9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

yet another insignificant programming notes... | HOME

TABLE OF CONTENTS HIDE

A Tutorial on Data
Representation
Integers, Floatingpoint
Numbers, and Characters

1.Number Systems
1.1Decimal Base 10 Number System
1.2Binary Base 2 Number System
1.3Hexadecimal Base 16 Number System
1.4Conversion from Hexadecimal to Binary

1.5Conversion from Binary to Hexadecimal

1.6Conversion from Base r to Decimal Base 10
1.7Conversion from Decimal Base 10 to Base

1.8General Conversion between 2 Base Systems

1.9Exercises Number Systems Conversion

2.Computer Memory & Data Representation

3.Integer Representation
3.1nbit Unsigned Integers
3.2Signed Integers

1.Number Systems
Human beings use decimal base 10 and duodecimal base 12 number systems
for counting and measurements probably because we have 10 fingers and two

3.3nbit Sign Integers in SignMagnitude Repre

3.4nbit Sign Integers in 1's Complement Repre
3.5nbit Sign Integers in 2's Complement Repre
3.6Computers use 2's Complement Representa

big toes. Computers use binary base 2 number system, as they are made from
binary digital components known as transistors operating in two states on
and off. In computing, we also use hexadecimal base 16 or octal base 8
number systems, as a compact form for represent binary numbers.

3.7Range of nbit 2's Complement Signed Integ

3.8Decoding 2's Complement Numbers
3.9Big Endian vs. Little Endian
3.10Exercise Integer Representation

4.FloatingPoint Number Representation

1.1Decimal Base 10 Number System

Decimal number system has ten symbols: 0, 1, 2, 3, 4, 5, 6, 7, 8, and 9, called
digits. It uses positional notation. That is, the leastsignificant digit rightmost
digit is of the order of 10^0 units or ones, the second rightmost digit is of the
order of 10^1 tens, the third rightmost digit is of the order of 10^2
hundreds, and so on. For example,
735=710^2+310^1+510^0

We shall denote a decimal number with an optional suffix D if ambiguity arises.

1.2Binary Base 2 Number System

Binary number system has two symbols: 0 and 1, called bits. It is also a positional
notation, for example,
10110B=12^4+02^3+12^2+12^1+02^0

We shall denote a binary number with a suffix B. Some programming languages

denote binary numbers with prefix 0b e.g., 0b1001000, or prefix b with the bits
quoted e.g., b'10001111'.
A binary digit is called a bit. Eight bits is called a byte why 8bit unit? Probably
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

4.1IEEE754 32bit SinglePrecision FloatingPo

4.2Exercises Floatingpoint Numbers
4.3IEEE754 64bit DoublePrecision FloatingP
4.4More on FloatingPoint Representation

5.Character Encoding

5.17bit ASCII Code aka USASCII, ISO/IEC 646

5.28bit Latin1 aka ISO/IEC 88591
5.3Other 8bit Extension of USASCII ASCII Exte
5.4Unicode aka ISO/IEC 10646 Universal Chara
5.5UTF8 Unicode Transformation Format 8b

5.6UTF16 Unicode Transformation Format 16

5.7UTF32 Unicode Transformation Format 32
5.8Formats of MultiByte e.g., Unicode Text Fi
5.9Formats of Text Files
5.10Windows' CMD Codepage
5.11Chinese Character Sets
5.12Collating Sequences for Ranking Character
5.13For Java Programmers java.nio.Charse
5.14For Java Programmers char and
5.15Displaying Hex Values & Hex Editors

6.Summary Why Bother about Data Represen

6.1Exercises Data Representation
1/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

because 8=23.

1.3Hexadecimal Base 16 Number System

Hexadecimal number system uses 16 symbols: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, A, B, C, D, E, and F, called hex digits. It is a positional
notation, for example,
A3EH=1016^2+316^1+1416^0

We shall denote a hexadecimal number in short, hex with a suffix H. Some programming languages denote hex numbers
with prefix 0x e.g., 0x1A3C5F, or prefix x with hex digit quoted e.g., x'C3A4D98B'.
Each hexadecimal digit is also called a hex digit. Most programming languages accept lowercase 'a' to 'f' as well as
uppercase 'A' to 'F'.
Computers uses binary system in their internal operations, as they are built from binary digital electronic components.
However, writing or reading a long sequence of binary bits is cumbersome and errorprone. Hexadecimal system is used as
a compact form or shorthand for binary bits. Each hex digit is equivalent to 4 binary bits, i.e., shorthand for 4 bits, as
follows:
0H(0000B)
(0D)

1H(0001B)
(1D)

2H(0010B)
(2D)

3H(0011B)
(3D)

4H(0100B)
(4D)

5H(0101B)
(5D)

6H(0110B)
(6D)

7H(0111B)
(7D)

8H(1000B)
(8D)

9H(1001B)
(9D)

AH(1010B)
(10D)

BH(1011B)
(11D)

CH(1100B)
(12D)

DH(1101B)
(13D)

EH(1110B)
(14D)

FH(1111B)
(15D)

1.4Conversion from Hexadecimal to Binary

Replace each hex digit by the 4 equivalent bits, for examples,
A3C5H=1010001111000101B
102AH=0001000000101010B

1.5Conversion from Binary to Hexadecimal

Starting from the rightmost bit leastsignificant bit, replace each group of 4 bits by the equivalent hex digit pad the left
most bits with zero if necessary, for examples,
1001001010B=001001001010B=24AH
10001011001011B=0010001011001011B=22CBH

It is important to note that hexadecimal number provides a compact form or shorthand for representing binary bits.

1.6Conversion from Base r to Decimal Base 10

Given a ndigit base r number: dn1dn2dn3...d3d2d1d0 base r, the decimal equivalent is given by:
dn1r^(n1)+dn2r^(n2)+...+d1r^1+d0r^0

For examples,
A1C2H=1016^3+116^2+1216^1+2=41410(base10)
10110B=12^4+12^2+12^1=22(base10)
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

2/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

1.7Conversion from Decimal Base 10 to Base r

Use repeated division/remainder. For example,
Toconvert261Dtohexadecimal:
261/16=>quotient=16remainder=5
16/16=>quotient=1remainder=0
1/16=>quotient=0remainder=1(quotient=0stop)
Hence,261D=105H

The above procedure is actually applicable to conversion between any 2 base systems. For example,
Toconvert1023(base4)tobase3:
1023(base4)/3=>quotient=25Dremainder=0
25D/3=>quotient=8Dremainder=1
8D/3=>quotient=2Dremainder=2
2D/3=>quotient=0remainder=2(quotient=0stop)
Hence,1023(base4)=2210(base3)

1.8General Conversion between 2 Base Systems with Fractional Part

1. Separate the integral and the fractional parts.
2. For the integral part, divide by the target radix repeatably, and collect the ramainder in reverse order.
3. For the fractional part, multiply the fractional part by the target radix repeatably, and collect the integral part in the
same order.

