This paper analyzes multi-step TD-learning algorithms within thedeadly triad'scenario,
characterized by linear function approximation, off-policy learning, and bootstrapping. In …