NeuralEnsemble/PyNN

What it solves

PyNN provides a simulator-independent language for building neuronal network models. It eliminates the need to rewrite model code when switching between different simulators or neuromorphic hardware systems, allowing researchers to write their model once and run it across multiple platforms without modification.

How it works

PyNN uses a high-level Python API to abstract the creation of populations of neurons, layers, columns, and their connections. It provides a library of standard neuron, synapse, and synaptic plasticity models that are verified to work consistently across supported simulators (such as NEURON, NEST, and Brian 2). It also includes built-in connectivity algorithms like all-to-all, random, and small-world, while allowing users to define their own custom connectivity in a simulator-independent way.

Who it’s for

Computational neuroscientists and researchers building neuronal network models who want to avoid being locked into a specific simulator or hardware system.

Highlights

• Simulator-independent API for consistent model execution across NEURON, NEST, and Brian 2.
• Support for neuromorphic hardware systems.
• High-level abstraction for populations, layers, and columns.
• Verified library of standard neuron and synapse models.
• Flexible connectivity algorithms for network structure.