Any #networks #rstats folks around? I have a problem that I need the right words to describe so I can search for solutions.

Say I have a set of nodes ('terminals') and one additional node ('source') in a network. I don't care how the source is connected to the terminals, but I want to minimize the total edge distance. The solution is most likely a hub-and-spoke model (limited to three or fewer tiers)...

Thanks folks, typing that out gave me an idea I need to try. Maybe this is a hierarchical clustering problem followed by a linear programming problem to calculate the new node locations.

MST was only the beginning. Things are getting interesting now.