Example 1:
Convert18.6875Dtobinary
IntegralPart=18D
18/2=>quotient=9remainder=0
9/2=>quotient=4remainder=1
4/2=>quotient=2remainder=0
2/2=>quotient=1remainder=0
1/2=>quotient=0remainder=1(quotient=0stop)
Hence,18D=10010B
FractionalPart=.6875D
.6875*2=1.375=>wholenumberis1
.375*2=0.75=>wholenumberis0
.75*2=1.5=>wholenumberis1
.5*2=1.0=>wholenumberis1
Hence.6875D=.1011B
Therefore,18.6875D=10010.1011B

Example 2:
Convert18.6875Dtohexadecimal
IntegralPart=18D
18/16=>quotient=1remainder=2
1/16=>quotient=0remainder=1(quotient=0stop)
Hence,18D=12H
FractionalPart=.6875D
.6875*16=11.0=>wholenumberis11D(BH)
Hence.6875D=.BH
Therefore,18.6875D=12.BH

1.9Exercises Number Systems Conversion

1. Convert the following decimal numbers into binary and hexadecimal numbers:
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

3/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

a. 108
b. 4848
c. 9000
2. Convert the following binary numbers into hexadecimal and decimal numbers:
a. 1000011000
b. 10000000
c. 101010101010
3. Convert the following hexadecimal numbers into binary and decimal numbers:
a. ABCDE
b. 1234
c. 80F
4. Convert the following decimal numbers into binary equivalent:
a. 19.25D
b. 123.456D

Answers: You could use the Windows' Calculator calc.exe to carry out number system conversion, by setting it to the
scientific mode. Run "calc" Select "View" menu Choose "Programmer" or "Scientific" mode.
1. 1101100B, 1001011110000B, 10001100101000B, 6CH, 12F0H, 2328H.
2. 218H, 80H, AAAH, 536D, 128D, 2730D.
3. 10101011110011011110B, 1001000110100B, 100000001111B, 703710D, 4660D, 2063D.
4. ??

2.Computer Memory & Data Representation

Computer uses a fixed number of bits to represent a piece of data, which could be a number, a character, or others. A nbit
storage location can represent up to 2^n distinct entities. For example, a 3bit memory location can hold one of these eight
binary patterns: 000, 001, 010, 011, 100, 101, 110, or 111. Hence, it can represent at most 8 distinct entities. You could use
them to represent numbers 0 to 7, numbers 8881 to 8888, characters 'A' to 'H', or up to 8 kinds of fruits like apple, orange,
banana; or up to 8 kinds of animals like lion, tiger, etc.
Integers, for example, can be represented in 8bit, 16bit, 32bit or 64bit. You, as the programmer, choose an appropriate
bitlength for your integers. Your choice will impose constraint on the range of integers that can be represented. Besides
the bitlength, an integer can be represented in various representation schemes, e.g., unsigned vs. signed integers. An 8bit
unsigned integer has a range of 0 to 255, while an 8bit signed integer has a range of 128 to 127 both representing 256
distinct numbers.
It is important to note that a computer memory location merely stores a binary pattern. It is entirely up to you, as the
programmer, to decide on how these patterns are to be interpreted. For example, the 8bit binary pattern "01000001B"
can be interpreted as an unsigned integer 65, or an ASCII character 'A', or some secret information known only to you. In
other words, you have to first decide how to represent a piece of data in a binary pattern before the binary patterns make
sense. The interpretation of binary pattern is called data representation or encoding. Furthermore, it is important that the
data representation schemes are agreedupon by all the parties, i.e., industrial standards need to be formulated and
straightly followed.
Once you decided on the data representation scheme, certain constraints, in particular, the precision and range will be
imposed. Hence, it is important to understand data representation to write correct and highperformance programs.

Rosette Stone and the Decipherment of Egyptian Hieroglyphs

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

4/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

Egyptian hieroglyphs nexttoleft were used by

the
ancient
Egyptians
since
4000BC.
Unfortunately, since 500AD, no one could longer
read the ancient Egyptian hieroglyphs, until the
rediscovery of the Rosette Stone in 1799 by
Napoleon's troop during Napoleon's Egyptian
invasion near the town of Rashid Rosetta in the
Nile Delta.
The Rosetta Stone left is inscribed with a decree
in 196BC on behalf of King Ptolemy V. The
decree appears in three scripts: the upper text is
Ancient Egyptian hieroglyphs, the middle portion
Demotic script, and the lowest Ancient Greek. Because it presents essentially the
same text in all three scripts, and Ancient Greek could still be understood, it
provided the key to the decipherment of the Egyptian hieroglyphs.
The moral of the story is unless you know the encoding scheme, there is no way that you can decode the data.
Reference and images: Wikipedia.

3.Integer Representation
Integers are whole numbers or fixedpoint numbers with the radix point fixed after the leastsignificant bit. They are contrast
to real numbers or floatingpoint numbers, where the position of the radix point varies. It is important to take note that
integers and floatingpoint numbers are treated differently in computers. They have different representation and are
processed differently e.g., floatingpoint numbers are processed in a socalled floatingpoint processor. Floatingpoint
numbers will be discussed later.
Computers use a fixed number of bits to represent an integer. The commonlyused bitlengths for integers are 8bit, 16bit,
32bit or 64bit. Besides bitlengths, there are two representation schemes for integers:
1. Unsigned Integers: can represent zero and positive integers.
2. Signed Integers: can represent zero, positive and negative integers. Three representation schemes had been proposed
for signed integers:
a. SignMagnitude representation
b. 1's Complement representation
c. 2's Complement representation
You, as the programmer, need to decide on the bitlength and representation scheme for your integers, depending on
your application's requirements. Suppose that you need a counter for counting a small quantity from 0 up to 200, you
might choose the 8bit unsigned integer scheme as there is no negative numbers involved.

3.1nbit Unsigned Integers

Unsigned integers can represent zero and positive integers, but not negative integers. The value of an unsigned integer is
interpreted as "the magnitude of its underlying binary pattern".

Example 1: Suppose that n=8 and the binary pattern is 0100 0001B, the value of this unsigned integer is 12^0 +
12^6=65D.

Example 2: Suppose that n=16 and the binary pattern is0001000000001000B, the value of this unsigned integer is
12^3+12^12=4104D.

Example 3: Suppose that n=16 and the binary pattern is0000000000000000B, the value of this unsigned integer is
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

5/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

0.
An nbit pattern can represent 2^n distinct integers. An nbit unsigned integer can represent integers from 0 to (2^n)1,
as tabulated below:

Minimum

Maximum

(2^8)1(=255)

16

(2^16)1(=65,535)

32

(2^32)1(=4,294,967,295)(9+digits)

64

(2^64)1(=18,446,744,073,709,551,615)
(19+digits)

3.2Signed Integers
Signed integers can represent zero, positive integers, as well as negative integers. Three representation schemes are
available for signed integers:
1. SignMagnitude representation
2. 1's Complement representation
3. 2's Complement representation
In all the above three schemes, the mostsignificant bit msb is called the sign bit. The sign bit is used to represent the sign
of the integer with 0 for positive integers and 1 for negative integers. The magnitude of the integer, however, is
interpreted differently in different schemes.

3.3nbit Sign Integers in SignMagnitude Representation

In signmagnitude representation:
The mostsignificant bit msb is the sign bit, with value of 0 representing positive integer and 1 representing negative
integer.
The remaining n1 bits represents the magnitude absolute value of the integer. The absolute value of the integer is
interpreted as "the magnitude of the n1bit binary pattern".

Example 1 : Suppose that n=8 and the binary representation is01000001B.

Sign bit is 0 positive
Absolute value is 1000001B=65D
Hence, the integer is +65D

Example 2 : Suppose that n=8 and the binary representation is10000001B.

Sign bit is 1 negative
Absolute value is 0000001B=1D
Hence, the integer is 1D

Example 3 : Suppose that n=8 and the binary representation is00000000B.

Sign bit is 0 positive
Absolute value is 0000000B=0D
Hence, the integer is +0D

Example 4 : Suppose that n=8 and the binary representation is10000000B.

Sign bit is 1 negative
Absolute value is 0000000B=0D
Hence, the integer is 0D

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

6/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

The drawbacks of signmagnitude representation are:

1. There are two representations 00000000B and 10000000B for the number zero, which could lead to inefficiency
and confusion.
2. Positive and negative integers need to be processed separately.

