300_pointneurons example in NeuroML

[pgleeson]

Good progress getting this running using https://github.com/NeuroML/NeuroMLlite/blob/master/neuromllite/SonataReader.py.

Plot below shows Sonata->NeuroML->Neuron (top) and BMTK->NEST (bottom) for a ~10 times reduced network (note colours of populations don't match).

Good match on subthreshold single cell behavour too.

Problems arise when scaling up though (seg fault in Neuron), possibly due to inefficient way I'm structuring applying the spike times in the generated Neuron code. Will try to optimise...

[pgleeson]

Full 300 cell network is producing a good match now (compare e.g. red population on left (Sonata->PyNN) to blue pop on right (Sonata->NeuroML->Neuron)):

Still painfully slow to set up/run in Neuron, some more optimisation of code required...

[pgleeson]

Have an optimised version of the export now with good subthreshold matching for the cells too (left (Sonata->PyNN->NEST), right (Sonata->NeuroML->NEURON)):

Working on getting this more integrated with NeuroMLlite/pyNeuroML for ease of use