Currently, we have DualizedNLPBlockModel, AugmentedNLPBlockModel, and ProxAugmentedNLPBlockModel in BlockNLPModels.jl. But these data structures are needed only for implementing certain decomposition algorithms, and they are not the crucial element for structured NLP's data structure. Since BlockNLPModels aims to be a minimal data structure for structured NLPs, I suggest we move these to BlockNLPAlgorithms. Another alternative is to keep them in a separate extension package, say, BlockNLPModelsExtensions.jl. Any thoughts, @grishabh147 @frapac @AnirudhSubramanyam?
Currently, we have
DualizedNLPBlockModel
,AugmentedNLPBlockModel
, andProxAugmentedNLPBlockModel
inBlockNLPModels.jl
. But these data structures are needed only for implementing certain decomposition algorithms, and they are not the crucial element for structured NLP's data structure. SinceBlockNLPModels
aims to be a minimal data structure for structured NLPs, I suggest we move these toBlockNLPAlgorithms
. Another alternative is to keep them in a separate extension package, say,BlockNLPModelsExtensions.jl
. Any thoughts, @grishabh147 @frapac @AnirudhSubramanyam?