Stability of coupled jump diffusions and applications (Q6140098)

In contrast to the existing works on stability of jump-diffusion processes (see, for example, \textit{I.-S. Wee}, Stochastic Processes Appl. 80, No. 2, 193--209 (1999; Zbl 0962.60046), \textit{R. Atar} and \textit{A. Budhiraja}, Electron. J. Probab. 7, Paper No. 22, 31 p. (2002; Zbl 1011.60061), \textit{D. Filipović} et al., Stochastics 82, No. 4--6, 475--520 (2010; Zbl 1230.60066), \textit{Z. Chao} et al., SIAM J. Control Optim. 55, No. 6, 3458--3488 (2017; Zbl 1374.60151), \textit{X. Chen} et al., Bernoulli 25, No. 2, 1045--1075 (2019; Zbl 1440.60075), \textit{X. Chen} et al., Appl. Math. Optim. 80, No. 2, 415--445 (2019; Zbl 1439.60072), \textit{G. Yin} and \textit{F. Xi}, SIAM J. Control Optim. 48, No. 7, 4525--4549 (2010; Zbl 1210.60089), \textit{X. Zong} et al., SIAM J. Control Optim. 52, No. 4, 2595--2622 (2014; Zbl 1390.34230)), this work focuses on stability and stabilization of a fully coupled system of jump diffusions where two jump-diffusion components interact with each other. The authors investigate the coupled jump-diffusion processes \(\mathbf{X}_1(t)\) and \( \mathbf{X}_2(t)\) in \({\mathbb{R}^{\ell_1}}\), \({\mathbb{R}^{\ell_2}}\), respectively, where \({\mathbb{R}^{\ell_1}}\) and \({\mathbb{R}^{\ell_2}}\) are two Euclidean spaces of dimensions \(\ell_1>0\) and \(\ell_2>0\). Specifically, the pair \((\mathbf{X}_1(t), \mathbf{X}_2(t))\) is the solution of the system of equations \[ \begin{cases} \displaystyle d\mathbf{X}_1(t)=b_1(\mathbf{X}_1(t), \mathbf{X}_2(t))dt +\sigma_1(\mathbf{X}_1(t), \mathbf{X}_2(t))d\mathbf{W}_1(t)+ \int_{{{{\mathbb{R}}}^{n_1}_*}} \gamma_1(\mathbf{X}_1(t-),\mathbf{X}_2(t-), {\boldsymbol{\phi}}) {{\widetilde{\mathbf{N_1}}}}(dt,d{\boldsymbol{\phi}}), \\ \displaystyle d\mathbf{X}_2(t)=b_2(\mathbf{X}_1(t), \mathbf{X}_2(t))dt+ \sigma_2(\mathbf{X}_1(t), \mathbf{X}_2(t))d\mathbf{W}_2(t)+ \int_{{{{\mathbb{R}}}^{n_2}_*}} \gamma_2(\mathbf{X}_1(t-),\mathbf{X}_2(t-),{\boldsymbol{\phi}}) {{\widetilde{\mathbf{N_2}}}}(dt,d{\boldsymbol{\phi}}),\\ \displaystyle \mathbf{X}_1(0)={{\mathbf{x}}}_1,\quad \mathbf{X}_2(0)={{\mathbf{x}}}_2, \end{cases} \] where, for \(i=1,2\), \({\mathbf{W}}_i(t)\) are standard \({\mathbb{R}}^{d_i}\)-valued Brownian motions; \(\mathbf{N}_i(dt, d{\boldsymbol{\phi}})\), \(i=1,2\), are corresponding to a random point process \(p(t)\) Poisson random measures independent of \({\mathbf{W}}_1(t)\), \({\mathbf{W}}_2(t)\), and \({{\widetilde{\mathbf{N_i}}}}(dt, d{\boldsymbol{\phi}}) = \mathbf{N}_i(dt, d{\boldsymbol{\phi}}) - {\boldsymbol{\nu_i}}(d{\boldsymbol{\phi}})dt\) are the compensated Poisson random measures on \([0, \infty) \times \mathbb R_*^{n_i}\) with \(\mathbb R_*^{n_i}:=\mathbb R^{n_i}\setminus\{\mathbf 0\}\); \(b_1:{\mathbb{R}^{\ell_1}}\times{\mathbb{R}^{\ell_2}}\to{\mathbb{R}^{\ell_1}}\) and \(b_2:{\mathbb{R}^{\ell_1}}\times{\mathbb{R}^{\ell_2}}\to{\mathbb{R}^{\ell_2}}\) are smooth functions; \(\sigma_1: {\mathbb{R}_1^{\ell_1}} \times {\mathbb{R}_2^{\ell_2}} \to {\mathbb{R}}^{\ell_1 \times d_1}\) and \(\sigma_2: {\mathbb{R}_1^{\ell_1}} \times {\mathbb{R}_2^{\ell_2}} \to {\mathbb{R}}^{\ell_2 \times d_2}\); \(\gamma_1:{\mathbb{R}^{\ell_1}}\times{\mathbb{R}^{\ell_2}} \times{{\mathbb{R}}}_*^{n_1}\to {{\mathbf{X}}}\), \(\gamma_2: {\mathbb{R}^{\ell_1}}\times{\mathbb{R}^{\ell_2}} \times{{\mathbb{R}}}_*^{n_2}\to{{\mathbf{Y}}}\) are measurable functions. It is assumed that \[ b_2({{\mathbf{x}}}_1,{{\mathbf{0}}})={{\mathbf{0}}},\ \sigma_2({{\mathbf{x}}}_1,{{\mathbf{0}}})={{\mathbf{0}}},\ \gamma_2({{\mathbf{x}}}_1,{{\mathbf{0}}},{\boldsymbol{\phi}})={{\mathbf{0}}}, \] where \({\mathbf{0}}\) denotes a zero vector with appropriate dimension. In this case \({{\mathbf{0}}}\) is an equilibrium point of \({{\mathbf{X}}}_2(t)\). The authors derived sufficient conditions under which the equilibrium point \({{\mathbf{0}}}\) or the trivial solution of \({{\mathbf{X}}}_2(t)\) is stable. Using the derived results, they investigate the stability of linearizable jump diffusions and fast-slow coupled jump diffusions. An algorithm for weak stabilization of weakly stabilizing a coupled system in which only one component can be controlled is proposed.