a C++/Python Agent-Based Modelling framework for large-scale distributed simulations

Pandora is a framework designed to create, execute and analyse agent-based models in high-performance computing environments. It has been programmed to allow the execution of large-scale agent-based simulations, and it is capable of dealing with thousands of agents developing complex actions. The users can choose to develop their code in Python (for fast prototyping) or C++ (complex models). Interfaces of both versions are identical, and share the same C++ base code (assuring compatibility and efficiency).

The framework has full Geographical Information System support, to cope with simulations in which spatial coordinates are relevant, both in terms of agent interactions and environment. The library also allows the researcher to design experiments containing thousands of different runs exploring the parameter sweep of the model. This can be done either by code (both C++ and Python), or using Cassandra.

Cassandra is a general purpose GUI tool that can be used to analyse the results generated by a simulation created with the library. Cassandra allows the user to plan, execute and visualize the complete execution of simulations using a combination of 2D and 3D graphics, as well as statistical figures.

Webpage: http://xrubio.github.io/pandora/

Mailing List/Google Group: http://groups.google.com/d/forum/pandora_users