Abstract
The probabilistic symbol is defined as the right-hand side derivative at time zero of the characteristic functions corresponding to the one-dimensional marginals of a time-homogeneous stochastic process. As described in various contributions to this topic, the symbol contains crucial information concerning the process. When leaving time-homogeneity behind, a modification of the symbol by inserting a time component is needed. In the present article, we show the existence of such a time-dependent symbol for non-homogeneous Itô processes. Moreover, for this class of processes, we derive maximal inequalities which we apply to generalize the Blumenthal–Getoor indices to the non-homogeneous case. These are utilized to derive several properties regarding the paths of the process, including the asymptotic behavior of the sample paths, the existence of exponential moments and the finiteness of p-variationa. In contrast to many situations where non-homogeneous Markov processes are involved, the space-time process cannot be utilized when considering maximal inequalities.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
The probabilistic symbol \(p(x,\xi )\) of a Markov process X is the function \(p:\mathbb {R}^d\times \mathbb {R}^d\rightarrow \mathbb {C}\) given by
if the limit exists and coincides for every \(R>0\), where
This symbol proves to be a crucial concept for deriving a wide range of properties of the stochastic process, such as conservativeness (cf. [16], Theorem 5.5), asymptotic behavior (cf. [20], Theorems 3.11 and 3.12), strong \(\gamma \)-variation (cf. [19] Corollary 5.10), Hausdorff-dimension (cf. [17], Theorem 4) and Hölder conditions [18]. For a survey on recent results, we refer to [4, 14]. By now, all of these results are restricted to the time-homogeneous case.
Proving some of these results, the symbol is utilized to derive maximal inequalities. Inequalities of this kind have been proved for Lévy processes (cf. [12]), certain Feller processes (cf. [18]) and homogeneous diffusions with jumps (cf. [20]). One can find a throughout discussion of maximal inequalities for various classes of stochastic processes in [8]. Formulating these maximal inequalities, Blumenthal–Getoor indices are used (cf. [18, 20]) which allow for a governance of the process’s paths by the behavior of the symbol in the variable \(\xi \), and, therefore, for the derivation of the properties stated above.
However, when leaving the time-homogeneity behind one does not expect the symbol, being the right-hand side derivative of the characteristic functions corresponding to the one-dimensional marginal at time zero, to yield any information regarding the entire process. To overcome this, a time component is added to the symbol or more precisely
Moreover, the existence of such a time-dependent probabilistic symbol was shown for rich càdlàg Feller evolution processes, i.e., non-homogeneous càdlàg Markov processes such that
for \(0\leqslant \tau \leqslant t\) and \(u \in B_b(\mathbb {R}^d)\) forms a strongly continuous evolution system. In addition, the domain of the infinitesimal generator \(A_\tau \)
given by
contains the test functions \(C_c(\mathbb {R}^d)\). Theorem 4.5 of [13] shows that the generator \(A_\tau |_{C_c^\infty }\) is a pseudo-differential operator with symbol \(-q(\tau ,x,\xi )\), i.e.,
for \(\tau \geqslant 0, x,\xi \in \mathbb {R}^d\) and \(\hat{f}\) is the Fourier transform of f. Moreover, it is shown that the symbol of the generator \(q(\tau ,x,\xi )\) and the time-dependent probabilistic symbol defined in (2) \(p(\tau ,x,\xi )\) coincide if the symbol is continuous in x.
In the present article, we prove the existence of the time-dependent symbol for non-homogeneous Itô processes (cf. Definition 2.2). We utilize this result to prove maximal inequalities, the existence of non-homogeneous generalizations of the Blumenthal–Getoor indices and an exemplary selection of properties of such processes. Before we do so, we fix some notations:
A family of \(\sigma \)-fields \((\mathcal {G}_t^\tau )_{0 \leqslant \tau \leqslant t}\) is called a two-parameter filtration if \(\mathcal {G}_s^\tau \subset \mathcal {G}_t^\tau \) for all \(0 \leqslant \tau \leqslant s \leqslant t\) and \(\mathcal {G}_t^{\tau _2} \subset \mathcal {G}_t^{\tau _1}\) for \(0\leqslant \tau _1 \leqslant \tau _2 \leqslant t\). The natural double filtration of X is denoted by \((\mathcal {F}^X)_t^\tau )_{0\leqslant \tau \leqslant t}\) and is defined as
Let \((\Omega ,\mathcal {M})\) be a measurable space equipped with the two-parameter filtration \( (\mathcal {M}_t^\tau )_{0\leqslant \tau \leqslant t}\). We call a stochastic process X adapted to the two-parameter filtration if for all \(0 \leqslant \tau \leqslant t\)
We tacitly assume that every stochastic process \(X:=(X_t)_{t \geqslant 0}\) is defined on a generic stochastic basis \((\Omega , \mathcal {A}, (\mathcal {A}_t)_{t \geqslant 0},\mathbb {P})\) takes values in \((\mathbb {R}^d, \mathcal {B}(\mathbb {R}^d))\) and is cádlág. Here, \(\mathcal {B}(\mathbb {R}^d)\) is the \(\sigma \)-field of Lebesgue sets. Moreover, we call \(\Delta X_t:= X_t - \lim _{s \uparrow t} X_s\) the jump of the process at time \(t\geqslant 0\), and for a stopping time \(\tau \) we call \(X^\tau := X1_{\llbracket 0, \tau \rrbracket }+ X_\tau 1_{\llbracket \tau , \infty \llbracket }\) the stopped process. The stochastic interval \(\llbracket \tau , \sigma \llbracket \) for two stopping times \(\tau , \sigma \) is defined by \(\{ (\omega , t) \in \Omega \times \mathbb {R}_+: \tau (\omega ) \leqslant t < \sigma (\omega )\}\). The stochastic intervals \(\llbracket \tau , \sigma \rrbracket \), \(\rrbracket \tau , \sigma \llbracket \), \(\rrbracket \tau , \sigma \rrbracket \) are defined alike. Additionally, we define for \(\omega \in \Omega \)
to be the integer-valued random measure on \(\mathbb {R}_+ \times \mathbb {R}^d\) associated with the jumps of the process X.
