[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-120056-en":3,"doc-seo-120056-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},120056,5909877438554,"Maeve","https://ap-avatar.wpscdn.com/avatar/5600025385ad2bf12a7?_k=1778553567797529272",8,"Research & Report","Generative model for multiple-purpose inverse design and forward prediction of disordered waveguides in linear and nonlinear regimes","Data-driven machine learning is used to efficiently explore the design space of disordered nanophotonic waveguides that are difficult to search with conventional simulation-based methods. A conditional generative adversarial network with Wasserstein distance and gradient penalty is applied for inverse design tasks including pattern optimization, geometry generation, and pattern reproduction. For forward prediction, a convolutional neural network serves as a fast solver to predict transmission spectra across telecommunication wavelengths, enabling linear-response extension to nonlinear regimes at higher input power.","Generative model for multiple-purpose inverse design and forward prediction of disordered waveguides in linear and nonlinear regimes  \nZiheng Guo 1, Zhongliang Guo2, Ognjen Arandjelović2, Andrea di Falco 1* 1 School of Physics and Astronomy, University of St. Andrews, Fife, KY16 9SS, United Kingdom 2 School of Computer Science, University of St. Andrews, Fife, KY16 9SS, United Kingdom  \nABSTRACT  \nData-driven machine learning framework has become a state-of-art technique to explore whole parameters design space for designing complex systems. In this work, we used conditional generative adversarial networks to inverse design three problems that we are interested in random nanophotonic systems: pattern optimization, geometry generation, and pattern reproduction. Meanwhile, automation convolutional neural networks group for forward prediction of the transmission spectra of disordered waveguides in linear and nonlinear regimes, at telecommunication wavelengths.  \n1. INTRODUCTION  \nUsing conventional libraries to search complex design spaces is extremely time consuming. At the same time, conventional inverse design problem is only able to solve single given type of applications, while it hard to extend it to other applications. For well-structured applications, for instance, well-structured waveguides, is easy using Finite different time domain or beam propagation method along with particle swarm optimization, genetic algorithms to achieve different design patterns. However, it requires hundreds of simulations to optimize one design. Therefore, datadriven machine learning to design complex system in nanophotonic is needed. This method enhances the design process and more efficient exploration of design spaces. There are many successful applications have been reported. For example, using deep neural network along with gaussian log likelihood loss function inverse design well-structured waveguide based on transmission and reflection. Furthermore, the generative adversarial network inverse design metasurface based on different absorption or frequency [1, 2, 3] have achieve some high accuracy results. However, in terms of inverse design more complex system, not many researchers has deep went through the design space.  \nIn our work, we present you the conditional generative adversarial network with Wasserstein distance loss with penalty to inverse design the disordered waveguide in linear regimes. Meanwhile, based our special figure of merit, input power vs transmission spectra, linear response can be extended to nonlinear regimes when input higher power. In such way, weare able to address three different inverse design applications. 1, inverse design known disordered structures to achieve the same performance. 2, inverse design known disordered structures to achieve upper and lower correlated behavior performance. 3, inverse design unknown spectra that give design pattern. Those applications open up the possibility to design the encryption waveguide, paired key design, all optical neural networks nonlinear activation function, as well as accelerate our inverse design process.  \n2. METHODOLOGY  \nIn this work, we describe nonlinear dispersive material by nonlinear Lorentz oscillator along with Yee’s grid. [4, 5] . FDTD generated 100K data, design pattern and its transmission spectra were paired generated. We train forward convolutional neural network using linear power of 1W, along with wavelength 1500 to 1550 mm. Design space is 64 x 64 pixels array inplace with random placed 200 air holes as binary images. CNN is used as fast FDTD solver. While for the inverse design applications, we use conditional generative adversarial network with altered loss that [6] from Equation 1 to Equation 2  \n􀝉􀝅􀝊􀯀 􀝉􀜽􀝔􀮽 􀜸 (􀜦 , 􀜩) = 􀥱􀯫~􀯉 􀳏􀳌􀳟􀳌(􀳣) [􀝈􀝋􀝃􀜦 (􀝔)] + 􀥱􀯭~􀯉􀳥(􀯭) (log (1 − 􀜦 (􀜩 (􀝖)) (1)  \n􀜽􀝎􀝃􀝉􀜽􀝔􀮼 􀥱 [􀜥 (􀝔, 􀝕 _ )] + 􀥱 (log (􀜥 (􀜩 (􀝕⨁􀝖), 􀝕_ ) + 􀟣􀥱 (||∇􀜥 (􀝕̂) ||2 − 1)2 (2)  \nWe employed a conditional Wasserstein generative adversarial network with gradient pen","cbCairRDcYUycpr0","https://ap.wps.com/l/cbCairRDcYUycpr0","pdf",672579,1,5,"English","en",105,"# Abstract\n# Introduction\n## Motivation and challenges\n## Proposed generative-learning approach\n# Methodology\n## Simulation data generation\n## Forward prediction using CNN\n## Inverse design using conditional Wasserstein GAN with gradient penalty\n# Results\n## Inverse design applications\n## Prediction and performance checks for unseen responses","[{\"question\":\"Why are conventional inverse-design searches too time-consuming for disordered waveguides?\",\"answer\":\"Conventional methods require extensive simulations to scan a complex parameter/design space, and typical inverse-design approaches are often limited to a single application type. This makes extending the design process to broader, more complex problems inefficient.\"},{\"question\":\"How does the forward prediction model estimate transmission spectra?\",\"answer\":\"A convolutional neural network is trained using paired FDTD-generated data. It predicts transmission spectra from input conditions in the linear regime and supports nonlinear response predictions at higher input power.\"},{\"question\":\"What is the role of the conditional Wasserstein GAN with gradient penalty in inverse design?\",\"answer\":\"The generator produces new disordered waveguide geometries conditioned on target transmission spectra, while the critic distinguishes generated geometries from real ones using an un-normalized spectral input. This adversarial setup enables multiple inverse-design objectives such as enhancement, reproduction, and new pattern generation.\"}]","Generative model for multiple-purpose inverse design and forward prediction of disordered waveguides in linear and nonlinear regimes | PDF",1785727908,13,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"generative-model-for-multiple-purpose-inverse-design-and-forward-prediction-of-disordered-waveguides-in-linear-and-nonlinear-regimes","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/generative-model-for-multiple-purpose-inverse-design-and-forward-prediction-of-disordered-waveguides-in-linear-and-nonlinear-regimes/120056/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why are conventional inverse-design searches too time-consuming for disordered waveguides?","Question",{"text":75,"@type":76},"Conventional methods require extensive simulations to scan a complex parameter/design space, and typical inverse-design approaches are often limited to a single application type. This makes extending the design process to broader, more complex problems inefficient.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the forward prediction model estimate transmission spectra?",{"text":80,"@type":76},"A convolutional neural network is trained using paired FDTD-generated data. It predicts transmission spectra from input conditions in the linear regime and supports nonlinear response predictions at higher input power.",{"name":82,"@type":73,"acceptedAnswer":83},"What is the role of the conditional Wasserstein GAN with gradient penalty in inverse design?",{"text":84,"@type":76},"The generator produces new disordered waveguide geometries conditioned on target transmission spectra, while the critic distinguishes generated geometries from real ones using an un-normalized spectral input. This adversarial setup enables multiple inverse-design objectives such as enhancement, reproduction, and new pattern generation.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":106,"show_sort_weight":107,"slug":108},"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":46,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":21,"slug":137},19,"General","general"]