Provable speed advantage of Double Q-learning

Determine whether Double Q-learning is provably faster than vanilla Q-learning in terms of convergence or computational complexity.

Background

The cited peer-review example discusses theoretical analyses of Double Q-learning, including non-asymptotic convergence guarantees for synchronous and asynchronous variants. The paper under review improves certain parameter dependencies but does not establish that Double Q-learning has a better overall theoretical performance than vanilla Q-learning.

The unresolved issue is whether the mitigation of overestimation bias provided by Double Q-learning translates into a provable speed or convergence advantage over vanilla Q-learning. The question is explicitly identified as unknown in the reproduced review text.