Estimation of Microstructural Properties of Wormlike Micelles Via a Multi-Scale Multi-Recommendation Batch Bayesian Optimization

초록

Microstructural properties of wormlike micelles (WLMs), which are employed in characterizing the system to predict rheological properties, have long been obtained via elusive experiments such as small-angle neutron scattering. Hence, in this work, a framework to explicitly obtain such properties from macroscopic rheology measurements was developed. Specifically, the parameters of a mesoscopic pointer-based algorithm, which can predict the linear rheology of WLMs were obtained with the aid of a multi-scale multi-recommendation (MSMR) batch Bayesian optimization (BO) methodology. From three case studies, it was observed that the MSMR batch BO was able to obtain a set of parameters, which showed high prediction accuracy, in comparison to a sequential BO. Specifically, it was found that microstructural properties such as persistent length and the diameter of WLMs were successfully predicted by the proposed framework. Hence, this framework can be utilized in characterizing various WLM systems from readily available macroscopic rheological measurements.

키워드