Fractional Poisson process. II (Q813596): Difference between revisions

Let \(H\in(1/2,1) \). For \(t>0\) the process \[ W_{H}(t) =\frac{1}{(H-1/2)} \int_{0}^{t} u^{1/2-H} \biggl(\int_{u}^{t}\tau ^{H- 1/2}(\tau-u)^{H-3/2}\,d\tau \biggr)\, dq(u) \] is called a fractional Poisson process, where \(q(u) =N(u)/\sqrt{\lambda }-\sqrt{\lambda }u\), and \(N(u)\) is a homogeneous Poisson process with the intensity \(\lambda >0,\) \(N(0) =0\) a.s. It is self-similar in a wide sense, has the wide-sense stationary increments, exhibits long-range dependence and has continuous paths. Its correlation function is \[ E(W_{H}(t) W_{H}(s) ) =\frac{1}{2}\frac{(2-2H) \cos \pi H}{(2H-1) \pi H}(| s| ^{2H}+| t| ^{2H}-| t-s| ^{2H}) \] and \(\text{Dim}(\operatorname{graph} W_{H})=2-H\) with probability one. The Vigner-Vile spectrum of the process \(W_{H}(t) \) is presented here. Denote \(\sigma _{H}(t) =\sqrt{\operatorname{Var}W_{H}(t) }.\) It is proved that \(W_H(t)/\sigma_H(t)\) converges to a normal \(N(0,1) \) limit as \(t\to +\infty \) or \(\lambda \to +\infty \). The skewness and excess kurtosis of \(W_{H}(t) \) are discussed as well. The process \(W_{H}(t) \) is different from that of the fractional Poisson process defined by Jumarie through the Liouville-Rieman fractional derivative. The list of references contains 24 positions. In the introduction the authors recall the similar properties of the fractional Brownian process and explain their interest in the process \( W_{H}(t) \) as caused by quantum physics and finance.