BrainPy: A general platform for computational neuroscience and brain-inspired computation
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
BrainPy b447e20e30 Merge pull request 'V1.0.2' (#17) from V1.0.2 into master 3 weeks ago
brainpy Remove support for Numba-CUDA backend 3 weeks ago
develop Remove support for Numba-CUDA backend 3 weeks ago
docs Remove support for Numba-CUDA backend 3 weeks ago
examples Remove support for Numba-CUDA backend 3 weeks ago
tests Remove support for Numba-CUDA backend 3 weeks ago
.gitignore Add changelog 3 weeks ago
.travis.yml Add Travis CI 1 month ago
LICENSE GPL v3 LICENSE 8 months ago
MANIFEST.in Update installation 5 months ago
README.md Add Travis CI 1 month ago
TODO.md update TODO 3 weeks ago
changelog.rst Remove support for Numba-CUDA backend 3 weeks ago
requirements-dev.txt Fix test requirements 1 month ago
requirements-doc.txt Add stability analysis for 3d system 2 months ago
requirements.txt Add systematical documents 2 months ago
setup.py All tests passed 1 month ago

README.md

Logo

LICENSE Documentation Conda PyPI version Build Status

Why to use BrainPy

BrainPy is an integrative framework for computational neuroscience and brain-inspired computation. Three core functions are provided in BrainPy:

  • General numerical solvers for ODEs and SDEs (support for DDEs and FDEs will come soon).
  • Neurodynamics simulation tools for brain objects, such like neurons, synapses and networks (support for soma and dendrites will come soon).
  • Neurodynamics analysis tools for differential equations, including phase plane analysis and bifurcation analysis (support for continuation analysis and sensitive analysis will come soon).

Moreover, BrainPy is designed to effectively satisfy your basic requirements:

  • Easy to learn and use, because BrainPy is only based on Python language and has little dependency requirements;
  • Highly flexible and transparent, because BrainPy endows the users with the fully data/logic flow control;
  • Simulation can be guided with the analysis, because the same code in BrainPy can not only be used for simulation, but also for dynamics analysis;
  • Efficient running speed, because BrainPy is compatible with the latest JIT compilers or any other accelerating framework you prefer (below we list the speed comparison based on Numba JIT).

Speed Comparison

BrainPy is a backend-independent neural simulator. Users can define models with any backend they prefer. Intrinsically, BrainPy supports the array/tensor-oriented backends such like NumPy, PyTorch, and TensorFlow, it also supports the JIT compilers such as Numba on CPU or CUDA devices. Extending BrainPy to support other backend frameworks you prefer is very easy. The details please see documents coming soon.

Installation

BrainPy is based on Python (>=3.7), and the following packages are required to be installed to use BrainPy:

  • NumPy >= 1.13
  • Matplotlib >= 3.3

Install BrainPy by using pip:

> pip install -U brainpy-simulator

Install BrainPy by using conda:

> conda install brainpy-simulator -c brainpy

Install BrainPy from source:

> pip install git+https://github.com/PKU-NIP-Lab/BrainPy
> # or
> pip install git+https://git.openi.org.cn/OpenI/BrainPy
> # or
> pip install -e git://github.com/PKU-NIP-Lab/BrainPy.git@V1.0.0

Let’s start

Here list several simple examples for neurodynamics simulation and analysis. Comprehensive examples and tutorials please see BrainModels.

HH Neuron Model

The Hodgkin–Huxley neuron model.

AMPA Synapse Model

The AMPA synapse model.

Gamma Oscillation Model

Implementation of the paper: Wang, Xiao-Jing, and György Buzsáki. “Gamma oscillation by synaptic inhibition in a hippocampal interneuronal network model.” Journal of neuroscience 16.20 (1996): 6402-6413.

E/I Balance Network

Implementation of the paper: Van Vreeswijk, Carl, and Haim Sompolinsky. “Chaos in neuronal networks with balanced excitatory and inhibitory activity.” Science 274.5293 (1996): 1724-1726.

Continuous-attractor Network

Implementation of the paper: Si Wu, Kosuke Hamaguchi, and Shun-ichi Amari. "Dynamics and computation of continuous attractors." Neural computation 20.4 (2008): 994-1025.

Phase Plane Analysis

Phase plane analysis of the INa,p+-IK model, where "input" is 50., and "Vn_half" is -45..

Codimension 1 Bifurcation Analysis

Codimension 1 bifurcation analysis of FitzHugh Nagumo model, in which "a" is equal to 0.7, and "Iext" is varied in [0., 1.].

Codimension 2 Bifurcation Analysis

Codimension 2 bifurcation analysis of FitzHugh Nagumo model, in which "a" is varied in [0.5, 1.0], and "Iext" is varied in [0., 1.].