A low-level repo for neural dyamics simulation in Cython.
Mainly composed of 4 modules:
neur: simulation of dynamical systems (neurons, networks, 2D oscillators, etc.)networks: build the abstract structure of the graph for networks (small-wold, Barabasi-Albert, etc.)stochastic: some stuff for implementing stochastic processes (thresholded OU, Poisson)chem: mainly Gillespie for ionic channel dynamics
Why not just use a simulator ?:
- simulators, though easier to use and multi-platform comaptible, are slow due to the high level of structures
- easier to define non-standard models
- easier to follow numerical subtleties
- more fun
Setup the environment using pipenv
pipenv install
pipenv shell
then compile with
make
or
make hardcore
(see here for more details on Cython compilation options).
Explicit Euler is roughly three times faster than RK4 and does not show great differences, apart from a slightly different timescale
