I will discuss the role of network architecture in the GLOnet inverse optimization platform, in which the global optimization process is reframed as the training of a generative neural network. I will show how a properly selected network architecture can smoothen the design space and how the architecture can be tailored based on the type and dimensionality of the design problem. I will also discuss new methods in which neural networks can serve as high speed surrogate Maxwell solvers capable of aiding the inverse design process. These hybrid physics- and data-driven concepts can apply to a broad range of nanophotonics systems.
{"title":"Disrupting the photonics innovation cycle with data- and physics-driven algorithms","authors":"Jonathan A. Fan","doi":"10.1117/12.2595667","DOIUrl":"https://doi.org/10.1117/12.2595667","url":null,"abstract":"I will discuss the role of network architecture in the GLOnet inverse optimization platform, in which the global optimization process is reframed as the training of a generative neural network. I will show how a properly selected network architecture can smoothen the design space and how the architecture can be tailored based on the type and dimensionality of the design problem. I will also discuss new methods in which neural networks can serve as high speed surrogate Maxwell solvers capable of aiding the inverse design process. These hybrid physics- and data-driven concepts can apply to a broad range of nanophotonics systems.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114752235","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
D. G. Pires, Jiannan Gao, Jane Peabody, N. Chandra, N. Litchinitser
In this talk we theoretically and experimentally investigate an interesting family of null solutions to Helmholtz equation in 3D free space - optical vortices, or zero lines of complex amplitude in a propagating light field, that are knotted or linked in a certain way. We design all-dielectric optical metasurfaces – nanostructures enabling unprecedented control over the amplitude, polarization and phase of optical fields, for generation of optical knots, and study their stability and evolution in engineered colloidal suspensions with saturable Kerr-like nonlinearity. These studies are further generalized to characterization of knot evolution in turbid linear and nonlinear media, such as clouds, fog, biological media, and undersea environments. Knotted electromagnetic fields may find applications in three-dimensional optical manipulations or could be considered as candidates for new information carriers in classical and quantum communication systems.
{"title":"Optical knots in engineered turbid media","authors":"D. G. Pires, Jiannan Gao, Jane Peabody, N. Chandra, N. Litchinitser","doi":"10.1117/12.2596430","DOIUrl":"https://doi.org/10.1117/12.2596430","url":null,"abstract":"In this talk we theoretically and experimentally investigate an interesting family of null solutions to Helmholtz equation in 3D free space - optical vortices, or zero lines of complex amplitude in a propagating light field, that are knotted or linked in a certain way. We design all-dielectric optical metasurfaces – nanostructures enabling unprecedented control over the amplitude, polarization and phase of optical fields, for generation of optical knots, and study their stability and evolution in engineered colloidal suspensions with saturable Kerr-like nonlinearity. These studies are further generalized to characterization of knot evolution in turbid linear and nonlinear media, such as clouds, fog, biological media, and undersea environments. Knotted electromagnetic fields may find applications in three-dimensional optical manipulations or could be considered as candidates for new information carriers in classical and quantum communication systems.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127995444","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Chalcogenide based materials are excellent candidates for implementing static and dynamic meta-optics as they possess very high permittivities and support large modulation of optical constants through various mechanisms such as, phase-change, photon-darkening, laser writing and anomalous thermo-optic effects. We present a study of various chalcogenide compositions used for static and active metasurfaces. We start with large area CVD grown amorphous Selenium nanoparticles on various substrates and show that their Mie-resonant response spans the entire mid-infrared range. By coupling Se Mie-resonators to ENZ substrates we demonstrate an order of magnitude increase in quality factor. Next, we investigate topological insulators Bi2Te3 metasurfaces and demonstrate that these high permittivity metasurfaces can yield very large absorption resonances that are tunable in the infrared range. Finally, we demonstrate ultra-wide dynamic tuning of PbTe metasurface resonators.
{"title":"Static and active chalcogenide based meta-optics","authors":"T. Lewi","doi":"10.1117/12.2597294","DOIUrl":"https://doi.org/10.1117/12.2597294","url":null,"abstract":"Chalcogenide based materials are excellent candidates for implementing static and dynamic meta-optics as they possess very high permittivities and support large modulation of optical constants through various mechanisms such as, phase-change, photon-darkening, laser writing and anomalous thermo-optic effects. We present a study of various chalcogenide compositions used for static and active metasurfaces. We start with large area CVD grown amorphous Selenium nanoparticles on various substrates and show that their Mie-resonant response spans the entire mid-infrared range. By coupling Se Mie-resonators to ENZ substrates we demonstrate an order of magnitude increase in quality factor. Next, we investigate topological insulators Bi2Te3 metasurfaces and demonstrate that these high permittivity metasurfaces can yield very large absorption resonances that are tunable in the infrared range. Finally, we demonstrate ultra-wide dynamic tuning of PbTe metasurface resonators.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"130 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116018713","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A central challenge in the development of nanophotonic structures and metamaterials is identifying the optimal design for a target functionality and understanding the physical mechanisms that enable the optimized device’s capabilities. In this talk, we will describe deep learning-driven strategies to both design complex nanophotonic structures, including across multiple device categories, as well as understand their behavior. We will highlight potential pathways to making deep learning a tool for global inverse design across multiple device categories, while also opening up the 'black box' of the machine learning algorithm to understand why a particular optimized design works well.