3.4nbit Sign Integers in 1's Complement Representation

In 1's complement representation:
Again, the most significant bit msb is the sign bit, with value of 0 representing positive integers and 1 representing
negative integers.
The remaining n1 bits represents the magnitude of the integer, as follows:
for positive integers, the absolute value of the integer is equal to "the magnitude of the n1bit binary pattern".
for negative integers, the absolute value of the integer is equal to "the magnitude of the complement inverse of
the n1bit binary pattern" hence called 1's complement.

Example 1 : Suppose that n=8 and the binary representation01000001B.

Sign bit is 0 positive
Absolute value is 1000001B=65D
Hence, the integer is +65D

Example 2 : Suppose that n=8 and the binary representation10000001B.

Sign bit is 1 negative
Absolute value is the complement of 0000001B, i.e., 1111110B=126D
Hence, the integer is 126D

Example 3 : Suppose that n=8 and the binary representation00000000B.

Sign bit is 0 positive
Absolute value is 0000000B=0D
Hence, the integer is +0D

Example 4 : Suppose that n=8 and the binary representation11111111B.

Sign bit is 1 negative
Absolute value is the complement of 1111111B, i.e., 0000000B=0D
Hence, the integer is 0D

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

7/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

Again, the drawbacks are:

1. There are two representations 00000000B and 11111111B for zero.
2. The positive integers and negative integers need to be processed separately.

3.5nbit Sign Integers in 2's Complement Representation

In 2's complement representation:
Again, the most significant bit msb is the sign bit, with value of 0 representing positive integers and 1 representing
negative integers.
The remaining n1 bits represents the magnitude of the integer, as follows:
for positive integers, the absolute value of the integer is equal to "the magnitude of the n1bit binary pattern".
for negative integers, the absolute value of the integer is equal to "the magnitude of the complement of the n1
bit binary pattern plus one" hence called 2's complement.

Example 1 : Suppose that n=8 and the binary representation01000001B.

Sign bit is 0 positive
Absolute value is 1000001B=65D
Hence, the integer is +65D

Example 2 : Suppose that n=8 and the binary representation10000001B.

Sign bit is 1 negative
Absolute value is the complement of 0000001B plus 1, i.e., 1111110B+1B=127D
Hence, the integer is 127D

Example 3 : Suppose that n=8 and the binary representation00000000B.

Sign bit is 0 positive
Absolute value is 0000000B=0D
Hence, the integer is +0D

Example 4 : Suppose that n=8 and the binary representation11111111B.

Sign bit is 1 negative
Absolute value is the complement of 1111111B plus 1, i.e., 0000000B+1B=1D
Hence, the integer is 1D

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

8/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

3.6Computers use 2's Complement Representation for Signed Integers

We have discussed three representations for signed integers: signedmagnitude, 1's complement and 2's complement.
Computers use 2's complement in representing signed integers. This is because:
1. There is only one representation for the number zero in 2's complement, instead of two representations in sign
magnitude and 1's complement.
2. Positive and negative integers can be treated together in addition and subtraction. Subtraction can be carried out
using the "addition logic".

Example 1: Addition of Two Positive Integers: Suppose thatn=8,65D+5D=70D

65D01000001B
5D00000101B(+
01000110B70D(OK)

Example 2: Subtraction is treated as Addition of a Positive and a Negative Integers: Suppose that
n=8,5D5D=65D+(5D)=60D
65D01000001B
5D11111011B(+
00111100B60D(discardcarryOK)

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html
Example 3: Addition of Two Negative Integers: Suppose that

9/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

Example 3: Addition of Two Negative Integers: Suppose that n=8, 65D 5D = (65D) + (5D) =
70D
65D10111111B
5D11111011B(+
10111010B70D(discardcarryOK)

Because of the fixed precision i.e., fixed number of bits, an nbit 2's complement signed integer has a certain range. For
example, for n=8, the range of 2's complement signed integers is 128 to +127. During addition and subtraction, it is
important to check whether the result exceeds this range, in other words, whether overflow or underflow has occurred.

Example 4: Overflow: Suppose thatn=8,127D+2D=129D overflow beyond the range

127D01111111B
2D00000010B(+
10000001B127D(wrong)

Example 5: Underflow: Suppose thatn=8,125D5D=130D underflow below the range

125D10000011B
5D11111011B(+
01111110B+126D(wrong)

The following diagram explains how the 2's complement works. By rearranging the number line, values from 128 to +127
are represented contiguously by ignoring the carry bit.

3.7Range of nbit 2's Complement Signed Integers

An nbit 2's complement signed integer can represent integers from 2^(n1) to +2^(n1)1, as tabulated. Take note
that the scheme can represent all the integers within the range, without any gap. In other words, there is no missing
integers within the supported range.

minimum

maximum

(2^7)(=128)

+(2^7)1(=+127)

16

(2^15)(=32,768)

+(2^15)1(=+32,767)

32

(2^31)(=2,147,483,648)

+(2^31)1(=+2,147,483,647)(9+digits)

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

10/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

64

(2^63)

+(2^63)1(=+9,223,372,036,854,775,807)

(=9,223,372,036,854,775,808)

(18+digits)

3.8Decoding 2's Complement Numbers

1. Check the sign bit denoted as S.
2. If S=0, the number is positive and its absolute value is the binary value of the remaining n1 bits.
3. If S=1, the number is negative. you could "invert the n1 bits and plus 1" to get the absolute value of negative
number.
Alternatively, you could scan the remaining n1 bits from the right leastsignificant bit. Look for the first occurrence
of 1. Flip all the bits to the left of that first occurrence of 1. The flipped pattern gives the absolute value. For example,
n=8,bitpattern=11000100B
S=1negative
Scanningfromtherightandflipallthebitstotheleftofthefirstoccurrenceof10111100B=60D
Hence,thevalueis60D

3.9Big Endian vs. Little Endian

Modern computers store one byte of data in each memory address or location, i.e., byte addressable memory. An 32bit
integer is, therefore, stored in 4 memory addresses.
The term"Endian" refers to the order of storing bytes in computer memory. In "Big Endian" scheme, the most significant
byte is stored first in the lowest memory address or big in first, while "Little Endian" stores the least significant bytes in
the lowest memory address.
For example, the 32bit integer 12345678H 221505317010 is stored as 12H 34H 56H 78H in big endian; and 78H 56H 34H
12H in little endian. An 16bit integer 00H 01H is interpreted as 0001H in big endian, and 0100H as little endian.

3.10Exercise Integer Representation

1. What are the ranges of 8bit, 16bit, 32bit and 64bit integer, in "unsigned" and "signed" representation?
2. Give the value of 88, 0, 1, 127, and 255 in8bit unsigned representation.
3. Give the value of +88, 88 , 1, 0, +1, 128, and +127 in 8bit 2's complement signed representation.
4. Give the value of +88, 88 , 1, 0, +1, 127, and +127 in 8bit signmagnitude representation.
5. Give the value of +88, 88 , 1, 0, +1, 127 and +127 in 8bit 1's complement representation.
6. [TODO] more.

Answers
1. The range of unsigned nbit integers is [0,2^n1]. The range of nbit 2's complement signed integer is [2^(n
1),+2^(n1)1];
2. 88(01011000), 0(00000000), 1(00000001), 127(01111111), 255(11111111).
3. +88(01011000), 88(10101000), 1(11111111), 0(00000000), +1(00000001), 128 (1000 0000),
+127(01111111).
4. +88(01011000), 88(11011000), 1(10000001), 0(00000000or10000000), +1(00000001), 127
(11111111), +127(01111111).
5. +88(01011000), 88(10100111), 1(11111110), 0(00000000or11111111), +1(00000001), 127
(10000000), +127(01111111).

