Home

Welcome to the combra_loihi wiki!

combra_loihi is a neuromorphic computing library for introducing astrocytes to Intel's Loihi chip. The library is developed by Computational Brain Lab (ComBra) at Rutgers University.

Version 0.1 (11/2018)

Prerequisites:

• python 3.5.2
• NxSDK 0.7

Download combra_loihi from GitHub and unzip the folder in any directory. To import the library api in python, do the following:

import sys
sys.path.append("directory to combra_loihi")
import combra_loihi.api as combra

Before using any module or function in the library, a NxNet object needs to be created. We can create an Astrocyte object by giving the NxNet and other parameters. More information about Astrocyte module in combra_loihi library can be found here.

import nxsdk.api.n2a as nx

net = nx.NxNet()
astro = combra.Astrocyte(net, other parameters)

The library also supports predefined Neural Astrocytic Networks (NAN). We can create a simple feedforward NAN by giving the NxNet and other parameters. More information about NAN module in combra_loihi library can be found here.

nan = combra.FeedforwardNAN(net, other parameters)

Upon the creation of the network, you can run the network on Loihi and provide a simulation time. Then disconnect the network to release it from the hardware.

net.run(simulation time)
net.disconnect()

The library supports a number of plotting functions that can be used to represent the SNN activity on Loihi. More information about the plotting functions can be found here.

For detailed examples on how to use combra_loihi see here.