# Physics-informed neural networks for inverse problems in nano-optics and metamaterials.

@article{Chen2020PhysicsinformedNN, title={Physics-informed neural networks for inverse problems in nano-optics and metamaterials.}, author={Yuyao Chen and Lu Lu and George Em Karniadakis and Luca Dal Negro}, journal={Optics express}, year={2020}, volume={28 8}, pages={ 11618-11633 } }

In this paper, we employ the emerging paradigm of physics-informed neural networks (PINNs) for the solution of representative inverse scattering problems in photonic metamaterials and nano-optics technologies. In particular, we successfully apply mesh-free PINNs to the difficult task of retrieving the effective permittivity parameters of a number of finite-size scattering systems that involve many interacting nanostructures as well as multi-component nanoparticles. Our methodology is fully…

## 109 Citations

Physics-informed neural network for inversely predicting effective electric permittivities of metamaterials

- Physics
- 2021

We apply a physics-informed neural network framework for inversely retrieving the effective material parameters of a two-dimensional metamaterial from its scattered field(s). We show that by…

Deep Learning the Electromagnetic Properties of Metamaterials—A Comprehensive Review

- Materials ScienceAdvanced Functional Materials
- 2021

Deep neural networks (DNNs) are empirically derived systems that have transformed traditional research methods, and are driving scientific discovery. Artificial electromagnetic materials…

WaveY-Net: physics-augmented deep-learning for high-speed electromagnetic simulation and optimization

- PhysicsOPTO
- 2022

We introduce WaveY-Net, a hybrid data- and physics-augmented convolutional neural network that can predict electromagnetic field distributions with ultra fast speeds and high accuracy for entire…

Hyper-parameter tuning of physics-informed neural networks: Application to Helmholtz problems

- Computer ScienceArXiv
- 2022

A hyper-parameter tuning procedure via Gaussian processes-based Bayesian optimization is applied to Helmholtz problems for bounded domains and a thorough study is conducted, focusing on performance and the applicability and necessity of the method.

Physics-informed Neural Network for Nonlinear Dynamics in Fiber Optics

- PhysicsArXiv
- 2021

The results report here show that the PINN is not only an effective partial differential equation solver, but also a prospective technique to advance the scientific computing and automatic modeling in fiber optics.

Physics-informed neural networks for imaging and parameter retrieval of photonic nanostructures from near-field data

- Physics
- 2021

Physics-informed neural networks for imaging and parameter retrieval of photonic nanostructures from near-field data Yuyao Chen1 and Luca Dal Negro1, 2, 3 1)Department of Electrical & Computer…

Physics-informed neural networks for solving parametric magnetostatic problems

- Computer ScienceArXiv
- 2022

The objective of this paper is to investigate the ability of physics-informed neural networks to learn the magnetic field response as a function of design parameters in the context of a two-dimensional (2-D) magnetostatic problem.

Solving the wave equation with physics-informed deep learning

- Geology
- 2020

We investigate the use of Physics-Informed Neural Networks (PINNs) for solving the wave equation. Whilst PINNs have been successfully applied across many physical systems, the wave equation presents…

Physics-informed neural networks with hard constraints for inverse design

- Computer ScienceSIAM J. Sci. Comput.
- 2021

This work proposes a new deep learning method—physics-informed neural networks with hard constraints (hPINNs)—for solving topology optimization and demonstrates the effectiveness of hPINN for a holography problem in optics and a fluid problem of Stokes flow.

Modified physics-informed neural network method based on the conservation law constraint and its prediction of optical solitons

- Physics
- 2021

Based on conservation laws as one of the important integrable properties of nonlinear physical models, we design a modified physics-informed neural network method based on the conservation law…

## References

SHOWING 1-10 OF 71 REFERENCES

Past Achievements and Future Challenges in 3D Photonic Metamaterials

- Physics
- 2011

Photonic metamaterials are man-made structures composed of tailored micro- or nanostructured metallo-dielectric sub-wavelength building blocks that are densely packed into an effective material. This…

Past achievements and future challenges in the development of three-dimensional photonic metamaterials

- Physics
- 2011

Photonic metamaterials are man-made structures composed of tailored micro- or nanostructured metallodielectric subwavelength building blocks. This deceptively simple yet powerful concept allows the…

Embedding Deep Learning in Inverse Scattering Problems

- Computer Science, MathematicsIEEE Transactions on Computational Imaging
- 2020

A novel convolutional neural network architecture is proposed, termed the contrast source network, that learns the noise space components of the radiation operator that helps in producing high resolution solutions without any significant increase in computational costs.

Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations

- Computer ScienceJ. Comput. Phys.
- 2019

Inverse design in nanophotonics

- PhysicsNature Photonics
- 2018

A cross-section of key developments in this emerging field of photonic optimization is outlined: moving from a recap of foundational results to motivation of applications in nonlinear, topological, near-field and on-chip optics.

Plasmonic cloaking of cylinders: Finite length, oblique illumination and cross-polarization coupling

- Physics
- 2010

Metamaterial cloaking has been proposed and studied in recent years by following several interesting approaches. One of them, the scattering-cancelation technique or plasmonic cloaking, exploits the…

Efficient and accurate inversion of multiple scattering with deep learning

- Computer ScienceOptics express
- 2018

This paper designs and training a deep convolutional neural network that can invert multiple scattered measurements to produce a high-quality image of the refractive index and shows that the proposed approach is substantially faster and achieves higher imaging quality compared to the state-of-the-art methods based on optimization.

Invisibility and Cloaking: Origins, Present, and Future Perspectives

- Physics
- 2015

The development of metamaterials, i.e., artificially structured materials that interact with waves in unconventional ways, has revolutionized our ability to manipulate the propagation of…

Effective medium theory for two-dimensional non-magnetic metamaterial lattices up to quadrupole expansions

- Physics
- 2015

We present a formulation for deriving effective medium properties of infinitely periodic two-dimensional metamaterial lattice structures beyond the conventional static and quasi-static limits. We…

REVIEW ARTICLE: Plasmonic and metamaterial cloaking: physical mechanisms and potentials

- Chemistry
- 2008

Some of the different cloaking mechanisms recently proposed in the literature, such as plasmonic cloaking based on scattering cancelation, coordinate-transformation cloaking and anomalous localized resonances for cloaking are overviewed, with particular attention to the physical phenomena involved.