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

11/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

4.FloatingPoint Number Representation

A floatingpoint number or real number can represent a very large 1.2310^88 or a very small 1.2310^88 value. It
could also represent very large negative number 1.2310^88 and very small negative number 1.2310^88, as well as
zero, as illustrated:

A floatingpoint number is typically expressed in the scientific notation, with a fraction F, and an exponent E of a certain
radix r, in the form of Fr^E. Decimal numbers use radix of 10 F10^E; while binary numbers use radix of 2 F2^E.
Representation of floating point number is not unique. For example, the number 55.66 can be represented as
5.56610^1, 0.556610^2, 0.0556610^3, and so on. The fractional part can be normalized. In the normalized form, there
is only a single nonzero digit before the radix point. For example, decimal number 123.4567 can be normalized as
1.23456710^2; binary number 1010.1011B can be normalized as 1.0101011B2^3.
It is important to note that floatingpoint numbers suffer from loss of precision when represented with a fixed number of
bits e.g., 32bit or 64bit. This is because there are infinite number of real numbers even within a small range of says 0.0
to 0.1. On the other hand, a nbit binary pattern can represent a finite 2^n distinct numbers. Hence, not all the real
numbers can be represented. The nearest approximation will be used instead, resulted in loss of accuracy.
It is also important to note that floating number arithmetic is very much less efficient than integer arithmetic. It could be
speed up with a socalled dedicated floatingpoint coprocessor. Hence, use integers if your application does not require
floatingpoint numbers.
In computers, floatingpoint numbers are represented in scientific notation of fraction F and exponent E with a radix of 2,
in the form of F2^E. Both E and F can be positive as well as negative. Modern computers adopt IEEE 754 standard for
representing floatingpoint numbers. There are two representation schemes: 32bit singleprecision and 64bit double
precision.

4.1IEEE754 32bit SinglePrecision FloatingPoint Numbers

In 32bit singleprecision floatingpoint representation:
The most significant bit is the sign bit S, with 0 for positive numbers and 1 for negative numbers.
The following 8 bits represent exponent E.
The remaining 23 bits represents fraction F.

Normalized Form
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

12/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

Let's illustrate with an example, suppose that the 32bit pattern is 11000000101100000000000000000000, with:
S=1
E=10000001
F=01100000000000000000000
In the normalized form, the actual fraction is normalized with an implicit leading 1 in the form of 1.F. In this example, the
actual fraction is 1.01100000000000000000000=1+12^2+12^3=1.375D.
The sign bit represents the sign of the number, with S=0 for positive and S=1 for negative number. In this example with
S=1, this is a negative number, i.e., 1.375D.
In normalized form, the actual exponent is E127 socalled excess127 or bias127. This is because we need to represent
both positive and negative exponent. With an 8bit E, ranging from 0 to 255, the excess127 scheme could provide actual
exponent of 127 to 128. In this example, E127=129127=2D.
Hence, the number represented is 1.3752^2=5.5D.

DeNormalized Form
Normalized form has a serious problem, with an implicit leading 1 for the fraction, it cannot represent the number zero!
Convince yourself on this!
Denormalized form was devised to represent zero and other numbers.
For E=0, the numbers are in the denormalized form. An implicit leading 0 instead of 1 is used for the fraction; and the
actual exponent is always 126. Hence, the number zero can be represented with E=0 and F=0 because 0.02^126=0.
We can also represent very small positive and negative numbers in denormalized form with E=0. For example, if S=1, E=0,
and F=01100000000000000000000. The actual fraction is 0.011=12^2+12^3=0.375D. Since S=1, it is a negative
number. With E=0, the actual exponent is 126. Hence the number is 0.3752^126 = 4.410^39, which is an
extremely small negative number close to zero.

Summary
In summary, the value N is calculated as follows:
For 1E254,N=(1)^S1.F2^(E127). These numbers are in the socalled normalized form. The sign
bit represents the sign of the number. Fractional part 1.F are normalized with an implicit leading 1. The exponent is
bias or in excess of 127, so as to represent both positive and negative exponent. The range of exponent is 126 to
+127.
For E=0,N=(1)^S0.F2^(126). These numbers are in the socalled denormalized form. The exponent
of 2^126 evaluates to a very small number. Denormalized form is needed to represent zero with F=0 and E=0. It can
also represents very small positive and negative number close to zero.
For E=255, it represents special values, such as INF positive and negative infinity and NaN not a number. This is
beyond the scope of this article.

Example 1: Suppose that IEEE754 32bit floatingpoint representation pattern is 010000000110000000000000

00000000.
SignbitS=0positivenumber
E=10000000B=128D(innormalizedform)
Fractionis1.11B(withanimplicitleading1)=1+12^1+12^2=1.75D
Thenumberis+1.752^(128127)=+3.5D

Example 2: Suppose that IEEE754 32bit floatingpoint representation pattern is 101111110100000000000000

00000000.
SignbitS=1negativenumber
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

13/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

E=01111110B=126D(innormalizedform)
Fractionis1.1B(withanimplicitleading1)=1+2^1=1.5D
Thenumberis1.52^(126127)=0.75D

Example 3: Suppose that IEEE754 32bit floatingpoint representation pattern is 101111110000000000000000

00000001.
SignbitS=1negativenumber
E=01111110B=126D(innormalizedform)
Fractionis1.00000000000000000000001B(withanimplicitleading1)=1+2^23
Thenumberis(1+2^23)2^(126127)=0.500000059604644775390625(maynotbeexactindecimal!)

Example 4 DeNormalized Form: Suppose that IEEE754 32bit floatingpoint representation pattern is 1
0000000000000000000000000000001.
SignbitS=1negativenumber
E=0(indenormalizedform)
Fractionis0.00000000000000000000001B(withanimplicitleading0)=12^23
Thenumberis2^232^(126)=2(149)1.410^45

4.2Exercises Floatingpoint Numbers

1. Compute the largest and smallest positive numbers that can be represented in the 32bit normalized form.
2. Compute the largest and smallest negative numbers can be represented in the 32bit normalized form.
3. Repeat 1 for the 32bit denormalized form.
4. Repeat 2 for the 32bit denormalized form.

Hints:
1. Largest positive number: S=0, E=11111110(254), F=11111111111111111111111.
Smallest positive number: S=0, E=000000001(1), F=00000000000000000000000.
2. Same as above, but S=1.
3. Largest positive number: S=0, E=0, F=11111111111111111111111.
Smallest positive number: S=0, E=0, F=00000000000000000000001.
4. Same as above, but S=1.

Notes For Java Users

You can use JDK methods Float.intBitsToFloat(intbits) or Double.longBitsToDouble(longbits) to create a
singleprecision 32bit float or doubleprecision 64bit double with the specific bit patterns, and print their values. For
examples,
System.out.println(Float.intBitsToFloat(0x7fffff));
System.out.println(Double.longBitsToDouble(0x1fffffffffffffL));

4.3IEEE754 64bit DoublePrecision FloatingPoint Numbers

The representation scheme for 64bit doubleprecision is similar to the 32bit singleprecision:
The most significant bit is the sign bit S, with 0 for positive numbers and 1 for negative numbers.
The following 11 bits represent exponent E.
The remaining 52 bits represents fraction F.

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

14/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

The value N is calculated as follows:

Normalized form: For 1E2046,N=(1)^S1.F2^(E1023).
Denormalized form: For E=0,N=(1)^S0.F2^(1022). These are in the denormalized form.
For E=2047, N represents special values, such as INF infinity, NaN not a number.

