twetto / iq-neuron

integer quadratic integrate-and-fire neuron

Integer Quadratic Integrate-and-Fire Neuron

Library for IQIF. A binary libiq for native C++ runtime and shared libraries libiq-network, libiz-network, and liblif-network are included. Please see below for running/installing instructions.

Dependencies

Runtime:

• OpenMP >= 4.5

• python3-matplotlib (plotting function)

Buildtime:

• gcc (C++11)

• cmake >= 3.9

• checkinstall (Debian-based packaging)

• base-devel (Arch-based packaging)

Compile & Install Shared Libraries:

libiq-network.so, libiz-network.so, and liblif-network.so can be used for bridging to python-iqif.

Universal installation

mkdir build && cd build
cmake .. (-DCMAKE_INSTALL_PREFIX=<your preferred directory>)
make -j
sudo make install

Debian-based installation

Instead of sudo make install you can use

sudo checkinstall --pkgname iq-neuron

Uninstall the package with sudo dpkg -r iq-neuron.

Arch-based installation

First download the PKGBUILD, go to the working directory, then

makepkg -si

Uninstall the package with sudo pacman -Rs iq-neuron.

Configuration & Usage

Please see the tutorial first.

You can change the synaptic weights in the Connection Table. The numbers in each lines are pre-synapse neuron index, post-synapse neuron index, weight, time constant respectively.

You can change the neuron parameters in the neuron parameter file. The parameters in each lines are neuron index, rest potential, threshold potential, reset potential, noise strength respectively.

It is recommended to use multithreading only when number of neurons is large (>100 for example).

I also have Izhikevich model and Leaky integrate-and-fire model for comparison. They are already in the shared libs. You need to change the main code to let it work in binary though.

Compile & Run

Run the IQIF directly using libiq binary. Use this method if you feel more comfortable with C++ instead of Python.

mkdir build && cd build
cmake ..
make -j
./libiq < ../inputs/session.txt (or use your custom session)
../utils/iq_plot.py