Description
It has become possible, with the advent of differentiable programming, to create models of experimental apparatus that include the stochastic data-generation processes, the full modeling of the reconstruction and inference procedures, and a suitably defined objective function, along with the cost of any given detector configuration, geometry and materials. This enables the end-to-end optimization of the instruments, by using techniques developed within computer science that are currently vastly exploited in fields such as fluid dynamics. The MODE Collaboration has started to consider the problem in its generality, to provide software architectures that may be useful for the optimization of experimental design. These models may be useful in a "human in the middle" system as they provide information on the relative merit of different configurations as a continuous function of the design choices. In this short contribution we summarize the studies that have been done so far and their potential in the long term.