Universal diagonal estimates for minimizers of the Levy-Lieb functional (Q6084681)

The authors of the article under review study properties of minimizers of the Levy-Lieb functional \[ \begin{aligned} F_{LL,\epsilon}(\varrho) &= \\ \min_{\Psi \in H^1_{\mathrm{sym}}(\mathbb{R}^{3N}), \varrho_{\Psi} = \varrho} &\left\{ \int_{\mathbb{R}^{3N}} \left( \epsilon | \nabla \Psi(x_1,...,x_N) |^2 + v_{ee}(x_1,...,x_N) | \Psi(x_1,...,x_N) |^2 \right) \mathrm{d} (x_1,...,x_N) \right\}, \end{aligned}\tag{1} \] where \(H^1_{\mathrm{sym}} (\mathbb{R}^{3N})\) denotes the closed linear subspace of \(H^1 (\mathbb{R}^{3N})\) consisting of those functions \(\Psi(x_1,...,x_{N})\) that are invariant under any permutation of the particle coordinates \(x_1,...,x_{N} \in \mathbb{R}^3\). Moreover, \[ \varrho_{\Psi}(x) = N \int_{\mathbb{R}^{3(N-1)}}| \Psi(x,q_1,...,q_{N-1}) |^2 \mathrm{d}(q_1,...,q_{N-1}) \] denotes the particle density related to \(\Psi\) and \(v_{ee}(x_1,...,x_N) = \sum_{1 \leq i < j \leq N} 1/|x_i-x_j|\) implements a repulsive Coulomb interaction between the particles. The minimization problem in Eq. (1) is a bosonic version of the standard Levy-Lieb functional, see [\textit{M. Levy}, Proc. Natl. Acad. Sci. U. S. A., 76, No. 12, 6062--6065 (1979; \url{doi:10.1073/pnas.76.12.6062}); \textit{M. Lewin} et al., ``Universal functionals in density functional theory, in: Density functional theory. Cham: Springer (2023); \textit{E. H. Lieb}, ``Density functionals for Coulomb systems'', Int. J. Quantum Chem. 24, 243--277 (1983)], where one minimizes over antisymmetric (or fermionic) wave functions. Let \[ D_{\delta} = \{ (x_1,...,x_N) \in \mathbb{R}^{3N} \ | \ \exists \ i \neq j \text{ s.t. } |x_i-x_j| \leq \delta \} \tag{2} \] denote the enlarged diagonal in position space and define \(\kappa(\varrho,r) = \sup_{x \in \mathbb{R}^3} \int_{ \{ |x-y| \leq r \}} \varrho(y) \mathrm{d} y/N\). The main result of the article under review is the following statement: assume that \(r\) is s.t. \(\kappa(\varrho,r) \leq 1/(4(N-1))\), let \(\alpha \leq r/(32 N)\), and suppose \(\epsilon N^2 \ll \alpha\). Then the minimizer \(\Psi_{\epsilon}\) of the problem in Eq. (1) satisfies \[ \int_{D_{\alpha}} | \Psi_{\epsilon} (x_1,...,x_N) |^2 \mathrm{d} (x_1,...,x_N) \leq \exp\left( - (1/24) \sqrt{\alpha/\epsilon} \right). \] The authors also prove a similar result for more general interaction potentials. In [\textit{U. Bindini} and \textit{L. De Pascale}, J. Éc. Polytech., Math. 4, 909--934 (2017; Zbl 1409.49010); \textit{C. Cotar} et al., Commun. Pure Appl. Math. 66, No. 4, 548--599 (2013; Zbl 1266.82057); \textit{C. Cotar} et al., Arch. Ration. Mech. Anal. 228, No. 3, 891--922 (2018; Zbl 1394.82015); \textit{M. Lewin}, C. R., Math., Acad. Sci. Paris 356, No. 4, 449--455 (2018; Zbl 1407.60002)] it has been shown that the problem in Eq. (1) converges to a multi-marginal optimal transport problem in the limit \(\epsilon \to 0\) (semiclassical limit). This problem has the property that minimizers vanish on the set \(D_{\delta}\) in Eq.(2) for some \(\delta > 0\), see [\textit{G. Buttazzo} et al., Appl. Math. Optim. 78, No. 1, 185--200 (2018; Zbl 1400.49051); \textit{M. Colombo} et al., SIAM J. Math. Anal. 51, No. 4, 2903--2926 (2019; Zbl 1419.35169); \textit{L. De Pascale}, ESAIM, Math. Model. Numer. Anal. 49, No. 6, 1643--1657 (2015; Zbl 1330.49048)]. The main result in the article under review is inspired by this result.