Correlating metasurface spectra with a generation-elimination framework

Abstract

Inferring optical response from other correlated optical response is highly demanded for vast applications such as biological imaging, material analysis, and optical characterization. This is distinguished from widely-studied forward and inverse designs, as it is boiled down to another different category, namely, spectra-to-spectra design. Whereas forward and inverse designs have been substantially explored across various physical scenarios, the spectra-to-spectra design remains elusive and challenging as it involves intractable many-to-many correspondences. Here, we first dabble in this uncharted area and propose a generation-elimination framework that can self-orient to the best output candidate. Such a framework has a strong built-in stochastically sampling capability that automatically generate diverse nominations and eliminate inferior nominations. As an example, we study terahertz metasurfaces to correlate the reflection spectra from low to high frequencies, where the inaccessible spectra are precisely forecasted without consulting structural information, reaching an accuracy of 98.77%. Moreover, an innovative dimensionality reduction approach is executed to visualize the distribution of the abstract correlated spectra data encoded in latent spaces. These results provide explicable perspectives for deep learning to parse complex physical processes, rather than “brute-force” black box, and facilitate versatile applications involving cross-wavelength information correlation.

Autonomous aeroamphibious invisibility cloak with stochastic-evolution learning

“…In addition, the latent variable z is introduced to build up the probabilistic relationship between x and y. 37,38 CVAE, composed of a recognition module, a latent space, and a reconstruction module, aims to acquire the intractable probability distribution and produce a family of candidates, rather than a single-valued mapping. For a given far-field pattern, the generation network serves as a producer to manufacture metasurface constellation, and the elimination network acts as an inspector to pick out the optimal one.…”

PISC-Net: A Comprehensive Neural Network Framework for Predicting Metasurface Infrared Emission Spectra

“…(1) Average accuracy: 59 × (1 e ) 100% ave (2) where e ave is defined as the average relative error between the predicted spectra and ground-truth spectra, that is…”

“…In addition, for the assessment of the network’s performance, we employ the following criteria to quantify the discrepancy between the target and predicted outcomes. Average accuracy: ( 1 − e a v e ) × 100 % where e ave is defined as the average relative error between the predicted spectra and ground-truth spectra, that is e v a e = 1 N ∑ i = 1 N false| y i − y i ′ false| / y i where y i ′ is the predicted value, y i is the true value, and N is the number of samples input to the model. Similarity: the correlation coefficient between two curves, defined as Similarity = ∑ i = 1 N ( y i − y̅ ) × ( y i ′ − y̅ ′ ) ∑ i = 1 N false( y i − y̅ false) 2 × ∑ i = 1 N …”

Creating Anti‐Chiral Exceptional Points in Non‐Hermitian Metasurfaces for Efficient Terahertz Switching

“…This unique placement grants THz waves the advantage of combining microwave and optical characteristics, resulting in promising applications in high‐speed communication, [ 1 , 2 , 3 ] imaging, [ 4 , 5 ] medical detection [ 6 , 7 ] and spectroscopy. [ 8 , 9 , 10 ] However, the available natural materials for utilization in the THz range are quite limited. In that case, metamaterials composed of subwavelength meta‐atoms have gradually become one of the primary implementation solutions in THz modulation devices due to their compact structures, easy fabrication, and diverse functionalities, such as polarization conversion, [ 11 , 12 ] frequency filtering, [ 13 , 14 ] phase manipulation, [ 15 , 16 , 17 ] holographic imaging, [ 18 , 19 , 20 ] beam focusing, [ 21 , 22 , 23 ] beam steering, [ 24 , 25 , 26 ] and perfect absorption.…”