Some designs has many parameters and sometimes it is not suitable to tune it.
Could you advise which optimizer better in terms of minimum time-efficiency?
There are six optimizer in HFSS.
1. Quasi Newton (Gradient)
2.Pattern Search (Search-based)
3.Genetic Algorithm (Random search)
4.Sequential Nonlinear Programming (Gradient)
5.Sequential Mixed Integer Nonlinear Programming (Gradient and Discrete)
6.MATLAB
i suggest u GA or any kind of random search methods such as ACO or PSO. if your optimization problem is complex and has many design parameters random optimization methods have more efficient performance.
I suggest you that if your programming problem is Quadratic in nature then you must follow Newton method for optimizing. Because newton method give the optimal solution for quadratic programming in one iteration. On the other hand, if your problem is discrete programming then i would like to suggest you to go with Dynamic programming or parametric Geometric programming. Finally, if your optimization function is highly complicated, non-differentiable and problem is NP hard problem then I suggest you to go with Evolutionary algorithms such as GA, PSO, ABC, ACO etc. depending upon the problem and the number of parameter required for the execution.
Best regards,
Harish Garg
i agree with harish
depending upon the problem you can choose your method.
Random search like GA and many recent bio inspired algorithms yield better results
- Badges
- Science topic