- Research
- Open access
- Published:
Small deviations for admixture additive & multiplicative processes
Journal of Inequalities and Applications volume 2018, Article number: 204 (2018)
Abstract
Define the admixture additive processes
and the admixture multiplicative processes
where \(\mathrm{t}=(t_{1},t_{2},t_{3},t_{4})\in\mathbb{R}_{+}^{\mathrm{4}},a_{1},a_{2},a_{3},a_{4}\) are finite constants, \(B(t_{1})\) is the standard Brownian motion, \(W_{\gamma}(t_{2})\) is the fractional integrated Brownian motion with index parameter \(\gamma>-1/2\), \(B_{H}(t_{3})\) is the fractional Brownian motion with Hurst parameter \(H\in(0,1)\), \(S_{\alpha}(t_{4})\) is the stable process with index \(\alpha\in(0,2]\), and they are independent of each other. The small deviation for \(\mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t})\) and the lower bound of small deviation for \(\mathbb{Y}_{\gamma,H,\alpha}(\mathrm{t})\) are obtained. As an application, limit inf type LIL is given for \(\mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t})\).
1 Introduction and main results
Let \(B= \{B(t_{1})\in\mathbb{R},t_{1}\in\mathbb{R}_{+} \}\) be a standard Brownian motion, \(W_{\gamma}= \{W_{\gamma}(t_{2})\in \mathbb{R}, t_{2}\in\mathbb{R}_{+} \}\) be a fractional integrated Brownian motion with index parameter \(\gamma>-1/2\), \(B_{H}= \{B_{H}(t_{3})\in \mathbb{R},t_{3}\in\mathbb{R}_{+} \}\) be a fractional Brownian motion with Hurst parameter \(H\in(0,1)\), \(S_{\alpha}= \{S_{\alpha}(t_{4})\in\mathbb{R},t_{4}\in\mathbb{R}_{+} \}\) be a stable process with index \(\alpha\in(0,2]\), and they are independent of each other. Define the admixture additive processes
and the admixture multiplicative processes
where \(\mathrm{t}=(t_{1},t_{2},t_{3},t_{4})\in\mathbb{R}_{+}^{\mathrm{4}}\) and \(a_{1},a_{2},a_{3},a_{4}\) are finite constants.
Remark 1.1
In particular, \(W_{0}(t)=B_{1/2}(t)=S_{2}(t)\) is the standard Brownian motion. Quite obviously, the process \(\mathbb{X}^{1, 1, 1, 1}_{0,1/2,2}(\mathrm{t})\) is the additive Brownian motions. For more details on the additive Brownian motions, the reader can refer to the monographs [1] and [2].
The object of study in this paper will be the small deviations for \(\mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t})\) and \(\mathbb{Y}_{\gamma,H,\alpha}(\mathrm{t})\) which are formally defined as follows:
Our standard reference is the monograph [3]. There are a number of papers concerned with the small deviations for various stochastic processes. For details, we refer to the monographs [3, 4]. If properly defined, the small deviation in some literatures is called the small ball probability (see, e.g., [5, 6]). There are various motivations for the study of some additive processes, and it has been actively investigated recently from different points of view, see Khoshnevisan, Xiao, and Zhong [2, 7] for a detailed discussion and the bibliography for further works in this area. First of all, additive processes play an important role in the study of other more interesting multiparameter processes since they locally resemble multiparameter processes, such as Brownian sheet, fractional Brownian sheet, and stable sheet, and also because they are more amenable to analysis. For example, locally and with time suitably rescaled, the Brownian sheet closely resembles an additive Brownian motion (see, e.g., [8, 9]). They also arise in the theory of intersection and self-intersection local times of Brownian processes (see, e.g.,[10, 11]).
Now, we briefly give the small deviation estimates for the admixture additive processes \(\mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t})\) and the admixture multiplicative processes \(\mathbb{Y}_{H,\gamma,\alpha}(\mathrm{t})\). The main results are the following.
