Because our system includes a number of nonlinear terms, the Levenberg-Marquardt is not optimal, as it may get stuck in local minima. In our implementation, we mitigated this issue with multiple starts to increase the likelihood of finding the global minimum. However, we also recommend future work to explore global optimization methods such as Simulated annealing or the Genetic Algorithm that may be more appropriate for this nonlinear system.
- Optimization methods we used (e.g. Levenberg - Marquardt optimization routine from scipy)