Explicit minimisers for anisotropic Riesz energies
R.L. FRank, J. Mateu, M.G. Mora, L. Rondi, L. Scardia, J. Verdera · 2025
Read Explicit minimisers for anisotropic Riesz energies by R.L. FRank, J. Mateu, M.G. Mora, L. Rondi, L. Scardia, J. Verdera (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-11644.txt.
Introduction
Rupert Frank and co-authors give exact forms for the best spread of points under a class of pull-and-push energies in a 2025 paper. The pull grows with squared gap, while the push fades with gap in a Riesz-like way and shifts by path. Such energies model crowds of charged or linked specks that must settle in some shape. The team names the winning shapes in closed form for this class. Exact winners are rare, so the result gives a test case for wider math work.
Catalog record
| CREATOR | R.L. FRank, J. Mateu, M.G. Mora, L. Rondi, L. Scardia, J. Verdera |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-11644
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Explicit minimisers for anisotropic Riesz energies R.L. FRank, J. Mateu, M.G. Mora, L. Rondi, L. Scardia, J. Verdera Abstract . In this paper we characterise the energy minimisers of a class of nonlocal interaction energies where the attraction is quadratic, and the repulsion is Riesz-like and anisotropic. In particular we show that, if the Fourier transform of the repulsive potential is positive, the minimiser is supported on a fully-dimensional ellipsoid, and its density is given by a Barenblatt-type profile. Our technique of proof is based on a Fourier representation of the potential of such measures, that extends a previous formula established by some of the authors in the Coulomb case. AMS 2010 Mathematics Subject Classification: 31A15 (primary); 49K20 (secondary) Keywords: nonlocal energy, potential theory, anisotropic interaction, Riesz potential 1. Introduction Motivated by applications in physics, biology and economics, in this paper we study a mean-field model of particles (or agents) interacting through a repulsive, anisotropic Riesz potential in a quadratic confinement. More precisely, we consider nonlocal energies of the form (1.1) I(\mu):=\int_{\mathbb{R}^{d}}(W\ast\mu)(x)\,d\mu(x)+\int_{\mathbb{R}^{d}}|x|^{2}\,d\mu(x), defined on probability measures \mu\in\mathcal{P}(\mathbb{R}^{d}) , for d\geq 2 . In ( 1.1 ) the potential W is of the form (1.2) W(x):=\frac{1}{|x|^{s}}\Psi\left(\frac{x}{|x|}\right),\quad s\in(0,d), for x\neq 0 and W(0)=+\infty , and the profile \Psi is even, strictly positive on \mathbb{S}^{d-1} , and such that both W and \widehat{W} are continuous on \mathbb{S}^{d-1} . Here \widehat{W} denotes the Fourier transform of W , see Section 2.3 . Explicit examples of anisotropic potentials of the form above can be found in [ 3 , 12 , 18 ] in the case s=d-2 , and are used to model interactions of defects in metals, see [ 15 ] . We also observe that small perturbations of the isotropic Riesz potential - namely, kernels of the form ( 1.2 ) with a profile \Psi=1+\varepsilon\Phi , where \Phi is even and smooth and \varepsilon is sufficiently small - automatically satisfy our assumptions (see [ 19 ] ). Alternatively, one could consider the fully nonlocal analogue of ( 1.1 ), \widetilde{I}(\mu):=\int_{\mathbb{R}^{d}}(\widetilde{W}\ast\mu)(x)\,d\mu(x), where the attractive/repulsive interaction potential \widetilde{W} is given by \widetilde{W}(x):=W(x)+\frac{1}{2}|x|^{2} . In fact, by translation invariance one can reduce the minimisation of \widetilde{I} to probability measures \mu with compact support and with \int_{\mathbb{R}^{d}}x\,d\mu(x)=0 , and for such measures the two functionals coincide. In what follows, we find it more convenient to work with I rather than \widetilde{I} and will focus on the formulation ( 1.1 ). We observe that for a constant profile \Psi , which without loss of generality can be assumed to be \Psi\equiv 1 , the potential W reduces to the classical (radially symmetric) Riesz potential. In this case it was proved in [ 4 ] that for s\in(\max\{d-4,0\},d) the unique minimiser of I over \mathcal{P}(\mathbb{R}^{d}) is given by the Barenblatt measure (1.3) \mu_{\text{iso},d}(x)=c_{d}\,(1-|x/r_{d}|^{2})^{\frac{s+2-d}{2}}\chi_{r_{d}\overline{B}}(x), where c_{d}>0 and r_{d}>0 are explicit constants depending on d , s , B=B_{1}(0) is the unit ball, \chi_{r_{d}\overline{B}} denotes the characteristic function of r_{d}\overline{B} , and where we identified the measure with its density \mu_{\text{iso},d}\in