Relaxation in time elapsed neuron network models in the weak connectivity regime (Q1731106)

The dynamics of the fully connected neuron network is given by the following nonlinear time elapsed evolution equation: \[\partial_tf=-\partial_xf-a(x,\epsilon m(t))f,\quad f(t,0)=p(t),\quad f(0,x)=f_o(x), \] where $f(x,t)$ denotes the density number of neurons which at time $t\geq 0$ are in the state $x\geq 0$. Here $a(x,\epsilon\mu)\geq 0$ represents the firing rate of a neuron in state $x$ for a network activity $\mu\geq 0$ and a network connectivity parameter $\epsilon\geq 0$. The function $m(t)=\int_0^\infty p(t-y)b(dy) $ describes the global neuronal activity at time $t$ resulting from earlier discharges. One defines also the delay distribution $b$ as a probability measure taking into account the persistence of the electric activity to the discharges in the network. Under some special assumptions on the firing rate $a(.,.)$, the delay distribution $b$ and initial data, the existence and uniqueness of weak solutions is assured and one proves a long-time asymptotic result on the solutions. Two separate cases are considered: the case without delay for $b=\delta_o$, $m(t)=p(t)$ and the case with delay when $b$ is a smooth function. However, the strategy of proof is the same.