[1]

Noda S, Chutinan A, Imada M. Trapping and emission of photons by a single defect in a photonic bandgap structure. Nature 2000;407:608–10. CrossrefGoogle Scholar

[2]

Akahane Y, Asano T, Song BS, Noda S. High-Q photonic nanocavity in a two-dimensional photonic crystal. Nature 2003;425:944–7. CrossrefGoogle Scholar

[3]

Song BS, Noda S, Asano T, Akahane Y. Ultra-high-Q photonic double-heterostructure nanocavity. Nat Mater 2005;4:207–10. CrossrefGoogle Scholar

[4]

Asano T, Song BS, Noda S. Analysis of the experimental Q factors (~1 million) of photonic crystal nanocavities. Opt Express 2006;14:1996–2002. CrossrefPubMedGoogle Scholar

[5]

Kuramochi E, Notomi M, Mitsugi S, Shinya A, Tanabe T,Watanabe T. Ultrahigh-Q photonic crystal nanocavities realized by the local width modulation of a line defect. Appl Phys Lett 2006;88:041112. CrossrefGoogle Scholar

[6]

Takahashi Y, Hagino H, Tanaka Y, Song BS, Asano T, Noda S. High-Q nanocavity with a 2-ns photon lifetime. Opt Express 2007;15:17206–13. CrossrefWeb of SciencePubMedGoogle Scholar

[7]

Kuramochi E, Taniyama H, Tanabe T, Shinya A, Notomi M. Ultrahigh-Q two-dimensional photonic crystal slab nanocavities in very thin barriers. Appl Phys Lett 2008;93:111112. CrossrefWeb of ScienceGoogle Scholar

[8]

Han Z, Checoury X, Néel D, David S, El Kurdi M, Boucaud P. Optimized design for 2×10^{6} ultra-high Q silicon photonic crystal cavities. Opt Commun 2010;283:4387–91. CrossrefGoogle Scholar

[9]

Sekoguchi H, Takahashi Y, Asano T, Noda S. Photonic crystal nanocavity with a *Q*-factor of ~9 million. Opt Express 2014;22:916–24. PubMedCrossrefWeb of ScienceGoogle Scholar

[10]

Asano T, Ochi Y, Takahashi Y, Kishimoto K, Noda S. Photonic crystal nanocavity with a Q factor exceeding eleven million. Opt Express 2017;25:1769–77. Web of ScienceCrossrefPubMedGoogle Scholar

[11]

Asano T, Noda S. Photonic crystal devices in silicon photonics. Proc IEEE 2018;106:1–13. Web of ScienceGoogle Scholar

[12]

Srinivasan K, Painter O. Momentum space design of high-Q photonic crystal optical cavities. Opt Express 2002;10:670–84. CrossrefPubMedGoogle Scholar

[13]

Englund D, Fushman I, Vucković J. General recipe for designing photonic crystal cavities. Opt Express 2005;13:5961–75. CrossrefPubMedGoogle Scholar

[14]

Tanaka Y, Asano T, Noda S. Design of photonic crystal nanocavity with *Q*-factor of ~10^{9}. J Light Technol 2008;26:1532–9. CrossrefGoogle Scholar

[15]

Lai Y, Pirotta S, Urbinati G, et al. Genetically designed L3 photonic crystal nanocavities with measured quality factor exceeding one million. Appl Phys Lett 2014;104:241101. CrossrefWeb of ScienceGoogle Scholar

[16]

Minkov M, Savona V. Automated optimization of photonic crystal slab cavities. Sci Rep 2015;4:5124. CrossrefWeb of ScienceGoogle Scholar

[17]

Nakamura T, Takahashi Y, Tanaka Y, Asano T, Noda S. Improvement in the quality factors for photonic crystal nanocavities via visualization of the leaky components. Opt Express 2016;24:9541–9. Web of SciencePubMedCrossrefGoogle Scholar

[18]

