Initial-boundary value problems for the Vlasov-Poisson equations in a half-space (Q2446243)

The author of this interesting paper considers the Vlasov-Poisson system of equations having the form: \[ -\triangle\varphi (x,t)=4\pi e\int\limits_{\mathbb{R}^{3}} \sum\limits_{\beta = \pm 1}\beta f^{\beta } (x,p,t)dp, \;\;x\in \mathbb{R}_{+}^3, \;\;0<t<T, \tag{*} \] \[ \begin{multlined} \partial f^{\beta }/\partial t+m_{\beta }^{-1}(p,\nabla_{x}f^{\beta })+\beta e (-\nabla_{x}\varphi + m_{\beta }^{-1}c^{-1}[p,B(x)], \nabla_{p}f^{\beta }) = 0, \\ x\in \mathbb{R}_{+}^3, \quad p\in \mathbb{R}^3, \quad 0<t<T, \quad \beta =\pm 1,\end{multlined}\tag{**} \] where \(\varphi (x,t)\) is the potential of the self-consistent electric field; \(f^{\beta }(x,p,t)\) is the distribution function of the density of positively charged ions if \(\beta =+1\), and of electrons if \(\beta = - 1\) at the point \(x\) with the impulse \(p\) at the moment \(t\); \(\nabla_{x}\), and \(\nabla_{p}\) are the gradients with respect to \(x\) and \(p\), respectively; \(m_{+1}\) and \(m_{-1}\) are the masses of an ion and electron, respectively; \(e\) is the charge of an electron; \(c\) is the speed of light; \(B(x)\) is the induction of an external magnetic field; \((\cdot , \cdot )\) is the scalar product in \(\mathbb{R}^3\); \( [\cdot , \cdot]\) is the vector product in \(\mathbb{R}^3\); \(\mathbb{R}_{+}^3\) is the set of all \(x\) in \(\mathbb{R}^3\) with \(x_1>0\). Suppose that following conditions hold: \[ f^{\beta }(x,p,t)|_{t=0}=f_0^{\beta }(x,p), \;x\in\overline{\mathbb{R}_{+}^3}, \, \;p\in\mathbb{R}^3, \, \, \beta =\pm 1, \] where \(f_{0}^{\beta }\) are given nonnegative functions; moreover, \(\varphi (x,t)|_{x_1=0}=0\) (\(x^{\prime }= (x_2,x_3)\in\mathbb{R}^2\), \(0\leq t<T\)). It is known that the system under consideration describes the evolution of densities for ions and electrons in a rarefied plasma. The main result is the existence and uniqueness of classical solutions with compact support under initial and boundary conditions -- Dirichlet, Neumann, and nonlocal conditions. Assume that the following initial conditions hold: \(f^{\beta } (x, p, t)|_{t=0}=f_{0}^{\beta }(x,p)\), \(x \in\overline{\mathbb{R}_{+}^3} \), \(p \in \mathbb{R}^3 \), \(\beta = \pm 1\), where \(f_0^{\beta }\) are given nonnegative functions; the Dirichlet boundary condition has the form \(\varphi (x, t)|_{x_1=0}=0\), \(x'= (x_2 , x_3 ) \in \mathbb{R}^2 \), \(0 \leq t < T\). A vector-valued function \((\varphi ,f^{\beta })\) is called a classical solution of the problem \((\ast )\), \((\ast \ast )\) under above stated initial and boundary conditions with \(\varphi \in C([0,T], C^{2+\sigma }(\overline{\mathbb{R}_{+}^3}))\) and \(f^{\beta }\in C^1(\overline{\mathbb{R}_{+}^3} \times\mathbb{R}^3\times [0,T])\) if \(\int_{\mathbb{R}^3}\sum_{\beta }\beta f^{\beta } (\cdot ,p,t)dp \in C([0,T], C_{\Omega }^{\sigma}(\overline{\mathbb{R}_{+}^3}))\); here \(\Omega \) is a bounded domain in \(\mathbb{R}^3\) and \(\overline{\Omega }\subset \mathbb{R}_{+}^{3}\); \(\varphi (x,t)\to 0\) as \(|x|\to \infty \) for any \(t\in [0,T]\), and \(\{\varphi ,f^{\beta }\}\) satisfy the system under consideration with the same initial and boundary data. The author states that if the vector-valued function \(B\) is two times differentiable, and for any \(T>0\) and all \(f_0^{\beta }\in \dot{C}^2(\mathbb{R}^6)\) satisfying the condition, \(\text{supp}f_0^{\beta } \subset (R_{\delta }^3\cap{B}_{\kappa } )\times B_{\rho }\), where \( R_{\delta }^3 = \{x \in \mathbb{R}^3 : \;\;x_1 > \delta \}\) and \(\delta ,\kappa , \rho > 0\}\) (\(B_{\rho } \) is an open ball with radius \(\rho \)) hold, and the norm \(\|f_{0}^{\beta}\|\) (\(\beta =\pm 1\)) is bounded, then there exists a unique classical solution of the problem under consideration satisfying the above stated initial and boundary data, and \(\text{supp}f^{\beta }(\cdot ,\cdot ,t)\) lies in an open set (\(t\in [0,T]\)). A generalization of this problem is considered, and an example with Neumann boundary condition illustrates the theory.