The Genetic Options section within the Portfolio Builder settings is available only when Genetic Optimization is selected. These options allow users to fine-tune the parameters of the genetic algorithm to optimize the portfolio building process.
Â
Genetic Options (for Genetic Optimization)
Â
Max Number of Generations
Description: This parameter defines the maximum number of generations the genetic algorithm will create before termination. A generation in this context refers to a complete cycle of selection, crossover, mutation, and evaluation.
- Purpose: Setting a higher number of generations allows the algorithm more opportunities to refine and improve the portfolio solutions. However, this also increases the computational time required.
- Use Case: Users should set the number of generations based on the complexity of the optimization problem and the computational resources available.
Â
Population Size per Island
Description: This parameter specifies the number of optimizing object instances (portfolios) created per island. The genetic algorithm in this module uses an island architecture, where multiple sub-populations (islands) evolve in parallel.
- Purpose: A larger population size increases the diversity of solutions but also requires more computational power and time for each generation.
- Use Case: Adjust the population size to balance between solution diversity and computational feasibility.
Â
Crossover Probability
Description: The probability that two optimizing object instances (portfolios) will exchange their genes (parameters). This process, known as crossover, combines the features of two parent portfolios to create new offspring portfolios.
- Purpose: Crossover introduces variability and helps explore new combinations of parameters, potentially leading to better solutions.
- Use Case: Set a high crossover probability to promote diversity and exploration, but not so high that it disrupts the convergence of good solutions.
Â
Mutation Probability
Description: The probability that the genes (parameters) of an optimizing instance (portfolio) will mutate. Mutation introduces random changes to some of the parameters, helping to maintain genetic diversity within the population.
- Purpose: Mutation helps prevent the algorithm from getting stuck in local optima by exploring new areas of the parameter space.
- Use Case: Set a moderate mutation probability to balance between maintaining diversity and converging towards optimal solutions.
Â
Island Number
Description: The number of islands in the genetic architecture. Each island is an isolated sub-population that evolves independently for a certain number of generations.
- Purpose: Using multiple islands helps maintain diversity across the entire population and allows for parallel processing.
- Use Case: Increase the number of islands to enhance diversity and parallelism, but be mindful of the increased computational requirements.
Â
Migration Interval
Description: The number of generations that should pass before optimization instances migrate between islands. Migration allows individuals from one island to move to another, promoting genetic diversity and the sharing of good solutions.
- Purpose: Regular migration helps avoid premature convergence and ensures that good solutions are propagated across the islands.
- Use Case: Set an appropriate migration interval to balance between isolated evolution and the sharing of solutions.
Â
Population Migration Rate
Description: The percentage of the island population that will migrate to another island during the migration process.
- Purpose: A higher migration rate increases the flow of genetic material between islands, promoting diversity. However, too high a rate can reduce the benefits of isolated evolution.
- Use Case: Adjust the migration rate to ensure a healthy exchange of solutions without overwhelming the unique evolutionary processes of each island.
Â
Summary
The Genetic Options section in the Portfolio Builder settings provides detailed control over the genetic algorithm's parameters, allowing users to tailor the optimization process to their specific needs. By adjusting these parameters, users can influence the balance between exploration and exploitation, manage computational resources, and enhance the overall effectiveness of the portfolio optimization process.