Minkov M, Savona V, Gerace D. Photonic crystal slab cavity simultaneously optimized for ultra-high Q/V and vertical radiation coupling. Appl Phys Lett 2017;111:131104. Web of ScienceCrossrefGoogle Scholar

[19]

Asano T, Noda S. Optimization of photonic crystal nanocavities based on deep learning. Opt Express 2018;26:32704–16. CrossrefPubMedWeb of ScienceGoogle Scholar

[20]

Molesky S, Lin Z, Piggott AY, Jin W, Vucković J, Rodriguez AW. Inverse design in nanophotonics. Nat Photonics 2018;12:659–70. Web of ScienceCrossrefGoogle Scholar

[21]

Lu J, Vuckovic J. Inverse design of nanophotonic structures using complementary convex optimization. Opt Express 2010;18:3793–804. PubMedCrossrefWeb of ScienceGoogle Scholar

[22]

Takezawa A, Kitamura M. Cross-sectional shape optimization of whispering-gallery ring resonators. J Lightwave Technol 2012;30:2776–82. CrossrefWeb of ScienceGoogle Scholar

[23]

Lin Z, Liang X, Lončar M, Johnson SG, Rodriguez AW. Cavity-enhanced second-harmonic generation via nonlinear-overlap optimization. Optica 2016;3:233–8. Web of ScienceCrossrefGoogle Scholar

[24]

Lin Z, Lončar M, Rodriguez AW. Topology optimization of multi-track ring resonators and 2D microcavities for nonlinear frequency conversion. Opt Lett 2017;42:2818–21. CrossrefWeb of SciencePubMedGoogle Scholar

[25]

Frei WR, Johnson HT, Choquette KD. Optimization of a single defect photonic crystal laser cavity. J Appl Phys 2008;103:033102. Web of ScienceCrossrefGoogle Scholar

[26]

Liang X, Johnson SG. Formulation for scalable optimization of microcavities via the frequency-averaged local density of states. Opt Express 2013;21:30812–41. Web of SciencePubMedCrossrefGoogle Scholar

[27]

Maeno K, Takahashi Y, Nakamura T, Asano T, Noda S. Analysis of high-Q photonic crystal L3 nanocavities designed by visualization of the leaky components. Opt Express 2017;25:367–76. CrossrefWeb of SciencePubMedGoogle Scholar

[28]

LeCun Y, Boser B, Denker JS, et al. Handwritten digit recognition with a back-propagation networks. In: Proceedings of Advances in Neural Information Processing Systems, Denver, CO, USA, 1990:396–404. Google Scholar

[29]

Glorot X, Bordes A, Bengio Y. Deep sparse rectifier neural networks. In: Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, Ft. Lauderdale, FL, USA, 2011:315–23. Google Scholar

[30]

Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R. Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res 2014;15:1929–58. Google Scholar

[31]

Krogh A, Hertz JA. A simple weight decay can improve generalization. In: Proceedings of Advances in Neural Information Processing Systems, Denver, CO, USA, 1991:950–7. Google Scholar

[32]

Rumelhart DE, Hinton GE, Williams RJ. Learning representations by back-propagating errors. Nature 1986;323:533–6. CrossrefGoogle Scholar

[33]

Polyak BT. Some methods of speeding up the convergence of iteration methods. USSR Comput Math Math Phys 1964;4:791–803. Google Scholar

[34]