A (strong) Markov process \((\Omega , \mathcal {M}, ( \mathcal {M}_t^\tau )_{0\leqslant \tau \leqslant t}, (X_t)_{t \geqslant 0}, \mathbb {P}^{\tau ,x})_{\tau \in \mathbb {R}_+, x \in \mathbb {R}^d}\) satisfies
for all \(\tau \leqslant s \leqslant t\) and all bounded Borel-measurable functions f. Moreover, every Markov process is normal, i.e., \(\mathbb {P}^{\tau ,x}(X_\tau =x)=1\). For more information on Markov processes see [5, 24]. We associate an evolution system \((T_{\tau ,t})_{0\leqslant \tau \leqslant t}\) of operators on \(B_b(\mathbb {R}^d)\) with every Markov process by setting
2 The Time-Dependent Probabilistic Symbol
In this section, \(X:=(\Omega , \mathcal {M}, ( \mathcal {M}_t^\tau )_{0\leqslant \tau \leqslant t}, (X_t)_{t \geqslant 0}, \mathbb {P}^{\tau ,x})_{\tau \in \mathbb {R}_+, x \in \mathbb {R}^d}\) denotes a Markov process. Before we start with the main topic of this section, we properly define the space-time process of a Markov process. That is due to the fact that various different definitions are used in the literature. The following definitions follows [3].
Let \(\hat{\Omega }:= \mathbb {R}_+ \times \Omega \) and the \(\sigma \)-field \(\hat{\mathcal {M}}:= \{ B \subset \hat{\Omega }: B_s \in \mathcal {M}\ \forall s \in \mathbb {R}_+\}\) where \(B_s\) denotes the s-slice of B for \(s \geqslant 0\). We define a process \(\hat{X}\) with values in \({ \hat{E}:=}\mathbb {R}_+ \times \mathbb {R}^d\) by
Moreover, we set
and
where \(\pi _0:\Omega \rightarrow \hat{\Omega }; \omega \mapsto (0, \omega )\), and \(B \in \hat{\mathcal {M}}\).
We call the homogeneous Markov process
the space-time process associated with X. The transition probability function of X is given by
and for any \(\hat{\mathcal {M}}\)-measurable random variable Y it holds true that
i.e., for \(\mathbb {P}^{(\tau ,x)}\)-almost all \((c,\omega ) \in \hat{\Omega }\) we have
That is due to
This will be used frequently throughout the following calculations.
Definition 2.1
We call a Markov process X non-homogeneous Markov semimartingale if for every \(\mathbb {P}^{\tau ,x}\), \(\tau \geqslant 0, x \in \mathbb {R}^d\) the process \((X_t)_{t \geqslant 0}\) is a semimartingale on \([\tau , \infty )\).
Definition 2.2
We call a non-homogeneous Markov semimartingale non-homogeneous Itô process if its characteristics \((B,C,\nu )\) are of the form
for \(t \geqslant \tau \geqslant 0, x \in { \mathbb {R}^d}\) and \(i,j \in \{1,...,d\}\).
Definition 2.3
Let X be a Markov process and let
be the first exit time from the ball of radius \(R>0\) after \(\tau \geqslant 0\), and \(\Vert \cdot \Vert \) the maximum norm. The function \(p:\mathbb {R}_+\times \mathbb {R}^d \times \mathbb {R}^d \rightarrow \mathbb {C}\) defined by
is called the time-dependent probabilistic symbol of the process, if the limit exists for every \(\tau \geqslant 0\) and \(x,\xi \in \mathbb {R}^d\) independently of the choice of R.
Example 2.4
-
1.
