Paper 3: Advanced Theory (Cambridge (CIE) A Level Computer Science): Revision Note

Exam code: 9618

Robert Hampton

Written by: Robert Hampton

Reviewed by: Bridgette Barrett

Updated on

Paper 3 overview

What is in Paper 3?

  • Paper 3 Advanced Theory is a 1 hour 30 minute written paper worth 75 marks

  • It is 25% of the A Level and is only sat by A Level students

  • It covers syllabus sections 13 to 20:

    • data representation (including floating-point numbers)

    • communication and internet technologies

    • hardware and virtual machines (including Boolean algebra and Karnaugh maps)

    • system software (including Reverse Polish Notation (RPN) and Backus-Naur Form (BNF))

    • security

    • artificial intelligence

    • computational thinking and problem-solving

    • further programming

  • You answer all questions, and calculators are not allowed

  • Many questions are multi-step methods where the working earns marks; examiner reports remind students that if a question asks for working, it must be shown to gain full credit

  • Questions that need pseudocode must follow the Cambridge pseudocode guide

  • For general 9618 advice and the Paper 1 skills that Paper 3 builds on, see:

Question types and how to answer them

How do you answer floating-point questions?

  • Typical stems include:

    • "Calculate the normalised floating-point representation of +54.8125 in this system. Show your working."

    • "Calculate the denary value of the given normalised binary floating-point number"

  • The system is given in the question; for example, 12 bits for the mantissa and 4 bits for the exponent, both in two's complement

  • Mark schemes credit each stage separately:

    • converting the number to binary

    • working out the exponent (how many places the binary point moves)

    • the correct final mantissa and exponent

  • Examiners report students losing marks on the exponent: errors in calculating it, or not showing how it was worked out and applied to the mantissa

  • Check that the result is normalised: a positive mantissa starts 01 and a negative mantissa starts 10

  • Fill every bit of the mantissa and exponent, padding with zeros where needed

  • For questions about range or precision, link the bits to the effect: more mantissa bits give greater precision and more exponent bits give a greater range; examiners credited answers that said how many bits the mantissa had, why that was too few, and that precision would be lost

How do you answer Boolean algebra and Karnaugh map questions?

  • Typical stems include:

    • "Write the Boolean expression that corresponds to the logic circuit as a sum-of-products"

    • "Complete the Karnaugh map (K-map) for the given truth table"

    • "Draw loop(s) around appropriate group(s) in the K-map to produce an optimal sum-of-products"

  • Fill every cell of a K-map with a 0 or a 1; examiners only award marks for a complete map using 0s and 1s

  • Loops must contain 1, 2, 4, 8 or 16 cells and can wrap around the edges; use the largest groups possible

  • Give the answer in the form asked for

    • If the question asks for a sum-of-products, do not simplify it further; examiners report many wrong answers that gave simplified expressions the question did not ask for

    • If the question asks you to simplify, show all the stages and name the law used where you can

    • When you apply De Morgan's laws, change the operator as well as splitting the bar: NOT (A AND B) becomes (NOT A) OR (NOT B); examiners report answers that split the bar but left AND as AND

How do you answer Reverse Polish Notation (RPN) questions?

  • Typical stems include:

    • "Evaluate this RPN expression"

    • "Show the changing contents of the stack as the RPN expression is evaluated"

    • Converting between infix and RPN

  • When showing the stack during RPN evaluation, only operands go on the stack; examiners report a common error of putting the operators on the stack

  • Show the stack after every push and pop, not only the final value

  • When converting RPN to infix, include brackets; examiners report that leaving them out changes the order of operations and makes the answer wrong

How do you answer BNF and syntax diagram questions?

  • Typical stems include "Complete the Backus-Naur Form (BNF) for the given syntax diagrams" and "Draw a syntax diagram for..."

  • Use exact BNF notation: <name> for each non-terminal, ::= for "is defined as" and | between alternatives

  • Examiners report that students often understood the syntax diagram but lost marks for BNF that was not fully correct

  • For recursion (for example, a code made of one or more digits), define the item in terms of itself, such as <code> ::= <digit> | <digit><code>

How do you answer graph algorithm questions?

  • Typical stems include "Calculate the shortest distance between the base and each of the other towns in the diagram using Dijkstra's algorithm" and "Describe the purpose of both the A* algorithm and Dijkstra's algorithm"

  • For a Dijkstra trace, show the working the examiners look for:

    • set the start node to 0 and every other node to infinity (∞)

    • update each distance as shorter routes are found

    • record which nodes have been visited

  • For "purpose" questions, the purpose of both algorithms is to find the shortest (optimum) path between two nodes; the A* algorithm also uses a heuristic to guide the search

How do you answer declarative programming questions?

