PyNN: a common interface for neuronal network simulators

Abstract

Computational neuroscience has produced a diversity of software for simulations of networks of spiking neurons, with both negative and positive consequences. On the one hand, each simulator uses its own programming or confi guration language, leading to considerable diffi culty in porting models from one simulator to another. This impedes communication between investigators and makes it harder to reproduce and build on the work of others. On the other hand, simulation results can be cross-checked between different simulators, giving greater confi dence in their correctness, and each simulator has different optimizations, so the most appropriate simulator can be chosen for a given modelling task. A common programming interface to multiple simulators would reduce or eliminate the problems of simulator diversity while retaining the benefi ts. PyNN is such an interface, making it possible to write a simulation script once, using the Python programming language, and run it without modifi cation on any supported simulator (currently NEURON, NEST, PCSIM, Brian and the Heidelberg VLSI neuromorphic hardware). PyNN increases the productivity of neuronal network modelling by providing high-level abstraction, by promoting code sharing and reuse, and by providing a foundation for simulator-agnostic analysis, visualization and data-management tools. PyNN increases the reliability of modelling studies by making it much easier to check results on multiple simulators. PyNN is open-source software and is available from http://neuralensemble.org/PyNN. Keywords: Python, interoperability, large-scale models, simulation, parallel computing, reproducibility, computational neuroscience, translation compiler standards and simulators develop. Another is that model source code is often not written with reuse and extension in mind, and so considerable rewriting to modularize the code is necessary. Probably the most important barrier is that code written for one simulator is not compatible with any other simulator.Although many computational models in neuroscience are written from the ground up in a general purpose programming language such as C++ or Fortran, probably the majority use a special purpose simulator that allows models to be expressed in terms of neuroscience-specifi c concepts such as neurons, ion channels, synapses; the simulator takes care of translating these concepts into a system of equations and of numerically solving the equations. A large number of such simulators are available (reviewed in Brette et al., 2007), mostly as open-source software, and each has its own programming language, confi guration syntax and/or graphical interface, which creates considerable diffi culty in translating models from one simulator to another, or even in understanding someone else's code, with obvious negative consequences for communication between investigators, reproducibility of others' models and building on existing models.However, the diversity of simulators also has a number of positive consequences: (i) it allows cross-checking -t...

Reliable Recall of Spontaneous Activity Patterns in Cortical Networks

“…All simulations were performed using the NEST simulator (Diesmann and Gewaltig, 2001) version 1.9 (http://www.nest-initiative. uni-freiburg.de), using the PyNN interface (Davison et al, 2008) (http://neuralensemble.org/PyNN). The code for the model is freely available from ModelDB (http://senselab.med.yale.edu/ModelDB/) and on the UNIC website (http://www.unic.cnrs-gif.fr).…”

Managing Burstiness and Scalability in Event-Driven Models on the SpiNNaker Neuromimetic System

“…Simulations for chip verification used ARM SoC Designer Simulator on a complete SystemC model of the SpiNNaker chip. The heterogeneous model tests ran on the physical SpiNNaker test chip, using PyNN [6] as a standard simulation front-end. Our simulations used a 4-chip system with 2 processors per chip (corresponding to the first test board).…”

“…During the simulation every pool emits 756 spikes, for a total of 3024 spikes per chip. The on-chip network implementations use the PyNN [6,10] multiplatform neural description environment, permitting direct comparison of the performance on-chip with a standard software simulator-in this case NEST [11]. On-chip simulations run in real-time, thus the 1,000 ms simulation time is the actual time to run.…”

Evolving spatio-temporal data machines based on the NeuCube neuromorphic framework: Design methodology and selected applications

“…PyNN (Davison et al, 2008) is a generic SNN simulation markup framework that allows the user to run arbitrary SNN models on a number of different simulation platforms, including software simulators PyNEST and Brian, and some neuromorphic hardware systems such as SpiNNaker and FACETS/BrainScaleS. It provides a "write once, run anywhere" (where "anywhere" is the list of simulators it supports) facility for the development of SNN simulations.…”