Adding Populations of Neurons to a ModLIN System

A ModLIN system consists of 'populations' of rate coded neurons, connected via 'projections'. In order to build a system you must add populations, and then join them with projections.

To add a population the nw_modAddPop() function is used. This function takes arguments that describe the population that it is creating:

sys

This is the name of the SystemML system to add the population to.

'Population_name'

A name for the population, this must not already be in use in the system. If it is then an appropriate error will be generated.

rate

The BRAHMS sample rate for the processes in the population.

dims

The dimensions of the pop in numbers of nodes. this sets the node count N=prod(dims), and organises the outputs of this layer into a matrix with dimensions dims. For a population with no spatial organisation, just pass a scalar node count as dims.

locs

If equal to dims, the location of each node will be its dimension index. for example, if dims=[2 5], the nodes will have locations [1 1], [2 1], [1 2], ...

Pass a row vector of length between 1 and length(dims) in a cell array to spread nodes on a grid spanning a rectangle locs units in size. for example, if dims=[10 5] and locs={[2 2]}, the fourth node will have location [0.7 0.2].

Pass locs as a {2xD vector} to specify the lower and upper bounds of a D-dimensional cuboid to spread the nodes over.

Alternatively, pass locs explicitly as an NxD location matrix (not in a cell).

'path/to/act'

The BRAHMS Namespace path of a ModLIN activation process.

actState

State data for the activation process.

'path/to/out'

The BRAHMS Namespace path of a ModLIN output function process.

outState

State data for the output function process.