For the first principles simulation phase we use 70 runs of the FDTD simulation of the 3D electric and magnetic fields (12 components) in the new 70 cavity structures. These runs can be processed in parallel depending on the computation resources. Each FDTD simulation comprises 80,000 steps of spatial derivative calculations for the PC nanocavity described by a mesh size of 480×320×170 cells. The training phase employs 10 runs (which can be processed in parallel), each comprising 50,000 steps of the forward and backward calculations of the multiplications of vectors and matrices with sizes of {vector, matrix}={50, 50×50}×9 (convolution), {450, 450×200}, {200, 200×50}, and {50, 50×1}. The structure generation phase uses 70 runs (which can be processed in parallel), each comprising the 20,000 steps of the same vector-matrix calculations to obtain the gradient of *Q* of each structure during candidate generation. The loss factor of type C involves an additional cost of calculating the distances between the present structure and all the structures in the training dataset, where the dimension of each structure vector is 100 and the number of structures in the training dataset changes from 1000 to 8000 in this work. This additional cost can be on the same order as the cost for the calculation of the gradient of *Q*. In total, the computational cost of the first principles calculations is larger than that for the learning and structure generation phases by a factor of roughly 100. Google Scholar

[35]

Pilozzi L, Farrelly FA, Marcucci G, Conti C. Machine learning inverse problem for topological photonics. Commun Phys 2018;1:57. Web of ScienceCrossrefGoogle Scholar

[36]

Peurifoy J, Shen Y, Jing L, et al. Nanophotonic particle simulation and inverse design using artificial neural networks. Sci Adv 2018;4:eaar4206. Web of SciencePubMedCrossrefGoogle Scholar

[37]

So S, Rho JM. Simultaneous inverse design of materials and structures via deep learning: demonstration of dipole resonance engineering using core-shell nanoparticles. ACS Appl Mater Interfaces 2019;11:24264–8. CrossrefWeb of ScienceGoogle Scholar

[38]

Liu D, Tan Y, Khoram E, Yu Z. Training deep neural networks for the inverse design of nanophotonic structures. ACS Photonics 2018;5:1365–9. CrossrefWeb of ScienceGoogle Scholar

[39]

Liu Z, Zhu D, Rodrigues SP, Lee KT, Cai W. Generative model for the inverse design of metasurfaces. Nano Lett 2018;18:6570–6. Web of SciencePubMedCrossrefGoogle Scholar

[40]

Long Y, Ren J, Li Y, Chen H. Inverse design of photonic topological state via machine learning. Appl Phys Lett 2019;114:181105. CrossrefWeb of ScienceGoogle Scholar

[41]

Melati D, Grinberg Y, Dezfouli MK, et al. Mapping the global design space of nanophotonic components using machine learning pattern recognition. Nat Commun 2019;10:4775. PubMedCrossrefWeb of ScienceGoogle Scholar

[42]

Jones DR, Schonlau M, Welch WJ. Efficient global optimization of expensive black-box functions. J Global Optim 1998;13:455–92. CrossrefGoogle Scholar

[43]

Shahriari B, Swersky K, Wang Z, Adams RP, De Freitas N. Taking the human out of the loop: a review of Bayesian optimization. In: Proc. IEEE 104, 2016:148–75. Google Scholar

[44]

Rana S, Li C, Gupta S, Nguyen V, Venkatesh S. High dimensional bayesian optimization with elastic gaussian process. In: Proc. of the 34th Int. Conf. on Mach. Learn. 70, 2017:2883–91. Google Scholar

[45]

Wang Z, Hutter F, Zoghi M, Matheson D, De Freitas N. Bayesian optimization in a billion dimensions via random embeddings. J Artif Intell Res 2016;55:361–7. CrossrefWeb of ScienceGoogle Scholar

[46]

Snoek J, Rippel O, Swersky K, et al. Scalable Bayesian optimization using deep neural networks. In: Proceedings of the 32nd International Conference on Machine Learning, PMLR 37, Lille, France, 2015:2171–80. Google Scholar

[47]

In Ref. [10], we reduced the resonant wavelength of the cavities in steps of about 3 nm by using surface oxidization and oxide removal. Theoretical calculations revealed that this wavelength change corresponds to a change in the radii of the air holes by about 0.5 nm (and a simultaneous change in the slab thickness by about 1 nm). Google Scholar

## Comments (0)

General note:By using the comment function on degruyter.com you agree to our Privacy Statement. A respectful treatment of one another is important to us. Therefore we would like to draw your attention to our House Rules.