Exponential function limit with error term

128 Views Asked by At

This question is in the context of the convergence of characteristic functions. If we have an i.i.d sequence of random variables $\{ X_n \}_{n = 1}^\infty$ each with mean $\mathbf{E} [X_n] = 0$ and finite variance $\mathbf{Var}[X_n] = \sigma^2 > 0$, and denote the partial sums by $S_n = \sum_{i = 1}^n X_i$ then the characteristic function of $$ Z_n = \frac{S_n}{\sigma \sqrt{n}} $$ will be $$ \mathbf{E} \left[ \exp{ \left( i t Z_n \right) } \right] = \mathbf{E} \left[ \exp{ \left( \frac{itX_1}{\sigma \sqrt{n}} \right) } \right]^n = \left(1 + \frac{i \mathbf{E}[X_1]}{\sigma \sqrt{n}} - \frac{t^2 \mathbf{E}[X_1^2]}{2 \sigma^2 n} + o \left( \frac{1}{n} \right) \right) = \left( 1 - \frac{t^2}{2n} + o \left( \frac{1}{n} \right) \right)^n $$ I'm not sure how to prove that the vanishing factor of $o(1/n)$ can be ignored so that the limit still goes to $e^{-t^2/2}$ as $n \to \infty$ as expected, could someone explain?

Edit: I think I've come up with a solution. We can follow along the same lines of the proof that the expression without the error term converges to the exponential. We have

$$ \lim_{n \to \infty} \log\left( 1 - \frac{t^2 + o(1)}{2n} \right)^n = \lim_{n \to \infty} n \log\left( 1 - \frac{t^2 + o(1)}{2n} \right) \\ = - \lim_{n \to \infty} \frac{\log\left( 1 - \frac{t^2 + o(1)}{2n} \right) - \log{1}}{ -\frac{t^2 + o(1)}{2n}} \lim_{n \to \infty} \frac{t^2 + o(1)}{2} \\ = - \frac{d}{dt} \log{t} \bigg\rvert_{t = 1} \lim_{n \to \infty} \frac{t^2 + o(1)}{2} = -\frac{t^2}{2} $$ And hence $$ \lim_{n \to \infty} \left( 1 - \frac{t^2 + o(1)}{2n} \right)^n = \exp{\left( \lim_{n \to \infty} \log\left( 1 - \frac{t^2 + o(1)}{2n} \right)^n \right)} = e^{-t^2/2} $$ Is all of this correct?