“…Instead of working on the entire search space, another decomposition method is to restrict the improvement region of each subproblem to be a local niche, also known as constrained decomposition (CD) [85]. This idea is similar to some TCH and PBI variants (e.g., [38,77,78,81,82]) that applied a Mating selection MOEA/D-DE [23], MOEA/D-TPN [90], MOEA/D-MRS [91] Neighborhood versus global ENS-MOEA/D [92], MOEA/D-ATO [93] Ensemble neighborhood MO-RCGA [94], SRMMEA [95], MOEA/D-EAM [96] Neighborhood of solutions Environmental selection MOEA/D-DE [23], EFR [97] Neighborhood versus global MOEA/DD [28], ISC-MOEA/D [98], g-DBEA [54], DDEA and DDEA+NS [54] Prioritizing isolated regions VaEA [99], OLS [100], ASEA [101] MOEA/D-STM [102], AOOSTM and AMOSTM [103], MOEA/D-IR [104] Matching solutions and subproblems MOEA/D-ARS [105], MOEA/D-CD and MOEA/D-ACD [85], MOEA/D-GR [85,106] Global replacement θ-DEA [107], MOEA/D-DU and EFR-RR [108], MOEA/D-CSM [109] Local niche competition dedicated parameterization on each subproblem. Specifically, a constrained optimization subproblem in CD is defined as: min…”