A new perspective on Wasserstein distances for kinetic problems (Q2123130)

The author introduces a new class of Wasserstein-type distances and proves boundedness properties associated to such distances in the context of Vlasov equation or Vlasov-Poisson system. If \(\mathcal{X}\) is either the \(d\)-dimensional torus \(\mathbb{T}^{d}\) or the Euclidean space \(\mathbb{R}^{d}\), \(\mu\),\(\nu\) are two probability measures on \(\mathcal{X}\times \mathbb{R} ^{d}\), and \(p\geq 1\), the \(p\)-Wasserstein distance between \(\mu\) and \(\nu\) is defined as \[ W_{p}(\mu ,\nu )=\left( \inf_{\gamma \in \Pi (\mu ,\nu )}\int_{(\mathcal{X}\times \mathbb{R}^{d})^{2}}(\left\vert x-y\right\vert |^{p}+\left\vert v-w\right\vert ^{p})d\gamma (x,v,y,w)\right) ^{1/p}, \] where \(\Pi (\mu ,\nu )\) is the set of all probability measures on \((\mathcal{X} \times \mathbb{R}^{d})^{2}\) with marginals \(\mu\) and \(\nu\). The author first proves that if \(f_{1},f_{2}\) are two solutions to the Vlasov equation on \(\mathcal{X}\), that is satisfying \(\partial _{t}f+v\cdot \nabla _{x}f+(\nabla K\ast \int fdv)\cdot \nabla _{v}f=0\), where the kernel \(K\) is smooth enough and satisfies \(\left\Vert D^{2}K\right\Vert _{\infty}=B<\infty\), then \[ W_{1}(f_{1}(t),f_{2}(t))\leq \min \{(1+t)e^{\frac{2}{3} B((1+t)^{3}-1)},e^{(1+2B)t}\}W_{1}(f_{1}(0),f_{2}(0)). \] This improves Dobrushin's estimate. The author then considers the Vlasov-Poisson system written as: \(\partial _{t}f_{\varepsilon}+v\cdot \nabla _{x}f_{\varepsilon}+E_{\varepsilon}\cdot \nabla _{v}f_{\varepsilon}=0\), \(E_{\varepsilon}=-\nabla _{x}U_{\varepsilon}\), \(-\varepsilon ^{2}\Delta _{x}U_{\varepsilon}=\int_{\mathbb{R}^{d}}f_{\varepsilon}dv-1\), with the initial condition \( f_{\varepsilon}\mid _{t=0}=f_{0,\varepsilon}\geq 0\), with \(\int_{\mathbb{T} ^{d}\times \mathbb{R}^{d}}f_{0,\varepsilon}dxdv=1\), where \(\varepsilon\) is a small parameter which involves the Debye length inside a plasma. The main result of the paper here proves a stability estimate for \( W_{2}(f_{1}(t),f_{2}(t))\), \(f_{1},f_{2}\) being two weak solutions to the Vlasov-Poisson system, if \(W_{2}(f_{1}(0),f_{2}(0))\) is small enough and satisfies boundedness hypotheses. This allows the description of the behavior of the solution to the Vlasov-Poisson system when \(\varepsilon\) goes to 0, through an almost optimal rate in the quasi-neutral limit.