On fractional tempered stable motion (Q2507646)

The authors define the fractional tempered stable motion (fTSm) as a process of stochastic integrals, i.e. \(\{\int_S f(t,s)dX_s^{TS}\): \(T\in \mathcal{T}\}\), where \(f:\mathcal{T}\times S\to\mathbb{R}\) is a deterministic function, and \(\{X_t^{TS}\): \(t\geq 0\}\) is a tempered stable Lévy process. More precisely, the Lévy measure of \(X_t^{TS}\) has the form \[ \nu(B)=\int_{\mathcal{R}_0^d}\int_0^\infty 1_B(sx)s^{-\alpha-1} e^{-s} ds\rho(dx),\quad B\in \mathcal{B}(\mathbb{R}^n), \] where \(\alpha\in (0,2)\), and the measure \(\rho\) is such that \[ \int_{\mathbb{R}^d_0}\| x\| ^\alpha \rho(dx)<+\infty,\quad \int_{\mathbb{R}^d_0}\| x\| \rho(dx)<+\infty,\;\text{if \(\alpha\in (0,1)\)}, \] and \[ \int_{\mathbb{R}^d_0}\| x\| (1+\ln^+\| x\| ) \rho(dx)<+\infty,\quad \text{if \(\alpha=1\)}. \] Then the fTSm in \(\mathbb{R}\) is given by \[ L_t^H:=\int_0^tK_{H,\alpha}(t,s)dX_s^{TS},\quad t\geq 0, \] where \(K_{H,\alpha}(t,s)\) is some Volterra kernel, depending on parameters \(\alpha\) and \(H\). The integral is defined in \(L^2(\Omega, \mathcal{F}, \mathbb{P})\)-sense. Further, the series representation of fSTm is presented, which is very useful for simulation and the studies the sample path properties. For some range of parameters, fSTm a) has Hölder continuous sample paths, b) is not a semimartingale, c) the sample paths of fSTm are nowhere bounded. Finally, the short and long time behaviour of fSTm is investigated. Under suitable normalization (i.e. short-time or long-time), the fSTm converges in finite-dimensional distributions to so-called fractional stable motion (short-time case), or to fractional Brownian motion (long-time case). Under some restrictions on \(H\) and on the measure \(\rho\) the weak convergence of measures in \(C([0,\infty),\mathbb{R})\) take place.