Limit theorems for random difference equations driven by mixing processes (Q2367099)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Limit theorems for random difference equations driven by mixing processes |
scientific article |
Statements
Limit theorems for random difference equations driven by mixing processes (English)
0 references
5 January 1994
0 references
Limit theorems for a sequence of stochastic processes determined by random difference equations are studied. Let \(\{\xi^ n_ k; n\in N, k\in N^*\}\) \((N=\{1,2,\ldots\},\;N^*=\{0,1,2,\ldots\})\) be an array of \(\mathbb{R}^ e\)-valued random variables. We suppose that \(\{\xi^ n_ k; k\in N^*\}\) is a stationary mixing process for each \(n\in N\). Let \(\{F^ n(x),G^ n(x); n\in N\}\) be a sequence of functions on \(\mathbb{R}^ d\). Then, for each \(n\in N\), we determine an \(\mathbb{R}^ d\)-valued stochastic process \(\{\varphi^ n_ k; k\in N^*\}\) inductively by the following random difference equation: \[ \begin{cases} \varphi^ n_ 0=x_ 0\in\mathbb{R}^ d,\\ \varphi^ n_ k-\varphi^ n_{k-1}=F^ n(\varphi^ n_{k-1})(\xi^ n_ k-a^ n)+(1/n)G^ n (\varphi^ n_{k-1})\quad\text{for } k=1,2,\ldots, \end{cases} \tag{1} \] where we set \(a^ n=E[\xi^ n_ 0I_{\{|\xi^ n_ 0|\leq r\}}]\) for some positive constant \(\tau\) and \(I_ A\) denotes the indicator function of the set \(A\). Further, we define \(\varphi^ n(t)\) by \[ \varphi^ n(t)=\varphi^ n_{[nt]},\qquad t\in [0,\infty), \tag{2} \] where \([t]\) denotes the integer part of \(t\). Then, we show that when \(n\to\infty\), the sequence of stochastic processes \(\{\varphi^ n(t)\}_ n\) of (2) converges in the sense of laws on \(d\)-dimensional Skorokhod's space to a solution of the following stochastic differential equation of jump type: \[ \begin{multlined} \varphi_ t=x_ 0+\int_{(0,t]}F(\varphi_{u-})dB_ u+\int_{(0,t]}\{C(\varphi_{u-})+ G(\varphi_{u-})\}du \\ +\int_{(0,t]}\int_{\{| z|\leq\tau\}}F(\varphi_{u-})z\tilde N(dudz)+\int_{(0,t]}\int_{\{| z|>\tau\}}F(\varphi_{u- })zN(dudz), \end{multlined} \tag{3} \] where \(B_ .\) is an \(e\)-dimensional centered Brownian motion, \(N(dudz)\) is a Poisson random measure on \((0,\infty)\times\mathbb{R}^ e\), and \(\tilde N(dudz)\) denotes the compensated measure. Also, \(C(x)\) is a correction function arising from the dependence of \(\{\xi^ n_ k\}_ k\) and the derivatives of the function \(F(x)\).
0 references
limit theorem
0 references
weak convergence
0 references
stochastic differential equation
0 references
random difference equations
0 references
stationary mixing process
0 references
stochastic differential equation of jump type
0 references
Brownian motion
0 references