Non-stationarity of objectives refers to situations where the goals, targets, or reward structures of a system change over time, rather than staying fixed. This concept appears often in machine learning, reinforcement learning, multi-agent systems, and real-world decision-making contexts. Here's a breakdown: In simple terms: A stationary objective means the goal stays the same throughout training or operation. A non-stationary objective means the goal changes — either gradually or abruptly.