used ROOT in the
TTree class. This is nothing like a spreadsheet like container for
ROOT. If you want to analyse your data using ROOT, having them
in the form of branches to a TTree object is the way to go. If you go
through the functions in this class, you will find that it offers attractive data analysis and plotting features. There are good examples
in the ROOT tutorials that you can try out to get a feeling what this
class is all about. If you want to have a in-depth analysis of your
data, especially in HEP, try using TTree. Beginners easily fall into the
trap of writing their data in simple ASCII format and then try thousands of plotting application in the web without being convinced by
any of them. Understanding TTree is a bit difficult especially if you
are new to ROOT but learning it once will save you from the trouble
of hunting for various applications and libraries (that are already
existing in ROOT). Let us begin with an example to illustrate how a
TTree object is created and how it is used for writing data.
1
2
3

float E,P;
TFile *f1 = new TFile("treefile.root","RECREATE","My root file1");
TTree *tr = new TTree("Tree","My Tree");

tr->Branch("Energy",&E,"E/F");

tr->Branch("Momentum",&P,"P/F");

6
7

TRandom *ran = new TRandom(450);

for(int ii=0;ii<10000;ii++){
E = ran->Gaus(0,50);

P = E*E;
tr->Fill();

9
10
11

12

tr->Draw("E");

13

//tr->Draw("E","E<10");

14

//tr->Draw("E:P");

15

gPad->Update();

104

16

cout<<"Press anything to quit..."<<endl;

17

cin>>a;

18

f1->Write();

19

f1->Close();

The picture you should have while using a TTree is that of a spreadsheet. In a spreadsheet we have columns, which in TTree is called
as a branch, the rows are called as entries. Let us fill our TTree with
two parameters E and P and try plot it using the TTree Draw function. We will store our TTree in a TFile to view it offline. The TTree
is created using its constructor which takes the name and the title
of the tree as arguments. The tree name can be used to retrieve the
tree from the file. To create branches, we use the Branch function.
This function takes the branch name, the address of the variable from
where the data to be written will be taken and the variable name and
its type in a special format. We will simply use the variable name E
for E and P for P. Since both are floats, the syntax is:
E/F and P/F
The other variable types and the syntax to be used are as follows:
C : a character string
B : an 8 bit signed integer
b : an 8 bit unsigned integer
S : a 16 bit signed integer
s : a 16 bit unsigned integer
I : a 32 bit signed integer
i : a 32 bit unsigned integer
F : a 32 bit floating point
D : a 64 bit floating point
L : a 64 bit signed integer
l : a 64 bit unsigned integer
O : [the letter o, not a zero] a boolean
Once the branches are created, the next step is to fill our data
using the Fill command. When this command is called, the data that
is stored in E and P (the addressess of which were used to create the
branches), at that point of time are written as an entry in the tree.
Therefore the number of entries stored in the tree will be the number
of times the Fill command was called.
Now once the data is stored we can use the tree object to plot them
directly. For example, to plot the histogram of E you can call:
tr->Draw(E);
Note here E is the variable name that we gave for the parameter
E. If you had given the variable name for E as energy then you can
call plot using:
tr->Draw(energy);

tree.cpp: How to create branches and

write to a TTree

105

If you want to plot with some conditions, for example if you do

not want to plot the histogram of E when E<10, you would the cut
argument of the Draw function:
tr->Draw(E,E<10);
If you want to plot the correlation (graph) between E and P then:
tr->Draw(E:P)
You can also use a condition in the above case.

Reading from a TTree

Reading from a TTree is similar to reading other object like histograms from a TFile. First, you should know the tree name and then
you should use the Get function, then typecast the returned pointer
to TTree. Using the tree pointer, you can do all other things.
TCanvas *c = new TCanvas("myCanvas","My Canvas");
// create file

1
2

TFile *f1 = new TFile("treefile.root","READ","My root file1");

TTree *tr = (TTree*)f1->Get("Tree1");
tr->Draw("E");

//tr->Draw("E","E<10");

//tr->Draw("E:P");

gPad->Update();

cout<<"Press anything to quit..."<<endl;

3
4

10

cin>>a;

11

f1->Close();

To view the contents of a tree stored in a TFile, you can follow the
method mentioned in the last chapter to view the contents of TFile.

