DDSNet: AI Model for Photonic Crystal Laser Design
Researchers propose DDSNet, a dual-domain symmetry-aware network for predicting properties of photonic crystal surface-emitting lasers (PCSELs). The model addresses computational bottlenecks in coupled-wave theory (CWT) by integrating translation-equivariant spectral filtering with symmetry-induced structural priors. This approach exploits spectral components and asymmetric structures in PhC unit-cell dielectric patterns, which are key factors governing device properties. DDSNet improves surrogate accuracy and screening reliability, especially in structure-sensitive regions, enabling efficient exploration of the photonic crystal lattice design space.
Key facts
- DDSNet stands for Dual-Domain Symmetry-Aware Network.
- It is designed for PCSEL property prediction.
- It integrates translation-equivariant spectral filtering with symmetry-induced structural prior.
- It addresses computational cost of coupled-wave theory (CWT).
- It exploits spectral components and asymmetric structures in PhC unit-cell dielectric patterns.
- It improves surrogate accuracy and screening reliability.
- The paper is available on arXiv with ID 2607.24785.
- The research focuses on photonic crystal surface-emitting lasers.
Entities
—