= Solution
For $Y=2/(1-U)$ with $U$ uniform on $(0,1)$,
$$
\Pr(Y\le y)=1-\frac2y\quad(y\ge2),\qquad q(y)=\frac2{y^2}\mathbf1_{\{y\ge2\}}.
$$
Thus $Y$ has a <Pareto distribution> and is the <importance sampling> proposal, concentrated on the required tail. The target tail integrand is $f(y)=1/(\pi(1+y^2))$, so its importance weight is
$$
w(y)=\frac{f(y)}{q(y)}=\frac{y^2}{2\pi(1+y^2)}.
$$
The function returns $\widetilde\theta=n^{-1}\sum_i w(Y_i)$. Since $\mathbb E_qw(Y)=\int_2^\infty f(y)\,dy=\theta$, it is an <unbiased estimator>. For its second moment,
$$
\mathbb E_qw(Y)^2=\frac1{2\pi^2}\int_2^\infty\frac{y^2}{(1+y^2)^2}\,dy.
$$
An antiderivative is $\tfrac12(\arctan y-y/(1+y^2))$, so the integral is $\tfrac12(\pi\theta+2/5)$. Independence of the proposals gives
$$
\boxed{\operatorname{Var}(\widetilde\theta)=\frac1n\left(\frac{\theta}{4\pi}+\frac1{10\pi^2}-\theta^2\right).}
$$
This is <importance sampling of a Cauchy tail>, with weights between $2/(5\pi)$ and $1/(2\pi)$. Their narrow range explains the much smaller <variance> than the indicator estimate, without appealing to nonexistent Cauchy moments.
Back to article page
