Relay Optimization Method

Insurance-linked securities portfolio with the VaR constraint optimization problem have a kind of weak dominance or ordering property, which enables us to reduce the variables’ dimensions gradually through exercising a genetic algorithm with randomly selected initial populations. This property also enables us to add boundary attraction potential to GA-MPC’s repair operator, among other modifications such as using Sobol sequence in the initial population selection for multiple runs, using WELL algorithm for RNG, adding orthogonal crossover to the mutation operator, using opposition-based starting population, using clockwise-shaped searching vectors instead of the original counterclockwise-shaped searching vectors, adding the catfish and pivot algorithms for combating the stagnation, and using slowly-changing betas in the searching vector coefficients, resulting in a hybridized GA-MPC-OX algorithm. Numerical experiment shows that the evolved GA-MPC-OX algorithm used in our relay optimization process gives the best objective values for our specific real world optimization problem, among more than twenty other algorithms.