Periodic Reporting for period 1 - WAVESIM (WAVESIM – The next generation electromagnetic wave solver)
Période du rapport: 2023-01-01 au 2024-06-30
WAVESIM: Lightning fast light simulations
Calculating how light propagates through complex structures is essential for developing faster internet, better camera lenses, more sensitive sensors, and more efficient light sources. Unfortunately, accurate simulations of large structures are very time consuming. As part of the EU funded project DEEP VISION, we developed a radically different simulation method that is up to a factor 1000 faster than state-of-the-art commercial solvers. Our method is not limited to light, it is can be used for many other types of waves, such as seismic waves, quantum wavefunctions, radio waves, or medical ultrasound. At the moment, however, our method is limited to structures that are small enough to fit on a single computer.
In the WAVESIM proof-of-concept project, we will distribute these computations over the cloud, allowing hundreds of computers to work in parallel, and allowing virtually unlimited structure sizes. Furthermore, we will investigate options for commercializing the software and integrating it in the design pipelines used in the photonics industry.
Calculating how light propagates through complex structures is essential for developing faster internet, better camera lenses, more sensitive sensors, and more efficient light sources. Unfortunately, accurate simulations of large structures are very time consuming. As part of the EU funded project DEEP VISION, we developed a radically different simulation method that is up to a factor 1000 faster than state-of-the-art commercial solvers. Our method is not limited to light, it is can be used for many other types of waves, such as seismic waves, quantum wavefunctions, radio waves, or medical ultrasound. At the moment, however, our method is limited to structures that are small enough to fit on a single computer.
In the WAVESIM proof-of-concept project, we will distribute these computations over the cloud, allowing hundreds of computers to work in parallel, and allowing virtually unlimited structure sizes. Furthermore, we will investigate options for commercializing the software and integrating it in the design pipelines used in the photonics industry.
We developed a domain-distribution technique that allows splitting a single large simulation over multiple computers or graphical processing units (GPU). In a proof of concept experiment, we were able to simulate wave propagation in a structures with sized exceeding 325 x 325 x 325 wavelengths in under 7 minutes, by distributing the computations over multiple GPUs.
Furthermore, we firmly established the validity of our method, proving it's convergence and accuracy both theoretically and experimentally.
Our was framework is implemented as open source software (see www.wavesim.org). It includes a series of automated tests and benchmarks to validate its correctness and efficiency.
In a scientific collaboration with prof. Pepijn Pinkse, we demonstrated the use of WAVESIM in understanding light-propagation in nanofabricated 3D photonic structures. The size of these structures far exceeded the volume tractable by conventional simulation tools. Our simulation results showed excellent agreement with experiments and with analytical theory
Furthermore, we firmly established the validity of our method, proving it's convergence and accuracy both theoretically and experimentally.
Our was framework is implemented as open source software (see www.wavesim.org). It includes a series of automated tests and benchmarks to validate its correctness and efficiency.
In a scientific collaboration with prof. Pepijn Pinkse, we demonstrated the use of WAVESIM in understanding light-propagation in nanofabricated 3D photonic structures. The size of these structures far exceeded the volume tractable by conventional simulation tools. Our simulation results showed excellent agreement with experiments and with analytical theory
Our solver sets a new standard for the speed and accuracy of scientific software for simulating for wave propagation.
Our method enables researchers and engineers to accurately simulate light propagation in structures with sizes that were previously unimaginable, allowing accurate predictions of light scattering in skin, brain, or blood; designing large-scale photonic circuits, optimizing light conversion in LEDs, or improving photovoltaics.
We established over a factor of over 1000 speed increase over the most commonly used state-of-the-art method (finite difference time domain), making it possible to simulate smaller structures in real time. This speed increase allows engineers to directly see the effect of changes in their designed photonic device or antenna, achieving video-rate feedback for simulations that would previously take minutes.
We have been building up a growing user community and collaborate with industrial partners to investigate the future commercialization of this technology. To ensure further uptake, it is vital to remain investing in the further professionalization of the software product, and in building and supporting an ever growing community of users.
Our method enables researchers and engineers to accurately simulate light propagation in structures with sizes that were previously unimaginable, allowing accurate predictions of light scattering in skin, brain, or blood; designing large-scale photonic circuits, optimizing light conversion in LEDs, or improving photovoltaics.
We established over a factor of over 1000 speed increase over the most commonly used state-of-the-art method (finite difference time domain), making it possible to simulate smaller structures in real time. This speed increase allows engineers to directly see the effect of changes in their designed photonic device or antenna, achieving video-rate feedback for simulations that would previously take minutes.
We have been building up a growing user community and collaborate with industrial partners to investigate the future commercialization of this technology. To ensure further uptake, it is vital to remain investing in the further professionalization of the software product, and in building and supporting an ever growing community of users.