Abstraction and Automation

• An algorithm is a set of instructions that can be followed to perform a task, and always terminates
• An algorithm could be expressed through a flow chart or pseudocode
• Pseudocode allows for working through the logic of a program without worrying about syntax
• We can hand-trace algorithms by taking them one line at a time and recording any changes in variables, as well as any outputs. This helps to work out what a program is doing.
• Good pseudocode can make it easier to program in a high level language
• It is important to account for all data a user might input
• Valid data - data we expect the user to input
• Invalid data - data we expect to generate an error (preferably caught or not accepted)
• Borderline data - data at the boundaries of acceptable values. Important to test to make sure it works as intended
• There may be other data we want to test for other reasons, for example checking compatibility with floats. Be prepared to come up with tests that are relevant to a given program
• We can accept feedback from users to improve and refine a program
• Abstraction is the process of hiding unnecessary detail.
• Representational abstraction means representing a problem in a simplified way with unnecessary detail removed
• Abstraction by generalisation means forming a hierarchical relationship by grouping common characteristics via 'is a kind of'
• Information hiding means hiding non-essential details so that it is easy to focus on what matters
• Procedural abstraction means abstracting any particular values in a problem to create a computation pattern or computational method; a procedure.
• We can abstract procedural abstraction further to achieve functional abstraction, where we hide the method so that we only see the input and output
• Data abstraction means specific details of how data is represented are hidden, in other words, how they are implemented in the code. This is the basis of abstract data types.
• Problem abstraction (or reduction) means we remove details from a problem we want to solve until it is reduced to a problem that has already been solved - for example plotting out a journey on a map abstracts into Dijkstra's algorithm
• Decomposition means breaking a problem down into smaller sub-problems until they are all individual tasks that are easier to solve
• Composition means combining existing procedures into compound procedures - for example by using libraries
• Composition can be seen in abstract data types, which are made up of smaller and simpler data types
• Automation means using abstracted models of real world situations to solve problems
• Computer science ultimately is about automating problems to simulate them and analyse them in a way we couldn't otherwise - whether this is modelling the growth of cancer cells or creating an interactive game for entertainment
• To automate, we design algorithms, implement them through code, implement abstracted models via data structures, and then execute everything