Testing: Refining algorithms

The idea

Testing and refining algorithms are distinct but related processes. Testing evaluates a program to find errors, using iterative testing during development or final testing at the end. A test plan uses normal, boundary, invalid, and erroneous data. Refining an algorithm improves its design, such as fixing errors, handling unexpected inputs robustly, or increasing efficiency. For example, if an algorithm crashes on negative numbers, refinement adds input validation. Trace tables support dry runs to track variable changes and locate logic errors (which produce incorrect results). Syntax errors break language rules and are found by translators. Effective refinement requires identifying the specific weakness and applying a targeted fix.

What earns the marks

• Refining an algorithm involves identifying weaknesses, such as logic errors or inefficiencies, and applying targeted fixes.
• Trace tables are used to dry-run algorithms, tracking variable values to pinpoint exactly where logic errors occur.
• Distinguishes logic errors, which produce incorrect results and are found by dry runs, from syntax errors, which break language rules.
• Refinements often include adding input validation, correcting loop conditions, or removing redundant instructions to improve efficiency.
• A suggested fix must directly address the identified error without altering the correct functionality of the original algorithm.
• Testing uses a test plan with normal, boundary, invalid and erroneous data, and distinguishes iterative testing from final/terminal testing.

Common mistakes and how to put them right

• Confusing syntax and logic errors.

  Put it right: Syntax errors break language rules (e.g. missing quotes), while logic errors produce incorrect outputs.

• Claiming a trace table finds syntax errors.

  Put it right: Trace tables track variable values to expose logic errors, while syntax errors are found by inspection or the compiler.

• Rewriting the entire algorithm instead of refining it.

  Put it right: Identify the specific weakness and change only the relevant lines of code.

• Failing to update variable values correctly in a trace table.

  Put it right: Carefully follow the algorithm step-by-step, updating the table only when a variable's value explicitly changes.

• Confusing iterative testing with final/terminal testing.

  Put it right: Iterative testing is repeated during development, while final/terminal testing is carried out at the end against the full requirements.

Exam tips

• When asked to refine an algorithm, identify the specific line or construct containing the error and write out the corrected code; if the question provides line numbers, refer to them.
• Use a trace table to systematically track variables during a dry run; this is the most reliable way to find hidden logic errors.
• Do not use a trace table to look for syntax errors; check the code against the language rules instead.
• When a question asks about testing, refer to a test plan and give examples of normal, boundary, invalid and erroneous test data.