Building Evolutionary Architectures, 2nd Edition

Chapter 2. Fitness Functions

The mechanics of evolutionary architecture cover the tools and techniques developers and architects use to build systems that can evolve. An important gear in that machinery is the protection mechanism called a fitness function, the architectural equivalent of a unit test for the domain part of an application. This chapter defines fitness functions and explains the categories and usage of this important building block.

An evolutionary architecture supports guided, incremental change across multiple dimensions.

As noted in our definition, the word guided indicates that some objective exists that architecture should move toward or exhibit. We borrow a concept from evolutionary computing called fitness functions, which are used in genetic algorithm design to define success.

Evolutionary computing includes a number of mechanisms that allow a solution to gradually emerge via mutation—​small changes in each generation of the software. The evolutionary computing world defines a number of types of mutations. For example, one mutation is called a roulette mutation: if the algorithm utilizes constants, this mutation will choose new numbers as if from a roulette wheel in a casino. For example, suppose a developer is designing a genetic algorithm to solve the traveling salesperson problem to find the shortest route between a number of cities. If the developer notices that smaller numbers supplied by the roulette mutation yield better results, they may build ...