On a class of distribution dependent stochastic differential equations driven by time-changed Brownian motions (Q6111021)

Let \(\mathcal P(\mathbb R^d)\) denote the space of Borel probability measures on \(\mathbb R^d\) equipped with the weak topology, let \(\mathcal P_\theta(\mathbb R^d)\) denote its subspace of measures with the finite \(\theta^{th}\) moment (equipped with the \(L^\theta\)-Wasserstein metric \(\mathbb W_\theta\)), let \(B\) be an \(m\)-dimensional Brownian motion, let \(D\) be an \(\alpha\)-stable adapted Lévy process with a Lévy stable index \(\alpha\in(1,2)\), \(D(0)=0\), with the Laplace transform \[ \mathbb Ee^{-\lambda D(t)}=e^{-t\phi(\lambda)},\qquad\lambda>0 \] for \[ \phi(\lambda)=\int_0^\infty(1-e^{-\lambda x})\,d\mu, \] where \(\mu\) is an infinite, \(\sigma\)-finite measure on \((0,\infty)\) such that \[ \int_0^\infty\min\{1,x\}\,d\mu<\infty \] and a continuous non-decreasing non-Markovian time change \[ E_t=\inf\,\{s>0:\,D(s)>t\} \] is defined. It is also assumed that \(B\) and \(D\) are independent. A stochastic differential equation \[dX_t=b(t,E_t,X_t,\operatorname{Law}X_t)\,dE_t+\sigma(t,E_t,X_t,\operatorname{Law}X_t)\,dB_{E_t}, \tag{1}\] where \(b:[0,T]\times[0,\infty)\times\mathbb R^d\times\mathcal P(\mathbb R^d)\to\mathbb R^d\) and \(\sigma:[0,T]\times[0,\infty)\times\mathbb R^d\times\mathcal P(\mathbb R^d)\to\mathbb R^d\otimes\mathbb R^m\) are measurable functions such that \[ b(t,E_t,0,\delta_0)=0,\qquad\sigma(t,E_t,0,\delta_0)=0, \] hence (1) has a trivial solution. Moreover it is assumed that \(b\) is bounded on bounded sets, \(b(t,E_t,\cdot,\cdot)\) is continuous on \(\mathbb R^d\times\mathcal P_2(\mathbb R^d)\) for every \(t\in[0,T]\), \[ |b(t_1,t_2,x,\mu)|^2+\|\sigma(t_1,t_2,x,\mu)\|^2\le K(1+|x|^2+\mathbb W_2^2(\mu,\delta_0)), \] \[ 2\langle x-y,b(t_1,t_2,x,\mu)-b(t_1,t_2,y,\nu)\rangle+\|\sigma(t_1,t_2,x,\mu)-\sigma(t_1,t_2,y,\nu)\|^2\le K(|x-y|^2+\mathbb W_2^2(\mu,\nu)) \] and \(b(t,E_t,X_t,\operatorname{Law}X_t)\) and \(\sigma(t,E_t,X_t,\operatorname{Law}X_t)\) are \(\mathcal F_{E_t}\)-adapted càglàd whenever \(X\) is \(\mathcal F_{E_t}\)-adapted càglàd. It is proved that if \(X_0\) is square integrable then there exists a unique solution \(X\) of (1) which belongs to \(L^2(\Omega;C([0,T];\mathbb R^d))\). Next, if an additional assumption \[ |b(t_1,t_2,x,\mu)-b(t_1,t_2,y,\nu)|^2+\|\sigma(t_1,t_2,x,\mu)-\sigma(t_1,t_2,y,\nu)\|^2\le K(|x-y|^2+\mathbb W_2^2(\mu,\nu)) \] is imposed, the authors establish an Itô formula for the solutions of (1) and, using suitable Lyapunov functionals, they provide sufficient conditions for stochastic stability, stability in probability, stochastic asymptotic stability and global stochastic asymptotic stability of the trivial solution. Finally, an averaging principle is proved for the equation (1).