source: FS-PRI / Utsa Chattopadhyay
|
|
|
Description
Machine learning (ML) methods are not only enabling great possibilities e.g. for image recognition and language translation, they provide excellent opportunies to accelerate advances in multiple areas of science and technology. In particular, so-called inverse problems covering challenges, which can be easily solved analytically or numerically in forward direction but which lack an inverse solution, are excellent examples, which can be tackled using ML methods. A promient inverse problem is the desing of optical thin-film coatings with desired amplitude and phase properties. Thin-film coatings do not only form the key building block for modern laser mirrors, they are used in various everday-life applications including for example antireflection coatings for eye glasses, windows and surfaces. This project targets new routes to optical coating design utilizing ML methods, promising to solve long-standing challenges for optical coating design such as design speed, the need to bring in expert knowledge to the design process as well as the difficulty, to find entrely new multi-layer designs.
Machine learning (ML) methods are not only enabling great possibilities e.g. for image recognition and language translation, they provide excellent opportunies to accelerate advances in multiple areas of science and technology. In particular, so-called inverse problems covering challenges, which can be easily solved analytically or numerically in forward direction but which lack an inverse solution, are excellent examples, which can be tackled using ML methods. A promient inverse problem is the desing of optical thin-film coatings with desired amplitude and phase properties. Thin-film coatings do not only form the key building block for modern laser mirrors, they are used in various everday-life applications including for example antireflection coatings for eye glasses, windows and surfaces. This project targets new routes to optical coating design utilizing ML methods, promising to solve long-standing challenges for optical coating design such as design speed, the need to bring in expert knowledge to the design process as well as the difficulty, to find entrely new multi-layer designs.
Selected Publications
Utsa Chattopadhyay, Florian Carstens, Andreas Wienke, Ingmar Hartl, Nihat Ay, Christoph Heyl and Henrik Tünnermann
EPJ Web Conf., 307 (2024) 04057. DOI: 10.1051/epjconf/202430704057
Funding
|
