With growing number of applications relying on complex simulations in areas ranging from science to engineering design, there is a need for algorithms and software implementations that can effectively optimize simulation output with minimal computational efforts. Our work in this context has two aspects:
We develop algorithms that rely solely on simulation output. This class of algorithms is called derivative-free optimization algorithms. We investigate computationally attractive yet accurate surrogate models to represent these simulations that drive efficient derivative-free optimization.
Because of complex interactions between entities in a supply chain, optimal decision making in a supply chain is a difficult task. With the help of derivative-free optimization algorithms, we optimize supply chain simulations that incorporate detailed supply chain dynamics for optimal inventory management.