Functional central limit theorem for super \(\alpha\)-stable processes (Q2386560)

Let \(\mathbb{M}_p(\mathbb{R}^d)\) be the space of Radon measures \(\mu\) on \(\mathbb{R}^d\) such that \(\langle\mu, f\rangle= \int f(x)\mu(dx)< \infty\). Let \[ C_p(\mathbb{R}^d)= \{f\in C(\mathbb{R}^d):|f(x)|\leq \text{const.}(1+ |x|^2)^{-p/2}\}\quad (x\in\mathbb{R}^d). \] Furthermore, let \(X= (X_t, Q_\mu)\) be a super \(\alpha\)-stable process. For the path-valued setting, define the numerical centered occupation time process \(Z^T_t(f)\) by \[ Z^T_t(f)= a^{-1}_d(T) \int^{T_t}_0 [\langle X_s, f\rangle- E\langle X_s, f\rangle]\,ds\quad (f\in C^+_p(\mathbb{R}^d)), \] where \(a_d(T)\) is some function of \(T\) which depends on \(d\). The author shows that in finite-dimensional distribution sense, \(Z^T_t(f)\to Z_t(f)\) in \(C([0,\infty),\mathbb{R})\), as \(T\to\infty\), where the finite-dimensional distribution of \((Z_t(f))_{t\geq 0}\) is characterized by some Laplace transformation which depends on \(d\).