Theorem 1.2
and
where \(\beta=\max \{2,\frac{2}{2\gamma+1},\frac{1}{H} \}\), \(\vartheta (\gamma, H, \alpha, a_{1}, a_{2}, a_{3}, a_{4} )= ( (\operatorname{sgn}(2-\beta)+1 ) (\frac {(a_{1}\pi)^{2}}{8} )^{\frac{1}{1+\beta}}+ (\operatorname{sgn}(2/ (2\gamma+1)- \beta)+1 ) (a_{2}^{2/(2\gamma+1)}k_{\gamma})^{\frac{1}{1+\beta}} + (\operatorname{sgn}(1/H-\beta)+1 ) (|a_{3}|^{1/H}C_{H} )^{\frac{1}{1+\beta}} + (\operatorname{sgn}(\alpha-\beta)+1 ) (|a_{4}|^{\alpha}A_{\alpha})^{\frac{1}{1+\beta}} )^{1+\beta},\nu (\gamma, H, \alpha )= (\operatorname{sgn}(2-\beta)+1 )\frac{\pi^{2}}{8} + (\operatorname{sgn}(2/(2\gamma+1)-\beta)+1 )k_{\gamma}+ (\operatorname{sgn}(1/H-\beta)+1 )C_{H} + (\operatorname{sgn}(\alpha-\beta)+1 )A_{\alpha}\). In particular, \(\vartheta (0,1/2,2,1,1,1,1 )=8\pi^{2}, \nu (0, 1/2, 2 )=-\frac{\pi^{2}}{2}\).
Remark 1.3
Unfortunately, we just obtain the lower bound of the small deviation for the admixture multiplicative processes. Certainly, it is easy to get the following upper and lower bounds of the small deviation in the special case \(\gamma=0,H=1/2,\alpha=2\),
and
We observe that the admixture multiplicative process \(\mathbb{Y}_{0,1/2,2}(\mathrm{t})\) has the same covariance function with \((4,1)\)-Brownian sheet \(\{X(\mathrm{t})\in\mathbb{R},\mathrm{t}\in\mathbb {R}_{+}^{\mathrm{4}} \}\). Naturally, are there some similar properties between them? In fact, the small deviation for \((4,1)\)-Brownian sheet is much more difficult to discuss. The fact that in [4] Wenbo V. Li and Qi-Man Shao only obtained the upper and lower bounds of the small ball probability for \((4,1)\)-Brownian sheet,
where \(0< K_{1},K_{2}<\infty\).
By Lemma 2.1 in [5] and (1.1) in Theorem 1.2, we deduce the following corollary.
Corollary 1.4
where \(\vartheta (\gamma, H, \alpha, a_{1}, a_{2}, a_{3}, a_{4} )\) is denoted as ϑ for convenience.
Remark 1.5
The result of above Corollary 1.4 seems to be a special case of the so-called de Bruijin’s exponential Tauberian theorem by Bingham et al. For details, it establishes the relationship between the asymptotic behavior of Laplace transform and the small deviation for the admixture additive processes \(\mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t})\).
As an application of the general results for Theorem 1.2, the below theorem shows the Chung-type law of the iterated logarithm (LIL) for the admixture additive processes \(\mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t})\).
Theorem 1.6
Remark 1.7
Since we have the lower bound of small deviation for \(\mathbb{Y}_{\gamma,H,\alpha}(\mathrm{t})\) only, the limit inf type LIL for \(\mathbb{Y}_{\gamma,H,\alpha}(\mathrm{t})\) cannot be obtained. In fact, using an argument similar to that given in the proof of (1.3), we can obtain
Furthermore, we also can consider a generalization of the admixture additive processes
and the admixture multiplicative processes
Obviously, there are some results in common with those given in Theorems 1.2, 1.6, and we omit the details.
The remainder of the paper is arranged as follows. We present some preliminaries and the basic lemmas for establishing the small deviations of the admixture additive processes \(\mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t})\) and the admixture multiplicative processes \(\mathbb{Y}_{H,\gamma,\alpha}(\mathrm{t})\) in Sect. 2. The proofs of the main results are given in Sect. 3.
2 Preliminaries
Firstly, we briefly recall the processes which are the compositions of constructing the admixture additive processes \(\mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t})\) and the admixture multiplicative processes \(\mathbb{Y}_{\gamma,H,\alpha}(\mathrm{t})\).