Let \((X_t)_{t \geqslant 0}\) be an additive process on \((\Omega , \mathcal {F}, (\mathcal {F}_t)_{t\geqslant 0},\mathbb {P})\) in the sense of Definition 1.6 of [15], i.e., X has independent increments, is stochastically continuous and càdlàg and starts in 0 at time \(t=0\). We define a family of probability measures on \((\Omega , \mathcal {F})\) by
$$\begin{aligned} \mathbb {P}^{\tau ,x}(X_t \in B):= \mathbb {P}( X_t -X_\tau \in B-x) \end{aligned}$$for \(B \in { \mathcal {B}(\mathbb {R}^d)}\) and \(\tau \in \mathbb {R}_+, x \in { \mathbb {R}^d}\). For this family, it holds true that \(\mathbb {P}^{\tau ,x}(X_\tau =x)=1\), and \((\Omega , \mathcal {F}, (\mathcal {F}_t)_{t\geqslant 0},(X_t)_{t \geqslant 0} ,\mathbb {P}^{\tau ,x})_{\tau \geqslant 0, x \in { \mathbb {R}^d}}\) is a Markov process. Let, in addition, X be a semimartingale for all \(\mathbb {P}^{\tau ,x}\) which is quasi-left-continuous, and possesses the characteristics \((B,C,\nu )\). Theorem II.4.15 of [7] provides the existence of a version of \((B,C,\nu )\) that is deterministic. Hence, in the following, we assume \((B,C,\nu )\) to be deterministic. By Corollary II.4.18 of [7], X has no fixed times of discontinuity, i.e.,
$$\begin{aligned} \{t\geqslant 0: \nu (\{t\} \times { \mathbb {R}^d})>0\} = \emptyset . \end{aligned}$$Therefore, when calculating the time-dependent probabilistic symbol of X, we derive with Theorem II.4.15 of [7] that
$$\begin{aligned}&p(\tau ,x,\xi )\\&\quad = -\lim _{h \downarrow 0} \mathbb {E}^{\tau ,x} \left( \frac{e^{i(X_{\tau +h}-x)'\xi }-1}{h} \right) \\&\quad = -\lim _{h \downarrow 0} \mathbb {E}\left( \frac{e^{i(X_{\tau +h}-X_\tau )'\xi }-1}{h} \right) \\&\quad = -\lim _{h \downarrow 0} \frac{e^{ i \xi (B_{\tau +h} -B_\tau ) - \frac{1}{2}\xi ' ( C_{\tau +h}-C_\tau )\xi + \int _{\mathbb {R}^d \setminus \{0\}}\left( e^{i\xi 'y}-1-i\xi 'y\chi (y)\right) \ \nu ((\tau ,\tau +h],\textrm{d}y) }-1}{h}. \end{aligned}$$This limit exists if and only if
$$\begin{aligned} -\lim _{h \downarrow 0} \frac{ i \xi ' (B_{\tau +h} -B_\tau ) - \frac{1}{2}\xi ' ( C_{\tau +h}-C_\tau )\xi + \int _{\mathbb {R}^d \setminus \{0\}} \left( e^{i\xi 'y}-1-i\xi 'y\chi (y) \right) \ \nu ((\tau ,\tau +h],\textrm{d}y)}{h} \end{aligned}$$exists. If \(B^{(i)}\) and \(C^{(ij)}\) are right-differentiable for all \(i,j \in \{1,...,d\}\) and if the function
$$\begin{aligned} \tau \mapsto \int _{\mathbb {R}^d \setminus \{0\}} \left( e^{i\xi 'y}-1-i\xi 'y\chi (y)\right) \ \nu ((0,\tau ],\textrm{d}y) \end{aligned}$$is right-differentiable the time-dependent symbol exists and is of the form
$$\begin{aligned} p(\tau ,x,\xi )=&\, -i \xi \partial _+ B_{\tau } + \frac{1}{2}\xi ' \partial _+ C_{\tau } \xi - \partial _+\\&\int _{0}^\tau \int _{\mathbb {R}^d \setminus \{0\}}\left( e^{i\xi 'y}-1-i\xi 'y\chi (y)\right) \ \nu (\textrm{d}s,\textrm{d}y), \end{aligned}$$where \(\partial _+\) denotes the right-derivative.
-
2.
Let \((X_t)_{t \geqslant 0}\) be a one-dimensional Brownian motion with variance function \(\sigma ^2(t)\) on \((\Omega , \mathcal {F}, (\mathcal {F}_t)_{t\geqslant 0},\mathbb {P})\). The process X is additive and a continuous semimartingale and
$$\begin{aligned} \mathbb {E}\left( e^{i(X_{t+\tau }-X_\tau )'\xi } \right) = \exp \left( \frac{1}{2}(\sigma ^2(t+\tau )-\sigma ^2(\tau ))\xi ^2 \right) . \end{aligned}$$By the previous example, the (non-homogeneous) probabilistic symbol exists if and only if the variance function is right-differentiable with right-derivative \(\partial _+ \sigma ^2\). In this case, we have
$$\begin{aligned} p(\tau ,x,\xi ) = \frac{1}{2}\xi ^2 \partial _+ \sigma ^2(\tau ). \end{aligned}$$
We have seen in (2) that under some mild conditions for a rich càdlàg Feller evolution process the symbol of the generator \(q(\tau ,x,\xi )\) and the time-dependent probabilistic symbol \(p(\tau ,x,\xi )\) coincide. Additionally, Corollary 3.5 of [3] states that the symbol of the generator of the homogeneous space-time process \(\hat{X}\) corresponding to X is given by
with \(\tau \geqslant 0, x \in \mathbb {R}^d, \xi \in \mathbb {R}^d\), \(\hat{x}=(\tau ,x),\hat{\xi }=(\xi _0,\xi ) \in \mathbb {R}^{d+1}\). Therefore, we expect the space-time process to be useful when calculating the symbol of a non-homogeneous Itô process, provided the characteristics of the space-time process are the ones of a homogeneous diffusion with jumps.
Hence, the following lemma states the characteristics of the space-time process associated to a non-homogeneous Markov semimartingale. Note that the proof is omitted since it is quite straightforward but needs tedious calculations.
Lemma 2.5
Let X be a non-homogeneous Markov semimartingale with characteristics \((B,C,\nu )\). In this case, the space-time process \(\hat{X}\) associated with X is semimartingale for all \(\mathbb {P}^{(\tau ,x)}, (\tau ,x)\in \mathbb {R}_+ \times \mathbb {R}^d\) and its characteristics \((\hat{B},\hat{C},\hat{\nu })\) are given by
for \(\tau \leqslant t\) and \((c,\omega ) \in \hat{\Omega }\).
