Explore SparseArrayKit.jl (DOK format, N-D tensors)

[Oblynx]

https://github.com/Jutho/SparseArrayKit.jl

N-D tensors

Synapses are by nature 2D (pre-, post-synaptic addresses). In case of multiple spatial dimensions, no reason to burden the synapse matrices with those but rather should calculate linear addresses as is currently done. However a CSC [pre x post] synapse matrix has 2 disadvantages:

• slow selection of [pre] rows, so it's useful for backpropagation instead of forward
• slow (+hacky) addition of new synapses src/dynamical_systems.jl#L319

Dict of Keys (DOK) format

This format provides O(1) access in both directions + insertion, so it's interesting to compare performance in practice.

Adopting this for DistalSynapses would also eliminate hacky code to resize the CSC with minimum loss.

[pitsianis]

The flow of information on a matrix is always from columns to rows, so it should be [post x pre], right?