4.4More on FloatingPoint Representation

There are three parts in the floatingpoint representation:
The sign bit S is selfexplanatory 0 for positive numbers and 1 for negative numbers.
For the exponent E, a socalled bias or excess is applied so as to represent both positive and negative exponent. The
bias is set at half of the range. For single precision with an 8bit exponent, the bias is 127 or excess127. For double
precision with a 11bit exponent, the bias is 1023 or excess1023.
The fraction F also called the mantissa or significand is composed of an implicit leading bit before the radix point
and the fractional bits after the radix point. The leading bit for normalized numbers is 1; while the leading bit for
denormalized numbers is 0.

Normalized FloatingPoint Numbers

In normalized form, the radix point is placed after the first nonzero digit, e,g., 9.8765D10^23D, 1.001011B2^11B. For
binary number, the leading bit is always 1, and need not be represented explicitly this saves 1 bit of storage.
In IEEE 754's normalized form:
For singleprecision, 1 E 254 with excess of 127. Hence, the actual exponent is from 126 to +127. Negative
exponents are used to represent small numbers < 1.0; while positive exponents are used to represent large numbers
> 1.0.
N=(1)^S1.F2^(E127)
For doubleprecision, 1E2046 with excess of 1023. The actual exponent is from 1022 to +1023, and
N=(1)^S1.F2^(E1023)
Take note that nbit pattern has a finite number of combinations =2^n, which could represent finite distinct numbers. It is
not possible to represent the infinite numbers in the real axis even a small range says 0.0 to 1.0 has infinite numbers. That
is, not all floatingpoint numbers can be accurately represented. Instead, the closest approximation is used, which leads to
loss of accuracy.
The minimum and maximum normalized floatingpoint numbers are:

Precision
Single

Double

NormalizedN(min)

NormalizedN(max)

00800000H

7F7FFFFFH

000000001
00000000000000000000000B

01111111000000000000000000000000B
E=254,F=0

E=1,F=0
N(min)=1.0B2^126
(1.1754943510^38)

N(max)=1.1...1B2^127=(2
2^23)2^127
(3.402823510^38)

0010000000000000H
N(min)=1.0B2^1022

7FEFFFFFFFFFFFFFH

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

N(max)=1.1...1B2^1023=(2
15/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

(2.2250738585072014
10^308)

2^52)2^1023
(1.797693134862315710^308)

Denormalized FloatingPoint Numbers

If E=0, but the fraction is nonzero, then the value is in denormalized form, and a leading bit of 0 is assumed, as follows:
For singleprecision, E=0,
N=(1)^S0.F2^(126)
For doubleprecision, E=0,
N=(1)^S0.F2^(1022)
Denormalized form can represent very small numbers closed to zero, and zero, which cannot be represented in normalized
form, as shown in the above figure.
The minimum and maximum of denormalized floatingpoint numbers are:

Precision
Single

DenormalizedD(min)

DenormalizedD(max)

00000001H

007FFFFFH

00000000000000000000000000000001B
E=0,F=00000000000000000000001B

000000000
11111111111111111111111B

D(min)=0.0...12^126=12^23
2^126=2^149
(1.410^45)

E=0,F=
11111111111111111111111B
D(max)=0.1...12^126=(1
2^23)2^126
(1.175494210^38)

Double

0000000000000001H
D(min)=0.0...12^1022=12^52
2^1022=2^1074

001FFFFFFFFFFFFFH
D(max)=0.1...12^1022=
(12^52)2^1022

(4.910^324)

(4.450147717014402310^308)

Special Values
Zero: Zero cannot be represented in the normalized form, and must be represented in denormalized form with E=0 and
F=0. There are two representations for zero: +0 with S=0 and 0 with S=1.
Infinity: The value of +infinity e.g., 1/0 and infinity e.g., 1/0 are represented with an exponent of all 1's E=255 for
singleprecision and E=2047 for doubleprecision, F=0, and S=0 for +INF and S=1 for INF.
Not a Number NaN: NaN denotes a value that cannot be represented as real number e.g. 0/0. NaN is represented with
Exponent of all 1's E=255 for singleprecision and E=2047 for doubleprecision and any nonzero fraction.

5.Character Encoding
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

16/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

In computer memory, character are "encoded" or "represented" using a chosen "character encoding schemes" aka
"character set", "charset", "character map", or "code page".
For example, in ASCII as well as Latin1, Unicode, and many other character sets:
code numbers 65D(41H) to 90D(5AH) represents 'A' to 'Z', respectively.
code numbers 97D(61H) to 122D(7AH) represents 'a' to 'z', respectively.
code numbers 48D(30H) to 57D(39H) represents '0' to '9', respectively.
It is important to note that the representation scheme must be known before a binary pattern can be interpreted. E.g., the
8bit pattern "01000010B" could represent anything under the sun known only to the person encoded it.
The most commonlyused character encoding schemes are: 7bit ASCII ISO/IEC 646 and 8bit Latinx ISO/IEC 8859x for
western european characters, and Unicode ISO/IEC 10646 for internationalization i18n.
A 7bit encoding scheme such as ASCII can represent 128 characters and symbols. An 8bit character encoding scheme
such as Latinx can represent 256 characters and symbols; whereas a 16bit encoding scheme such as Unicode UCS2
can represents 65,536 characters and symbols.

5.17bit ASCII Code aka USASCII, ISO/IEC 646, ITUT T.50

ASCII American Standard Code for Information Interchange is one of the earlier character coding schemes.
ASCII is originally a 7bit code. It has been extended to 8bit to better utilize the 8bit computer memory organization.
The 8thbit was originally used for parity check in the early computers.
Code numbers 32D(20H) to 126D(7EH) are printable displayable characters as tabulated:

Hex

SP

"

&

'

<

>

Code number 32D(20H) is the blank or space character.

'0' to '9': 30H39H (0011 0001B to 0011 1001B) or (0011 xxxxB where xxxx is the equivalent integer
value)
'A' to 'Z': 41H5AH(01010001Bto01011010B) or (010xxxxxB). 'A' to 'Z' are continuous without gap.
'a' to 'z': 61H7AH(01100001Bto01111010B) or (011xxxxxB). 'A' to 'Z' are also continuous without
gap. However, there is a gap between uppercase and lowercase letters. To convert between upper and lowercase,
flip the value of bit5.
Code numbers 0D(00H) to 31D(1FH), and 127D(7FH) are special control characters, which are nonprintable non
displayable, as tabulated below. Many of these characters were used in the early days for transmission control e.g.,
STX, ETX and printer control e.g., FormFeed, which are now obsolete. The remaining meaningful codes today are:
09H for Tab '\t'.
0AH for LineFeed or newline LF, '\n' and 0DH for CarriageReturn CR, 'r', which are used as line delimiter aka
line separator, endofline for text files. There is unfortunately no standard for line delimiter: Unixes and Mac use
0AH "\n", Windows use 0D0AH "\r\n". Programming languages such as C/C++/Java which was created on
Unix use 0AH "\n".

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

17/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

In programming languages such as C/C++/Java, linefeed 0AH is denoted as '\n', carriagereturn 0DH as '\r',
tab 09H as '\t'.

DEC

HEX

Meaning

DEC

HEX

00

Meaning

NUL

Null

17

11

DC1

Device
Control1

01

SOH

Startof
Heading

18

12

DC2

Device
Control2

02

STX

StartofText

19

13

DC3

Device
Control3

03

ETX

EndofText

20

14

DC4

Device
Control4

04

EOT

Endof

21

15

NAK

Negative

Transmission

Ack.

