Modularity and scalability of simulation models has become an important topic in the world of agricultural simulations. A lot of simulation affords have proven to be a one off event. Linking of existing models is generally a tedious job. Different researchers are currently taken different approaches to these problems. These different approaches arise partly because of different problem definitions. Below I have tried to make a comprehensive list of identified problems and solutions.
For the purpose of SimRoot, I have written my own simulation environment. This choice is at odds with the long list of simulation environments that are already available. Since, even though a simulation environment may address many of the below identified problems, incompatibility of it's modules with other simulation environments would make linking and reuse of modules still a difficult task. However, currently, none of the existing simulation environments have gained authoritative status. Standards are lacking. Besides, none of the Simulation environments that are currently available meet SimRoot's requirements to model spatial and hierarchical data. If any reader can prove me wrong on this, I would be pleased to receive a comment.
Credit should go to Fritz Evers, who got me thinking about modularity and scalability of simulation models.
Related projects/efforts:
Seamless (is an EU project that tries to spin-off an generic modeling environment)
Issues with modularity and scalability of simulation models:
- Technical issues, different programming languages, different operating systems, different platforms.
- Unit conversion issues
- Programming style issues
- The timeloop, timestep dependencies, separation between rates, integration, and communication
