In this paper, we theoretically design a dual-band graphene-based terahertz (THz) absorber combining the magnetic resonance with a THz cold mirror without any metallic loss. The absorption spectrum of the all-dielectric THz absorber can be actively manipulated after fabrication due to the tunable conductivity of graphene. After delicate optimization, two ultra-narrow absorption peaks are achieved with respective full width at half maximum (FWHM) of 0.0272 THz and 0.0424 THz. Also, we investigate the effect of geometric parameters on the absorption performance. Coupled mode theory (CMT) is conducted on the dual-band spectrum as an analytic method to confirm the validity of numerical results. Furthermore, physical mechanism is deeply revealed with magnetic and electric field distributions, which demonstrate a totally different principle with traditional plasmonic absorber. Our research provides a significant design guide for developing tunable multi-resonant THz devices based on all-dielectric configuration.

Although neural networks have been applied in many fields since they were first introduced, the feasibility of applying it to predict the solution of Maxwell's equations remains open. In this letter, we investigate the feasibility of utilizing the long- and short-term memory (LSTM) neural network to solve the time-domain electromagnetic (TDEM) forward problems. With ground truth datasets being generated from the finite-difference time-domain (FDTD) method, a novel LSTM-TDEM model structure is proposed, and trained by a new feasible algorithm specifically designed for time-domain simulation. The strong approximation ability of the LSTM-TDEM method can accurately predict the electromagnetic field distributions for topologies of different geometries, materials, and excitations at different locations. The effectiveness of the proposed LSTM-TDEM method has been validated by several numerical experiments. The average relative error can be as small as 0.63% for 2-D case and 0.35% for 3-D case. It is worth noting that the training data are only 5% (198/4087) and 30% (660/2189) in 2-D and 3-D cases, respectively. Meanwhile, compared with the traditional FDTD method, the proposed method greatly reduces the calculation time, with its speedup ratio more than 1800 and 44 000 times over the FDTD method in 2-D and 3-D cases, respectively.

An active toroidal switch is numerically presented in a terahertz metamaterial. Transmission of toroidal dipole can be tuned by the conductivity of vanadium dioxide, and its modulation depth can be up to 32%. The resonance of toroidal dipole is verified by the distribution of electric current. Due to the unique field distribution of the toroidal mode, the proposed design may find potential applications in enhanced absorption and sensing.

In this article, an electromagnetic–thermal coupling algorithm based on the discontinuous Galerkin time-domain (DGTD) and spectral-element time-domain (SETD) methods is proposed, which can adopt two independent grids for thermal simulation and electrical simulation, respectively. It employs time-domain solvers for both the electrical and thermal fields. Compared with the conventional cosimulation scheme (e.g., time-domain solvers for the thermal field and frequency-domain solver for the electrical field), it can provide better simulation precision for problems characterized with small size or drastic field changes. By comparing the results of electromagnetic–thermal coupling simulation with the results of electromagnetic simulation, the effect of temperature change on electromagnetic field propagation is analyzed.

One fundamental difficulty in multiphysics numerical simulation is the complex interactions between different physics domains leading to plenty of computational costs. Although neural networks have recently been introduced in multiphysics simulations, the modeling complexity and the enormous amount of training data required may still pose significant challenges to researchers. In this work, we introduce a low-cost, electromagnetic-centric, multiphysics modeling approach to simulate microwave filters. With ground-truth datasets being generated from the finite element method, a novel deep hybrid neural network (DHNN) model structure is introduced, which uses the sigmoid and the ReLU functions as activators to mimic the diversity of biological neurons. A new, more feasible training algorithm is proposed for the efficient development of the DHNN model. The algorithm adopts the design-of-experiment (DOE) sampling technique and is specifically designed for the simulation of multiphysics responses. The strong approximation ability of the DHNN can lead to high-accuracy modeling with fewer training data and less resource consumption. Another advantage of this approach is that the modeling process is more concise and easier to apply compared with other modeling technologies. Numerical examples show that the DHNN can achieve higher accurate results with much less training data compared to traditional ANNs. The advantages of the proposed method in computational efficiency are more pronounced, especially when the amount of input data increases.

Two-dimensional (2D) materials, due to their unique electronic, optical and structural properties, have attracted extensive attention of researchers in the world. However, most of 2D materials have low optical absorption efficiencies in the visible and near-infrared regimes, which leads to the weak light–matter interaction and limits their further applications in optoelectronic devices. Thus, enhancing the light–matter interaction of various 2D materials in the visible and near-infrared regimes, has been a key topic for many optoelectronic equipment and related applications. In this topical review, we summarized the recent developments of the 2D materials-based optical absorbers in the visible and near infrared regimes, focusing mainly on the methods and relevant physical mechanisms of several typical perfect absorbers, such as narrowband perfect absorbers, dual-band perfect absorbers, and broadband perfect absorbers. Finally, several prospective research directions from our perspectives are presented at the end.