# Why is the step-size $\alpha_t = 1/t$ not appropriate for non-stationary function approximation?

Sutton-Barto (Section: Selecting Step-Size Parameters Manually, page: 222)

The classical choice $$\alpha_t = 1/t$$, which produces sample averages in tabular MC methods, is not appropriate for TD methods, for nonstationary problems, or for any method using function approximation.

Why is the choice $$\alpha_t = 1/t$$, which produces sample averages in tabular MC methods? Why is it not appropriate for TD methods, for nonstationary problems, or for any method using function approximation?

• I think there's a typo in your first question. You're asking "why is the choice x", but then you don't finish the sentence. Or are you asking why that step size produces sample averages? Having said that, it seems that you're asking more than one specific question in this post. Please, only one question per post.
– nbro
Apr 11 at 8:44