Nonlinear dynamics & physical computing
Dynamics, feedback and computation
We study nonlinear behaviour in semiconductor lasers, systems with delayed feedback and reconfigurable electronic circuits. Experiments and modelling help us understand their dynamics and investigate how those dynamics can be used for information processing and computation.
Our work connects fundamental questions about stability, bifurcations and chaos with photonic reservoir computing and analog optimization. The physical behaviour of a system is both an object of study and a possible computational resource.
Research in practice
Controlling light. Computing with dynamics.
Semiconductor lasers · Optical feedback
Reproducible wavelength tuning
This laser-tuning work is a collaboration with B-PHOT, Vrije Universiteit Brussel (VUB).
R. de Mey et al., IEEE Photonics Technology Letters 38, 378–381 (2026).Open full-size figure (opens in a new tab)
Optical feedback changes the states available to a semiconductor laser. By coordinating the feedback strength and phase, dynamic targeting guides the laser along a chosen path.
The free-space experiment demonstrated tuning over 2.1 GHz. Forward and backward measurements show how the optical frequency follows the control sweep.
Group contributor: Alexandre Locquet.
Read the 2026 study →AATLAS · Analog–digital computing
Analog cores, digital coordination
Programmable analog networks use their physical dynamics to search for solutions to optimization problems. Digital coordination supports their configuration, communication and interaction with the application.
AATLAS investigates how to extend this approach across multiple analog cores, including the partitioning of larger problems and the balance between solution quality, computation time and energy consumption.
The diagram shows the project architecture concept; multicore implementation and scaling are research objectives.
Group contributors: Alexandre Locquet · Muhammad Usman · Harikrishna Kayitha · Corentin Bounon.
Explore AATLAS →Experimental and modelling capabilities
What we can investigate
- Laser and delay-system dynamics. Time-series analysis, experiments and modelling to identify oscillatory regimes, bifurcations and feedback-dependent behaviour.
- Programmable analog networks. Implementation and characterization of nonlinear networks on FPAAs, including their transients, steady states and comparison with dynamical models.
- Computational performance. Studies of how network configuration and problem mapping affect solution quality, computation time and energy use.
Circuit resources, connectivity and device characteristics constrain the networks that can be implemented. Multicore scaling and the partitioning of larger optimization problems remain active research objectives.
Discuss a system or computational problem →Research themes
From laser feedback to programmable circuits
Laser dynamics and routes to chaos
We investigate how optical and optoelectronic feedback change the behaviour of semiconductor lasers. Bifurcation diagrams, time-series analysis and models help characterize periodic, quasiperiodic and chaotic regimes, including crisis routes to chaos.
Studies of synchronization, time-delay signatures and chaotic signal generation connect these questions to optical information processing and chaos-based communication.
Browse laser-dynamics papers →Pulse patterns and delayed feedback
Delayed feedback can generate square waves, multiplet pulse sequences and complex oscillatory patterns. We study how these regimes emerge and depend on system parameters, including their relationship to optical frequency-comb dynamics.
This work provides a basis for investigating signal generation and the use of nonlinear pulse dynamics in information processing.
Read the frequency-comb study →Photonic reservoir computing
Reservoir computing uses the response of a dynamical system to transform input signals into states that can be used for a computational task. We investigate how feedback, filtering and delay affect the memory and performance of laser-based reservoirs.
Our recent work includes reservoir computing with state-dependent delay.
Read the reservoir-computing study →Reconfigurable analog computing
Field-programmable analog arrays (FPAAs) provide an electronic platform for implementing nonlinear networks. We investigate Hopfield networks and Ising machines that use physical dynamics to search for solutions to combinatorial optimization problems.
Within AATLAS, this research connects analog computing cores with FPGA-based digital coordination and the study of larger, multicore architectures.
Explore the AATLAS project →Related project
AATLAS: hybrid analog–digital optimization
AATLAS brings together Georgia Tech–CNRS, Institut Jean Lamour and CentraleSupélec to develop a reconfigurable optimization platform combining FPAA computing cores and FPGA coordination. The project studies network implementation, multicore scaling and the trade-offs between solution quality, computation time and energy consumption.
Our FPAA work includes collaboration with Jennifer Hasler at Georgia Tech. Meet the AATLAS partners and collaborators →
Explore the AATLAS projectSelected publications
Recent studies
- Agile Laser Wavelength Tuning Using Dynamic Targeting
R. de Mey, S. W. Jolly, A. Locquet and M. Virte · IEEE Photonics Technology Letters 38(6), 378–381 (2026). - Reservoir computing with state-dependent time delay
G. O. Danilenko et al. · Physical Review E 111, 035313 (2025). - Busse Balloon for Optical Frequency Combs From a Semiconductor Laser With Time-Delayed Optoelectronic Feedback
M. S. Islam et al. · Journal of Lightwave Technology 42(21), 7628–7633 (2024).
Teaching
Nonlinear dynamics in student projects
In ECE 6564 — Nonlinear Dynamics and Chaos, students apply these ideas to electronic circuits, robotic control, fluid mixing, biological rhythms and space systems through simulations, experiments and modelling.
View the student project showcase