Abstract

In the first part of this talk, I will present our recent advances in the coherent perfect-absorber and laser (CPAL) for electromagnetic and elastodynamic waves. This intriguing effect is enabled by parity-time (PT)-symmetry breaking condition. We first propose to accomplish ultra-sensitivity and robustness to noise by demonstrating a CPAL -locked radiofrequency sensor to detect extremely small-scale reactive or conductive perturbations. The results show that the sensitivity and resolvability of the CPAL -locked sensor may go well beyond the traditional rf sensors. Moreover, we extend the concept for flexural waves, and show that the lasing operation of CPAL device can first observed theoretically and validated numerically (3D COMSOL full elasticity simulations) in elastic thin-plates.  In the second part, I will shortly discuss a generative deep learning approach for shape recognition of an arbitrary object from its acoustic scattering amplitudes. The model exploits adversarial learning and variational inference to predict the unique shape of the object. The nonunique solution space is overcome by the multiangle and multifrequency phaseless far-field patterns.