Improved Berezin-Li-Yau inequality and Kröger inequality and consequences 0 0 footnotetext: 2020 Mathematics Subject Classification . Primary 35P15; Secondary 35P20, 42B37. Key words and phrases: Dirichlet eigenvalue, Neumann eigenvalue, Berezin-Li-Yau inequality, Kröger inequality
Zaihui Gan, Renjin Jiang & Fanghua Lin · 2025
Read Improved Berezin-Li-Yau inequality and Kröger inequality and consequences 0 0 footnotetext: 2020 Mathematics Subject Classification . Primary 35P15; Secondary 35P20, 42B37. Key words and phrases: Dirichlet eigenvalue, Neumann eigenvalue, Berezin-Li-Yau inequality, Kröger inequality by Zaihui Gan, Renjin Jiang & Fanghua Lin (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-2507-20330.txt.
Introduction
Zaihui Gan, Renjin Jiang, and Fanghua Lin sharpen two classic eigenvalue inequalities in a two thousand twenty-five math paper. The Berezin-Li-Yau and Kröger inequalities bound sums of Dirichlet and Neumann eigenvalues of domains.
The authors prove quantitative improvements to both inequalities in Euclidean space of dimension at least two. The Kröger improvement resolves an open question Weidl raised in two thousand six. Consequences follow for Pólya's famous conjecture that eigenvalue bounds are exact for tiling domains. Infinitely many Dirichlet eigenvalues satisfy the conjecture in dimension three and above, and infinitely many Neumann eigenvalues do in dimension five and above when the spectrum is discrete.
Catalog record
| CREATOR | Zaihui Gan, Renjin Jiang & Fanghua Lin |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-20330
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Improved Berezin-Li-Yau inequality and Kröger inequality and consequences 0 0 footnotetext: 2020 Mathematics Subject Classification . Primary 35P15; Secondary 35P20, 42B37. Key words and phrases: Dirichlet eigenvalue, Neumann eigenvalue, Berezin-Li-Yau inequality, Kröger inequality Zaihui Gan, Renjin Jiang & Fanghua Lin Abstract Abstract . We provide quantitative improvements to the Berezin-Li-Yau inequality and the Kröger inequality, in \mathbb{R}^{n} , n\geq 2 . The improvement on Kröger’s inequality resolves an open question raised by Weidl from 2006. The improvements allow us to show that, for any open bounded domains, there are infinite many Dirichlet eigenvalues satisfying Pólya’s conjecture if n\geq 3 , and infinite many Neumann eigenvalues satisfying Pólya’s conjecture if n\geq 5 and the Neumann spectrum is discrete. 1 Introduction Given an open bounded domain \Omega in {{{\mathbb{R}}}^{n}} , n\geq 2 . Let 0<\lambda_{1}(\Omega)\leq\lambda_{2}(\Omega)\leq\lambda_{3}(\Omega)\leq\cdots be eigenvalues to the Dirichlet Laplacian on \Omega . When dealing with the Neumann Laplacian on \Omega , we shall assume that -\Delta_{N} on \Omega has a discrete spectrum as 0=\gamma_{1}(\Omega)\leq\gamma_{2}(\Omega)\leq\gamma_{3}(\Omega)\leq\cdots. We shall simply write \lambda_{k},\gamma_{k} instead of \lambda_{k}(\Omega),\gamma_{k}(\Omega) when there is no confusion. In 1954, Pólya [ 26 ] conjectured that, it should hold on arbitrary domain that \frac{|\Omega|\omega(n)}{(2\pi)^{n}}\gamma_{k+1}^{n/2}\leq k\leq\frac{|\Omega|\omega(n)}{(2\pi)^{n}}\lambda_{k}^{n/2}, where \omega(n) denotes the volume of unit ball in {{{\mathbb{R}}}^{n}} . Pólya himself [ 27 ] proved this conjecture for tiling domains (regular tiling domains for the Neumann case) in the plane. His method extends to high dimensions, and so the conjecture is true for tiling domains in high dimensions too. See Kellner [ 17 ] for removing the regular assumption regarding Neumann eigenvalues and simplification. The Pólya conjecture has been wide open since then. Other than tiling domains, Pólya’s conjecture was known to be true on on product domains \Omega_{1}\times\Omega_{2}\subset{\mathbb{R}}^{n_{1}+n_{2}} , n_{1}\geq 2 , n_{2}\geq 1 provided Pólya’s conjecture holds on \Omega_{1} , by the work of Laptev [ 21 ] . Filonov et al. [ 5 , 6 ] verified Pólya’s conjecture on balls for Dirichlet eigenvalues for all n\geq 2 , and for Neumann eigenvalues for n=2 , and on annuli for Dirichlet eigenvalues for n=2 . Recently, it was proved in [ 16 ] that on a Lipschitz domain \Omega , for any \epsilon\in(0,1) it holds for all \lambda_{k}\geq\Lambda(\Omega,\epsilon) that k\leq(1+\epsilon)\frac{|\Omega|\omega(n)}{(2\pi)^{n}}\lambda_{k}^{n/2}, where \Lambda(\Omega,\epsilon) depending on the geometry of \Omega is given explicitly. Moreover, [ 16 ] also provides in all dimensions n\geq 2 classes of irregular shaped domains that satisfy Pólya’s conjecture on the Dirichlet eigenvalue. For general domains \Omega , it was known that Pólya’s conjecture holds for the first two Dirichlet eigenvalues and the first two non-trivial Neumann eigenvalues, see [ 13 ] for the Rayleigh-Faber-Krahn inequality ( \lambda_{1}(\Omega) ), the Krahn-Szegö inequality ( \lambda_{2}(\Omega) ) and the Szegö-Weinberger inequality ( \gamma_{2}(\Omega) ), and also recent breakthroughs [ 3 , 12 ] for \gamma_{3}(\Omega) . In higher dimensions, n\geq 3 , Freitas [ 10 ] proved that the first b(n) Dirichlet eigenvalues satisfying Pólya conjecture, where b(n)\geq 4 and grows as dimension increases. For arbitrary bounded open domains, Berezin [ 2 ] and Li-Yau [ 24 ] independently proved that (1.1) \frac{|\Omega|^{2/n}\omega(n)^{2/n}}{(2\pi)^{2}}\sum_{i=1}^{k}\lambda_{i}\geq\frac{n}{n+2}k^{\frac{n+2}{n}}, which implies that (1.2) \frac{|\Omega|^{2/n}\omega(n)^{2/n}}{(2\pi)^{2}}\lambda_{k}\geq\frac{n}{n+2}k^{\frac{2}{n}}. Melas [ 25 ] later gave a quantitative improvement of Li-Yau’s estimate as (1.3) \sum_{i=1}^{k}\lambda_{i}\geq\frac{n}{n+2}\frac{(2\pi)^{2}}{|\Omega|^{2/n}\omega(n)^{2/n}}k^{\frac{n+2}{n}}+C(n)k\frac{|\Omega|}{I(\Omega)}, where I(\Omega) is the moment of inertia of \Omega , i.e., (1.4) \displaystyle