AQA Computer Science A-Level

4.5.4 Binary number system

Advanced Notes

www.pmt.education
Specification:

4.5.4.1 Unsigned binary:

Know the difference between unsigned binary and signed binary
Know that in unsigned binary the minimum and maximum values for a
given number of bits, ​n​, are 0 and 2​n​ -1 respectively

4.5.4.2 Unsigned binary arithmetic:

Be able to:
● add two unsigned binary integers
● multiply two unsigned binary integers

4.5.4.3 Signed binary using two’s complement:

Know that signed binary can be used to represent negative integers and
that one possible coding scheme is two’s complement.
Know how to:
● represent negative and positive integers in two’s complement
● perform subtraction using two’s complement
● calculate the range of a given number of bits, ​n​.

4.5.4.4 Numbers with a fractional part:

Know how numbers with a fractional part can be represented in:
● fixed point form in binary in a given number of bits
● floating point form in binary in a given number of bits
Be able to convert for each representation from:
● decimal to binary of a given number of bits
● binary to decimal of a given number of bits

4.5.4.5 Rounding errors:

Know and be able to explain why both fixed point and floating point
representation of decimal numbers may be inaccurate.

www.pmt.education
4.5.4.6 Absolute and relative errors:
Be able to calculate the absolute error of numerical data stored and
processed in computer systems.
Be able to calculate the relative error of numerical data stored and
processed in computer systems.
Compare absolute and relative errors for large and small magnitude
numbers, and numbers close to one.

4.5.4.7 Range and precision:

Compare the advantages and disadvantages of fixed point and floating
point forms in terms of range, precision and speed of calculation.

4.5.4.8 Normalisation of floating point form:

Know why floating point numbers are normalised and be able to
normalise un-normalised floating point numbers with positive or negative
mantissas.

4.5.4.9 Underflow and overflow:

Explain underflow and overflow and describe the circumstances in
which they occur.

www.pmt.education
 Signed and unsigned binary

1011 
The string of ​0​
s and ​1​
s above is a ​binary number​. It could be either signed or unsigned,
and there’s ​no way to work that out​. A computer ​has to be told​ whether a number is signed
or unsigned.

Unsigned binary numbers can only represent ​positive numbers​. Signed binary allows for
the representation of ​negative numbers​ using binary.

If the number above were unsigned, it could be ​converted ​to

decimal by assigning place values to each of the digits and
adding up the total of the values under which a ​1​falls,
resulting in the decimal value ​11​
.

Range of unsigned numbers

The ​range ​of numbers that can be represented by an unsigned binary number depends on
the number of bits​ available. With one bit, the decimal numbers 0 and 1 can be
represented. With two bits, the decimal numbers 0, 1, 2 and 3 can be represented.

There is a ​pattern ​to the range of numbers that can be represented by a given number of
bits. For n bits, there are ​2n​
​ possible ​permutations ​of the bits, giving a ​range ​of decimal
numbers from ​0​to ​2​-1​
n​
.

For example, with ​eight ​bits, the decimal numbers ​0​to ​255​(= ​28​
​-1​
) can be represented.

Unsigned binary arithmetic

Adding two unsigned binary integers

When adding unsigned binary numbers, there are ​four important rules​ to remember:

1. 0 + ​
0 + 0 = 0 
2. 0 + ​
0 + 1 = 1 
3. 0 + ​
1 + 1 = 10 
4. 1 + 1 + 1 = 11 

Make sure you understand these rules and adding binary numbers will be easy.

www.pmt.education
Example:​ Add the unsigned binary integers ​1011​and ​1110​
.

Place the two binary numbers above each other so that the
  1  0  1  1 
digits line up​.
+  1  1  1  0 

Starting from the least significant bits (the right hand side),
  1  0  1  1 
add the values in each column​ and place the total below.
+  1  1  1  0  For the first column (highlighted), rule 2 from above
applies.
        1 

  1  0  1  1  Move on to the next column. This time rule 3 applies. In the

case that the result of addition for a single column is ​more
+  1  1  1  0  than one digit​, place the first digit of the result in ​small
    1 
0  1  writing​ under the ​next most significant​ bit’s column.

  1  0  1  1  On to the next column, where there is a ​0​ , a 1 and a small

1. In this case, rule 3 applies again. Therefore the result is
+  1  1  1  0 
10​. Because ​10​is​ two digits long​, the 1 is written in small
  1 
0​
1 
0  1  writing under the next most significant bit’s column.