Before we state the main theorem of this section, the following lemma provides that non-homogeneous Itô process are indeed a generalization of rich Feller evolution processes.
Lemma 2.6
Let \((X_t)_{t\geqslant 0}\) be a rich Feller evolution process on \(C_\infty (\mathbb {R}^d)\) with generator \(A_s\) and time-dependent symbol \(p: \mathbb {R}_+ \times \mathbb {R}^d \times \mathbb {R}^d \rightarrow \mathbb {C}\) such that \(p(\cdot , x,\xi )\) is continuous for all \(x, \xi \in \mathbb {R}^d\). Then, X is a non-homogeneous Itô process.
Proof
Let \((X_t)_{t\geqslant 0}\) be a rich Feller evolution system on \(C_\infty (\mathbb {R}^d)\). Analogously to the proof of Theorem 3.1 of [23], but under usage of the non-homogeneous version of Dynkin’s formula as mentioned in [13], we show that \((X_t)_{t\geqslant 0}\) is a semimartingale with characteristics \((B,C,\nu )\). Since it is well-known that X is a Markov process, it suffices to show that the characteristics are of the form mentioned in Definition 2.2:
By Theorem 3.2 and Lemma 3.7 of [3] the space-time process \(\hat{X}\) associated to X is a rich Feller process on \(C_\infty (\mathbb {R}_+ \times \mathbb {R}^d)\). Hence, Theorem 3.10 of [23] provides that the characteristics \((\hat{B},\hat{C},\hat{\nu })\) of \(\hat{X}\) are of the form
and, therefore, we conclude with equations (9) to (11) that
for \(t \geqslant \tau \), where all equalities are meant \(\mathbb {P}^{(\tau ,x)}\text {-a.s.}\). Transition from \( \mathbb {P}^{(\tau ,x)}\) to \( \mathbb {P}^{\tau ,x}\) yields the statement. \(\square \)
Theorem 2.7
Let X be a non-homogeneous Itô process and let \(\ell = (\ell ^{(j)})_{1 \leqslant j \leqslant d}\) and \(Q= (q^{(ik)})_{1 \leqslant j,k\leqslant d}\) be continuous, N be such that the function
is continuous. In this case, the time-dependent symbol exists and equals
Proof
Let X be a non-homogeneous Itô process, and let \(\hat{X}\) be the associated space-time process. The characteristics of \(\hat{X}\) are given by
for \(i,j \in \{2,...,d+1\}\). All of the equations above are meant \(\hat{\mathbb {P}}^{(\tau ,x)} \text {-a.s.}\)
For \(T \in \mathcal {B}(\mathbb {R}_+)\) and \(B \in \mathcal {B}(\hat{E})\) we have
where \(\hat{N}((c,\omega ),B)=N_c(\omega ,B_0)\) is a transition kernel from \(\hat{\Omega }\times \hat{E}\) to \(\mathbb {R}^d\). Consequently, \(\hat{X}\) is a Itô process and we denote by \(\hat{p}(\hat{x},\hat{\xi })\) its (homogeneous) probabilistic symbol.
For the stopping time \(\hat{\sigma }:= \inf \{ h \geqslant 0: \Vert \hat{X}_h -(\tau ,x) \Vert > R \}\) it holds true that
We compute for \(\hat{\xi }=(0,\xi )\), \(\hat{\ell }=(1,\ell ), \hat{c}=(\hat{c}^{(i,j)}{ )_{i,j \in \{1,...,d+1\}}}\) with \(\hat{c}^{(1,j)}=\hat{c}^{(i,1)}=0\) and \(\hat{c}^{(i,j)}=c^{((i-1),(j-1))}\) for \(i,j \in \{2,...,d+1\}\):
\(\square \)
We have observed, for instance in Example 2.4 that a process does not necessary need to be a non-homogeneous Itô process to possess a time-dependent symbol. However, the processes considered up to this point are quasi-left continuous. We will see in the subsequent example that, when leaving quasi-left continuity behind, the time-dependent symbol does not contain the same information on the process as before. This is not unexpected, as we have encountered similar situations in the homogeneous case.
Example 2.8
Let us consider the following (deterministic!) process:
where \(\tau \geqslant 0\) and \(x \in \mathbb {R}.\) Here, we write \(X^{\tau ,x}\) for the process \((X, \mathbb {P}^{\tau ,x})\) (cf. [21, 22]). The following figure depicts the path of \(X^{0,0}\).
![figure a](http://media.springernature.com/lw685/springer-static/image/art%3A10.1007%2Fs10959-023-01308-y/MediaObjects/10959_2023_1308_Figa_HTML.png)
The process X is a non-homogeneous Markov process, and considering the cut-off function \(\chi (x)=1_{\{|x| \leqslant 0.5\}}\) we receive a non-homogeneous Markov semimartingale with characteristics \(B \equiv 0, C \equiv 0\) and \(\nu (\textrm{d}t,\textrm{d}x)= \delta _1(\textrm{d}t)\delta _1(\textrm{d}x)\). In this case, the time-dependent symbol exists and is given by
Nevertheless, the symbol does not provide any information regarding the jumps of the process.
