PyNN is a project written in python that aims at interfacing a handful of neural network simulators. A unified high level interface means that a model can be configured easily and only once and then it can be deployed on all supported simulators. Different simulators provide diverse features and also might use various numerical approaches and approximations leading to possibly different results. Having an option to compare data from several simulators without extra work is a great feature PyNN offers.

GeNN is an efficient neural network simulator written in C++. The characteristic feature of this simulator is the ability to run simulation on GPU thus greatly decreasing required computational time.

Interfacing GeNN from PyNN would be a valuable acquisition for both projects. PyNN will have another, faster simulator at hand and GeNN will get a python interface. Python is much more widespread among the neuroscientists. Having such an interface will help to make GeNN more popular and hopefully will reduce the time scientists spent on their simulations.



Anton Komissarov


  • Jamie Knight
  • Thomas Nowotny