Looks good @grishabh147, and thanks @frapac for the code review!
Just one suggestion: I think it would be better to separate the dual decomposition solver from BlockNLPModels since we want to keep BlockNLPModels minimal. Eventually, we need to store it in a separate repository, but if that's too complicated for now, then we can keep it out of the src directory. E.g., we store other source code in the example directory. In that way, we can make the BlockNLPModels' dependency minimal.
Looks good @grishabh147, and thanks @frapac for the code review!
Just one suggestion: I think it would be better to separate the dual decomposition solver from
BlockNLPModelssince we want to keepBlockNLPModelsminimal. Eventually, we need to store it in a separate repository, but if that's too complicated for now, then we can keep it out of thesrcdirectory. E.g., we store other source code in theexampledirectory. In that way, we can make theBlockNLPModels' dependency minimal.