L^{1}(\mathbb{R}^{d}) (see also [ 2 ] ). For 0<s\leq d-4 , instead, it was proved in [ 11 ] that the unique minimiser is given by the uniform probability measure on a sphere whose radius depends on d . Note that in some of these previous works the equivalent minimisation problem for \widetilde{I} was considered. The values of the constants c_{d} and r_{d} can be extracted from [ 11 ] . For a related result in dimension d=1 , see [ 10 ] . In the present paper we are interested in the anisotropic case where \Psi is not necessarily constant. We show that for dimension d=3 and for the full range s\in(0,3) , if \widehat{W}>0 , the unique minimiser \mu_{\text{min}} of the energy I over \mathcal{P}(\mathbb{R}^{d}) is supported on a fully-dimensional ellipsoid E\subset\mathbb{R}^{3} , and its density is given by a Barenblatt-type profile. Roughly speaking, as long as \widehat{W}>0 , the effect of any anisotropic potential of the form ( 1.2 ) on the minimiser is simply that of ‘deforming’ the support of ( 1.3 ) from a ball into a suitable ellipsoid. Our proof extends to higher dimensions for a partial range of Riesz homogeneities. To be precise, our main result reads as follows. Theorem 1.1 . Let s\in[d-3,d)\cap(0,5] , and let W be as in ( 1.2 ) with \Psi even. Assume that W and \widehat{W} are strictly positive and continuous on \mathbb{S}^{d-1} . Then there exists a unique minimiser \mu_{0} of I over \mathcal{P}(\mathbb{R}^{d}) . It is given by the push-forward of the measure ( 1.3 ) through the function x\mapsto RD(a/r_{d})x , for some rotation R\in SO(d) and some a\in\mathbb{R}^{d} with a_{i}=a\cdot e_{i}>0 . More precisely, (1.4) \mu_{0}(x)=\frac{c_{d}}{\prod_{j=1}^{d}(a_{j}/r_{d})}\left(1-\left|D\big(\tfrac{1}{a}\big)R^{T}x\right|^{2}\right)^{\frac{s+2-d}{2}}\chi_{E}(x), where E is the ellipsoid RD(a)\overline{B} , and c_{d} and r_{d} are the constants from ( 1.3 ). In the statement above D(a) denotes the diagonal matrix such that (D(a))_{ii}=a_{i} , and D\big(\tfrac{1}{a}\big) denotes the diagonal matrix such that (D(\tfrac{1}{a}))_{ii}=1/a_{i} . In the two-dimensional case the result of Theorem 1.1 has been proved by Carrillo and Shu in [ 5 ] . They also considered the three-dimensional case in [ 6 ] , but only under additional symmetry assumptions on the potential, which essentially made the problem two-dimensional. In [ 20 ] a new strategy of proof was developed, which allowed to successfully remove the additional assumptions in [ 6 ] in the special case of Coulomb singularity s=d-2 in three dimensions (corresponding to s=1 ). In [ 21 ] this method was adapted to the case of logarithmic interactions in two dimensions, which correspond loosely speaking to ‘ s=0 ’. In the paper [ 6 ] Carrillo and Shu raised the question of whether the methods of [ 20 , 21 ] could be extended, in the three-dimensional case, to the full range s\in(0,3) of homogeneity, beyond the Coulomb singularity s=1 . In this paper we give a positive answer to that question, and characterise the minimisers in the full range s\in(0,3) , without the additional symmetry assumptions on \Psi required in [ 6 ] . Moreover, since the range [d-3,d)\cap(0,5] considered here includes the Coulomb exponent s=d-2 for 3\leq d\leq 7 , our result extends those of [ 20 , 21 ] to dimension d\leq 7 . The strategy of proof of our main result consists in showing that there exist a rotation R\in SO(d) and a diagonal matrix D(a) with positive entries such that the candidate \mu_{\text{min}} , that is, the push-forward of \mu_{\text{iso},d} through the function x\mapsto RD(a/r_{d})x , satisfies the Euler-Lagrange conditions (1.5) \displaystyle\left(W\ast\mu_{\text{min}}\right)(x)+\frac{1}{2}|x|^{2} \displaystyle=C\quad\text{for }x\in E, (1.6) \displaystyle\left(W\ast\mu_{\text{min}}\right)(x)+\frac{1}{2}|x|^{2} \displaystyle\geq C\quad\text{for }x\in\mathbb{R}^{d}, where C is a constant, and E is the ellipsoid RD(a)\overline{B} . To this aim we follow the methodology developed in our paper [ 20 ] which requires, as a first step, to write the potential W\ast\mu_{\text{min}} in Fourier space. This computation is considerably more involved than in the Coulomb case, since the Fourier transform of \mu_{\text{min}} involves Bessel and hypergeometric functions, and only works for s\in(d-4,d)\cap(0,5] , where the further restriction s\in(0,5] is needed to deal with the lack of integrability of \widehat{W\ast\mu_{\text{min}}} . In the isotropic case, similar computations can be found in [ 1 ] , for