{"id":"arxiv/2603.17033","url":"https://arxiv.org/abs/2603.17033","tags":["math.OC"],"score":0,"title":"From Non-Identifiability to Goal-Integrated Decision-Making in Parametric Inverse Optimization","domain":"math.OC","source":"arxiv","authors":["Farzin Ahmadi","Fardin Ganjkhanloo","Kimia Ghobadi"],"reviews":0,"subject":"math.OC","summary":"Inverse optimization seeks to recover unknown objective parameters from observed decisions, yet fundamental questions about when recovery is possible have received limited formal treatment. This paper develops a comprehensive theoretical framework for inverse optimization in parametric convex models. We first establish that non-identifiability is the generic case: even with normalization and multiple observations, the parameter set compatible with data is generically multi-dimensional, and regularization does not resolve this. We derive necessary and sufficient conditions for…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-03-17","reactions":{}}