Initialize neuron network
new.network.RdThis function initializes a new network object with specified parameters. Networks are used to simulate two-dimensional cortical patches (of layers and columns) using Growth Transform dynamical systems.
Details
Mathematically, networks are points (representing neurons) connected by directed edges. Within the growth-transform (GT) model framework, these edges are transconductance values representing synaptic connections between neurons.
Point types: Points can be grouped by types, which affect their behavior and connectivity. Within the GT model framework, these types each have their own temporal modulation constants (determining, e.g., whether the cell bursts or fires singular spikes) and valence (excitatory or inhibitory).
Global structure: Modelling the mammalian cortex, networks are assumed to divide into a coarse-grained two-dimensional coordinate system of layers (rows) and columns (columns). Each point is assigned to a layer-column coordinate (called a "node"), having both local x-y coordinates within that node and a global x-y coordinate within the network.
Local structure: Each layer-column coordinate defines a "node" containing a number of points determined by layer and type. Connections (edges) within a node are determined by a local recurrence factor matrix determining the transconductance between points of each type. These edges are called "local".
Long-range projections: Connections (edges) between points in different nodes are determined by a long-range projection motif and labelled with the same of that motif.