Large-scale coherent optical matrix multiplications
Liquid Crystal Deep Neural Architectures
Recent studies in machine learning have shown that large neural networks can dramatically improve the network performance, however, large networks are problematic in terms of computation time and memory usage for von Neumann computing architectures. On the other hand, modern computers are 104 − 108 times more power-hungry and less effective than the human brain for a wide range of tasks including perception, communication, and decision making. Therefore, developing computers that combine, learn, and analyze vast amounts of information quickly and efficiently is becoming increasingly important. The main goal of LC-DNA is to realize a scalable, power-efficient, and computationally fast analog neural network using a nonlinear liquid crystal (LC) as an analog hidden layer in our optical architecture. To demonstrate the scalability, speed, and power efficiency of our approach, we will compare it with state-of-the-art CPU and GPU technologies on challenging benchmark tasks. We believe that our results within this project will pave a way for computing technologies toward numerous large-scale complex problems that are not achievable yet.
Optical smart imaging through turbulent media
In real life, it is usually required to provide imaging through dynamic scattering environments, such as atmospheric turbulence, fog, and biological tissues. As known, such environments strongly limit the imaging resolution, and quite often the distorted images are hard to identify visually. Adaptive optics (AO) is one of the state-of-the-art techniques to deal with that. In traditional AO, the distorted wavefront of the object is measured and a deformable mirror is used to compensate for wavefront perturbations. However, AO is getting problematic for complex objects and strong turbulence due to the lag time in the feedback process. Recently, we have proposed an optical computing approach that can predict large chaotic systems with a performance achievable by large supercomputers only. This proposal aims to use optical computing in AO to overcome the challenge of imaging through the strong turbulence, as new changes necessary on the mirror can be predicted ahead of time. The obtained results will have a strong impact in both fields of optical imaging and optical computing. If successful, our new technique will be one of the most advanced real-world applications of optical computing and can be implemented on various types of flying objects in the future.
Large-scale coherent optical matrix multiplications
Nowadays, computing technologies play a huge role in our society. In this context, linear algebra is the most common, and most resource-intensive field in modern computational algorithms, its widespread use makes it possible to build networks of large neurons, optimize large systems, study graphs with multiple nodes, complex connections, and so on. The aim of our project is to create a new optical computing technology, which in the future will be able to perform large-scale and low energy-consuming linear algebraic operations compared to existing CPU, GPU and TPU technologies in this field.
Patterned chiral liquid crystals towards spatiotemporal and spectral beam shaping
We propose smart solutions towards on-demand spatial and spectral beam shaping peculiarities from patterned chiral liquid crystals integrated with a defect layer. Our preliminary theoretical simulations show that electrically tuning the retardance of the defect layer can generate various kinds of well-known complex inhomogeneous beams, such as vector beams, optical vortex beams, full Poincare beams, and azimuthally varying beams. We emphasize that inhomogeneous beams in terms of spatial degrees, polarization, phase, and wavelength are the cornerstone of nowadays information and communication technologies. Our approach is unique since it can provide beam shaping wide range tuning capabilities, sub-wavelength resolution as well as time modulation beyond GHz frequencies.
