A Gladyshev theorem for trifractional Brownian motion and \(n\)-th order fractional Brownian motion (Q2064839)

The paper under review proves limit theorems for the weighted quadratic variation of tri-fractional Brownian motion and \(n\)-th order fractional Brownian motion. \textit{G. Baxter} [Proc. Am. Math. Soc. 7, 522--527 (1956; Zbl 0070.36304)] initiated the limiting theorems for the weighted quadratic variation of Gaussian processes, and \textit{E. G. Gladyshev} [Theory Probab. Appl. 6, 52--61 (1962; Zbl 0107.12601); translation from Teor. Veroyatn. Primen. 6, 57--61 (1961)] established a similar result under more general assumptions to the fractional Brownian motion. Gladyshev's theorem has been extended in various directions by many authors, but existing techniques do not apply to the tri-fractional Brownian motion (tri-fBm) and the \(n\)-th order fractional Brownian motion (\(n\)-fBm), where a tri-functional Bronwian motion \(Z_{H, K} = \{Z_{h, K}(t)\}_{t\ge 0}\) with parameters \(H\in (0, 1)\) and \(K\in (0, 1)\) is centered Gaussian process with covariance function \(C_{H, K}(s, t) = t^{2HK} + s^{2HK} - (t^{2H} + s^{2H})^K\) for \(t, s \in [0, \infty)\), and an \(n\)-th order fractional Brownian motion \(B_{H, n} = \{B_{H, n}(t)\}_{t\ge 0}\) is a centered Gaussian process with the covariance function \(G_{H, n}(s, t)\) in Page 2. Section 2 presents main results Theorem 2.1 (a) for the tri-fractional Brownian motion with \(HK < 1/2\) and (b) for the tri-fractional Brownian motion with \(HK \ge 1/2\) to have limiting behavior for the critical exponent \(\gamma = 2HK\) and \(\gamma =1\) respectively, and Theorem 2.2 for the \(n\)-th order fractional Brownian motion to have the limiting behavior for the critical exponent \(\gamma =1\) for \(m\ge 3\). Critical cases for the limiting behavior are given in Proposition 2.3. The proof of Proposition 2.3 follows from Theorem 2.1 of \textit{L. Viitasaari} [Probab. Surv. 16, 62--98 (2019; Zbl 1451.60039)] to reduce the existence of a double limit for a centered Gaussian processes, and the double sequence is uniformly bounded by Cauchy-Schwarz inequality and a condition for the Proposition 2.3 with monotone non-decreasing for one parameter. Theorem 2.4 gives convergence results at the critical cases for the tri-functional Brownian motion and the \(n\)-th order fractional Brownian motion, and Theorem 2.5 indicates the \(L^p\)-convergence of the \(m\)-th order fractional Brownian motion \(B_{H,m}\) for \(m\ge 2\), \(H\in (m-1, m), T\in (0, \infty)\). Section 3 lays out proofs of main results in Section 2 with technical lemmas. The key is to obtain proper upper bounds of \[ \phi_{j,k}^{(m,n)} = E\left[ (X(\frac{j}{2^m}) -X(\frac{j-1}{2^m})) (X(\frac{k}{2^m}) -X(\frac{k-1}{2^m})) \right], \] for \(1\le j \le 2^m\) and \(1\le k \le 2^n\). Lemma 3.1 first shows the upper bounds of \(\phi_{j,k}^{(m,n)}\) for the tri-functional Brownian motion which is suffice to reach the convergence in the main theorem with \(T=1\). The proof ot Theorem 2.1 follows from Isserlis' theorem \[ E(S_n^{\alpha}) = 2^{\alpha n} \sum_{k=1}^{2^n} \phi_{k, k}^{(n)}, \ \ Var (S_n^{\alpha}) = 2^{2\alpha n +1} \sum_{j, k=1}^{2^n} (\phi_{j, k}^{(n)})^2, \] and the Lei-Nualart process \(X_K(t)\) relation with \(Z_{H, K}(t)\) and case by case of Theorem 2.1. The proof of Theorem 2.2 follows from Taylor series expansion with \(z= 1/t\), Cauchy-Schwartz inequality and the similar method in the proof of Theorem 2.1. The proof of Theorem 2.4 follows by the same token and Proposition 2.3. Theorem 2.5 follows from a direct corollary of Proposition 2.3 and H-self-similar property.