On the reflected Ornstein–Uhlenbeck process with catastrophes
Introduction
Reflected Ornstein–Uhlenbeck (OU) process is widely used in various applied fields, such as neuroscience, mathematical biology, finance and queueing theory. Specifically, the unrestricted OU process is the most frequently proposed stochastic model for the single neuronal activity (see, for instance, [8], [18], [21], [24], [36], [38], [46]). In such context, the membrane potential evolution can be described by focusing the attention on the OU process confined by a lower reflecting boundary that can be interpreted as the neuronal reversal hyperpolarization potential (cf. [23], [39]). The OU process with reflection at the origin also arises as an approximating process for population dynamics (cf. [1], [37]) and for queueing systems (see, for instance, [47], [48]). Furthermore, the diffusion limits of a number of classical birth and death processes asymptotically approach the OU process (cf. [22], [27], [33], [37]). In particular, in [22] the reflected OU process arises as a heavy-traffic approximation for queueing systems characterized by linear arrival and service rates. Recently, the reflected OU process has been applied to the so-called regulated financial market (see, for instance, [4], [5], [6]). In various types of instances, first-passage time densities are invoked to describe events such as firing times in neuronal modeling, extinction in population dynamics and busy period in queueing systems (cf. for instance, [2], [3], [9], [17], [34], [40], [44]). References to other applications in economics, in finance and in queueing systems can be found in [29].
During the last three decades, great attention has been paid to the description of biological, physical and engineering systems subject to various types of catastrophes, a catastrophe being defined as a random event resulting in the extinction of all individuals or customers in the system. The usual framework is that the system evolves according to the dynamics of some continuous-time Markov chain and it is influenced by catastrophes that occur at exponential rate (cf. [7], [10], [13], [14], [31], [32], [35], [41], [42], [43]). These works are concerned with various quantities, such as the transient and the stationary probabilities, the time of extinction and the first occurrence time of effective catastrophe. The results obtained for continuous-time Markov chains have suggested the possibility of deriving corresponding results for diffusion processes bounded by one reflected boundary and subject to catastrophes that occur at exponential rate ξ. Indeed, in [12], [13], [15], [16] some general results for the transient and steady-state probability density functions (pdf’s) of diffusion processes in the presence of catastrophes have been obtained. In particular, in [12] analytical and computational results for transient and steady-state pdf’s for the Wiener process in the presence of catastrophes have been determined. Instead, in [13] a heavy-traffic approximation for the queue in the presence of catastrophes is given that is seen to be equivalent to a Wiener process subject to randomly occurring jumps. Furthermore, in [15] time-non-homogeneous processes in the presence of catastrophes that occur with a time-varying intensity function are considered. Instead, in [16] a system performing a continuous-time random walk on the integers, subject to catastrophes occurring at constant rate and followed by exponentially-distributed repair times, is studied. A heavy-traffic approximation of this system is a Wiener process with jumps. Jump-diffusion processes as models for neuronal activity are also considered in [24], [25], [28].
The aim of the present paper is to provide some quantitative informations on the reflected OU process subject to catastrophes that occur at exponential rate reducing the state of the system instantaneously to zero. The paper is organized as follows. In Section 2 we obtain various relations between functions characterizing the process with catastrophes and those of the same process in the absence of catastrophes. In Sections 3 Reflected OU process with catastrophes, 4 First-visit time for OU process with catastrophes we focus on the OU process restricted to by a reflecting boundary in zero state under the assumption that random catastrophes occur with exponential rate. In particular, we determine the transition and steady-state pdf’s and their moments (Section 3) and we analyze the problem of the first visit time to zero state (Section 4). Extensive numerical computations with MATHEMATICA have been performed to show the role played by the involved parameters. Moreover, in Section 5 an application of the OU process with catastrophes is considered. Finally, in Appendix A some classical results concerning the reflected OU process are given.
Section snippets
Diffusion process with catastrophes
Let be a regular one-dimensional time-homogeneous diffusion process with drift and infinitesimal variance restricted to the interval by a reflecting boundary in zero state (cf. [11]). For this process we denote with the transition pdf, with . We construct a new stochastic process defined in I as follows. Starting from the state at the initial time , the process evolves according to the process until
Reflected OU process with catastrophes
Let be an OU process restricted to the interval by a reflecting boundary in zero state, characterized by drift and infinitesimal varianceThe boundary is nonattracting-natural if and attracting-natural if , whereas x = 0 is a regular boundary (see, for instance, [30] for the classification of the boundaries). Some classical results on the reflected OU process are given in Appendix A.
