Get Started#
brainpy.state is the point-neuron modeling layer of the BrainX ecosystem:
one differentiable substrate that serves two audiences from the same code.
If you come from computational neuroscience, you build biophysical spiking networks, run them efficiently, and analyze their dynamics.
If you come from brain-inspired computing / SNN-ML, the very same models are differentiable — you train them with surrogate gradients and scale them with linear-memory online learning.
These three pages take you from a clean install to a running network to the mental model that makes the rest of the documentation click.
Get a working install for CPU, GPU (CUDA), or TPU — or pull the whole BrainX ecosystem in one command.
Build and run an excitatory–inhibitory balanced network, then plot a
spike raster. The fastest way to feel what brainpy.state does.
The four big ideas — state, units, composition, and transform — plus
the “two worlds, one substrate” framing that runs through the whole site.
Where to go next#
After Get Started, the Core Concepts spine explains why the pieces fit together — most importantly the keystone AlignPre / AlignPost — the keystone design. From there, pick your track: BrainPy-style Modeling for native modeling and training, or the API Reference to look up a specific model.