Moving on to the most significant column where there are

  1  0  1  1 
three 1s. Rule 4 applies, so the result for this column is ​11​
.
+  1  1  1  0  The first digit of the result is written under the next most
significant bit’s column, but it can be written full size as
1  1​
1 
0​
1 
0  1 
there are no more columns to add.

Finally, the result is ​read off from the full size numbers​ at
1  1  0  0  1  the bottom of each column. In this case, ​1011 + 1110 = 
11001​ .

After carrying out binary addition, it’s a good idea to ​check your answer ​by converting to
decimal if you have time. In this example, ​1011​ 2​ (​11​
10​) + ​1110​
2​ (​14​
10​) = ​11001​
2​ (​25​
10​) so
we haven’t made any mistakes.

www.pmt.education
Multiplying two unsigned binary integers
When ​multiplying ​unsigned binary numbers, write out one of the two numbers starting
under ​each occurrence of a 1​ in the second number and then ​add ​the contents of the
columns.

Example:​ Multiply the binary numbers ​1011​and ​1010​

.

Choose one of the two numbers as your

Guide: 1  0  1  1 
guide ​and write it out in columns.
             
Write out ​the second number​ under each
      1  0  1  0  occurrence of a 1 in the guide number,
aligning​ the least significant bit​ of the
    1  0  1  0  0  second number with​ the position of the 1 ​in
the guide number.
1  0  1  0  0  0  0 

Now perform​ binary addition​ on the

        1  0  1  0  columns, ​excluding the guide​, using the
technique explained earlier.
      1  0  1  0  0 

+  1  0  1  0  0  0  0 

  1  1​
1 
0  1  1  1  0 

The result of the addition can now be ​read

  1  1  
0  1  1  1  0  off from the full size numbers​ at the bottom
of each column. In this case,
1011 × 1010 = 1101110 

As with binary addition, binary multiplication can be ​checked ​by converting to decimal. In
this example, 1011​2​ (11​10​) × 1010​2​ (10​10​) = 1101110​2 (110​
​ 10​) so the multiplication has
worked correctly.

www.pmt.education
 Signed binary with two’s complement

There are a few different ​coding schemes ​that can be used for signed binary. AQA uses
one called​ two’s complement​ to allow for the representation of ​both positive and negative
numbers​ in binary.

When using two’s complement, the ​most significant bit​ of a number is ​given a negative
place value​. For example, with four bits, the place values would be:

-8 4 2 1 
This allows ​negative numbers​ to represented like so:

-8  4  2  1 

1 0 1 1 
 
-8​+ 2 + 1 = ​-5 

Subtraction using two’s complement

Computers ​work by adding numbers​. In order to perform subtraction, computers in fact
add negative numbers​. For example, if a computer had to work out the value of ​14 - 6​, it
would actually work out ​14 + (-6)​.

Example:​ Subtract 12 from 8.

In five bit two’s complement, 8​10​ is ​01000​
2​and
-12​10​ is ​10100​. ​Five is the minimum​ number of bits
2​
required in order to represent ​-12​ .
  -16  8  4  2  1 

  0  1  0  0  0  The two’s complement numbers are then ​added

using the same technique for adding that is
+  1  0  1  0  0  explained above before the result can be read off
as ​11100​ .
2​
  1  1  1  0  0 
Checking the result, ​-16 + 8 + 4 = -4​so the
calculation is correct.

www.pmt.education
 Range of two’s complement numbers
Given a ​certain number of bits​, the ​range ​of a two’s
-8  4  2  1 
complement signed binary number includes both ​positive
and negative ​values.
0 1 1 1 
For example, with four bits, the ​largest positive value ​that
 
can be represented is 7 and the ​most negative number​ that 4 + 2 + 1 = 7 
can be represented is -8 as shown on the right hand side of
the page.

More generally, with ​n​ bits, the range of a two’s complement -8  4  2  1 
signed binary number is from ​2​ -1​to ​-2​
n-1​ n-1​
.
1 0 0 0 
 
= -8 

Numbers with a fractional part

Binary can be used to represent ​numbers with a fractional part​. There are​ two ways​ of
doing this, one uses​ fixed point ​form and the other uses ​floating point ​form.

Fixed point binary

When using fixed point with a given number of bits, a specified number of bits are placed
before​ a ​binary point​ and the remaining bits fall ​behind ​the binary point.