05

ENQ

Enquiry

22

16

SYN

Sync.Idle

06

ACK

Acknowledgment

23

17

ETB

Endof
Transmission

07

BEL

Bell

24

18

CAN

Cancel

08

BS

BackSpace

25

19

EM

Endof

'\b'

Medium

09

HT

HorizontalTab
'\t'

26

1A

SUB

Substitute

10

0A

LF

LineFeed'\n'

27

1B

ESC

Escape

11

0B

VT

VerticalFeed

28

1C

IS4

File
Separator

12

0C

FF

FormFeed'f'

29

1D

IS3

Group
Separator

13

0D

CR

Carriage

30

1E

IS2

Return'\r'

Record
Separator

14

0E

SO

ShiftOut

31

1F

IS1

Unit
Separator

15

0F

SI

ShiftIn

16

10

DLE

Datalink

127

7F

DEL

Delete

Escape

5.28bit Latin1 aka ISO/IEC 88591

ISO/IEC8859 is a collection of 8bit character encoding standards for the western languages.
ISO/IEC 88591, aka Latin alphabet No. 1, or Latin1 in short, is the most commonlyused encoding scheme for western
european languages. It has 191 printable characters from the latin script, which covers languages like English, German,
Italian, Portuguese and Spanish. Latin1 is backward compatible with the 7bit USASCII code. That is, the first 128
characters in Latin1 code numbers 0 to 127 7FH, is the same as USASCII. Code numbers 128 80H to 159 9FH are not
assigned. Code numbers 160 A0H to 255 FFH are given as follows:

Hex

NBSP

SHY

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

18/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

ISO/IEC8859 has 16 parts. Besides the most commonlyused Part 1, Part 2 is meant for Central European Polish, Czech,
Hungarian, etc, Part 3 for South European Turkish, etc, Part 4 for North European Estonian, Latvian, etc, Part 5 for
Cyrillic, Part 6 for Arabic, Part 7 for Greek, Part 8 for Hebrew, Part 9 for Turkish, Part 10 for Nordic, Part 11 for Thai, Part 12
was abandon, Part 13 for Baltic Rim, Part 14 for Celtic, Part 15 for French, Finnish, etc. Part 16 for SouthEastern European.

5.3Other 8bit Extension of USASCII ASCII Extensions

Beside the standardized ISO8859x, there are many 8bit ASCII extensions, which are not compatible with each others.
ANSI American National Standards Institute aka Windows1252, or Windows Codepage 1252: for Latin alphabets used
in the legacy DOS/Windows systems. It is a superset of ISO88591 with code numbers 128 80H to 159 9FH assigned to
displayable characters, such as "smart" singlequotes and doublequotes. A common problem in web browsers is that all
the quotes and apostrophes produced by "smart quotes" in some Microsoft software were replaced with question marks
or some strange symbols. It it because the document is labeled as ISO88591 instead of Windows1252, where these
code numbers are undefined. Most modern browsers and email clients treat charset ISO88591 as Windows1252 in
order to accommodate such mislabeling.

Hex

EBCDIC Extended Binary Coded Decimal Interchange Code: Used in the early IBM computers.

5.4Unicode aka ISO/IEC 10646 Universal Character Set

Before Unicode, no single character encoding scheme could represent characters in all languages. For example, western
european uses several encoding schemes in the ISO8859x family. Even a single language like Chinese has a few
encoding schemes GB2312/GBK, BIG5. Many encoding schemes are in conflict of each other, i.e., the same code number
is assigned to different characters.
Unicode aims to provide a standard character encoding scheme, which is universal, efficient, uniform and unambiguous.
Unicode standard is maintained by a nonprofit organization called the Unicode Consortium @ www.unicode.org.
Unicode is an ISO/IEC standard 10646.
Unicode is backward compatible with the 7bit USASCII and 8bit Latin1 ISO88591. That is, the first 128 characters are
the same as USASCII; and the first 256 characters are the same as Latin1.
Unicode originally uses 16 bits called UCS2 or Unicode Character Set 2 byte, which can represent up to 65,536
characters. It has since been expanded to more than 16 bits, currently stands at 21 bits. The range of the legal codes in
ISO/IEC 10646 is now from U+0000H to U+10FFFFH 21 bits or about 2 million characters, covering all current and ancient
historical scripts. The original 16bit range of U+0000H to U+FFFFH 65536 characters is known as Basic Multilingual Plane
BMP, covering all the major languages in use currently. The characters outside BMP are called Supplementary Characters,
which are not frequentlyused.
Unicode has two encoding schemes:
UCS2 Universal Character Set 2 Byte: Uses 2 bytes 16 bits, covering 65,536 characters in the BMP. BMP is
sufficient for most of the applications. UCS2 is now obsolete.
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

19/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

UCS4 Universal Character Set 4 Byte: Uses 4 bytes 32 bits, covering BMP and the supplementary characters.

5.5UTF8 Unicode Transformation Format 8bit

The 16/32bit Unicode UCS2/4 is grossly inefficient if the document contains mainly ASCII characters, because each
character occupies two bytes of storage. Variablelength encoding schemes, such as UTF8, which uses 14 bytes to
represent a character, was devised to improve the efficiency. In UTF8, the 128 commonlyused USASCII characters use
only 1 byte, but some lesscommonly characters may require up to 4 bytes. Overall, the efficiency improved for document
containing mainly USASCII texts.
The transformation between Unicode and UTF8 is as follows:

Bits

Unicode

00000000

UTF8Code
0xxxxxxx

0xxxxxxx

Bytes
1
(ASCII)

11

00000yyy
yyxxxxxx

110yyyyy10xxxxxx

16

zzzzyyyy
yyxxxxxx

1110zzzz10yyyyyy
10xxxxxx

21

000uuuuu
zzzzyyyy

11110uuu10uuzzzz
10yyyyyy10xxxxxx

yyxxxxxx
In UTF8, Unicode numbers corresponding to the 7bit ASCII characters are padded with a leading zero; thus has the same
value as ASCII. Hence, UTF8 can be used with all software using ASCII. Unicode numbers of 128 and above, which are less
frequently used, are encoded using more bytes 24 bytes. UTF8 generally requires less storage and is compatible with
ASCII. The drawback of UTF8 is more processing power needed to unpack the code due to its variable length. UTF8 is the
most popular format for Unicode.
Notes:
UTF8 uses 13 bytes for the characters in BMP 16bit, and 4 bytes for supplementary characters outside BMP 21bit.
The 128 ASCII characters basic Latin letters, digits, and punctuation signs use one byte. Most European and Middle
East characters use a 2byte sequence, which includes extended Latin letters with tilde, macron, acute, grave and other
accents, Greek, Armenian, Hebrew, Arabic, and others. Chinese, Japanese and Korean CJK use threebyte sequences.
All the bytes, except the 128 ASCII characters, have a leading '1' bit. In other words, the ASCII bytes, with a leading
'0' bit, can be identified and decoded easily.
Example: (Unicode:60A8H597DH)
Unicode(UCS2)is60A8H=0110000010101000B
UTF8is111001101000001010101000B=E682A8H
Unicode(UCS2)is597DH=0101100101111101B
UTF8is111001011010010110111101B=E5A5BDH
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

20/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

5.6UTF16 Unicode Transformation Format 16bit

UTF16 is a variablelength Unicode character encoding scheme, which uses 2 to 4 bytes. UTF16 is not commonly used.
The transformation table is as follows:

Unicode

UTF16Code

Bytes

xxxxxxxxxxxxxxxx

SameasUCS2no
encoding

000uuuuuzzzzyyyy
yyxxxxxx
(uuuuu0)

110110wwwwzzzzyy110111yy
yyxxxxxx
(wwww=uuuuu1)

Take note that for the 65536 characters in BMP, the UTF16 is the same as UCS2 2 bytes. However, 4 bytes are used for
the supplementary characters outside the BMP.
For BMP characters, UTF16 is the same as UCS2. For supplementary characters, each character requires a pair 16bit
values, the first from the highsurrogates range, \uD800\uDBFF, the second from the lowsurrogates range \uDC00
\uDFFF.

