PyNNLess is yet another abstraction layer on top of PyNN. It aims at providing a simple and stable API for experiments with relatively small spiking neural networks. It should work with all backends used in the Human Brain Project (HBP).
Backends include the software simulator NEST (versions 2.2 and 2.4), the SpiNNaker multicore system (NMMC1) developed at Manchester University, the HICANN physical modell (NMPM1) developed at Heidelberg University, its emulation, the ESS and the Spikey chip also developed at Heidelberg.
PyNNLess provides a common API for PyNN versions 0.6, 0.7 and 0.8 and works around the bugs in the hardware backend bindings. Eventually, at some point in the future these bugs will be fixed and PyNNLess will be obsolete.
Both network descriptions and recorded results are provided in a JSON-like object format, making it very easy to use PyNNLess but rendering it impractical for larger networks and long simulation times.
You might find PyNNLess interesting if you want to simulate fairly small networks (both network description and recorded results have to fit into main memory) and run them on multiple backends or with different PyNN versions.
Download the most recent version of PyNNLess using the following command:
git clone https://github.com/hbp-sanncs/pynnless.gitPyNNLess depends on PyNN in either version 0.7 or 0.8. Examples on how to
use PyNNLess can be found in the
examples
folder. You can simply execute the examples, you do not need to globally install
PyNNLess.
If you want to install PyNNLess on your system you can do so using the following command (execute from the directory into which you have downloaded PyNNLess):
sudo pip install pynnlessIt can be uninstalled with the following command:
sudo pip uninstall pynnless- Information on how to use the HBP Neuromorphic Platform: HBP Guidebook
- PyNN to SpiNNaker Wrapper: sPyNNaker
This project has been tailored to the use-cases required in our own work. If you'd like to expand the functionality please send a pull request on GitHub. Feel free to open an issue on GitHub if you think you've found a bug.
This project has been initiated by Andreas Stöckel in 2015 as part of his Masters Thesis at Bielefeld University in the Cognitronics and Sensor Systems Group which is part of the Human Brain Project, SP 9.
This project and all its files are licensed under the GPL version 3 unless explicitly stated differently.
The "tomahawk" logo has been adapted from a drawing by OpenClipart user Firkin.