Standard binary place values are used for columns ​before ​the binary point, and the
columns ​behind ​the binary point start at ½, then ¼ and ⅛ etc.

For example, with 8 bits ​split evenly ​with four bits before and after the binary point, the
number 11.3125​10​ can be represented in binary as ​10110101​ .

8  4  2  1  •  1
2
  1
4
  1
8
  1
16
 

1 0 1 1    0 1 0 1 
 
1 1
8 + 2 + 1 + 4
+ 16
= 11.3125  

www.pmt.education
Floating point binary (binary to decimal)
Floating point binary is comparable to ​scientific notation​ in that a number is represented as
a ​mantissa ​and an ​exponent​. In scientific notation, the number 3,100,000 could be written
as 3.1×10​6​ where 3.1 is the mantissa and 6 is the exponent.

In ​exam questions​ on floating point numbers, both the mantissa and exponent will be
represented using ​two's complement signed binary​.

In floating point binary, a number of bits are allocated to the mantissa the remaining bits
form the exponent. For example, with 5 bits for the mantissa and three for the exponent:

0 1 1 0 1  0 1 1 
Mantissa Exp.

In order to convert from floating point form to decimal, first ​convert the exponent to
decimal​. In this example, ​011​2​ = 3​ .
10​

Next, treating the number ​as if there were a binary point ​between the ​first and second
digits of the mantissa​, move the binary point the ​number of positions specified ​by the
exponent. In this case, the binary point moves three positions.

Now treat the mantissa as a​ fixed point​ binary number, with the binary point fixed in the
position specified by the exponent.

− 8  4  2  1  •  1
2
 

0 1 1 0    1 
1
4 + 2 + 2
= 6.5  

As shown above, the result of converting the floating point binary number ​01101011​to
decimal (with a five bit mantissa and three bit exponent) is ​6.5​
.

www.pmt.education
Floating point binary (decimal to binary)
In order to convert a decimal number to a floating point binary number, you must first
convert your decimal number to fixed point binary​.

Example:​ Convert the decimal number 14.625 to floating point binary.

14.625 is smaller than 16, so we need columns before the binary point for -16, 8, 4 and 2.
0.625 = ½ + ⅛ so we need columns after the binary point for ½, ¼ and ⅛.

− 16 8   4  2  1  •  1
2
  1
4
  1
8
 

0 1 1 1 0    1 0 1 
Now that we have a ​fixed point representation ​of our decimal
number, we have to ​normalise ​the number. In order to be
normalised, a floating point number ​must start with​ ​01​(for a
positive number) or ​10​(for a negative number). As explained
above, when converting from floating point to decimal, the
binary point is ​assumed to be between the first two digits​ in
the mantissa. In this case, we must ​add a leading ​0​and
move the binary point four positions to the left.

Our exponent must therefore be ​four​, and ​positive ​because we moved the decimal point ​to
the left​. If the point had moved to the right, our exponent would be ​negative​. Converting
this exponent into binary gives us ​0100​ .

We now have our mantissa as ​01110101​and our exponent as ​0100​ . Therefore 14.625​10
= ​011101010100​2​in floating point notation with an eight bit mantissa and a four bit
exponent.

www.pmt.education
 Rounding errors

There are some decimal numbers that ​cannot possibly be represented exactly ​in binary,
even with the use of fixed point or floating point notation. A bit like ⅓, which can only be
represented in decimal as ​0.3333​ …, there are some numbers which binary can only
approximately ​represent.

There are many numbers that binary ​cannot accurately represent​, one of which is ​0.1​10
which is ​0.00011001100110011​ … in binary. For this reason, both fixed point and
floating point representations of decimal numbers ​may be inaccurate​.

Absolute and relative errors

You can calculate ​absolute ​and ​relative ​errors in order to see how close a particular
number is to an​ actual value​.

Absolute error calculation

An absolute error is ​the actual amount ​by which a value is inaccurate and can be
calculated by finding the ​difference ​between the given value and the actual value.

Example:​ The number 14.6​10​ is represented in fixed point binary as ​1110.1​

2​. Calculate
the absolute error.

The binary ​1110.1​is equal to ​14.5​ , so the absolute error can be calculated like so:
10​
 
14.6​
10​ - 14.5​
10​ = 0.1​ .
10​

www.pmt.education
Relative error calculation
A ​relative error​ is a ​measure of uncertainty​ in a given value ​compared to the actual value
which is ​relative to the size of the given value​. A relative error can be calculated using the
formula:

absolute error
relative error = actual value

This formula gives a relative error as a decimal, but can give a ​percentage ​if the result is
multiplied by 100​.

