This is the code for my paper: Adaptive materials design: Exploitation and exploration strategies
Adaptive materials design integrates surrogate model inference, exploitation and exploration strategy, and experimental feedback into a closed-loop framework, which can navigate efficiently the huge chemical design space of high-entropy alloys and hence significantly accelerate the discovery of novel materials. Of particular interest, physical properties can be optimized by exploring unknown compositional regions with the number of validating experiments minimized. Till now, successful discovery of materials through adaptive materials design, based on different datasets, surrogate models and utility functions, insufficiently detailed the underlying strategies. Here, we focus on the exploitation and exploration strategies applied to the Invar alloy dataset, and by training on the same dataset but testing with different strategies, we demonstrate why exploitation and exploration strategies are an important process in adaptive design. We discuss how to balance the exploration and exploitation strategies and elucidate the effects of relevant hyperparameters. We believe this work provides essential guidance for future adaptive materials design. Keywords Adaptive design; Exploitation and exploration strategies; High-entropy alloys; Invar alloys