Using a C structure for a branch

Sometimes, you might for the sake of convenience store your data in
a structure. In such cases you have to use tree in a slightly different
way:
typedef struct {

1
2

int LayerNumber;

int StripNumber;

float noiseRate;
} RPCNOISERATE;

5
6

int main()

treeread.cpp: Reading a TTree from a

file.

106

// app instance

TApplication app("my_app",0,0);

10

RPCNOISERATE nr;

11

int a;

12

// create canvas*

13

TCanvas *c = new TCanvas("myCanvas","My Canvas");

14
15
16
17
18
19
20

// create file
TFile *f1 = new TFile("treefile.root","RECREATE","My root file1");
TTree *tr = new TTree("Tree","My Tree");
tr->Branch("RPCNR",&nr,"L/I:S/I:N/F");
TRandom *ran = new TRandom(450);
for(int ii=0;ii<12;ii++){
for(int jj=0;jj<32;jj++){

21
22

nr.LayerNumber = ii;

23

nr.StripNumber = jj;

24

nr.noiseRate = ran->Gaus(0,50);

25

tr->Fill();
}

26
27

28

tr->Draw("L:N");

29

//tr->Draw("E","E<10");

30

//tr->Draw("E:P");

31

gPad->Update();

32

cout<<"Press anything to quit..."<<endl;

33

cin>>a;

34

f1->Write();

35

f1->Close();

In the above example, we have created a structure called RPCNOISERATE which contains the variables LayerNumber, StripNumber and
the noiseRate. To create a branch this structure we call the Branch
function like this:
1

tr->Branch("RPCNR",&nr,"L/I:S/I:N/F");

RPCNR is the name of the branch and nr is the address of the

structure. The last argument is a string which is a concatenation of
the variable names and their respective types. In the structure the
first variable in LayerNumber for which we give the variable name L
and since its type is int we append /I to it and similarly we give the
variable name S to the second variable in the structure and append
/I to it as it is also an integer. The last variable is name N and a /F

treeStruct.cpp: Creating a tree branch

using a C structure.

107

is appended since its a float. All these string are concatenated with a
:. This is the syntax to be followed if a branch has to be made like a
structure. The filling part is similar to the previous example.

Getting the values at a particular entry

Often you want read the value of a particular entry in the tree. For
that, you need to first set the branch address appropriately as shown
below:
1
2
3
4
5
6
7

float E,P;
// create file
TFile *f1 = new TFile("treefile.root","READ","My root file1");
TTree *tr = (TTree*)f1->Get("Tree");
tr->SetBranchAddress("Energy",&E);
for(int ii=0;ii<tr->GetEntries();ii++){
tr->GetEntry(ii);

cout<<E<<endl;

In the above example, we read a tree named Tree from a file (the file
we created in the first example in this chapter) and we know that
it has a branch named Energy. To read the value of it, you have to
supply the address where the value will be read into. This is done
using the SetBranchAddress function wherein we supply the branch
name and the address. We now loop over all the entries and use the
GetEntry function which takes the entry number as the argument
and stores the value of that entry in the address given in the SetBranchAddress (i.e in E). The GetEntry function does the opposite of
Fill.

Skeleton Code Generation

Sometimes it would be tedious to set branch address for all the
branches because the structure of the tree is too complicated or you
might not know the branch structure of your tree. In that case you
should use the MakeClass function which generated a class for your
tree. It writes a .C and .h file for your tree which will set the branch
addresses for you. All you need to do is to include the .C and .h file
in you analysis code and access the tree using the constructor of the
class that you generated. This is the most efficient way to analyze a
tree.

getEntry.cpp: Reading the values at a

particular entry.