{"id":"arxiv/2606.15501","url":"https://arxiv.org/abs/2606.15501","tags":["math.OC","cs.DS"],"score":0,"title":"Modern Primal-Dual Frameworks for Prior-Free Online Resource Allocation","domain":"math.OC","source":"arxiv","authors":["Rad Niazadeh","Rajan Udwani"],"reviews":0,"subject":"math.OC","summary":"Linear-programming (LP)-based primal-dual methods are fundamental for designing and analyzing algorithms in adversarial (prior-free) online resource allocation. This chapter provides a tutorial on two modern primal-dual frameworks, emphasizing recent developments and contemporary models in operations research. Part~I develops an LP-based convex-programming framework where solving a regularized convex program at each arrival captures the tradeoff between greediness and hedging, yielding a dual certificate via Karush-Kuhn-Tucker (KKT) conditions. Because standard LP relaxations can be weak or…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-06-13","reactions":{}}