Particle approximation of nonlocal interaction energies

From MaRDI portal





This paper is concerned with \(n\)-particle interaction energies of the form \N\[\N\mathcal E_n(x_1,\dots,x_n) = \frac{1}{n^2} \sum_{\substack{1\le i,j\le n\\\Ni\neq j}} g(x_i-x_j), \tag{1.1} \N\]\Nwhere \(g:\mathbb R^N\to\mathbb R\) is a pairwise interaction kernel. These energies arise naturally in particle systems and can be viewed as discrete counterparts of the continuous interaction energy\N\[\N\mathcal E(\mu) = \int_{\mathbb R^N}\int_{\mathbb R^N} g(x-y)\,d\mu(x)\,d\mu(y), \tag{1.2}\N\]\Ndefined on probability measures \(\mu\in\mathcal P(\mathbb R^N)\).\N\NThe analysis is carried out under the following basic assumption \((H_1)\): \(g\) is bounded from below, lower semicontinuous, and belongs to \(L^1_{\mathrm{loc}}(\mathbb R^N)\).\N\NThe paper studies Riesz-type nonlocal energies associated with general interaction kernels and their discretizations. Its main result establishes that the discrete energies \(\mathcal E_n\) \(\Gamma\)-converge, with respect to the weak-\(^*\) topology, to the limiting functional \(\mathcal E\) on the space of probability measures. More precisely, the authors prove the following statement.\N\N\textbf{Theorem 5 (\(\Gamma\)-convergence).} Suppose that \(g\) satisfies assumption \((H_1)\). Then \(\mathcal E_n \xrightarrow{\Gamma} \mathcal E\). In particular, if \(\mathcal E(\mu)<+\infty\), then any recovery sequence is necessarily given by \(n\)-discrete probability measures, at least for \(n\) sufficiently large.\N\NIn addition, under general assumptions, the paper addresses the minimization problem for the discrete energies and proves the existence of minimizing particle configurations.











This page was built for publication: Particle approximation of nonlocal interaction energies

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6872597)