{"title":"Deep learning to explain and design complex nanophotonic structures","authors":"A. Raman","doi":"10.1117/12.2595477","DOIUrl":"https://doi.org/10.1117/12.2595477","url":null,"abstract":"A central challenge in the development of nanophotonic structures and metamaterials is identifying the optimal design for a target functionality and understanding the physical mechanisms that enable the optimized device’s capabilities. In this talk, we will describe deep learning-driven strategies to both design complex nanophotonic structures, including across multiple device categories, as well as understand their behavior. We will highlight potential pathways to making deep learning a tool for global inverse design across multiple device categories, while also opening up the 'black box' of the machine learning algorithm to understand why a particular optimized design works well.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116236377","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Structured light carrying spin and orbital angular momentum brings about new light-matter interactions in optical nanostructures. We demonstrate the possibility of using structured light beams carrying orbital angular momentum (OAM) to access resonant modes of all-dielectric meta-atoms that cannot be excited by the conventional Gaussian beam or by a plane wave. We use multipole decomposition approach to match extinction resonances with high-order multipole excitation. These results can find applications in sensing, spectroscopy, and enable new regimes of nonlinear optical interactions.
{"title":"Excitation of all-dielectric meta-atoms with structured light beams","authors":"P. Terekhov, N. Chandra, N. Litchinitser","doi":"10.1117/12.2594870","DOIUrl":"https://doi.org/10.1117/12.2594870","url":null,"abstract":"Structured light carrying spin and orbital angular momentum brings about new light-matter interactions in optical nanostructures. We demonstrate the possibility of using structured light beams carrying orbital angular momentum (OAM) to access resonant modes of all-dielectric meta-atoms that cannot be excited by the conventional Gaussian beam or by a plane wave. We use multipole decomposition approach to match extinction resonances with high-order multipole excitation. These results can find applications in sensing, spectroscopy, and enable new regimes of nonlinear optical interactions.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126913679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Y. Kiarashi, Mohammadreza Zandehshahvar, Muliang Zhu, H. Maleki, Sajjad Abdollahramezani, Tyler Brown, Reid Barton, A. Adibi
A new deep-learning approach based on dimensionality reduction techniques for the design and knowledge discovery in nanophotonic structures will be presented. It is shown that reducing the dimensionality of the response and design spaces in a class of nanophotonic structures can provide new insight into the physics of light-matter interaction in such nanostructures while facilitating their inverse design. These unique features are achieved while considerably reducing the computation complexity through dimensionality reduction. It is also shown that this approach can enable an evolutionary design method in which the initial design can be evolved intelligently into an alternative with favorable specification like less complexity, more robustness, less power consumption, etc. In addition to providing the details about the fundamental aspects of the latent learning approach, its application to design of reconfigurable metasurfaces will be demonstrated.
{"title":"Latent learning for design and knowledge discovery in nanophotonics","authors":"Y. Kiarashi, Mohammadreza Zandehshahvar, Muliang Zhu, H. Maleki, Sajjad Abdollahramezani, Tyler Brown, Reid Barton, A. Adibi","doi":"10.1117/12.2595199","DOIUrl":"https://doi.org/10.1117/12.2595199","url":null,"abstract":"A new deep-learning approach based on dimensionality reduction techniques for the design and knowledge discovery in nanophotonic structures will be presented. It is shown that reducing the dimensionality of the response and design spaces in a class of nanophotonic structures can provide new insight into the physics of light-matter interaction in such nanostructures while facilitating their inverse design. These unique features are achieved while considerably reducing the computation complexity through dimensionality reduction. It is also shown that this approach can enable an evolutionary design method in which the initial design can be evolved intelligently into an alternative with favorable specification like less complexity, more robustness, less power consumption, etc. In addition to providing the details about the fundamental aspects of the latent learning approach, its application to design of reconfigurable metasurfaces will be demonstrated.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124895983","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Supervised neural networks are rising as an algorithm of choice for surrogate models in photonics, because they are versatile, fast to evaluate, easily differentiable, and perform well in high-dimensional problems. However, the drawback of this black box approach is that it requires a lot of data. Unfortunately in the context of photonics, data is generated through expensive full solves of Maxwell’s equations. This talk will present ways to open the black box for better data efficiency and performance of deep surrogate models. The first part of this talk will present how active learning can reduce the need for data by at least an order of magnitude by adapting the data generation to the model learning. The second part will present how information about the physics can be incorporated into the neural network for more efficiency.