5.7UTF32 Unicode Transformation Format 32bit

Same as UCS4, which uses 4 bytes for each character unencoded.

5.8Formats of MultiByte e.g., Unicode Text Files

Endianess or byteorder : For a multibyte character, you need to take care of the order of the bytes in storage. In
big endian, the most significant byte is stored at the memory location with the lowest address big byte first. In little
endian, the most significant byte is stored at the memory location with the highest address little byte first. For example,
with Unicode number of 60A8H is stored as 60A8 in big endian; and stored as A860 in little endian. Big endian, which
produces a more readable hex dump, is more commonlyused, and is often the default.

BOM Byte Order Mark : BOM is a special Unicode character having code number of FEFFH, which is used to
differentiate bigendian and littleendian. For bigendian, BOM appears as FEFFH in the storage. For littleendian, BOM
appears as FFFEH. Unicode reserves these two code numbers to prevent it from crashing with another character.
Unicode text files could take on these formats:
Big Endian: UCS2BE, UTF16BE, UTF32BE.
Little Endian: UCS2LE, UTF16LE, UTF32LE.
UTF16 with BOM. The first character of the file is a BOM character, which specifies the endianess. For bigendian, BOM
appears as FEFFH in the storage. For littleendian, BOM appears as FFFEH.
UTF8 file is always stored as big endian. BOM plays no part. However, in some systems in particular Windows, a BOM is
added as the first character in the UTF8 file as the signature to identity the file as UTF8 encoded. The BOM character
FEFFH is encoded in UTF8 as EFBBBF. Adding a BOM as the first character of the file is not recommended, as it may be
incorrectly interpreted in other system. You can have a UTF8 file without BOM.

5.9Formats of Text Files

Line Delimiter or EndOfLine EOL : Sometimes, when you use the Windows NotePad to open a text file
created in Unix or Mac, all the lines are joined together. This is because different operating platforms use different
character as the socalled line delimiter or endofline or EOL. Two nonprintable control characters are involved: 0AH
LineFeed or LF and 0DH CarriageReturn or CR.
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

21/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

Windows/DOS uses OD0AH CR+LF, "\r\n" as EOL.

Unixes use 0AH LF, "\n" only.
Mac uses 0DH CR, "\r" only.

EndofFile EOF : [TODO]

5.10Windows' CMD Codepage

Character encoding scheme charset in Windows is called codepage. In CMD shell, you can issue command "chcp" to
display the current codepage, or "chcpcodepagenumber" to change the codepage.
Take note that:
The default codepage 437 used in the original DOS is an 8bit character set called Extended ASCII, which is different
from Latin1 for code numbers above 127.
Codepage 1252 Windows1252, is not exactly the same as Latin1. It assigns code number 80H to 9FH to letters and
punctuation, such as smart singlequotes and doublequotes. A common problem in browser that display quotes and
apostrophe in question marks or boxes is because the page is supposed to be Windows1252, but mislabelled as ISO
88591.
For internationalization and chinese character set: codepage 65001 for UTF8, codepage 1201 for UCS2BE, codepage
1200 for UCS2LE, codepage 936 for chinese characters in GB2312, codepage 950 for chinese characters in Big5.

5.11Chinese Character Sets

Unicode supports all languages, including asian languages like Chinese both simplified and traditional characters,
Japanese and Korean collectively called CJK. There are more than 20,000 CJK characters in Unicode. Unicode characters
are often encoded in the UTF8 scheme, which unfortunately, requires 3 bytes for each CJK character, instead of 2 bytes in
the unencoded UCS2 UTF16.
Worse still, there are also various chinese character sets, which is not compatible with Unicode:
GB2312/GBK: for simplified chinese characters. GB2312 uses 2 bytes for each chinese character. The most significant bit
MSB of both bytes are set to 1 to coexist with 7bit ASCII with the MSB of 0. There are about 6700 characters. GBK is
an extension of GB2312, which include more characters as well as traditional chinese characters.
BIG5: for traditional chinese characters BIG5 also uses 2 bytes for each chinese character. The most significant bit of
both bytes are also set to 1. BIG5 is not compatible with GBK, i.e., the same code number is assigned to different
character.
For example, the world is made more interesting with these many standards:

Simplified

Traditional

Standard

Characters

Codes

GB2312

BACDD0B3

USC2

548C8C10

UTF8

E5928CE8B090

BIG5

A94DBFD3

UCS2

548C8AE7

UTF8

E5928CE8ABA7

Notes for Windows' CMD Users : To display the chinese character correctly in CMD shell, you need to choose the
correct codepage, e.g., 65001 for UTF8, 936 for GB2312/GBK, 950 for Big5, 1201 for UCS2BE, 1200 for UCS2LE, 437 for the
original DOS. You can use command "chcp" to display the current code page and command "chcpcodepage_number" to
change the codepage. You also have to choose a font that can display the characters e.g., Courier New, Consolas or Lucida
Console, NOT Raster font.
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

22/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

5.12Collating Sequences for Ranking Characters

A string consists of a sequence of characters in upper or lower cases, e.g., "apple", "BOY", "Cat". In sorting or comparing
strings, if we order the characters according to the underlying code numbers e.g., USASCII characterbycharacter, the
order for the example would be "BOY", "apple", "Cat" because uppercase letters have a smaller code number than
lowercase letters. This does not agree with the socalled dictionary order, where the same uppercase and lowercase letters
have the same rank. Another common problem in ordering strings is "10" ten at times is ordered in front of "1" to "9".
Hence, in sorting or comparison of strings, a socalled collating sequence or collation is often defined, which specifies the
ranks for letters uppercase, lowercase, numbers, and special symbols. There are many collating sequences available. It is
entirely up to you to choose a collating sequence to meet your application's specific requirements. Some caseinsensitive
dictionaryorder collating sequences have the same rank for same uppercase and lowercase letters, i.e., 'A', 'a' 'B', 'b'
... 'Z', 'z'. Some casesensitive dictionaryorder collating sequences put the uppercase letter before its lowercase
counterpart, i.e., 'A' 'B' 'C'... 'a' 'b''c'.... Typically, space is ranked before digits '0' to '9', followed
by the alphabets.
Collating sequence is often language dependent, as different languages use different sets of characters e.g., , , a, with
their own orders.

