QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning, Tabish Rashid, Mikayel Samvelyan, Christian Schroeder de Witt, Gregory Farquhar, Jakob Foerster, Shimon Whiteson, 2018International Conference of Machine Learning 2018, Vol. 80DOI: 10.48550/arXiv.1803.11485 - This paper introduces QMIX, a value decomposition method that utilizes a mixing network with a monotonicity constraint, enabling more complex state-dependent value factorization than VDN.