{"id":"arxiv/2609.08567","url":"https://arxiv.org/abs/2609.08567","tags":["math.OC"],"score":0,"title":"Structured Positive-Definite Optimal Control Synthesis of Closed-Loop Recommendation Systems over Social Networks","domain":"math.OC","source":"arxiv","authors":["Simone Mariano","Paolo Frasca"],"reviews":0,"subject":"math.OC","summary":"We study the design of feedback recommendation policies for networked multi-topic opinion dynamics. The design of recommendations is formulated as an infinite-horizon linear-quadratic optimal control problem that favors suggestions aligned with each agent's current opinion, thereby using opinion alignment as a proxy for engagement. At the same time, the performance index penalizes polarization, deviation from the uncontrolled equilibrium of the opinion dynamics, and from the current agents' opinion, recommendation effort, and favors coherence among neighbors' recommendations. An agent-wise…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-08","reactions":{}}