Neural Engineering

Transformative Technologies

Work Package: WP4
Programme: P9
Deliverable: Deliverable 9.1: “Relating structural coupling with dynamical correlation in laser networks”

Deliverable due date: month 30

This document reports on the progress of work on Deliverable 9.1. All the planned tasks related to this Deliverable have been accomplished.
Obtained results have been published in two open access journals
1) New Journal of Physics 17, 093002 (2015)
2) Optics Express 23, 5571 (2015)
and are accessible with no restrictions under the following links:
1)
http://iopscience.iop.org/article/10.1088/1367-2630/17/9/093002
2)
https://www.osapublishing.org/oe/fulltext.cfm?uri=oe-23-5-5571&id=312350
These publications are a scientific part of the Deliverable 9.1 and this “Project Deliverable Report” document describes the relation with NETT research goals.
Excitable systems are ubiquitous in nature, and in these systems, noise, couplings, and correlations play crucial roles in their information processing capabilities, enhancing, for example, detection of and response to very weak (subthreshold) external signals. In the first aforementioned publication (NJP 2015) we investigated the role of noise, both dynamical and observational noise, by using a symbolic analysis tool known as permutation entropy. This entropy provides information about temporal correlations present in a sequence of events, such as neuronal spikes or optical spikes. We analyzed the influence of noise and characterized the transition from the weak noise regime (where the system encodes information about weak external signals in sequences of correlated spikes), to the strong noise regime (where the systems loses its capability of encoding subthreshold signals in a correlated spiking output).
In the second publication (OE 2015) we investigated the role of direct coupling of weak external inputs, specifically, we modulated experimentally the laser pump current with a sinusoidal signal of varying frequency, and analyzed the role of the frequency in the dynamical correlations induced by this direct structural (hardware) coupling. We performed a detailed analysis of the sequence of correlated spikes that were generated. We determined practical limits for encoding information in a spiking optical output. We found that, if the external signal is too fast, the coupling is not effective, and it does not affect the intrinsic temporal correlations present the natural laser spiking output. In consequence, the weak external input, which is coupled to the laser via “structural” hardware (direct current modulation), is functionally uncoupled: it is not encoded in the laser spiking output. On the contrary, if the external signal is sufficiently slow, the laser encodes the information in its spiking output: the probabilities of the most probable and less probable symbolic patterns vary with the frequency of the signal. In this way, the input signal is functionally coupled to the laser spiking output.
In both publications we did a detailed comparison experiments-model simulation, which not only confirmed the robustness of the experimental findings, but also, allowed to explore parameter regions that are not accessible in the experiments due to the limited bandwidth of the detection equipment.
The work also covers milestone 13.

Contributors: Carlos Quintero Quiroz, S Pigolotti, MC Torrent, Cristina Masoller, Taciano Sorrentino, Andres Aragoneses, Jordi Garcia-Ojalvo, Antonio J Pons (UPC)