{"title":"Opening the black box for data efficiency and inverse design in photonics","authors":"R. Pestourie, Steven G. Johnson","doi":"10.1117/12.2592805","DOIUrl":"https://doi.org/10.1117/12.2592805","url":null,"abstract":"Supervised neural networks are rising as an algorithm of choice for surrogate models in photonics, because they are versatile, fast to evaluate, easily differentiable, and perform well in high-dimensional problems. However, the drawback of this black box approach is that it requires a lot of data. Unfortunately in the context of photonics, data is generated through expensive full solves of Maxwell’s equations. This talk will present ways to open the black box for better data efficiency and performance of deep surrogate models. The first part of this talk will present how active learning can reduce the need for data by at least an order of magnitude by adapting the data generation to the model learning. The second part will present how information about the physics can be incorporated into the neural network for more efficiency.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"113 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124748622","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Willie J Padilla, Yang Deng, Simiao Ren, Jordan M. Malof
Deep neural networks are empirically derived systems that have transformed research methods and are driving scientific discovery. Artificial electromagnetic materials, such as electromagnetic metamaterials, photonic crystals, and plasmonics, are research fields where deep neural network results evince the data driven approach; especially in cases where conventional computational and optimization methods have failed. We propose and demonstrate a deep learning method capable of finding accurate solutions to ill-posed inverse problems, where the conditions of existence and uniqueness are violated. A specific example of finding the metasurface geometry which yields a radiant exitance matching the external quantum efficiency of gallium antimonide is demonstrated.
{"title":"Deep learning and inverse design of artificial electromagnetic materials","authors":"Willie J Padilla, Yang Deng, Simiao Ren, Jordan M. Malof","doi":"10.1117/12.2593081","DOIUrl":"https://doi.org/10.1117/12.2593081","url":null,"abstract":"Deep neural networks are empirically derived systems that have transformed research methods and are driving scientific discovery. Artificial electromagnetic materials, such as electromagnetic metamaterials, photonic crystals, and plasmonics, are research fields where deep neural network results evince the data driven approach; especially in cases where conventional computational and optimization methods have failed. We propose and demonstrate a deep learning method capable of finding accurate solutions to ill-posed inverse problems, where the conditions of existence and uniqueness are violated. A specific example of finding the metasurface geometry which yields a radiant exitance matching the external quantum efficiency of gallium antimonide is demonstrated.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128670970","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Z. Kudyshev, S. Bogdanov, Zachariah Olson, Xiaohui Xu, D. Sychev, A. Kildishev, V. Shalaev, A. Boltasseva
Discovering novel, unconventional optical designs in combination with advanced machine-learning assisted data analysis techniques can uniquely enable new phenomena and breakthrough advances in many areas including imaging, sensing, energy, and quantum information technology. It demonstrated that compared to other inverse-design approaches that require extreme computation power to undertake a comprehensive search within a large parameter space, machine learning assisted topology optimization can expand the design space while improving the computational efficiency. This talk will highlight our most recent findings on 1) merging topology optimization with artificial-intelligence-assisted algorithms and 2) integrating machine-learning based analysis with photonic design and quantum optical measurements.
{"title":"Advancing photonic design and measurements with artificial intelligence","authors":"Z. Kudyshev, S. Bogdanov, Zachariah Olson, Xiaohui Xu, D. Sychev, A. Kildishev, V. Shalaev, A. Boltasseva","doi":"10.1117/12.2594790","DOIUrl":"https://doi.org/10.1117/12.2594790","url":null,"abstract":"Discovering novel, unconventional optical designs in combination with advanced machine-learning assisted data analysis techniques can uniquely enable new phenomena and breakthrough advances in many areas including imaging, sensing, energy, and quantum information technology. It demonstrated that compared to other inverse-design approaches that require extreme computation power to undertake a comprehensive search within a large parameter space, machine learning assisted topology optimization can expand the design space while improving the computational efficiency. This talk will highlight our most recent findings on 1) merging topology optimization with artificial-intelligence-assisted algorithms and 2) integrating machine-learning based analysis with photonic design and quantum optical measurements.","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130918800","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abantika Ghosh, Mohannad Elhamod, Jie Bu, Wei‐Cheng Lee, A. Karpatne, V. Podolskiy
{"title":"Physics-guided machine learning for Maxwell's equations","authors":"Abantika Ghosh, Mohannad Elhamod, Jie Bu, Wei‐Cheng Lee, A. Karpatne, V. Podolskiy","doi":"10.1117/12.2594575","DOIUrl":"https://doi.org/10.1117/12.2594575","url":null,"abstract":"","PeriodicalId":389503,"journal":{"name":"Metamaterials, Metadevices, and Metasystems 2021","volume":" 24","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120834503","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：