Full-Wave Harmonic Balance Framework for Dispersive Time-Varying Photonic Structures

This paper introduces a full-wave harmonic-balance framework for modeling dispersive time-varying photonic structures. The method solves for the steady-state Floquet response directly in the frequency domain by representing time-varying components as induced secondary sources: volumetric polarization densities for bulk media and surface current densities for conductive sheets. Material dispersion and radiation operators are evaluated at each harmonic frequency, while temporal modulation enters as off-diagonal convolutional coupling in Floquet space. The framework is validated using two representative examples: a parametrically pumped rolled graphene cylinder, where numerical results agree with an analytical Floquet scattering solution and resolve harmonic-specific scattering, absorption, and near fields; and an ITO-based epsilon-near-zero (ENZ) space-time metasurface, enabling shape optimization of a reflective device that converts an optical input into sidebands and redirects them into selected spatial diffraction orders. The approach establishes a general platform for modeling and designing dispersive active photonic structures with harmonic-resolved field, power, absorption, and diffraction observables.

Improvements to AI Systems:

1. Physics-Informed Neural Solver for Time-Varying Dispersive Media

• Train a neural network to predict steady-state Floquet harmonics directly from material dispersion profiles and modulation waveforms, bypassing iterative harmonic-balance loops.

• The improved AI can output harmonic-resolved fields, absorption, and scattering for arbitrary 2D/3D geometries in milliseconds, enabling real-time design exploration.

2. Generative Inverse Design for Space-Time Metasurfaces

• Use a conditional generative model (e.g., diffusion or GAN) that takes target sideband frequencies, diffraction orders, and efficiency as input, and generates optimal spatial patterns of time-varying conductivity or permittivity.

• The improved AI can propose manufacturable ITO or graphene metasurface layouts for optical frequency conversion, beam steering, or dynamic wavefront shaping—without manual optimization.

3. Harmonic-Aware Surrogate for Multi-Physics Co-Simulation

• Build a surrogate model that maps modulation parameters (pump amplitude, phase, frequency) to harmonic coupling matrices in Floquet space, learned from the paper’s full-wave solver.

• The improved AI can accelerate system-level simulations (e.g., integrated photonic circuits with time-varying components) by replacing expensive full-wave blocks with fast, differentiable surrogates—enabling gradient-based co-design of electronics and photonics.

4. Automatic Anomaly Detection in Dispersive Active Structures

• Train an autoencoder on harmonic-resolved near-field and absorption patterns from the framework.

• The improved AI can flag unexpected sideband generation or absorption resonances in experimental measurements, identifying fabrication errors or nonlinear instabilities in real time.

5. Cross-Scale Transfer Learning for New Material Systems

• Pre-train a transformer on the paper’s examples (graphene cylinder, ITO ENZ metasurface) and fine-tune for other dispersive materials (e.g., TMDs, phase-change materials) using sparse simulation data.

• The improved AI can predict harmonic responses for novel material stacks with minimal retraining, accelerating discovery of tunable photonic devices.

Time-varying photonic structures redistribute electromagnetic energy among Floquet harmonics, enabling frequency conversion, nonreciprocity, parametric gain, and dynamic wavefront control. Accurate modeling of realistic platforms remains challenging when temporal modulation occurs in strongly dispersive materials, because each harmonic experiences a distinct material response while remaining coupled to all others through the modulation. This work introduces a full-wave harmonic-balance framework for dispersive time-varying photonic structures. The formulation solves the steady-state Floquet response in the frequency domain by representing modulated components as induced secondary sources: volumetric polarization densities for bulk media and surface current densities for conductive sheets. Material dispersion and radiation operators are evaluated at each harmonic frequency, whereas temporal modulation enters as off-diagonal convolutional coupling in Floquet space. The framework is validated for a parametrically pumped rolled graphene cylinder, where numerical results agree with an analytical Floquet scattering solution and resolve harmonic-specific scattering, absorption, and near fields. It is further applied to an ITO-based epsilon-near-zero (ENZ) space-time metasurface, enabling shape optimization of a reflective device that converts an optical input into sidebands and redirects them into selected spatial diffraction orders. The approach establishes a general platform for modeling and designing dispersive active photonic structures with harmonic-resolved field, power, absorption, and diffraction observables.