Asymptotic expansions of densities of sums of random vectors without third moment (Q1871219)

Let \((X_1,Y_1),\ldots ,(X_n,Y_n)\) be i.i.d. random vectors with common density function \(f(x,y)\), \(EX_1=EY_1=0\), \(EX_1^2=EY_1^2=1\), \(EX_1Y_1=\rho \in (-1,1)\). Then \[ \lim_{n\to \infty }P\left(\sum_{j=1}^n X_j/\sqrt{n}\leq x,\sum_{j=1}^n Y_j/\sqrt{n}\leq y\right)=\int_{-\infty }^x\int_{-\infty }^y\Phi (u,v) du dv, \] where \[ \Phi (u,v)=\frac{1}{2\pi \sqrt{1-\rho ^2}}\exp \left\{-\frac{1}{2(1-\rho ^2)}(u^2-2\rho uv+v^2)\right\}. \] Let \[ f_n(x,y)={\partial^2\over \partial x\partial y}P\left(\sum_{j=1}^n X_j/\sqrt{n}\leq x,\sum_{j=1}^n Y_j/\sqrt{n}\leq y\right). \] The paper concerns the case \(E(X_1^2+Y_1^2)^{3/2}=\infty \). Assuming \(\lim_{t\to \infty }f(tx,ty)/ f(t,t)=l(x,y)\) uniformly on \(S=\{(x,y):x^2+y^2=1\}\) and \(l(ax,ay)=a^{-4-\alpha }l(x,y)\) for all \(a>0\), \(\alpha \in (0,1)\), \(\sup _{x^2+y^2}l(x,y)<\infty \) and \(f\) bounded, the paper proves that as \(n\to \infty\), \((f_n(x,y)-\Phi (x,y))(n^2f(\sqrt{n},\sqrt{n}))^{-1}\to L(x,y)\) uniformly on \(R^2\), where \[ L(x,y)=(2\pi)^{-2}\iint e^{-it_1x-it_2y}L^*(t_1,t_2)e^{-r_1/2-r_2/2-t_1t_2\rho } dt_1 dt_2 \] and \[ L^*(t_1,t_2)=\iint \left\{e^{it_1x+it_2y}-1-(it_1x+it_2y)-\tfrac{1}{2}(it_1x+it_2y)^2\right\}l(x,y) dx dy. \]