Inverse Design of Metamaterials via Deep Learning for Electromagnetically Induced Transparency

Abstract

The physical limitations of metamaterial structures cannot be solved under the conditions of high time cost and complex algorithms in metamaterial inverse engineering in the past. This paper proposes limiting the value range of metamaterial structural parameters through a single structural parameter acquisition method (SSPAM) for the first time, which will meet the expected values of our predictions and obtain high-quality and effective data in a relatively short time. This is the first attempt to use this method to solve the problem of physical limitations in the inverse design of metamaterials effectively. Furthermore, it is a further improvement of inverse design, enhances the reliability of metamaterial inverse design once again, and realizes the idea of on-demand design. The mean squared error of our best deep learning model is 0.00075 and 0.00026 in the training set and validation set, respectively, and 3.0×10−5 in the test set. We input three specific points of the EIT spectrum into our optimal model to predict the corresponding EIT structural parameters inversely, verified by numerical simulation calculation, and obtained satisfactory results. This work can provide new ideas and methods for the inverse design of metamaterials for other models.

Inverse Design of Micro Phononic Beams Incorporating Size Effects via Tandem Neural Network

“…The phrase “deep learning” (DL) was first coined by Hinton [ 26 ], by which remarkable achievements have been obtained in the fields of computer vision [ 27 ], speech recognition [ 28 ], decision making [ 29 ] and so on, showing a promising future for dealing with the problems of the inverse design of realistic structures and materials, as the underlying nature steps away from the data-driven path. Stimulated by this interactive feature without concurrent numerical simulations, the DL model based on a neural network has been widely applied to study the electromagnetic response for given structures [ 30 , 31 ], the constitutive of solid materials [ 32 ], the manipulation of low-frequency acoustic waves [ 33 ], the photonic and phononic topological state [ 34 , 35 ], the electric and magnetic dipoles [ 36 ] and so on, which further promotes the development of the optimization of PCs and MMs for anticipated band gap properties. Regarding one-dimensional PnCs and elastic MMs, various DL NNs based on the multilayer perceptron (MLP) have been established.…”