README_TwoPopulationNetworkPlastic_PyNEST.md

TwoPopulationNetworkPlastic - PyNEST implementation

Simulation script

The model is defined in model.py and parameter_dicts.py, and can be run by executing run_model.py. The resulting spike and and connectivity data is stored in the data_path defined in parameter_dicts.py. The data is plotted by executing plot_data.py.

Simulation details

By default, this implementation is based on the iaf_psc_alpha neuron and the stdp_pl_synapse_hom synapse models provided in NEST. Alternatively, the user may choose a NESTML description of the dynamics (see iaf_psc_alpha and stdp_pl_synapse) by setting pars['neuron_model']='iaf_psc_alpha_nestml' in parameter_dicts.py. To enable usage of the NESTML models, the script build_nestml_models.py needs to be executed first.

The network is connected according to the fixed_indegree connection rule in NEST.

The neuron dynamics is propagated in time using exact integration (Rotter & Diesmann (1999)) with a simulation step size $\Delta{}t$. The synapse dynamics is updated in an event-based manner as described by Morrison et al. (2007).

The model implementation runs with NEST 3.6 and NESTML 5.3.0.

Simulation parameters

Name	Value	Description
$T$	$10000\,\text{ms}$	total simulation time
$\Delta{}t$	$2^{-3}=0.125\,\text{ms}$	duration of simulation step
tics_per_step	$2^7=128$	number of tics per simulation step $\Delta{t}$ (time resolution)

References

• Linssen CAP, Babu PN, He J, Eppler JM, Rumpe B, Morrison A (2022). NESTML 5.1.0. Zenodo.

• Morrison et al. (2007). Spike-timing dependent plasticity in balanced random networks. Neural Computation 19:1437-1467

• Rotter & Diesmann (1999). Exact digital simulation of time-invariant linear systems with applications to neuronal modeling. Biological Cybernetics 81(5-6):381-402. doi:10.1007/s004220050570

• Spreizer et al. (2022). NEST 3.3. Zenodo. doi:10.5281/zenodo.6368024