The standard Brownian motion \(B= \{B(t)\in\mathbb{R},t\in \mathbb{R}_{+} \}\) with \(B(0)=0\) specifies
The fractional integrated Brownian motion \(W_{\gamma}= \{W_{\gamma}(t)\in\mathbb{R},t\in\mathbb{R}_{+} \}\) is defined by
where \(\Gamma(z)=\int_{0}^{\infty}x^{z-1}e^{-x}\,\mathrm{d}x\) is the gamma function and \(B(s)\) is a real-valued standard Brownian motion.
The fractional Brownian motion \(B_{H}= \{B_{H}(t)\in \mathbb{R},t\in\mathbb{R}_{+} \}\) with \(B_{H}(0)=0\) is a Gaussian process which has mean zero and the following covariance function:
The stable process \(S_{\alpha}= \{S_{\alpha}(t)\in\mathbb{R},t\in\mathbb{R}_{+} \}\) with index \(\alpha\in(0,2]\) has the characteristic function \(\mathbb{E} (e^{\mathrm{i}(u,S_{\alpha}(t))} )=e^{-t\psi(u)}\) whose exponent \(\psi(u)\) has the following form:
where \(\widetilde{W}_{\alpha}(u,\theta)\) satisfies
Specifically, if \(\psi(u)=\lambda|u|^{\alpha}\), the process \(S_{\alpha}(t)\) is called the symmetric stable process. For more details on the above processes, the reader can refer to the monographs [12, 13].
In order to establish the estimates of the small deviations conveniently, we present some lemmas at first. In fact, Lemmas 2.1–2.4 show the relation between the small deviation estimates of sup-norm and the range for the Brownian motion, the fractional integrated Brownian motion, the fractional Brownian motion, and the stable process, respectively.
Lemma 2.1
Let \(\{B(t)\in\mathbb{R},t\in\mathbb{R}_{+} \}\) be a standard Brownian motion, then for any given finite constant \(a_{1}\in\mathbb{R}\),
Then, for the range \(R_{1}=\sup_{s,t\in[0,1]} \vert B(t)-B(s) \vert \), we have
Lemma 2.2
Let \(\{W_{\gamma}(t)\in\mathbb{R},t\in\mathbb{R}_{+} \}\) be a fractional integrated Brownian motion, then for any given finite constant \(a_{2}\in\mathbb{R}\),
where \(k_{\gamma}\in(0,\infty)\) is given by
Then, for the range \(R_{2}=\sup_{s,t\in[0,1]} \vert W_{\gamma}(t)-W_{\gamma}(s) \vert \), we have
Lemma 2.3
Let \(\{B_{H}(t)\in \mathbb{R},t\in\mathbb{R}_{+} \}\) be a fractional Brownian motion, then for any given finite constant \(a_{3}\in\mathbb{R}\),
where \(C_{H}\in(0,\infty)\) is given by
Then, for the range \(R_{3}=\sup_{s,t\in[0,1]} \vert B_{H}(t)-B_{H}(s) \vert \), we have
Lemma 2.4
Let \(\{S_{\alpha}(t)\in\mathbb{R},t\in \mathbb{R}_{+} \}\) be a stable process, then for any given finite constant \(a_{4}\in\mathbb{R}\),
where \(A_{\alpha}>0\) is the principle Dirichlet eigenvalue for the fractional Laplacian operator associated with \(S_{\alpha}(t)\) in the interval \([-1,1]\). Then, for the range \(R_{4}=\sup_{s,t\in[0,1]} \vert S_{\alpha}(t)-S_{\alpha}(s) \vert \), we have
Remark 2.5
It seems to be little known about the explicit value of \(A_{\alpha}\) in Lemma 2.4, \(0<\alpha<2\), although sometimes this constant appears in some other problems. The best known bounds of \(A_{\alpha}\) for a symmetric stable process whose characteristic function exponent \(\psi(u)=K|u|^{\alpha}\) are
and it is a challenge to find more explicit expression for \(A_{\alpha}\) than the well-known variation one.
Remark 2.6
Lemmas 2.1–2.4 are proved easily by the well-known results in [3]. In the case of the Brownian motions, it is well known that \(k_{0}=C_{1/2}=A_{2}=\frac{ \pi^{2}}{8}\).
3 Proof of the main results
We give the proof of the main results in this section. An inspection of our arguments reveals that the special structures of admixture additive and multiplicative processes play a very important role in the following derivations. A key ingredient of our approach is with reference to Chen and Li [3].
