Nt1330 Unit 3 Assignment 1 Reference Set

This figure assumes that the main reference set, covers the indicated circle of sections A, B, C. The solution 1 is created from a convex combination of reference solutions A, B that is added to the reference set as the only solution. In a similar way, combining of convex and non-convex reference of new and original solutions are created points 2, 3 and 4. The complete reference set are including 7 solutions (members) that is shown in the figure above. In genetic algorithm, two solutions are selected randomly from the population and a crossover operator used for the production of one or more children. GA are including a sample population of 100 elements that are selected randomly to create crossover. But in scatter search, two or more of the reference set in a systematic approach in order to produce new
Update of reference set: To create and update reference set two fundamental points should be considered:
1. This collection should include solutions that better meet the objectives, in terms of quality.
2. In addition to solution with high quality, other solutions that are causing scattering in search paths should also be considered in this set.
Therefore, the main goal is using a smart updating method for a reference set, according to produced solutions in two previous stages and existence solution in reference set.
4. Create subset: At this stage using an intelligent approach and non-coincidentally, answer that need to be combined are identified from the members of reference set. One advantage of this step is creating a subset that unlike genetic algorithms are not necessarily only has two members. This approach allows the solution space simultaneously affect more compounds for searching. So the probability of reaching better solutions increases.
5. The combination of selected subsets: After forming subsets by a combination convex and nonconvex, possible solutions are combined to obtain new solutions.

5. The structure of proposed scatter search