Example:​ The number 12.4 is represented in fixed point binary as ​1100.011​

2​. Calculate
the relative error as a percentage to four significant figures.

The binary ​1100.011​is equal to 12.375​10​. The ​absolute ​error is therefore 0.025. Using
the formula, we can calculate the relative error.

0.025
relative error % = 12.4 × 100 = 0.2016 to 4 s.f .

Errors in relation to magnitude

An absolute error of 0.1cm in a measurement of 50m results in a ​very small relative error
of 0.002% but ​the same absolute error​ of 0.1cm in a measurement of 1cm results in a
much larger relative error​ of 10%.

Fixed point vs floating point

Although both fixed point and floating point ​perform the same function​ of representing
numbers with fractional parts in binary, they each have their own ​relative advantages and
disadvantages​.

Floating point allows for the representation of a ​greater range of numbers​ with a ​given
number of bits ​than fixed point. This is because floating point can take advantage of an
exponent which can be ​either positive or negative​. The ​number of bits allocated ​to each
part of a floating point number affects the numbers that can be represented. A​ large
exponent​ and a ​small mantissa ​allows for a​ large range​ but​ little precision​. In contrast, a
small exponent ​and a ​large mantissa ​allows for ​good precision​ but only a ​small range​.

In a similar way, the ​placement of the binary point ​in fixed point notation determines the
range and precision of the numbers that can be represented. A binary point ​close to the
left ​of a number gives ​good precision​ but only a​ small range ​of numbers. However, move
the binary point ​to the right​ and the ​range is increased​ while ​decreasing precision​.

www.pmt.education
 Normalisation

Floating point numbers are ​normalised ​in order to ​provide the maximum level of precision
for a given number of bits​. Normalisation involves ensuring that the a floating point
numbers ​starts with​ ​01​(for a positive number) or ​10​(for negative numbers).

Example:​ Normalise the binary number ​000110100101​which is a floating point number

with an ​eight bit mantissa ​and a four ​bit exponent​.

First, split the number into mantissa and exponent.

0 0 0 1 1 0 1 0  0 1 0 1 
Mantissa Exponent

Next, ​adjust the mantissa​ so that it starts ​01​or ​10​

. In this case, because we’re dealing
with a positive number, we will move all of the bits ​two places to the left​ and ​add zeros to
the end ​of the mantissa. Our new mantissa is ​01101000​ .

Because we’ve made the mantissa ​bigger ​by shifting the bits two positions to the left, we
must ​reduce the exponent ​by two so as to ​ensure the same number is still represented​.
The current exponent is 5​10​ so, ​subtracting two​, the new exponent must be 3​10​ which is
0011​2​ in binary.

0 1 1 0 1 0 0 0  0 0 1 1 
Mantissa Exponent

We now have a mantissa that starts with the digits ​01​

. A ​positive normalised ​number.

www.pmt.education
 Underflow and overflow

Underflow and overflow are two types of ​error ​that can occur when working with binary.

Underflow
Underflow occurs when ​very small numbers​ are to be
represented but there ​are not enough bits available​. For
example, the number ​0.015625​ 10​ can be represented in
fixed point binary as ​0000001​ 2​ with one bit ​before ​the binary
point and six bits ​after ​the binary point. However, if only 5 bits
are available after the binary point, the number would be
represented as ​000000​ 2​ which is ​010​
​.

Overflow
Overflow occurs when a number is​ too large to be represented​ with the available bits.
Overflow is particularly important when using signed binary.

Example:​ Using 8 bit two’s complement signed binary, perform the operation ​127 + 1​
.

First, ​convert ​each number to two’s complement binary. ​127​

10​ is ​01111111​
2​and 1​10​ is
00000001​ . Next, perform ​binary addition​.
2​

  -128  64  32  16  8  4  2  1 

  0  1  1  1  1  1  1  1 

+  0  0  0  0  0  0  0  1 

  1​
1
  0​
1
  0​
1
  0​
1
  0​
1
  0​
1
  0​
1 
0 

As the working above shows, the result of ​01111111​+ ​00000001​in two’s complement
signed binary is ​10000000​
. Converting these numbers to decimal, we have that ​127 + 1 
= -128​which obviously ​isn’t right​. ​Overflow ​has occurred.

www.pmt.education