Proof of (1.1) in Theorem 1.2
We follow similar approaches and steps in the proof of Theorem 5.2 as those in [3] and Lemma 2 in [5]. For any given finite constants \(a_{1},a_{2},a_{3},a_{4}\in\mathbb{R}\), using the triangle inequality, we have
Thus
Without loss of generality, we argue the following condition: \(\gamma+1/2< H\leq 1/2\leq 1/\alpha\), in this case \(\beta=2/(2\gamma+1)\).
On the one hand, using (3.1) for fixed \(\delta>0\) small enough, we have
where \(C(a_{2},k_{\gamma},\delta)=a_{2}^{2/(2\gamma+3)}k_{\gamma}^{(2\gamma +1)/(2\gamma+3)} +4\delta\).
Then, combining (2.1), (2.3), (2.5), and (2.7) with (3.2), we get
Taking \(\delta\rightarrow0\) in (3.3), we obtain that
On the other hand, we observe that
Hence
Combining (2.4) with (3.4), we obtain
The proof of results in other conditions for Theorem 1.2 (1.1) follows from a similar approach given in the above argument, and we omit the details here. □
Proof of (1.2) in Theorem 1.2
Fix \(0<\varepsilon<1\), we note that
Thus, combining (2.1), (2.3), (2.5), and (2.7)with (3.5), we obtain
where \(\nu (\gamma, H, \alpha )= (\operatorname{sgn}(2-\beta)+1 )\frac{\pi^{2}}{8} + (\operatorname{sgn}(2/(2\gamma+1)-\beta)+1 )k_{\gamma}+ (\operatorname{sgn}(1/H- \beta)+1 )C_{H} + (\operatorname{sgn}(\alpha-\beta)+1 )A_{\alpha}\). □
Proof of (1.3) in Theorem 1.6
As a matter of convenience, we take \(\vartheta (\gamma, H, \alpha, a_{1}, a_{2}, a_{3}, a_{4} )\) by abbreviated notation ϑ. Let \(T_{n}=\theta^{n^{2}},\theta>1,n=1,\ldots \) . For any \(\lambda<\vartheta^{1/\beta}\), using the scaling property and (1.1), we have
Then by the Borel–Cantelli lemma, we have
For any \(T_{n}\leq T\leq T_{n+1}\),
Therefore
Thus the lower bound is proved by \(\lambda\rightarrow\vartheta^{1/\beta}\) and \(\theta\rightarrow1\) in (3.6).
Inspiration for the approach of the proof for the upper bound comes from Kuelbs [14] and Talagrand [15]. Taking \(T_{k}=2^{k},k=1,\ldots \) . Let \(\lambda>\vartheta^{1/\beta}\) and fix \(\delta>0\). Then choose \(j\geq 1\) independent of k such that \(T_{k+j}\geq \delta^{-1}T_{k}\), and for every \(k=1,\ldots \) ,
Define the events
where \(\phi(T)= (T^{-1}\log\log T )^{1/\beta}\).
According to the definition of \(\mathrm{D}_{k}\) and (3.7), we have
where \(\mathrm{T_{k}}= (T_{k},T_{k},T_{k},T_{k} )\).
Observe that
and
Therefore, using (3.8)–(3.10) and the scaling property, by the Gaussian correlation inequality [16] with any \(0<\eta<1\), we obtain
where \(P_{k}((1-\eta^{2})^{1/2}\lambda)=\mathbb{P} (\phi(T_{k})\sup_{\mathrm{t}\in[0,T_{k}]^{4}} \vert \mathbb{X}^{a_{1}, a_{2}, a_{3}, a_{4}}_{\gamma,H,\alpha}(\mathrm{t}) \vert \leq (1-\eta ^{2})^{1/2}\lambda )\).
