Spins-B is probably the most well known code for photonics inverse design in the community. Therefore we chose to do our work on gradient-free optimization inside it. The present fork is aimed at gathering branches which include gradient-free optimization.

Spins-b (as forked in the present repository) is based on optimization as follows: 1- random initialization 2- optimization by BFGS: more precisely this is L-BFGS-B 3- discretization enforced by sigmoid transformation. In the case of grating, there is an additional optimization step 4- by SLSQP, using a parametrization. Here the optimization is continuous (the parametrization is continuous) but the design is discrete (there are only two permittivities).

We add one more step termed NG, at the end of 3: - Lengler's method (paper) - equipped with a smoothing operator

The smoothing operator is detailed `here < https://github.com/facebookresearch/nevergrad/blob/8403d6c9659f40fec2a3cf7f474b3d8610f0f2e4/nevergrad/optimization/optimizerlib.py#L388>`_.

We are very grateful to Spins-B for providing us with this great code, central for our experiments.

For example this branch contains code for running Lengler+smoothing directly in Spins-B for the Bend90 case.

1+2+3+NG is better than 1+2+3 because the discretization by our discrete optimization methods works better than enforcing discretization through sigmoids. NG alone (as opposed to 1+2+3+NG) outperforms numerically 1+2+3 in some cases, in particular for large budgets. However, without the initial BFGS step, NG sometimes provides designs which are not smooth enough: the initial point provided by L-BFGS-B as included in Spins-B is essential for ensuring a good smoothness, which is better for buildability.

Our code uses Nevergrad. We are intensive Nevergrad users and we are happy to chat in the user group.

SPINS-B is the open source version of SPINS, a framework for gradient-based (adjoint) photonic optimization developed over the past decade at Jelena Vuckovic's Nanoscale and Quantum Photonics Lab at Stanford University. The full version can be licensed through the Stanford Office of Technology and Licensing (see FAQ).

The overall architecture is explained in our paper Nanophotonic Inverse Design with SPINS: Software Architecture and Practical Considerations.

Documentation is continually updated over time.

• Gradient-based (adjoint) optimization of photonic devices
• 2D and 3D device optimization using finite-difference frequency-domain (FDFD)
• Support for custom objective functions, sources, and optimization methods
• Automatically save design methodology and all hyperparameters used in optimization for reproducibility

We are protoyping the next version of SPINS-B. This version of SPINS-B will support these new features:

• Co-optimization of multiple device regions simulataneously
• Integration with FDTD and other electromagnetic solvers
• Easier to use and extend

Traditional nanophotonic design typically relies on parameter sweeps, which are expensive both in terms of computation power and time, and restrictive in their parameter space. Likewise, completely blackbox optimization algorithms, such as particle swarm and genetic algorithms, are also highly inefficient. In both these cases, the computational costs limit the degrees of the freedom of the design to be quite small. In contrast, by leveraging gradient-based optimization methods, our nanophotonic inverse design algorithms can efficiently optimize structures with tens of thousands of degrees of freedom. This enables the algorithms to explore a much larger space of structures and therefore design devices with higher efficiencies, smaller footprint, and novel functionalities.

Any publications resulting from the use of this software should acknowledge SPINS-B and cite the following papers:

For general device optimization:

• Su et al. Nanophotonic Inverse Design with SPINS: Software Architecture and Practical Considerations. arXiv:1910.04829 (2019).

For grating coupler optimization:

• Su et al. Fully-automated optimization of grating couplers. Opt. Express (2018).
• Sapra et al. Inverse design and demonstration of broadband grating couplers. IEEE J. Sel. Quant. Elec. (2019).