Speaker
Description
Many phenomena and functional devices in optics and photonics rely on discrete objects, called scatterers, that interact with light in a predefined way. The optical properties of these scatterers are entirely described by the T-matrix. The T-matrix is computed for a given scatterer from a larger number of solutions to the Maxwell equations. Still, once known, various photonic materials made from these scatterers can be semi-analytically studied within a multi-scattering formalism. These photonic materials can consist of periodically or many, i.e., up to millions, a-periodically arranged objects with known T-matrices. Such a usage scenario points to the importance of storing the T-matrices for future exploitation since the computation of the T-matrix is demanding. Recalculating a T-matrix is detrimental in terms of financial expenses spent on computational resources and energy consumption, which should be reduced for ecological reasons. Therefore, there is a need to reuse these T-matrices once they have been calculated, and a fundamental request from the community concerns a standard data format that contains the T-matrix and unambiguous information about the corresponding object in terms of metadata. To respond to this demand, we describe here our efforts in the frame of DAPHONA project funded by BMBF to establish a data format and how to capitalize on it, using an infrastructure to archive and share T-matrices. Following the FAIR principles, we perceive a standard in HDF5 format, dedicate a database on a large-scale data facility, and provide search functionality available via the dedicated web server. Besides saving monetary and economic resources, this structure allows for a data-driven approach in this research field. It constitutes the first step in solving forward and inverse design problems based on the correspondence between the T-matrix and the geometry of the object with the help of machine learning.