3 Maximal Inequalities and Time-Dependent Blumenthal–Getoor Indices
As we have pointed out before, for homogeneous processes like multivariate \(\alpha \)-stable processes [1], more general Lévy processes [2], Feller processes satisfying some mild conditions [18] and homogeneous diffusion with jumps [20], there exists a set of indices, called Blumenthal–Getoor indices, which utilize the symbol to derive maximal inequalities. For a historical overview we refer to [14]. Equally, we now want to consider maximal inequalities for non-homogeneous processes with the help of the time-dependent symbol. In this framework, the non-homogeneous growth and sector condition of the symbol play an important role:
for every \(s\geqslant 0, x,\xi \in \mathbb {R}^d\) and \(c,c_0 >0\).
Specifically, we want to use the maximal inequalities to examine the paths of the process, including the asymptotic behavior of the sample paths, the p-variation of the paths and the existence of the exponential moments of the process. The following indices generalize the Blumenthal–Getoor indices as defined in Definitions 4.2 and 4.5 of [18] or Definition 3.8 of [20] to the non-homogeneous case.
Definition 3.1
The quantities
are called time-dependent indices of X in the starting point, where
with \(\kappa =(4 \arctan (1/2 c_0))^{-1}\) where \(c_0\) comes from the sector condition (IS).
Definition 3.2
Let \(\tau \in \mathbb {R}_+, x\in \mathbb {R}^d\) and \(R>0\). The quantities
are the time-dependent indices of X at infinity, where
with \(\kappa =(4 \arctan (1/2 c_0))^{-1}\) where \(c_0\) comes from the sector condition (IS).
Example 3.3
-
1.
Let \((X_t)_{t \geqslant 0}\) be a one-dimensional Brownian motion with right-differentiable variance function \(\sigma ^2(t)\) such that \(\partial _+ \sigma ^2(t)\) is bounded (cf. Definition I.4.9 of [7]). In this case, the time-dependent symbol is given by
$$\begin{aligned} p(\tau ,x,\xi ) = \frac{1}{2}\xi ^2 \partial _+ \sigma ^2(\tau ). \end{aligned}$$Additionally, we have
$$\begin{aligned} H(R)&= \frac{1}{2}\frac{1}{R^2} \sup _{s \geqslant 0} \left| \partial _+ \sigma ^2(s) \right| ,\\ H(s,x,R)&= \frac{1}{2}\frac{1}{R^2}\left| \partial _+ \sigma ^2(s) \right| ,\\ h(R)&= \frac{1}{2}\frac{1}{4^2 \kappa ^2 R^2} \inf _{s \geqslant 0} (\partial _+ \sigma ^2(s)), \\ h(\tau ,x,R)&= \frac{1}{2}\frac{1}{4^2 \kappa ^2 R^2} (\partial _+ \sigma ^2(s)), \end{aligned}$$and, therefore,
$$\begin{aligned} \beta _0 =\underline{\beta _0}= \beta _\infty ^{\tau ,x}= \underline{\beta _\infty ^{\tau ,x}}= 2 =\delta _0=\overline{\delta _0} = \delta ^{\tau ,x}_\infty = \overline{\delta ^{\tau ,x}_\infty }. \end{aligned}$$ -
2.
Let \((X_t)_{t \geqslant 0}\) the unique solution of the SDE
$$\begin{aligned} dX_t = \mu (X_t,t)\ \textrm{d}t + \sigma (X_t,t) \ dB_t, \quad t \geqslant 0, \end{aligned}$$where \(\mu ,\sigma \in C^{1,1}_b(\mathbb {R}\times \mathbb {R}_+)\) non-decreasing, \(\sigma >0\) and \((B_t)_{t \geqslant 0}\) is a standard Brownian motion. By Lemma 4.6 of [13], the time-dependent symbol of X is given by
$$\begin{aligned} p(\tau ,x,\xi )=-i\mu (x,\tau )\xi + \frac{1}{2}\sigma (x,\tau )^2\xi ^2. \end{aligned}$$Hence,
$$\begin{aligned} H(R)&= \frac{1}{R}\sup _{s \geqslant 0} \sup _{x \in \mathbb {R}} \left| \frac{1}{2R} \sigma (x,s)^2+i\mu (x,s) \right| \\ h(R)&=\frac{1}{R^2}\inf _{s \geqslant 0} \inf _{x \in \mathbb {R}} \frac{1}{2}\sigma (x,s)^2\\ H(s,x,R)&= \frac{1}{R} \left| \frac{1}{2R}\sigma (x+2R,s)^2+i \mu (x+2R,s) \right| \\ h(s,x,R)&= \frac{1}{R^2} \frac{1}{2}\sigma (x+2R,s)^2, \end{aligned}$$and, therefore,
$$\begin{aligned} \beta _0&= \underline{\beta _0}=1,\\ \overline{\delta _0}&=\delta _0= 2,\\ \beta ^{\tau ,x}_\infty&= \underline{\beta ^{\tau ,x}_\infty }= \left( 2-\sup \left\{ \lambda> 0 : \lim _{R\rightarrow 0} \frac{\sigma (x+2R,\tau )^2}{R^{\lambda }} =0 \right\} \right) \wedge \\&\quad \hspace{1.1cm} \left( 1-\sup \left\{ \lambda> 0 : \lim _{R\rightarrow 0} \frac{\mu (x+2R,\tau ) =0}{R^{\lambda }} \right\} \right) , \\ \overline{\delta ^{\tau ,x}_\infty }&= \delta ^{\tau ,x}_\infty = 2- \sup \left\{ \lambda > 0 : \lim _{R\rightarrow 0} R^{-\lambda }\sigma (x+2R,\tau )^2 =0 \right\} . \end{aligned}$$
The proofs of the previous section crucially rely on the space-time process to transfer properties of homogeneous Markov processes to the non-homogeneous framework. However, when deriving properties like the asymptotic behavior of sample paths or maximal inequalities we do not expect the space-time process to be of much use. This is due to the fact that if we utilize the space-time process, a deterministic drift with slope 1 is added. This obscures the path-behavior of the original process. Additionally, the indices of the space-time process do, in general, not allow for a calculation of the time-dependent indices of the underlying process. Considering the symbol of the space-time process
with \(\tau \geqslant 0, x \in \mathbb {R}^d, \xi \in \mathbb {R}^d\), \(\hat{x}=(\tau ,x),\hat{\xi }=(\xi _0,\xi ) \in \mathbb {R}^{d+1}\) and the definition of the indices, one easily sees that when calculating the (homogeneous) indices of the space-time process a linear term (in R) is added to the functions H resp. h along with a supremum in time for the indices at infinity of the space-time process.
