Any design process can be mathematically modelled as an optimisation process and can then be solved using a computer. Optimisation methods can be constrained or unconstrained and local or global in scope. The goal of a design process is to minimise/maximise a cost function. Typically, a cost function is formulated to reflect the level of proximity of the system’s performance to the desired targets. The design variables are updated iteratively through carefully devised steps specific to each optimisation mechanism that ensure a convergence to a local or global optimum.
Key concepts covered include:
- Definition of a maximisation/minimisation problem
- Constrained, unconstrained optimisation
- Local optimisation methods based on gradient-search techniques
- Intelligent global search techniques
- Travelling salesman problem as an example of discrete optimisation
- Function optimisation methods, implementations in photonics problems
- Topological and shape optimisation for fabrication processes
- Coding in Python/Matlab and simulation training