Slides 14-15: combine DECOA-style decomposition-based OA (Muts, Nowak & Hendrix, JOGO 77:75-96 2020, https://doi.org/10.1007/s10898-020-00888-x) with DeCODe community detection (Allman, Tang & Daoutidis, OptEng 20:1067-1084 2019, https://doi.org/10.1007/s11081-019-09450-5). Six years later: contrib.community_detection is merged but has zero solver coupling (dormant since 2024; generate_structured_model() is the ready-made hook), Decogo appears dormant, DeCODe was never released, and contrib.benders is likewise unconnected. The Daoutidis group's 2024 "Taking the human out of decomposition-based optimization" line (C&ChE) confirms the direction is alive.
Scope: use detected communities to structure OA subproblems (per-community linearization/NLP solves with linking-variable coordination) or to auto-configure contrib.benders; prototype on gdplib/MINLPLib instances with block structure.
Part of roadmap epic #2 (item C5).
Slides 14-15: combine DECOA-style decomposition-based OA (Muts, Nowak & Hendrix, JOGO 77:75-96 2020, https://doi.org/10.1007/s10898-020-00888-x) with DeCODe community detection (Allman, Tang & Daoutidis, OptEng 20:1067-1084 2019, https://doi.org/10.1007/s11081-019-09450-5). Six years later:
contrib.community_detectionis merged but has zero solver coupling (dormant since 2024;generate_structured_model()is the ready-made hook), Decogo appears dormant, DeCODe was never released, andcontrib.bendersis likewise unconnected. The Daoutidis group's 2024 "Taking the human out of decomposition-based optimization" line (C&ChE) confirms the direction is alive.Scope: use detected communities to structure OA subproblems (per-community linearization/NLP solves with linking-variable coordination) or to auto-configure
contrib.benders; prototype on gdplib/MINLPLib instances with block structure.Part of roadmap epic #2 (item C5).