Inverse calculation of liquid crystal parameters using scientific machine learning

Liquid crystals are fundamental in all modern opto-electronic devices. They have the peculiar property of being intermediate between liquids and solids: in fact, the optic axis of the liquid crystal can be changed locally by applying external perturbations and stimuli [1]. An accurate estimate of the electro-optical properties is of primary importance for both for applications as well as fundamental science [2]. For example, slight discrepancies in the elastic constants can predict huge variation in the phase retardation imparted on the optical field, when used as a tunable waveplate. Scientific machine learning (SciML) has recently found application in all areas of science. Here we use SciML [3] to characterize a liquid crystal cell and go beyond the standard techniques which usually have no access to the point by point optic axis orientation inside the cell.