5.13For Java Programmers java.nio.Charset

JDK 1.4 introduced a new java.nio.charset package to support encoding/decoding of characters from UCS2 used
internally in Java program to any supported charset used by external devices.
Example: The following program encodes some Unicode texts in various encoding scheme, and display the Hex codes of
the encoded byte sequences.
importjava.nio.ByteBuffer;
importjava.nio.CharBuffer;
importjava.nio.charset.Charset;

publicclassTestCharsetEncodeDecode{
publicstaticvoidmain(String[]args){
//Trythesecharsetsforencoding
String[]charsetNames={"USASCII","ISO88591","UTF8","UTF16",
"UTF16BE","UTF16LE","GBK","BIG5"};

Stringmessage="Hi,!";//messagewithnonASCIIcharacters
//PrintUCS2inhexcodes
System.out.printf("%10s:","UCS2");
for(inti=0;i<message.length();i++){
System.out.printf("%04X",(int)message.charAt(i));
}
System.out.println();

for(StringcharsetName:charsetNames){
//GetaCharsetinstancegiventhecharsetnamestring
Charsetcharset=Charset.forName(charsetName);
System.out.printf("%10s:",charset.name());

//EncodetheUnicodeUCS2charactersintoabytesequenceinthischarset.
ByteBufferbb=charset.encode(message);
while(bb.hasRemaining()){
System.out.printf("%02X",bb.get());//Printhexcode
}
System.out.println();
bb.rewind();
}
}
}
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

23/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

UCS2:00480069002C60A8597D0021[16bitfixedlength]
Hi,!
USASCII:48692C3F3F21[8bitfixedlength]
Hi,??!
ISO88591:48692C3F3F21[8bitfixedlength]
Hi,??!
UTF8:48692CE682A8E5A5BD21[14bytesvariablelength]
Hi,!
UTF16:FEFF00480069002C60A8597D0021[24bytesvariablelength]
BOMHi,![ByteOrderMarkindicatesBigEndian]
UTF16BE:00480069002C60A8597D0021[24bytesvariablelength]
Hi,!
UTF16LE:480069002C00A8607D592100[24bytesvariablelength]
Hi,!
GBK:48692CC4FABAC321[12bytesvariablelength]
Hi,!
Big5:48692CB17AA66E21[12bytesvariablelength]
Hi,!

5.14For Java Programmers char and String

The char data type are based on the original 16bit Unicode standard called UCS2. The Unicode has since evolved to 21
bits, with code range of U+0000 to U+10FFFF. The set of characters from U+0000 to U+FFFF is known as the Basic
Multilingual Plane BMP. Characters above U+FFFF are called supplementary characters. A 16bit Java char cannot hold a
supplementary character.
Recall that in the UTF16 encoding scheme, a BMP characters uses 2 bytes. It is the same as UCS2. A supplementary
character uses 4 bytes. and requires a pair of 16bit values, the first from the highsurrogates range, \uD800\uDBFF, the
second from the lowsurrogates range \uDC00\uDFFF.
In Java, a String is a sequences of Unicode characters. Java, in fact, uses UTF16 for String and StringBuffer. For BMP
characters, they are the same as UCS2. For supplementary characters, each characters requires a pair of char values.
Java methods that accept a 16bit char value does not support supplementary characters. Methods that accept a 32bit
int value support all Unicode characters in the lower 21 bits, including supplementary characters.
This is meant to be an academic discussion. I have yet to encounter the use of supplementary characters!

5.15Displaying Hex Values & Hex Editors

At times, you may need to display the hex values of a file, especially in dealing with Unicode characters. A Hex Editor is a
handy tool that a good programmer should possess in his/her toolbox. There are many freeware/shareware Hex Editor
available. Try google "Hex Editor".
I used the followings:
NotePad++ with Hex Editor Plugin: Opensource and free. You can toggle between Hex view and Normal view by
pushing the "H" button.
PSPad: Freeware. You can toggle to Hex view by choosing "View" menu and select "Hex Edit Mode".
TextPad: Shareware without expiration period. To view the Hex value, you need to "open" the file by choosing the file
format of "binary" ??.
UltraEdit: Shareware, not free, 30day trial only.
Let me know if you have a better choice, which is fast to launch, easy to use, can toggle between Hex and normal view,
free, ....
The following Java program can be used to display hex code for Java Primitives integer, character and floatingpoint:
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

24/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

publicclassPrintHexCode{

publicstaticvoidmain(String[]args){
inti=12345;
System.out.println("Decimalis"+i);//12345
System.out.println("Hexis"+Integer.toHexString(i));//3039
System.out.println("Binaryis"+Integer.toBinaryString(i));//11000000111001
System.out.println("Octalis"+Integer.toOctalString(i));//30071
System.out.printf("Hexis%x\n",i);//3039
System.out.printf("Octalis%o\n",i);//30071

charc='a';
System.out.println("Characteris"+c);//a
System.out.printf("Characteris%c\n",c);//a
System.out.printf("Hexis%x\n",(short)c);//61
System.out.printf("Decimalis%d\n",(short)c);//97

floatf=3.5f;
System.out.println("Decimalis"+f);//3.5
System.out.println(Float.toHexString(f));//0x1.cp1(Fraction=1.c,Exponent=1)

f=0.75f;
System.out.println("Decimalis"+f);//0.75
System.out.println(Float.toHexString(f));//0x1.8p1(F=1.8,E=1)

doubled=11.22;
System.out.println("Decimalis"+d);//11.22
System.out.println(Double.toHexString(d));//0x1.670a3d70a3d71p3(F=1.670a3d70a3d71E=3)
}
}

In Eclipse, you can view the hex code for integer primitive Java variables in debug mode as follows: In debug perspective,
"Variable" panel Select the "menu" inverted triangle Java Java Preferences... Primitive Display Options Check
"Display hexadecimal values byte, short, char, int, long".

6.Summary Why Bother about Data Representation?

Integer number 1, floatingpoint number 1.0 character symbol '1', and string "1" are totally different inside the
computer memory. You need to know the difference to write good and highperformance programs.
In 8bit signed integer, integer number 1 is represented as 00000001B.
In 8bit unsigned integer, integer number 1 is represented as 00000001B.
In 16bit signed integer, integer number 1 is represented as 0000000000000001B.
In 32bit signed integer, integer number 1 is represented as 00000000000000000000000000000001B.
In 32bit floatingpoint representation, number 1.0 is represented as 0 01111111 0000000 00000000 00000000B,
i.e., S=0, E=127, F=0.
In 64bit floatingpoint representation, number 1.0 is represented as 0 01111111111 0000 00000000 00000000
00000000000000000000000000000000B, i.e., S=0, E=1023, F=0.
In 8bit Latin1, the character symbol '1' is represented as 00110001B or 31H.
In 16bit UCS2, the character symbol '1' is represented as 0000000000110001B.
In UTF8, the character symbol '1' is represented as 00110001B.
If you "add" a 16bit signed integer 1 and Latin1 character '1' or a string "1", you could get a surprise.
https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html
6.1Exercises Data Representation

25/26

9/10/2015

ATutorialonDataRepresentationIntegers,Floatingpointnumbers,andcharacters

6.1Exercises Data Representation

For the following 16bit codes:
0000000000101010;
1000000000101010;

Give their values, if they are representing:

1. a 16bit unsigned integer;
2. a 16bit signed integer;
3. two 8bit unsigned integers;
4. two 8bit signed integers;
5. a 16bit Unicode characters;
6. two 8bit ISO88591 characters.
Ans: 1 42, 32810; 2 42, 32726; 3 0, 42; 128, 42; 4 0, 42; 128, 42; 5 '*'; ''; 6 NUL, '*'; PAD, '*'.

REFERENCES & RESOURCES

1. FloatingPoint Number Specification IEEE 754 1985, "IEEE Standard for Binary FloatingPoint Arithmetic".
2. ASCII Specification ISO/IEC 646 1991 or ITUT T.501992, "Information technology 7bit coded character set for
information interchange".
3. LatinI Specification ISO/IEC 88591, "Information technology 8bit singlebyte coded graphic character sets Part
1: Latin alphabet No. 1".
4. Unicode Specification ISO/IEC 10646, "Information technology Universal MultipleOctet Coded Character Set
UCS".
5. Unicode Consortium @ http://www.unicode.org.

Last modified: January, 2014

Feedback, comments, corrections, and errata can be sent to Chua HockChuan (ehchua@ntu.edu.sg) | HOME

https://www3.ntu.edu.sg/home/ehchua/programming/java/DataRepresentation.html

26/26