The Gaussian core model in high dimensions (Q1615253)

Let \(f:(0\to\infty)\to\mathbb{R}\) be any function. The lower \(f\)-energy \(E_f(\mathcal{C})\) of a discrete subset \(\mathcal{C}\subset\mathbb{R}^n\) is \(\liminf_{r\to\infty} \frac{1}{\sharp (\mathcal{C}\cap B^n_r(0))}\sum_{x,y\in \mathcal{C}\cap B^n_r(0), x\neq y}f(|x-y|)\). Assume \(E_f(\rho,n)=\min_{\mathcal{C}} E_f(\mathcal{C})\) where the minimum is taken over all sets with the density \(\rho\). Suppose that \(f(t)=e^{-\alpha t^2}\) and \(\rho\) is fixed. Then we have \(E_f(\rho,n)=(\rho+o(1))(\pi/\alpha)^{n/2}\) for \(0<\alpha<4\pi/e\) and \(E_f(\rho,n)\geq \rho(\frac{1}{2}e^{1-\alpha e/(8\pi)}+o(1))^n\) for \(\alpha\geq 4\pi/e\). Also similar results for \(f(t)=1/t^{n+s}\), \(s>0\) are stated.