In this section, we assume all characteristics to be encountered to be with respect to the truncation function \(id\cdot \chi \), where \(\chi \in C_c^\infty (\mathbb {R}^d)\) is a symmetric cut-off function with
where \(R >0\).
Theorem 3.4
Let X be a non-homogeneous Itô process with characteristics as in Theorem 2.7. In this case, we have
for \(t\geqslant 0\), \(R>0\) and a constant \(c_d>0\) which only depends on the dimension d. If, in addition, (IS) holds true, we have
where \(c_k>0\) only depends on \(c_0\) of the sector condition (IS).
Remark 3.5
Throughout the following proof, we often make use of Lemma 5.2 of [20] although it is a statement for time-homogeneous symbols. A closer look shows that an analogous statement holds for the time-dependent probabilistic symbol and the proof also works analogously.
Proof
The proof of (16) closely follows the proof of Proposition 3.10 of [20]. We omit the proof of (17) since it is generalized from the proof of Lemma 6.3 of [18] analogously to the following generalization. However, let us mention that one has to use Dynkin’s formula for non-homogeneous processes as stated in [13] when Corollary 3.6 is utilized in [18].
In order to prove (16), let X be a non-homogeneous Itô process such that the differential characteristics \((\ell ,Q,\nu )\) of X are locally bounded and continuous. At first, we show that for S, R and \(\sigma := \inf \{ t \geqslant \tau : \Vert X_t -x \Vert > S \}\) as above we have
where \(c_d=4d+16\widetilde{c}_d\) and \(t \geqslant \tau \). We introduce the stopping time
as the first time the jumps of \(X^\sigma \) exceed R, and estimate the following
We deal with the terms on the right-hand side one after another, starting with the first one.
Again we separate the first term (19) in order to get control over the big jumps. For \(t \geqslant \tau \) let
The process \(\check{X}\) is a special semimartingale on \([\tau , \infty )\) with characteristics
Now let \(u=(u_1,...,u_d)':\mathbb {R}^d \rightarrow \mathbb {R}^d\) be such that \(u_j \in C_b^2(\mathbb {R}^d)\) is 1-Lipschitz continuous, \(u_j\) depends only on \(x^{(j)}\) and is zero in zero for \(j=1,...,d\). We define the auxiliary process for \(t \geqslant \tau \):
where
\(\check{M}\) is a \(\mathbb {P}^{\tau ,x}\)-local martingale on \([\tau ,\infty )\) by [7] Theorem II.2.42. Applying Lemma 3.7 of [23] we have under (IG)
since \(u_j\in C_b^2(\mathbb {R}^d)\). Let us mention, that although Lemma 3.7 of [23] considers homogeneous diffusion with jumps only the proof is alike for non-homogeneous Itô processes. In particular, since \(\check{M}\) is uniformly bounded it is an \(L^2\)-martingale on \([\tau ,t]\). We define
and obtain for \(t \geqslant \tau \):
For the left summand of (21), we estimate for \(t \geqslant \tau \):
where we used Doob’s inequality for the martingale \(\check{M}^\sigma \) and the Lipschitz property of u in combination with Corollary II.3 of [11]. Since
we obtain
where we have used Lemma 5.2 of [20] on the second term. By choosing a sequence \((u_n)_{n\in \mathbb {N}}\) of functions of the type described above which tends to the identity in a monotonous way we obtain
Now let us consider the term \(\mathbb {P}^{\tau ,x}{( D^c)}\) of (21). The Markov inequality provides
Again we chose a sequence \((u_n)_{n\in \mathbb {N}}\) of functions as we described in (20), but this time it is important that the first and second derivatives are uniformly bounded. Since the \(u_n\) converge to the identity, the first partial derivatives tend to 1 and the second partial derivatives to 0. In the limit (\(n\rightarrow \infty \)), we obtain
For term (23), we get
and for term (24)
where we have used again Lemma 5.2 of [20] on the second term, and the inequalities
It remains to deal with the second term of (19). Let \(\delta >0\) be fixed (at first) and \(m:\mathbb {R}\rightarrow ]1,1+\delta [\) a strictly monotone increasing auxiliary function. Since \(m\geqslant 1\) and since we have at least one jump of size \(>R\) on \(\{\tau _R\leqslant t\}\) we obtain
because \(m(\left\| z \right\| ) 1_{[ [0 , \sigma [ [}(s) 1_{B_R(0)^c} (z)\) is in class \(F_p^1\) of Ikeda and Watanabe (see [6], Section II.3). Since \(\delta \) can be chosen arbitrarily small we obtain by Lemma 5.2 of [20]
Plugging together (22), (25), (26) and (27) we obtain (18). For the particular case \(\sigma ={\sigma _{3\widetilde{R}}^{\tau ,x}}\), we have
and therefore, for every \(\widetilde{R}>0\)
Setting \(\widetilde{R}:=(1/2)R\) we obtain (16). \(\square \)
The maximal inequalities (16) and (17) allow for statements of the asymptotic behavior of the sample paths: First, we state a result concerning the behavior near the starting point \(x \in \mathbb {R}^d\) at time \(\tau \geqslant 0\) with respect to the measure \(\mathbb {P}^{\tau ,x}\). The second statement, treats the same behavior at infinity. The proof of both statements is inspired by Theorem 4.3 and Theorem 4.6 of [18] but takes the time-component \(\tau \) into account. Let us mention, that for a function \(f:\mathbb {R}_+ \rightarrow \mathbb {R}_+\) we denote by \(f(t^+):= \limsup _{s \downarrow t} f(s)\) and \(f(t_+):= \liminf _{s \downarrow t} f(s)\).
