Jumping SDEs: absolute continuity using monotonicity. (Q1766067)

Let \(G\subseteq \mathbb R\) be an open set and \(N\) a Poisson measure on \([0,T]\times G\times \mathbb R_ {+}\) whose intensity measure equals to \(\varphi (z)\, ds\, dz\, du\), \(\varphi \in C^ 1(G)\) strictly positive. Let us consider a stochastic differential equation \[ X_ {t} = x + \int ^ {t}_ {0}b(X_ {s-})\, ds + \int ^ {t}_ {0}\int _ {G}\int ^ \infty _ {0}\sigma (X_ {s-})\eta (z) \mathbf 1_ {\{u\leq \zeta (X_ {s-})\}}N( ds, dz, du), \tag{1} \] where \(b,\sigma ,\zeta :\mathbb R\to \mathbb R\) are locally Lipschitz continuous, \(\zeta \geq 0\) and the function \(| b| + | \sigma | \zeta \) has at most a linear growth. Suppose that \(\eta \in L^ 1(G,\varphi (z) \, dz)\). Then there exists a unique càdlàg adapted solution to (1). However, to prove that the law of \(X_ {t}\) admits a density by standard methods seems to be difficult since the diffusion coefficient contains an indicator function. Therefore, the author proposes an alternative argument based on some monotonicity considerations. Namely, let \(\sigma \), \(\zeta \) be increasing and strictly positive on \(\mathbb R\), let \(\eta \in C^ 2(G)\) be such that \(\eta \geq 0\) and \(\eta ^ {\prime \prime }\) is bounded, and let a non-degeneracy condition \(\int _ {G} \mathbf 1_ {\{\eta ^ \prime (z) \neq 0\}}\varphi (z)\, dz = \infty \) hold. Then the law of \(X_ {t}\) is absolutely continuous with respect to Lebesgue measure for every \(t\in \mathopen ] 0,T\mathclose ]\).