Microscopic derivation of Vlasov equations with singular potentials

The Vlasov-Poisson equation is a classical example of an effective equation which shall describe the coarse-grained time evolution of a system consisting of a large number of particles which interact by Coulomb or Newton’s gravitational force. Although major progress concerning a rigorous justification of such an approach was made recently, there are still substantial steps necessary to obtain a completely convincing result. The main goal of this work is to yield further progress in this regard. \ To this end, we consider on the one hand $N$-dependent forces $f^N$ (where $N$ shall denote the particle number) which converge pointwise to Coulomb or alternatively Newton`s gravitational force. More precisely, the interaction fulfills $f^N(q)=\pm \frac{q}{\mathrm{\mid }q{\mathrm{\mid }}^3}$ for $\mathrm{\mid }q\mathrm{\mid }>N^{-\frac{7}{18}+ϵ}$ and has a cut-off at $\mathrm{\mid }q\mathrm{\mid }=N^{-\frac{7}{18}+ϵ}$ where $ϵ>0$ can be chosen arbitrarily small. We prove that under certain assumptions on the initial density $k_0$ the characteristics of Vlasov equation provide typically a very good approximation of the $N$-particle trajectories if their initial positions are i.i.d. with respect to density $k_0$. Interestingly, the cut-off diameter is of smaller order than the average distance of a particle to its nearest neighbor. Nevertheless, the cut-off is essential for the success of the applied approach and thus we consider additionally less singular forces scaling like $\mathrm{\mid }f(q)\mathrm{\mid }=\frac{1}{\mathrm{\mid }q{\mathrm{\mid }}^{\alpha }}$ where $\alpha \in (1,\frac{4}{3}]$. In this case we are able to show a corresponding result even without any regularization. Although such forces are distinctly less interesting than for instance Coulomb interaction from a physical perspective, the introduced ideas for dealing with forces where even the related potential is singular might still be helpful for attaining comparable results for the arguably most interesting case $\alpha =2$.