Almost all important situations in nature and in laboratory face the modeler with the challenge of multiphysics. Multiple processes with wildly different scales in space and time develop simultaneously, often in overlapping regions. Each process is modeled by well known equations and well develop algorithms and modeling software often exist for each process. But to model the complexity of the multiplicity of concurrent processes what should we attempt. Should we federate the existing models and wire together the existing codes using perhaps some advanced computer science infrastructure to best handle the concurrency? Or should we unify our understanding into one single theoretical mathematical framework where the various processes are modeling by instances of the same general methodology?
Both approaches have their appeal. Federation promises to cash on the long accumulated expertise and to minimize the need to develop new methods and new codes. However the experience is often that this promise is betrayed in the quagmire of maddening complexity in getting incompatible algorithms to integrate properly. Unification has the promise of avoiding the need to integrate incompatible algorithms but perhaps at the cost of wasting established methods and codes.
We discuss this duality within the scope of the European Space Weather projects currently funded by the European Space Agency and the European Commission.