PyNN and NeuroML are independently developed approaches to allow portability of models across simulators. These reflect 2 differing approaches to model specification:
- Declarative specification: the structure of the model is explicitly specified in a structured model exchange format. XML is well suited as a basis for a language in this format, and is used by NeuroML as well as SBML and CellML.
- Procedural specification: the function calls or procedures for building a model are standardised. This is the case with PyNN, where Python scripts can be used to create simulations on multiple simulators.
These approaches are complementary, and a number of options are available to allow interaction between PyNN and model components in NeuroML (in particular NeuroML v2.0).
|Information on the latest developments towards greater interaction between PyNN & NeuroML2 can be found here here and examples of the conversions can be found on Open Source Brain.|