Our bound improves on currently known bounds in the regime when the approximation error $\epsilon$ is not too small and is optimal up to logarithmic factors when $\epsilon$ is a constant.
We obtain our results by analyzing the log-Sobolev constants of appropriate Markov chains rather than their spectral gaps.