After a new set of offspring has been created through the application of genetic operators, two final decisions must be made to complete the generational cycle: which individuals will survive to form the next population, and when the entire evolutionary process should stop.

7.1 Survivor Selection Policies

Survivor selection, also known as replacement, determines which individuals from the combined pool of parents and offspring are kept for the next generation. The policy should preserve fit individuals while maintaining diversity. A common strategy used in conjunction with survivor selection is Elitism, which guarantees that the fittest individual(s) from the current generation are always carried over to the next, ensuring that the best-found solution is never lost.

7.2 Termination Conditions

A well-defined termination condition is essential to prevent a GA from running indefinitely, especially after its performance has plateaued and further computation yields diminishing returns. The algorithm should stop when a satisfactory solution has been found or when further progress is unlikely.

Common termination criteria include:

• No improvement in the population for X iterations. A practical implementation of this involves a counter that increments each time no better offspring are found and resets to zero whenever an improved solution appears. The algorithm terminates when the counter reaches a predefined limit.
• Reaching a maximum number of generations.
• The objective function value reaching a predefined target.

With the core mechanics of a standard GA defined, we can now turn to more advanced strategies for handling the complexities of real-world problems.