About

Motivation

Electromagnetic simulation is essential for metasurface design. More broadly, computational Physics underpins almost all modern technology development. Simulations are no longer just for validation purposes but are a critical phase of the design process, enabling exploration and optimization of systems that previously would have been impossible.

While there has been considerable progress, existing simulators are far from meeting the computational demands of many science and engineering problems that society will need to confront in the coming decades (e.g., in electromagnetics, pharmaceuticals, genetics, nuclear fusion processes, and Earth system/climate models). Metasurface simulation in particular suffers from significant accuracy limitations and tradeoffs as optical size increases to relevant length scales (i.e., on the order of millimeters to multiple centimeters, where full-wave tools such as FDTD are infeasible, even with GPUs).

This substantially limits the design space and the commercial viability of metasurface technologies due to large performance gaps between simulated and fabricated devices, driving costly empirical design loops. At commercially relevant length scales, existing design tools fail to accurately model stray light and calculate efficiencies, and also fall short in providing accurate models for more advanced, multi-element systems required in use cases such as AR/VR, high-resolution imaging, and information processing.

Vision

At EdgeDyne, we believe that both current software and electronic hardware are inadequate to meet the simulation demands of the 21st century. We believe that optics offers a potential path forward to augment but not replace electronics, and that metamaterial-based computing is a promising candidate that offers significant advantages in terms of parallelism and computation through free-space propagation.

There’s unlikely to be a one-size-fits-all solution, wherein optics fully replaces electronics, but there may be specific applications where optics could benefit in use cases where electronics are beginning to run into limits (e.g., in edge-deployed machine vision sensors, combinatorial optimization, very large matrix-matrix operations, solving high-dimensional PDEs, and AI training). The system designs and architectures for these different use cases are still open research questions at this stage and they’ll need to contend with key challenges including memory bandwidth, reconfigurability, nonlinear processing, and manufacturability.

Our near-term focus is advancing metasurface design and simulation capabilities in order to catalyze R&D in metasurface technologies and metasurface-enhanced computing architectures. Without the design tools and methodology to accurately capture the physics of metasurfaces at multi-centimeter length scales where we can directly exploit massive parallelism using free space optics, we believe that efforts to advance metasurface technology and computing will be constrained for years to come.

Longer term, we aim to leverage these simulation capabilities to develop the algorithms, components, and systems needed to create ultra-fast, computational physics processors to help address the scientific and engineering challenges of the 21st century. Achieving this will require collaboration across disciplines, industries, and the broader metasurface ecosystem.

Founder

Shane Colburn, PhD, MBA
Founder & Lead Engineer
LinkedIn
Google Scholar

Shane has over a decade of experience with photonic metamaterials. He has conducted extensive research on meta-optics for imaging, information processing, reconfigurability, and beamshaping. His experience includes nanofabrication and experimental optics, and a strong background in electronics, including embedded systems + FPGAs, signal/image processing, and electromagnetics. He is the author of the open-source metasurface design framework rcwa_tf