Furthermore, let be the reflected OU process, defined in
First-visit time for OU process with catastrophes
In this Section we focus our attention on the random variable that describes the first-visit time of from to the zero state. Proposition 4.1 The transition pdf of in the presence of an absorbing boundary at 0 is:where is given in (A.2). Furthermore, the FVT pdf of is:with given in (A.8). Proof Eq. (31) can be easily obtained by virtue of (4), (A.7). Furthermore, making
A heavy-traffic approximation to a queueing system
Let be a birth–death process with catastrophes, describing a queueing system such that the transitions occur according to the following scheme: (i) with rate ; (ii) with rate ; (iii) with rate ; (iv) with rate . The effect of each catastrophe is to make the queue instantly empty. For all and the transition probabilities satisfy the following system of forward equations:
Acknowledgments
This work has been performed under partial support by MIUR (PRIN 2008). We acknowledge the constructive criticism of an anonymous reviewer on an earlier version of this paper.
References (48)
- et al.
A note on birth-death processes with catastrophes
Stat. Prob. Lett.
(2008) A result on the first-passage time of an Ornstein–Uhlenbeck process
Stat. Prob. Lett.
(2007)- et al.
Jump-diffusion processes as models for neuronal activity
BioSystems
(1997) - et al.
Effects of random jumps on a very simple neuronal diffusion model
BioSystems
(2002) Optimal control of a simple immigration–emigration process through total catastrophes
Eur. J. Oper. Res. (Stochastic and Statistics)
(2004)- et al.
Survival models based on the Ornstein–Uhlenbeck process
Lifetime Data Anal.
(2004) On a reflected Ornstein–Uhlenbeck process with an application
Bull. Aust. Math. Soc.
(1991)- et al.
On the first passage times of reflected O–U processes with two-sided barriers
Queueing Syst.
(2006) - L. Bo, G. Ren, Y. Wang, X. Yang, Modeling the exchange rates in a target zone by a reflected diffusion, 2009,...
- et al.
On the conditional default probability in a regulated market: a structural approach
Quant. Finance
(2011)
Some integral functionals of reflected SDEs and their applications in finance
Quant. Finance
The extinction time of a birth, death and catastrophe process and of a related diffusion model
Adv. Appl. Prob.
On a stochastic leaky integrate-and-fire neuronal model
Neural Comput.
The first passage time problem for Gauss-diffusion processes: algorithmic approaches and applications to LIF neuronal model
Methodol. Comput. Appl. Probab.
Transient analysis of immigration birth-death processes with total catastrophes
Prob. Eng. Inform. Sci.
The Theory of Stochastic Processes
Diffusion Processes Subject to Catastrophes
On the queue with catastrophes and its continuous approximation
Queueing Syst.
On time non-homogeneous stochastic processes with catastrophes
A double-ended queue with catastrophes and repairs and a jump-diffusion approximation
Methodol. Comput. Appl. Probab.
Estimation of the input parameters in the Ornstein–Uhlenbeck neuronal model
Phys. Rev. E
Spiking Neuron Models: Single Neurons Populations Plasticity
Cited by (16)
Generating random variates from PDF of Gauss–Markov processes with a reflecting boundary
2018, Computational Statistics and Data AnalysisCitation Excerpt :Generating random variates from a specific distribution playsa fundamental role in mathematical models of phenomena in many fields of science, engineering, finance and economics (see, for instance, Glasserman, 2004; Iacus, 2008; Kroese et al., 2011). In many contexts, there is a need to sample from densities for reflected stochastic processes, as in population dynamics with immigration (cf., Renshaw, 2011; Ricciardi et al., 1999), in queueing systems (see, Di Crescenzo et al., 2003; Pender, 2015; Ward and Glynn, 2005), in financial applications (cf., Han et al., 2016; Linetsky, 2005), in neuronal modeling (cf., Buonocore et al., 2015; D’Onofrio and Pirozzi, 2016; Inoue and Doi, 2007; Lánský and Ditlevsen, 2008) and in more general applied fields (see, Abundo, 2014; Di Crescenzo et al., 2016; Giorno et al., 2011; Giorno et al., 2012; Headrick and Mugdadi, 2006; Ricciardi and Sacerdote, 1987; Wonho, 2009). The problem of creating efficient algorithms for sampling from densities with reflection can be non-trivial, even if the density of the process without reflection is of a standard type; various techniques, such as the inverse transform method and the acceptance–rejection method, may be required (see, for instance, Devroye, 1986; Ross, 2013).
Skew Ornstein-Uhlenbeck processes and their financial applications
2015, Journal of Computational and Applied MathematicsExact solutions and asymptotic behaviors for the reflected Wiener, Ornstein-Uhlenbeck and Feller diffusion processes
2023, Mathematical Biosciences and EngineeringModeling COVID-19 pandemic with financial markets models: The case of Jaén (Spain)
2023, Mathematical Biosciences and EngineeringContinuous-time multi-type Ehrenfest model and related Ornstein–Uhlenbeck diffusion on a star graph
2022, Mathematical Methods in the Applied Sciences