An official website of the United States government
Abstract HiLAB(Hybridinverse‐design withLatent‐space learning,Adjoint‐based partial optimizations, andBayesian optimization), is a new paradigm for inverse design of nanophotonic structures. Combining early‐terminated topological optimization (TO) with a Vision Transformer–based variational autoencoder (VAE) and a Bayesian search, HiLAB addresses multifunctional device design by generating diverse freeform configurations at reduced simulation costs. Shortened adjoint‐driven TO runs, coupled with randomized physical parameters, produce robust initial structures. These structures are compressed into a compact latent space by the VAE, enablingBayesian optimizationtoco‐optimize geometry and physical hyperparameters. Crucially, the trained VAE can be reused for alternative objectives or constraints by adjusting only the acquisition function. Compared to conventional TO pipelines prone to local optima, HiLAB systematically explores near‐global optima with considerably fewer electromagnetic simulations. Even after accounting for training overhead, the total number of full electromagnetic simulations decreases by an order of magnitude, accelerating the discovery of fabrication‐friendly devices. Demonstrating its efficacy, HiLAB is used to design an achromatic beam deflector for red, green, and blue wavelengths, achieving balanced diffraction efficiencies of ∼25% while mitigating chromatic aberrations, a performance surpassing existing demonstrations. Overall, HiLAB provides a flexible platform for robust, multi‐parameter photonic designs and rapid adaptation to next‐generation nanophotonic challenges.
more »
« less
Fast Inverse Design of 3D Nanophotonic Devices Using Boundary Integral Methods
Recent developments in the computational automated design of electromagnetic devices, otherwise known as inverse design, have significantly enhanced the design process for nanophotonic systems. Inverse design can both reduce design time considerably and lead to high-performance, nonintuitive structures that would otherwise have been impossible to develop manually. Despite the successes enjoyed by structure optimization techniques, most approaches leverage electromagnetic solvers that require significant computational resources and suffer from slow convergence and numerical dispersion. Recently, a fast simulation and boundary-based inverse design approach based on boundary integral equations was demonstrated for two-dimensional nanophotonic problems. In this work, we introduce a new full-wave three-dimensional simulation and boundary-based optimization framework for nanophotonic devices also based on boundary integral methods, which achieves high accuracy even at coarse mesh discretizations while only requiring modest computational resources. The approach has been further accelerated by leveraging GPU computing, a sparse block-diagonal preconditioning strategy, and a matrix-free implementation of the discrete adjoint method. As a demonstration, we optimize three different devices: a 1:2 1550 nm power splitter and two nonadiabatic mode-preserving waveguide tapers. To the best of our knowledge, the tapers, which span 40 wavelengths in the silicon material, are the largest silicon photonic waveguiding devices to have been optimized using full-wave 3D solution of Maxwell’s equations.
more »
« less
- PAR ID:
- 10382209
- Date Published:
- Journal Name:
- ACS Photonics
- ISSN:
- 2330-4022
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract In recent years, wave-based analog computing has been at the center of attention for providing ultra-fast and power-efficient signal processing enabled by wave propagation through artificially engineered structures. Building on these structures, various proposals have been put forward for performing computations with waves. Most of these proposals have been aimed at linear operations, such as vector-matrix multiplications. The weak and hardly controllable nonlinear response of electromagnetic materials imposes challenges in the design of wave-based structures for performing nonlinear operations. In the present work, first, by using the method of inverse design we propose a three-port device, which consists of a combination of linear and Kerr nonlinear materials, exhibiting the desired power-dependent transmission properties. Then, combining a proper arrangement of such devices with a collection of Mach–Zehnder interferometers (MZIs), we propose a reconfigurable nonlinear optical architecture capable of implementing a variety of nonlinear functions of the input signal. The proposed device may pave the way for wave-based reconfigurable nonlinear signal processing that can be combined with linear networks for full-fledged wave-based analog computing.more » « less
-
Abstract The radiative characteristics of the radiofrequency receive coils dictate the signal-to-noise ratio (SNR) of magnetic resonance images. Despite the crucial importance of RF coils, the practical coil design process has remained a largely empirical one. This work introduces a novel optimization framework for rational coil design, which relies on a fully automated pipeline that combines rapid electromagnetic simulations, shape optimization and coil meshing. The objective function iteratively maximizes SNR performance in a target region of interest with respect to the ultimate intrinsic SNR, which is the theoretically highest SNR independent from any particular coil design. The forward simulation employs a fast electromagnetic solver based on coupled surface and volume integral equations. The coils are represented as B-spline curves with an associated width, and automatically meshed for EM simulation. We implemented a new method to tune and decouple coils at each iteration without manual user intervention. The algorithm optimizes the size and position of a given number of coils with a combination of grid search and a line search. We demonstrated the framework by designing receive arrays of increasing complexity that yield optimal SNR for different target regions inside a numerical head model. SNR simulation time ranged from 15 s for a 3-coil configuration to 32 s for a 12-coil array, constrained to a helmet-like surface, including tuning and decoupling. The optimized 12-coil geometry yielded 9% higher average SNR performance in the brain at 3 T. This work represents the first automated coil optimization framework that uses full-wave electromagnetic simulations and ultimate performance benchmarks. This novel approach enables the systematic design of coils for magnetic resonance imaging with significantly improved SNR performance, potentially transforming coil development from empirical design to physics-driven optimization.more » « less
-
From higher computational efficiency to enabling the discovery of novel and complex structures, deep learning has emerged as a powerful framework for the design and optimization of nanophotonic circuits and components. However, both data-driven and exploration-based machine learning strategies have limitations in their effectiveness for nanophotonic inverse design. Supervised machine learning approaches require large quantities of training data to produce high-performance models and have difficulty generalizing beyond training data given the complexity of the design space. Unsupervised and reinforcement learning-based approaches on the other hand can have very lengthy training or optimization times associated with them. Here we demonstrate a hybrid supervised learning and reinforcement learning approach to the inverse design of nanophotonic structures and show this approach can reduce training data dependence, improve the generalizability of model predictions, and significantly shorten exploratory training times. The presented strategy thus addresses several contemporary deep learning-based challenges, while opening the door for new design methodologies that leverage multiple classes of machine learning algorithms to produce more effective and practical solutions for photonic design.more » « less
-
Consider the scattering of a time-harmonic acoustic plane wave by a bounded elastic obstacle which is immersed in a homogeneous acoustic medium. This paper is concerned with an inverse acoustic-elastic interaction problem, which is to determine the location and shape of the elastic obstacle by using either the phased or phaseless far-field data. By introducing the Helmholtz decomposition, the model problem is reduced to a coupled boundary value problem of the Helmholtz equations. The jump relations are studied for the second derivatives of the single-layer potential in order to deduce the corresponding boundary integral equations. The well-posedness is discussed for the solution of the coupled boundary integral equations. An efficient and high order Nyström-type discretization method is proposed for the integral system. A numerical method of nonlinear integral equations is developed for the inverse problem. For the case of phaseless data, we show that the modulus of the far-field pattern is invariant under a translation of the obstacle. To break the translation invariance, an elastic reference ball technique is introduced. We prove that the inverse problem with phaseless far-field pattern has a unique solution under certain conditions. In addition, a numerical method of the reference ball technique based nonlinear integral equations is proposed for the phaseless inverse problem. Numerical experiments are presented to demonstrate the effectiveness and robustness of the proposed methods.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript
- Journal Article:
- https://doi.org/10.1021/acsphotonics.2c01072
