What are the factors to be considered while choosing one among the multi-objective optimization solvers - paretosearch() and gamultiobj()?

What are the merits and demerits of gamultiobj() and paretosearch()?

How does the designer choose among these?

How do these behave when single objective functions are provided?

Answer 1

[Gayatri]

Hi Dhanesh,

When deciding between 'paretosearch' and 'gamultiobj' for solving multi-objective optimization problems, consider the following factors:

1. 'gamultiobj' is based on genetic algorithms. It is well-suited for complex and non-differentiable objective functions. 'paretosearch' uses a pattern search method and might be more efficient for problems where gradient information is absent but the problem is less complex.
2. 'gamultiobj'can provide a broader diversity of solutions in the Pareto front due to its genetic algorithm nature, which explores a wider solution space through mechanisms like crossover and mutation. 'paretosearch' may not explore the solution space as thoroughly as 'gamultiobj', potentially leading to less diversity in the Pareto front.
3. 'gamultiobj' might require more computational resources due to its evolutionary approach than the 'paretosearch'.
4. 'paretosearch' can be more efficient than 'gamultiobj' for certain types of problems, particularly with fewer local optima.

Please refer the below documentation for 'Compare paretosearch and gamultiobj': https://www.mathworks.com/help/gads/compare-paretosearch-gamultiobj.html

The choice between 'paretosearch' and 'gamultiobj' largely depends on the specific characteristics of the optimization problem at hand.

Please refer the below documentations for 'paretosearch' and 'gamultiobj':

• https://www.mathworks.com/help/gads/paretosearch.html
• https://www.mathworks.com/help/gads/gamultiobj.html

I hope it helps!

[Dhanesh R]

Thank you @Gayatri