Theorem 3.6
Let X be a non-homogeneous Itô process such that the differential characteristics of X are locally bounded and continuous. Then, we have
If the symbol \(p(\tau ,x,\xi )\) of the process X satisfies (IS), then we have in addition
All these limits are meant \(\mathbb {P}^{\tau ,x}\)-a.s with respect to every \(\tau \in \mathbb {R}_+\) and \(x\in \mathbb {R}^d\).
Proof
Here, we only prove (29) and (31) and omit the proofs of (30) and (32) since (29) and (32) and (30) and (31) are very similar, respectively.
Let \(\varepsilon >0\), \(\tau \in \mathbb {R}_+\) and \(x\in \mathbb {R}^d\). We start with proving (29):
Let \(\lambda > \sup _{\tau < s \leqslant \tau + \varepsilon } \beta _\infty ^{s,x}\) and choose \(\sup _{\tau< s \leqslant \tau + \varepsilon } \beta _\infty ^{s,x}< \alpha _2< \alpha _1 <\lambda \). For \(t<T_0^\varepsilon \) with \(T_0^\varepsilon \) sufficiently small, for (16) one obtains:
Now let \(t_k:=(1/2)^k\) for \(k\in \mathbb {N}\). Since
where \(k_0^\varepsilon \) depends on \(T_0^\varepsilon \), the Borel-Cantelli Lemma derives
and, hence, \(\sup _{\tau \leqslant s \leqslant \tau +t_k} \Vert X_s-x \Vert < (t_k)^{1/\alpha _1}\) for all \(k\geqslant k_1^\varepsilon (\omega )\) on a set of probability one. For fixed \(\omega \) in this set and \(t \in [t_{k+1}, t_k]\) and \(k\geqslant k^\varepsilon _1(\omega ) \geqslant k^\varepsilon _0\), we have
and since \(\lambda >\alpha _1\)
which converges \(\mathbb {P}^{\tau ,x}\)-a.s to zero as \(t\downarrow 0\). Since \(\lambda >\sup _{\tau < s \leqslant \tau + \varepsilon } \beta _\infty ^{s,x}\) and \(\varepsilon >0\) arbitrary, \(\varepsilon \downarrow 0\) provides the statement.
In order to prove (31), we derive the following:
Let \( \inf _{\tau < s \leqslant \tau +\varepsilon } \overline{\delta _\infty ^{s,x}}> \lambda ' > \lambda \). Moreover, let \((t_k)_{k \in \mathbb {N}}\) be a sequence such that
Hence, the maximal inequality (17) provides for k sufficiently large
Fatou’s lemma implies
Hence,
and, therefore,
Since \(\lambda < \lambda '\) we observe, that
Since \(\varepsilon >0\) arbitrary, the statement follows. \(\square \)
The proof of the following theorem parallels Lemma 3.6, and hence, we omit details of the proof.
Theorem 3.7
Let X be a non-homogeneous Itô process such that the differential characteristics of X are locally bounded and continuous. Then, we have
If the symbol \(p(\tau ,x,\xi )\) of the process X satisfies (IS) then we have in addition
All these limits are meant \(\mathbb {P}^{\tau ,x}\)-a.s with respect to every \(\tau \in \mathbb {R}_+\) and \(x\in \mathbb {R}^d\).
The time homogeneous version of the following result is Lemma 5.11 of [4].
Lemma 3.8
Let X be a non-homogeneous Itô process such that the differential characteristics of X are continuous, and let (IG) hold true. Then
for all \(t \geqslant \tau \geqslant 0\) and \(x,\xi \in \mathbb {R}^d\).
Proof
At first, let us reconsider the stopping time \(\sigma \) as defined in (7):
and let \(p(\tau ,x,\xi )\) be the symbol of X given by
In order to apply Gronwall’s Lemma, we estimate the following:
(IG) provides (see Lemma 3.3 of [23]) the finiteness of the constant
Application of Gronwall’s Lemma provides
and with Fatou’s Lemma follows
where the first inequality follows by Proposition 3.4. \(\square \)
Finally, we generalize Theorem 2.10 of [10] which is a nice and applicable criterion for the finiteness of p-variation.
