Geometric properties of solutions to elliptic PDE’s in Gauss space And related Brunn-Minkowski type inequalities

Andrea Colesanti, Lei Qin, Paolo Salani · 2025

Read Geometric properties of solutions to elliptic PDE’s in Gauss space And related Brunn-Minkowski type inequalities by Andrea Colesanti, Lei Qin, Paolo Salani (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-2502-00184.txt.

Introduction

Andrea Colesanti, Lei Qin, and Paolo Salani prove a shape result for equations in Gauss space in a 2025 math paper. Gauss space weights points by a bell shaped curve, which changes how size and volume work. The authors study the first main vibration value of a weighted operator on bounded domains with a fixed edge value.

They prove an inequality of Brunn-Minkowski type for that value, a rule that links the value for a blended domain to the values for its parts. They also prove that the matching positive solution has a concave log form when the domain is convex. The results extend classic shape rules to the weighted setting.

Catalog record

CREATORAndrea Colesanti, Lei Qin, Paolo Salani
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-00184

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Geometric properties of solutions to elliptic PDE’s in Gauss space And related Brunn-Minkowski type inequalities Andrea Colesanti, Lei Qin, Paolo Salani Abstract . We prove a Brunn-Minkowski type inequality for the first (nontrivial) Dirichlet eigenvalue of the weighted p p -operator − Δ p , γ u = − div ( | ∇ u | p − 2 ∇ u ) + ( x , ∇ u ) | ∇ u | p − 2 , -\Delta_{p,\gamma}u=-\text{div}(|\nabla u|^{p-2}\nabla u)+(x,\nabla u)|\nabla u|^{p-2}, where p > 1 p>1 , in the class of bounded Lipschitz domains in ℝ n \mathbb{R}^{n} . We also prove that the corresponding positive eigenfunctions are log-concave if the domain is convex. 1. Introduction The classical Brunn-Minkowski inequality has a deep impact on both geometry and analysis, see the beautiful survey paper by Gardner [ 20 ] for more information. Its tentacles extend in many directions and, noticeably, Brunn-Minkowski type inequalities have been proved for several variational functionals, see for instance [ 3 , 4 , 5 , 8 , 9 ] . In particular, the Brunn-Minkowski inequality for the first Dirichlet eigenvalue of the Laplace operator [ 5 ] has been extended to the cases of the p -Laplace operator [ 10 ] , of the Monge-Ampère operator [ 41 ] , of the k -Hessian operators [ 36 , 42 ] , and of the principal frequency of fully non-linear homogeneous elliptic operators [ 14 ] . Moreover, very recently, a Brunn-Minkowski inequality has been established for the principal frequency of the Laplace operator in the Gauss space, see [ 11 ] . Roughly speaking, the main goal of this paper is to extend the latter result to the p -Laplace operator in the Gauss space, for p>1 . Noticeably, every Brunn-Minkowski type inequality mentioned above is strictly connected to a suitable convexity property of the minimizing functions of the variational functionals at hand. As far as we know, in the known cases of an eigenvalue/principal frequency, this property turns out to be log-concavity. Ultimately, the same happens for the principal frequency of the Gaussian p -Laplacian, indeed here we also prove the log-concavity of the associated eigenfunctions in convex domains. Now, let us describe more precisely the results of this paper. Let \Omega be a bounded domain in \mathbb{R}^{n} . For p>1 , the p -th Rayleigh quotient in Gauss space is given by (1) \frac{\int_{\Omega}|\nabla u(x)|^{p}d\gamma}{\int_{\Omega}|u(x)|^{p}d\gamma}. Here \gamma denotes the Gaussian probability measure in \mathbb{R}^{n} , given by \gamma(A)=\frac{1}{(2\pi)^{n/2}}\int_{A}e^{-|x|^{2}/2}dx\,,\quad\text{for any measurable set }A\,. We consider the following minimum problem: \lambda_{p,\gamma}:=\inf\left\{\frac{\int_{\Omega}|\nabla u|^{p}d\gamma}{\int_{\Omega}|u|^{p}d\gamma}\colon u\in W^{1,p}_{0}(\Omega,\gamma),\,\int_{\Omega}|u|^{p}d\gamma>0\right\}. The corresponding Euler-Lagrange equation is given by (2) \left\{\begin{aligned} &-\Delta_{p,\gamma}u=\lambda_{p,\gamma}|u|^{p-2}u\ \ &\mathrm{in}\ \ \Omega,\\ &\ u=0\ \ \ \ &\mathrm{on}\ \ \partial\Omega\,,\end{aligned}\right. where -\Delta_{p,\gamma}u:=-\text{div}(|\nabla u|^{p-2}\nabla u)+(x,\nabla u)|\nabla u|^{p-2}\,. Any (weak) solution of problem ( 2 ) is called a first (Dirichlet) eigenfunction. For the existence, we can refer, for instance, to [ 19 , section 6] . By standard arguments (based on the ones of [ 34 ] ), it follows that all first eigenfunctions coincide