Hence, for every \(N=1,\ldots \) , if only \(T_{k}\geq N\), i.e., \(k\geq \log N\),
On the other hand, it is easy to see that the occurrence number of \(\{{D_{k};k\geq 1}\}\) is no more than j by (3.7) and the definition of \(D_{k}\), so there is
Therefore
Then, for any \((1-\eta^{2})^{1/2}\lambda>\vartheta^{1/\beta}\), by scaling
Accordingly, for every \(N=1,\ldots \) , we have
Hence
The upper bound follows from (3.11) by \(\delta\rightarrow 0\), \(\eta\rightarrow1\) and \(\lambda\rightarrow\vartheta^{1/\beta}\). □
4 Concluding remarks
We end this paper with the following comment: it has been found that the small deviation estimate has close connection with various approximation quantities of compact sets and operators and has a variety of applications in studies of Hausdorff dimensions, rate of convergence in Strassen’s law of the iterated logarithm, and empirical processes (see, e.g., [17, 18]). Intuitively, we expect the results which were obtained in this paper can enrich the above relative fields. Moreover, there are scant papers related to the admixture additive processes and the admixture multiplicative processes. The behaviors of the admixture additive processes and the admixture multiplicative processes deserve to be investigated extensively.
References
Khoshnevisan, D., Shi, Z.: Brownian sheet and capacity. Ann. Probab. 27(3), 1135–1159 (1999)
Khoshnevisan, D., Xiao, Y., Zhong, Y.: Local times of additive Lévy processes. Stoch. Process. Appl. 104, 193–216 (2003)
Chen, X., Li, W.V.: Small Deviation Estimates for Some Additive Processes. High Dimensional Probability III. Springer, Basel (2003)
Li, W.V., Shao, Q.M.: Gaussian processes: inequalities, small ball probabilities and applications. Handbook of Statist. 19, 533–597 (2001)
Li, W.V.: Small ball probabilities for Gaussian Markov processes under the \(L_{p}\)-norm. Stoch. Process. Appl. 92, 87–102 (2001)
Dereich, S., Fehringer, F., Matoussi, A., et al.: On the link between small ball probabilities and the quantization problem for Gaussian measures on Banach spaces. J. Theor. Probab. 16, 249–265 (2003)
Khoshnevisan, D., Xiao, Y., Zhong, Y.: Measuring the range of an additive Lévy process. Ann. Probab. 31, 1097–1141 (2003)
Dalang, R.C., Walsh, J.B.: Geography of the level sets of the Brownian sheet. Probab. Theory Relat. Fields 96, 153–176 (1993)
Dalang, R., Mountford, T.: Non-independence of excursions of the Brownian sheet and of additive Brownian motion. Trans. Am. Math. Soc. 355, 967–985 (2003)
Khoshnevisan, D., Xiao, Y.: Level sets of additive Lévy processes. Ann. Probab. 30, 62–100 (2002)
Chen, X., Li, W.V.: Large and moderate deviations for intersection local times. Probab. Theory Relat. Fields 128, 213–254 (2004)
Mörters, P., Peres, Y.: Brownian Motion. Cambridge University Press, Cambridge (2010)
Kyprianou, A.E.: Lévy Processes and Applications, Fluctuations of Lévy Processes with Applications. Springer, Berlin (2014)
Kuelbs, J.: When is the cluster set of \(S_{n}/a_{n}\) empty? Ann. Probab. 9, 377–394 (1981)
Talagrand, M.: Lower classes for fractional Brownian motion. J. Theor. Probab. 9, 191–213 (1996)
Li, W.V.: A Gaussian correlation inequality and its applications to small ball probabilities. Electron. Commun. Probab. 4, 111–118 (1999)
Theiler, J.: Estimating fractal dimension. J. Opt. Soc. Am. A, Opt. Image Sci. Vis. 7, 1055–1073 (1990)
Shorack, G.R., Wellner, J.A.: Empirical Processes with Applications to Statistics. Society for Industrial and Applied Mathematics, Philadelphia (1986)
Acknowledgements
We thank the referees for their time and comments.
Funding
This work was supported by the National Science Foundation of Fujian Province (No. 2017J01567), Science Foundation of the Education Department of Fujian Province (No. JA15476), and the Program for Fujian Normal University Innovation Team “Probability and Statistics: Theory and Application” (No. IRTL1704).
Author information
Authors and Affiliations
Contributions
All authors jointly worked on the results and they read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Liang, M., Wu, B. Small deviations for admixture additive & multiplicative processes. J Inequal Appl 2018, 204 (2018). https://doi.org/10.1186/s13660-018-1798-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13660-018-1798-4