In the proof im reading for the CLT, they seem to be using the following claim:
If $h(t)=o(t^2)$ then if $$g(n)\stackrel{\text{def}}{=}h\Big(\frac{t}{\sigma\sqrt{n}}\Big)$$ we have that $g(n)=o(\frac{1}{n})$.
More explicitly, I have a function $h$ that satisfies $$\lim_{t \rightarrow \infty} \frac{h(t)}{t^2} = 0$$ (and I seem to know nothing else about $h$) and then they are telling me that $$\lim_{n \rightarrow \infty} h\Big(\frac{t}{\sigma\sqrt{n}}\Big) n = 0 .$$ But I don't see how I can say anything about $\lim_{x \rightarrow 0} h(x)$ just from knowing $h(t)=o(t^2)$.
Is the claim true? How can I show it?
They likely meant to say $h(t)\in o(t^2)$ as $t\to0$ rather than $t\to\infty$, since otherwise the claim is false: take $h(t)=t$ (so that $h(t)\in o(t^2)$ as $t\to\infty$), then $g(n) = \frac t{\sigma\sqrt{n}}$ is not in $o(\frac1n)$ as $n\to\infty$ since $\lim_{n\to\infty}\frac{tn}{\sigma\sqrt n}=\infty$.
However, if $h(t)\in o(t^2)$ as $t\to0$, then the claim is indeed true since $$ \lim_{n\to\infty}h\left(\frac t{\sigma\sqrt n}\right)n = \frac{t^2}{\sigma^2}\lim_{s\to0}\frac{h(s)}{s^2} = 0 $$ using the change of variables $s = \frac t{\sigma\sqrt n}$, showing that $g(n)\in o(\frac1n)$ as $n\to\infty$.