Using Wavesim to investigate Mirror Symmetry in Three-Dimensional Multiple-Scattering Media
An artistic representation showing the different response of the interference pattern depending on the polarization of light relative to the symmetry axis, while simultaneously illustrating the connection between mirror phenomena and the analogy to the Drexhage experiment on emitter distance from a mirror surface.
Recently, a study at the University of Twente used Wavesim to simulate the effects of symmetrical light scattering in their experimental designs and complement the experimental results. The original publication, ‘Mirror Symmetry in three-dimensional Multiple-Scattering Media’ was published in Physical Review Letters.
​Evangelos Marakis from the Foundation for Research and Technology – Hellas in Greece is a co-author and he described how Wavesim was used in the project, and how it helped run the huge number of simulations required for the study.
Could you elaborate on the advantages of WaveSim over FDTD for this specific study?
Certainly. Wavesim shines in its ability to converge quickly in cases with moderate refractive index contrasts, but even higher contrast cases are quite fast in terms of speed.
It offers scalability, allowing me to test and refine workflows at lower accuracy levels before ramping up to maximum precision.This flexibility saved hours of computation time that could otherwise be wasted on trivial coding errors or misalignments.
It is also worth mentioning that in this publication the interference pattern created was quite sensitive and required a good alignment of the structure with the source. WaveSim's speed allowed this to be done in minutes rather than hours. Its compatibility with our homemade functions also saved us from having to rework existing tools for integration into an FDTD solver.
With WaveSim, I could simulate real-world
experimental conditions and align fabrication with theory, achieving what was once impractical with traditional methods.
What made you choose WaveSim for this study? How does it compare to traditional methods like finite-difference time-domain (FDTD)?
The primary reason was its ease of use and efficiency without sacrificing accuracy. WaveSim is exceptionally fast and doesn't require a specialized computer setup- I could conduct a full-fledged scientific study on a modest PC.​​
In contrast, FDTD has a steep learning curve, particularly with open-source platforms, requiring significant coding expertise. Commercial FDTD solvers, while faster and more robust, are expensive and often lack integration with native programming languages. This makes WaveSim a far more accessible and flexible option.
The output intensity field, despite being a random distribution, displays a symmetrical peak at the center, parallel to the symmetry line.
Using WaveSim, we explored the near field of the structure and investigated how imperfections influence the phenomenon, providing insights into the robustness of the pattern against fabrication errors.
Were there any challenges or limitations with these methods?
Of course, no method is perfect. However, for the purpose of my study I didn’t have any direct limitations using WaveSim. GPU memory limitations were the main consideration. However, thanks to its scalable resolution and efficient sampling, this never became a critical issue.
FDTD, on the other hand, struggles with long computation times, smaller simulation domains, and inherent numerical dispersion. While boundary conditions are well-established in both methods, WaveSim delivers significant speed advantages in this regard.
What did WaveSim enable you to accomplish that would have been difficult or impossible with FDTD?
One standout advantage was the ability to perform statistical ensemble analysis. This study required simulations of multiple structures with similar statistics—an impractical task with FDTD, which would have required weeks of computation and multiple workstations, in addition to significantly scaling down the problem.
With WaveSim, I could perform the entire analysis on a single computer. This included simulations of experimental conditions, bringing my work in lithography and light simulation together in perfect harmony.
What do you think the field can learn from combining such numerical methods with experiments?
I love that we are able to study the actual physical system. WaveSim bridges the gap between theoretical models and real-world systems. Historically, optical systems were approximated due to the high costs of precise simulations. Manufacturing imperfections were often ignored or blindly optimized through trial and error.
Now, we can study actual physical systems with their imperfections in detail, making optimization more reliable and less costly. I believe WaveSim will help photonic fabrication mature to rival semiconductor production in precision, efficiency, and scalability.
In our study, we explore fundamental symmetries not commonly found in nature. A scattering structure with mirror symmetry is analyzed in detail along the symmetry axis.