  • Typical stems describe a set of facts (for example, students and the subjects they choose) and ask you to write extra clauses, or to give the result of a goal

  • Examiners report that the highest marks went to students who remembered that:

    • every clause ends with a full stop

    • facts use lower-case letters only

    • clauses are written in the correct order with no spelling errors

  • When giving the output of a goal, do not add capital letters or a full stop to the list of results

How do you answer object-oriented programming questions?

  • Typical stems include "Outline the structure of a class", "Give three differences between an object and a class" and writing pseudocode for a class with a constructor

  • In pseudocode, close every class with ENDCLASS; examiners report that missing ENDCLASS was a typical error

  • Use the identifiers already given in the question for attributes and parameters; examiners report marks lost for using different variable names in declarations or the constructor

  • For definitions, be specific: "inheritance is when a new class derives attributes and methods from a parent class" scores, where a vague phrase does not

How do you answer user-defined data type questions?

  • Typical stems include "Write pseudocode statements to declare the record data type...", or to declare an enumerated, set or pointer type

  • Declare every user-defined type with the keyword TYPE; a record type ends with ENDTYPE. Examiners report marks lost for starting the definition with DECLARE, or for not beginning it with TYPE

  • Use the type names, field names and data types given in the question

How do you answer "purpose" questions?

  • Some Paper 3 questions ask for the purpose of something, such as asymmetric key cryptography or quantum cryptography

  • Examiners report that describing how it works or listing its features does not answer a purpose question; say what it is for and why it is used

Showing working on multi-step questions

What should your working include?

  • In Paper 3, the method often earns more marks than the final answer

Question type

What your working should show

Floating-point to binary

The number in binary, the number of places the point moves, then the final mantissa and exponent

Floating-point to denary

The exponent in denary, where the point moves to, then the denary value

K-map

Every cell filled with 0 or 1, the loops drawn, then one product term per loop

Boolean simplification

Every stage of the simplification

RPN evaluation

The stack after every push and every operation

Dijkstra's algorithm

Starting values (0 and ∞) and each updated distance

  • If you run out of time on a multi-step question, write down the stages you have completed; examiners report that students earned credit for correct methods even when the final answer was wrong

Worked Example

Real numbers are stored in a computer using floating-point representation with:

  • 12 bits for the mantissa

  • 4 bits for the exponent

  • two's complement form for both the mantissa and the exponent

Calculate the normalised floating-point representation of +27.375 in this system. Show your working.

[3]

Answer:

  • 27.375 converted to binary: 11011.011 [1 mark]

  • The binary point moves 5 places to the left to give 0.11011011, so the exponent is 5 [1 mark]

  • Mantissa: 011011011000, Exponent: 0101 [1 mark]

The mantissa starts with 01, which shows it is a normalised positive number; the remaining bits are padded with zeros to fill all 12 bits.

Examiner Tips and Tricks

  • Write the exponent as a denary number in your working before converting it to binary, because examiners report marks lost on the exponent step

Common mistakes in Paper 3

  • Getting the exponent wrong, or not showing how it was calculated, in floating-point questions

  • Leaving K-map cells blank or using symbols other than 0 and 1

  • Simplifying a sum-of-products when the question did not ask for it

  • Putting operators on the stack when evaluating RPN

  • Leaving out brackets when converting RPN to infix

  • Using capital letters in facts or in the output of a goal, or missing the full stop at the end of a clause

  • Describing how something works when the question asks for its purpose

Unlock more, it's free!

Join the 100,000+ Students that ❤️ Save My Exams

the (exam) results speak for themselves:

Robert Hampton

Author: Robert Hampton

Expertise: Curriculum Expert

Rob has over 16 years' experience teaching Computer Science and ICT at KS3 & GCSE levels. Rob has demonstrated strong leadership as Head of Department since 2012 and previously supported teacher development as a Specialist Leader of Education, empowering departments to excel in Computer Science. Beyond his tech expertise, Robert embraces the virtual world as an avid gamer, conquering digital battlefields when he's not coding.

Bridgette Barrett

Reviewer: Bridgette Barrett

Expertise: Development Editor

After graduating with a degree in Geography, Bridgette completed a PGCE over 30 years ago. She later gained an MA Learning, Technology and Education from the University of Nottingham focussing on online learning. At a time when the study of geography has never been more important, Bridgette is passionate about creating content which supports students in achieving their potential in geography and builds their confidence.