Relevant publications
2022
- Mushegh Rafayelyan, Henrik Melkonyan, Arman Tigranyan, and Etienne Brasselet, «Optical cloaking of macroscopic objects by geometric-phase vortex processing», Journal of Optics, 24(9), p 094005. PDF
- Gianni Jacucci, Louis Delloye, Davide Pierangeli, Mushegh Rafayelyan, Claudio Conti, and Sylvain Gigan, «Tuneable spin-glass optical simulator based on multiple light scattering», Physical Review A, 105(3), p. 033502. PDF
2021
- Davide Pierangeli, Mushegh Rafayelyan, Claudio Conti, and Sylvain Gigan, «Scalable spin-glass optical simulator», Physical Review Applied, v15, 3, p. 034087. PDF
- Tatevik M Sarukhanyan, Hermine Gharagulyan, Mushegh S Rafayelyan, Sergey S Golik, Ashot H Gevorgyan, and Roman B Alaverdyan, «Multimode Robust Lasing in a Dye-Doped Polymer Layer Embedded in a Wedge-Shaped Cholesteric», Molecules, v.26, 19, p 6089. PDF
- Ashot H Gevorgyan, Sergey S Golik, Nikolay A Vanyushkin, Ilya M Efimov, Mushegh S Rafayelyan, Hermine Gharagulyan, Tatevik M Sarukhanyan, Meruzhan Z Hautyunyan, and Gvidon K Matinyan, «Magnetically Induced Transparency in Media with Helical Dichroic Structure»,Materials, v.14, 9, p. 2172. PDF
2020
- Mushegh Rafayelyan, Jonathan Dong, Yongqi Tan, Florent Krzakala, and Sylvain Gigan, «Large-scale optical reservoir computing for spatiotemporal chaotic systems prediction», Physical Review X, v10, 4, p. 041037. PDF
- Jonathan Dong, Ruben Ohana, Mushegh Rafayelyan, and Florent Krzakala, «Reservoir Computing meets Recurrent Kernels and Structured Transforms», Advances in Neural Information Processing Systems, v.33. PDF
2019
- Jonathan Dong, Mushegh Rafayelyan, Florent Krzakala, and Sylvain Gigan, «Optical reservoir computing using multiple light scattering for chaotic systems prediction», IEEE Journal of Selected Topics in Quantum Electronics, 26, no. 1, p. 1-12. PDF
- M. S. Rafayelyan, H. Gharagulyan, T. M. Sarukhanyan, A. H. Gevorgyan, R. S. Hakobyan, and R. B. Alaverdyan, «Light energy accumulation by cholesteric liquid crystal layer at oblique incidence», Liquid Crystals, 46, no. 7, p. 1079-1090. PDF
up to 2018
- Mushegh Rafayelyan, Gonzague Agez, and Etienne Brasselet. «Ultrabroadband gradient-pitch Bragg-Berry mirrors», Physical Review A 96, no. 4, p. 043862 (2017). PDF
- Mushegh Rafayelyan and Etienne Brasselet. «Bragg-Berry mirrors: reflective broadband q-plates», Optics letters, 41, no 17 p. 3972-3975(2016). PDF
- Mushegh Rafayelyan, Georgiy Tkachenko, and Etienne Brasselet. «Reflective spin-orbit geometric phase from chiral anisotropic optical media», Physical review letters, 116, no. 25, p. 253902 (2016). PDF
Conference presentations
2022
- Tatevik Sarukhanyan, H. Melkonyan, M. Rafayelyan,«Nonlinear liquid crystals as a resource towards scalable and ultraefficient optical neural networks», the 9th International Symposium “Optics & its applications 2022” (OPTICS-2022)․
- Arman Tigranyan, A. Sargsyan, M. Rafayelyan, «Programmable optical linear operations based on multiple light scattering through disordered medium», the 9th International Symposium “Optics & its applications 2022” (OPTICS-2022)․
2021
- T.M. Sarukhanyan, H. Gharagulyan, M.S. Rafayelyan, S.S. Golik, A.H. Gevorgyan, R.B.Alaverdyan, «The role of the geometric phase in cholesteric with a defect layer toward multimodal robust lasing», International Conference Laser Physics 2021, Book of Abstracts, Ashtarak, Armenia, p. 46.
- A. Tigranyan, N. Badalyan, A. Sargsyan, M. Rafayelyan, «Optical scalable matrix-vector multiplication», International Conference Laser Physics 2021, Book of Abstracts, Ashtarak, Armenia, p. 41.
- M. Rafayelyan, D. Pierangeli, C. Conti, and S. Gigan, «Large-Scale Optical Simulator for Spin Glasses», International Conference Laser Physics 2021, Book of Abstracts, Ashtarak, Armenia, p.30.
- Mushegh Rafayelyan, Invited talk at Laser Physics 2021, Institute for Physical research, Ashtarak, Armenia, “Large-Scale Optical Simulator for Spin Glasses”, 09/2021
- Mushegh Rafayelyan, Invited talk at Frontiers in Optics & Photonics summer school, Yerevan, Armenia, “Optical Large-scale matrix multiplications”, 09/2021
- Mushegh Rafayelyan, Invited seminar at A. Alikhanyan National Laboratory, Yerevan, Armenia, “Large scale optical simulations”, 09/2021
- Mushegh Rafayelyan, Invited seminar at Russian-Armenian University, Yerevan, Armenia, “Large-scale neural networks in optics”, 09/21