up to scalar multiplication, and they don’t change sign inside \Omega , so we will consider only nonnegative eigenfunctions, if not differently specified. More details can be found in [ 15 ] and [ 19 ] , where the authors give basic introductory material for problem ( 2 ). As already said, one of our main results is a Brunn-Minkowski type inequality for \lambda_{p,\gamma} , in the class of open bounded domains with Lipschitz boundary. To avoid useless complications, throughout the paper we will consider only connected sets, even if not explicitly specified. Theorem 1.1 . Fix p>1 and t\in[0,1] . Let \Omega_{0},\Omega_{1} and \Omega_{t}=(1-t)\Omega_{0}+\Omega_{1}:=\{(1-t)x_{0}+tx_{1}:\ x_{0}\in\Omega_{0},\ x_{1}\in\Omega_{1}\}, be open bounded Lipschitz domains in \mathbb{R}^{n} , n\geq 2 . Then \lambda_{p,\gamma}(\Omega_{t})\leq(1-t)\lambda_{p,\gamma}(\Omega_{0})+t\lambda_{p,\gamma}(\Omega_{1}). Remark 1.2 . Note that, if \Omega_{0} and \Omega_{1} belong to the class \mathcal{A}^{n} of Lipschitz domains in \mathbb{R}^{n} with positive reach, then \Omega_{t} belongs to the same class as well, as observed in [ 14 ] . The Lipschitz regularity of the involved domains assures that the related solutions of problem ( 2 ) are continuous up to the boundary. For this, of course, we could assume weaker conditions, in particular a suitable Wiener’s type condition, see [ 21 ] . On the other hand, in the proof of Theorem 1.1 the Lipschitz regularity of \partial\Omega_{t} will be also used to prove that a viscosity sub-solution of problem ( 2 ), once extended as 0 outside of \Omega_{t} , remains a viscosity sub-solution. For this property, as well, we could weaken the assumptions, however, we prefer to give a clear and simple statement and to leave the investigation in this direction to a possible future research. Remark 1.3 . We do not attempt to give an account of the vast literature concerning regularity for solutions of quasilinear elliptic equations. We just point out (see Proposition 2.1 ) that, for any bounded open domain \Omega , a weak solution u of problem ( 2 ) is in fact of class C^{1,\alpha}_{\mathrm{loc}}(\Omega)\cap C^{2}(\Omega\setminus\bar{\mathcal{C}}) for some \alpha\in(0,1) , where \mathcal{C}=\{x\in\Omega:\ \nabla u(x)=0\} ; moreover, if \partial\Omega is of class C^{1,\alpha} , then u\in C^{1,\beta}(\overline{\Omega})\cap C^{2}(\Omega\setminus\bar{\mathcal{C}}) (some \beta\in(0,1) ). Hence, if \Omega_{0} and \Omega_{1} are of class C^{1,\alpha} , taking into account also the previous Remark 1.2 , the proofs of Theorem 1.1 and of our other subsequent results would be simpler. As it is now well understood, when a Brunn-Minkowski type inequality holds, it can be used to derive an Urysohn’s type inequality for the involved functional (see for instance [ 2 ] , and [ 43 ] for a related general theory). In order to state the property that we obtain in this case, let us recall that, given an open bounded convex set \Omega and a direction y\in{{\mathbb{S}}^{n-1}} , the width of \Omega in the direction y , that we denote by w(\Omega,y) , is the distance between the supporting hyperplanes to \Omega with outer unit normals y and -y . The mean width w(\Omega) of \Omega is then defined as w(\Omega)=\frac{1}{\mathcal{H}^{n-1}({{\mathbb{S}}^{n-1}})}\int_{{{\mathbb{S}}^{n-1}}}w(\Omega,y)\mbox{\rm d}{\mathcal{H}}^{n-1}(y), where {\mathcal{H}}^{n-1} is the (n-1) -dimensional Hausdorff measure. If \Omega is bounded, but not necessarily convex, we refer to its mean width w(\Omega) as the mean width of its convex hull. Similarly to [ 11 , Corollary 1.5] , we obtain the following result. Corollary 1.4 . Let p>1 and \Omega be an open bounded Lipschitz domain in \mathbb{R}^{n} , n\geq 2 . Then (3) \lambda_{p,\gamma}(\Omega)\geq\lambda_{p,\gamma}(\Omega^{\sharp}), where \mbox{$\Omega^{\sharp}$ is a ball centered at $0$ such that}\,\,w(\Omega^{\sharp})=w(\Omega). We recall that in the plane mean width coincides with the Euclidean perimeter of the convex hull. Then the above corollary, for n=2 , tells that the disk centered at the origin has the minimum principal Gauss p -frequency among sets with given Euclidean perimeter. Hence, on one hand, Corollary 1.4 (also for n larger than 2 ) may appear somewhat surprising, since minimizers in the Faber-Krahn inequality in the Gauss space (like isoperimetric sets) are half-spaces (see [ 16 ] , and, for instance, [ 1 , 7 ] for more references). On the other hand, the mean width of a half-space is infinity, moreover the width is a metric quantity and does not depend

Read the complete text