Open
Description
Summary
PR #1961 has raised a broader issue related to how the decomposition in pyomo.network works. The decomposition utilizes deactivated Arcs to determine the structure of the problem. However, this assumes the Arc has been expanded. What if someone actually wanted to deactivate the Arc without expanding it? There is also a potential issue working with GDP transformations. The BigM and Hull transformations deactivate the Disjuncts, which is where the deactivated Arcs live (the deactivated arcs do not get copied). I'm not sure the sequential decomposition would work after applying a BigM or Hull transformation.