I(\Omega):=\inf_{a\in{{{\mathbb{R}}}^{n}}}\int_{\Omega}|x-a|^{2}\,dx. Above C(n) can be taken as \frac{1}{24(n+2)} , see Cheng et al. [ 4 , p. 36] , and a better constant for n=2 , C(2)=\frac{1}{32} , from [ 19 , (16)] . Melas’s result have been further improved by Kovařík, Vugalter and Weidl [ 19 ] for n=2 , where a term of order k^{\frac{3}{2}-\epsilon} can be added to the RHS of ( 1.3 ) on domains with some smoothness. See also [ 11 , 15 , 29 ] for further generalisations. Kröger [ 20 ] proved for bounded domains with piecewisely smooth boundary that (1.5) \displaystyle\frac{|\Omega|^{2/n}\omega(n)^{2/n}}{(2\pi)^{2}}\sum_{j=1}^{k+1}\gamma_{j}\leq\frac{n}{n+2}k^{1+2/n}. and (1.6) \displaystyle\frac{|\Omega|^{2/n}\omega(n)^{2/n}}{(2\pi)^{2}}\gamma_{k+1}\leq\left(\frac{n+2}{2}\right)^{2/n}k^{2/n}. Li and Tang [ 23 ] showed the above inequality ( 1.5 ) is strict by multiplying on the right hand side an implicit factor C(k,n,\Omega)<1 which tends to 1 as k goes to infinity. We note Filonov [ 7 ] recently improved the estimate for Neumann eigenvalues on planar convex domains. Laptev [ 21 ] showed that it holds for the Riesz means of Dirichlet eigenvalues and Neumann eigenvalues that (1.7) \sum_{k:\,\lambda_{k}<\lambda}(\lambda-\lambda_{k})\leq\frac{2}{n+2}\frac{\omega(n)|\Omega|}{(2\pi)^{n}}\lambda^{1+\frac{n}{2}} and (1.8) \sum_{k:\,\gamma_{k}<\gamma}(\gamma-\gamma_{k})\geq\frac{2}{n+2}\frac{\omega(n)|\Omega|}{(2\pi)^{n}}\gamma^{1+\frac{n}{2}}. He then used these two inequalities to give a new proof of ( 1.2 ) and ( 1.6 ). Note that Laptev [ 21 ] used the Riesz means to verify Pólya’s conjecture on product domains \Omega_{1}\times\Omega_{2}\subset{\mathbb{R}}^{n_{1}+n_{2}} provided Pólya’s conjecture holds on \Omega_{1} as we recalled above. In recent works [ 8 , 9 ] , Frank and Larson have launched the study of two term expansion of the Riesz means for both Dirichlet/Neumann eigenvalues. In particular, asymptotic behavior have been established for Lipschitz domains by [ 8 , 9 ] . Very recently Frank and Larson [ 8 ] obtained improvements of ( 1.7 ) and ( 1.8 ) via the uncertainty principle, as (1.9) \sum_{k:\,\lambda_{k}<\lambda}(\lambda-\lambda_{k})\leq\frac{2}{n+2}\frac{\omega(n)|\Omega|}{(2\pi)^{n}}\lambda^{1+\frac{n}{2}}\left(1-ce^{-c^{\prime}(\Omega)\sqrt{\lambda}}\right) and (1.10) \sum_{k:\,\gamma_{k}<\gamma}(\gamma-\gamma_{k})\geq\frac{2}{n+2}\frac{\omega(n)|\Omega|}{(2\pi)^{n}}\gamma^{1+\frac{n}{2}}\left(1+ce^{-c^{\prime}(\Omega)\sqrt{\lambda}}\right), where the constants c,c^{\prime} are implicit. For a given bounded open domain \Omega , let (1.11) \displaystyle J_{1}:=\frac{|\Omega|^{1/2}}{2|I(\Omega)|^{1/2}}, and R_{\Omega}>0 be such that |\Omega|=|B(0,R_{\Omega})| . Note that \displaystyle\frac{n}{n+2}|\Omega|R_{\Omega}^{2}\leq I(\Omega)=\inf_{a\in{{{\mathbb{R}}}^{n}}}\int_{\Omega}|x-a|^{2}\,dx\leq\frac{1}{4}(\mathrm{diam}(\Omega))^{2}|\Omega|, the quantity J_{1} has lower and upper bounds as (1.12) \displaystyle\frac{1}{\mathrm{diam}(\Omega)}\leq J_{1}\leq\sqrt{\frac{n+2}{n}}\frac{1}{2R_{\Omega}}. Adopting the elegant method of using the moment of inertia by Males [ 25 ] , we obtain the following estimates for the Riesz means of Dirichlet eigenvalues. Theorem 1.1 . Let \Omega be a bounded open set in {{{\mathbb{R}}}^{n}} , n\geq 2 . Then it holds for \lambda>0 that (1.13) \displaystyle\sum_{k:\,\lambda_{k}<\lambda}(\lambda-\lambda_{k}) \displaystyle\leq\frac{2}{n+2}\lambda^{1+\frac{n}{2}}\frac{|\Omega||\omega(n)|}{(2\pi)^{n}}\left(1-C_{1}(n)\lambda^{-1}|J_{1}|^{2}\right)_{+}, where a_{+}=\max\{0,a\} and (1.14) \displaystyle C_{1}(n)=\begin{cases}\frac{157}{480},\ &n=2,\\ \frac{5n(n+2)}{96}\left(1-\frac{\sqrt{n+2}}{8\sqrt{n}\Gamma(1+\frac{n}{2}))^{2/n}}\right)^{n-1}\left(2-\frac{\sqrt{n+2}}{8\sqrt{n}\Gamma(1+\frac{n}{2}))^{2/n}}\right),\ &n\geq 3.\end{cases} Theorem 1.1 will follow from a sharper estimate below (Theorem 2.4 ). With the aid of Stirling’s formula we see that in C_{1}(n) the factor involving Gamma function have a uniformly positive lower bound, about 1.014, for all n\geq 3 , see Remark 2.6 below. Theorem 1.1 also implies an eigenvalue estimate (cf. Corollary 2.7 below), which shows that the constant C_{1}(n) improves on previous version of Melas’ improvement ( 1.3 ). Recall that Weyl’s conjecture on sharper asymptotic behavior of Dirichlet eigenvalue states for domains with piecewise smooth boundary that (1.15) \displaystyle\mathcal{N}^{D}_{\Omega}(\lambda)=\frac{|\Omega|\omega(n)}{(2\pi)^{n}}\lambda^{n/2}-\frac{\mathcal{H}^{n-1}(\partial\Omega)}{2^{n+1}\pi^{\frac{d-1}{2}}\Gamma(\frac{n+1}{2})}\lambda^{\frac{n-1}{2}}+o(\lambda^{\frac{n-1}{2}}), where \mathcal{N}^{D}_{\Omega}(\lambda) is the counting function \mathcal{N}^{D}_{\Omega}(\lambda):=\#\{\lambda_{k}(\Omega):\,\lambda_{k}(\Omega)<\lambda\}. The conjecture has been proved by Ivrii [ 14 ] under the condition that the set of all periodic geodesic billiards in \Omega has measure zero, and proved in [ 28 ] for convex analytic domains and polygons. From ( 1.15 ), one might hope to replace (1-C_{1}(n)\lambda^{-1}|J_{1}|^{2}) in R.H.S. of ( 1.13 ) by something like (1-C_{1}(n)\lambda^{-1/2}|J_{1}|) . However, since we didn’t assume any regularity of \partial\Omega , \mathcal{H}^{n-1}(\partial\Omega) might be infinite, Weyl’s sharper asymptotic formula cannot hold in such general settings, see however [ 19 , Theorem 2] for domains with certain smoothness on {\mathbb{R}}^{2} . Theorem 1.1 implies the following qualitative result. Corollary 1.2 . Let \Omega be a bounded open set in {{{\mathbb{R}}}^{n}} , n\geq 3 . Then there exist infinitely many eigenvalues \{\lambda_{k_{j}}\}_{j\in\mathbb{N}} satisfying Pólya’s conjecture, i.e., (1.16) \displaystyle\frac{|\Omega||\omega(n)|}{(2\pi)^{n}}\lambda_{k_{j}}^{\frac{n}{2}}\geq k_{j}. The corollary will follow from Theorem 2.8 below, where we in fact show that for a certain length