Dissertação de Mestrado #601 – Luisa Fernanda Ramirez Ochoa – 15/06/2018

The objective of our work is to apply the concepts of statistical mechanics to the virtual retina of a vertebrate, in order to verify some of the assertions made in Ref [1]. By means of a whole simulation of the vertebrate retina network [2], we obtained data from the ganglion layer as a discrete pattern of activity (spike), or silence from each ganglion neuron. We used maximum entropy criteria to find the probability distribution that describes the system, and we used inverse Monte Carlo simulations for fitting the parameters to within a precision by means of a gradient descent algorithm. Since our purpose is to make a thermodynamic analysis of this system, we found this distribution for different groups of neurons with sizes ranging from 9 to 49 cells each, and we constructed the curves of specific heat. We found that these curves have a peak increasing with the system length at the effective temperature of the retina, suggesting the existence of criticality. In addition, our results agree with the experiments carried out in Ref [1] and strengthen the hypothesis that biological systems may work in a critical state.