Theorem 3.9
Let X be a non-homogeneous Itô process such that the differential characteristics of X are continuous. For every \(t \geqslant \tau \geqslant 0\) and \(p > \sup _{\tau \geqslant 0} \beta _\infty ^{\tau ,x}\) the following holds true
where the supremum is taken over all finite partitions \(\pi _n= (t_i)_{i=1,...,n}\) with \(\tau =t_0< t_1< \cdots <t_n=t\). I.e., the p-variation of the paths of X is \(\mathbb {P}^{\tau ,x}\)-a.s finite for \(p > \sup _{\tau \geqslant 0} \ \beta _\infty ^{\tau ,x}\).
Proof
Let \(t, r >0\) and \(\lambda >p\). Then, Theorem 3.4 provides the following estimation
for r small enough and \(K>0\). Hence, Theorem 1.3 of [9] provides the statement. \(\square \)
Data Availability
All data generated or analyzed during this study are included in this published article.
References
Blumenthal, R.M., Getoor, R.K.: Some theorems on stable processes. Trans. Am. Math. Soc. 95, 263–273 (1960)
Blumenthal, R.M., Getoor, R.K.: Sample functions of stochastic processes with stationary independent increments. J. Math Mech. 10, 493–516 (1961)
Böttcher, B.: Feller evolution systems: generators and approximation. Stoch Dyn 4, 239–241 (2018)
Böttcher, B., Schilling, R., Wang, J.: Lévy Matters III. Lévy-Type Processes: Construction, Approximation and Sample Path Properties. Lecture Notes in Mathematics, vol. 2099. Springer, Berlin (2013)
Gulisashvili, A., van Casteren, J.A.: Non-autonomous Kato Classes and Feynman–Kac Propagators. World Scientific, Singapore (2006)
Ikeda, N., Watanabe, S.: Stochastic Differential Equations and Diffusion Processes. North-Holland Publishing Company, Tokio (1981)
Jacod, J., Shiryaev, A.: Limit Theorems for Stochastic Processes. Springer, Berlin (1987)
Kühn, F., Schilling, R.: Maximal inequalities and applications. Probab. Surv. 20, 382–485 (2023)
Manstavičius, M.: p-variation of strong Markov processes. Ann. Probab. 32, 2053–2066 (2004)
Manstavičius, M., Schnurr, A.: Criteria for the finiteness of the strong p-variation for Lévy-type processes. Stoch. Anal. Appl. 35(5), 873–899 (2017)
Protter, P.: Stochastic Integration and Differential Equations. Version 2.1. Springer, Berlin (2005)
Pruitt, W.E.: The Hausdorff dimension of the range of a process with stationary independent increments. Indiana J. Math. 19, 371–378 (1969)
Rüschendorf, L., Schnurr, A., Wolf, V.: Comparison of time-inhomogeneous Markov processes. Adv. Appl. Probab. 48, 1015–1044 (2016)
Schnurr, A., Rickelhoff, S.: From Markov processes to semimartingales. Probab. Surv. 20, 568–607 (2023)
Sato, K.: Lévy Processes and Infinitely Divisible Distributions. Cambridge Studies in Advanced Mathematics. Cambridge (1999)
Schilling, R.L.: Conservativeness and extensions of Feller semigroups. Positivity 2, 239–256 (1998)
Schilling, R.L.: Feller processes generated by pseudo-differential operators: on the Hausdorff dimension of their sample paths. J. Theor. Probab. 11, 303–330 (1998)
Schilling, R.L.: Growth and Hölder conditions for the sample paths of Feller processes. Probab. Theory Relat. Fields 112, 565–611 (1998)
Schilling, R., Schnurr, A.: The symbol associated with the solution of a stochastic differential equation. Electr. J. Probab. 15, 1369–1393 (2010)
Schnurr, A.: Generalization of the Blumenthal–Getoor Index to the class of homogeneous diffusions with jumps and some applications. Bernoulli 19(5A), 2010–2032 (2013)
Schnurr, A.: On deterministic Markov processes: expandability and related topics. Markov Process. Relat. Fields 19, 693–720 (2013)
Schnurr, A.: A classification of deterministic hunt processes with some applications. Markov Process. Relat. Fields 17(2), 259–276 (2011)
Schnurr, A.: The Symbol of a Markov Semimartingale. PhD thesis, TU Dresden, 2009. https://nbn-resolving.org/urn:nbn:de:bsz:14-ds-1244626491491-70401
van Casteren, J.A.: Markov Processes, Feller Semigroups and Evolution Equations. World Scientific, Singapore (2010)
Funding
Open Access funding enabled and organized by Projekt DEAL. The research has been supported by the DFG (German Science Foundation, Grant No. SCHN 1231/2-1).
Author information
Authors and Affiliations
Contributions
SR and AS wrote the main manuscript text and SR prepared the figure. All authors reviewed the manuscript.
Corresponding author
Ethics declarations
Conflict of interest
The authors declare 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 licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Rickelhoff, S., Schnurr, A. The Time-Dependent Symbol of a Non-homogeneous Itô Process and Corresponding Maximal Inequalities. J Theor Probab (2023). https://doi.org/10.1007/s10959-023-01308-y
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s10959-023-01308-y
Keywords
- Non-homogeneous Markov process
- Semimartingale
- Time-dependent symbol
- Generalized indices
- Path properties
- Diffusion with jumps