Past Research
During my PhD, we discovered that the driven Chua’s chaotic circuit presented a transition to chaos via destruction of a two-frequency torus. Though this route was known, the description of topological changes suffered by the chaotic attractor was not. Later on, I work to define the phase of chaotic oscillators. Though the phase of a periodic signal is a well-known quantity, for chaotic trajectories, it is not. Our work has proposed a mathematical framework to define and measure it.
To secure communication systems based on chaos, the chaotic wave signal encoding information should not be transmitted. Instead, one could transmit the trajectory’s Poincaré first return times (PFRT). This understanding resulted in the cipher today one of the founding references for the area of chaos-based cryptography. We have proved (see http://hdl.handle.net/10216/65184) a conjecture that the distribution of PFRTs can be analytically calculated by the eigenvalues of the unstable periodic orbits of the chaotic attractor, and shown that a very important and popular quantity to characterise chaotic behaviour - the Kolmogorov-Sinai entropy – can be estimated from PFRTs.
Typically, chaotic behaviour in a nonlinear system is replaced by periodic behaviour even when arbitrarily small parameter alterations are made. I worked on experiments to demonstrate this and validate works that explain why periodic behaviour is ubiquitous in nature, came in Ref. [42,90].
My work in neural networks with neurons connected by both electrical and chemical synapses has provided a rigorous ground to explain the regulatory effect that electrical synapses have in the brain. The study of such networks, known as multiplex networks, only became a topic of attention by the scientific community by 2013. We have proposed a formula for the mutual information in terms of Lyapunov exponents (that quantify chaos) and used to provide support for the Infomax theory that suggests that the brain evolve by maximising information. We have discovered the phenomenon of Collective Almost Synchronization, ubiquitous in complex networks, and that has been shown to be crucial for the chaos enhancement of machine learning methods to model and predict EEG signals. I have also shown that the cause-effect relationship between variables, or causality (and that is usually characterised by temporal quantities), is a spatial-temporal phenomenon. Finally, we have shown that a power grid without control systems can be designed to be reliable by properly connecting generators and consumers.