[논문 리뷰] Comparison results for the $p$-torsional rigidity on convex domains

Cristian Enache, Mihai Mihăilescu외 1명|arXiv (Cornell University)|2026. 03. 13.

Nonlinear Partial Differential Equations인용 수 0

한 줄 요약

논문은 다면적 도형에서 정규화된 p-비틀림 강성 T(p;Ω)를 차원의 의존성과 반지름 의존성에 대한 예리한 비교 원리를 제시하고, 서로 다른 내접원 반지름을 갖는 볼록 도메인들 간의 보편 상수 gamma_{D,p}를 통해 T(p;Ω_b) ≤ T(p;Ω_a)가 성립하는 조건을 규명합니다.

ABSTRACT

For each open, bounded and convex domain $Ω\subset \mathbb{R}^{D},$ $D\geq 2$, and each real number $p>1,$ we denote by $u_{p}$ the $p$\emph{-torsion function} on $Ω$, i.e. the solution of the \emph{torsional creep problem} $Δ_{p}u=-1$ in $Ω$, $u=0$ on $\partial Ω$, where $Δ_{p}u:=\operatorname{div}( \left\vert abla u ight\vert ^{p-2} abla u) $ is the $p$-Laplacian. Let $T_p(Ω)$ be the $p$\emph{-torsional rigidity} on $Ω$, defined as $T_{p}\left( Ω ight) :=\int_{Ω}u_{p}dx$. Define $T\left( p;Ω ight) :=\left\vert Ω ight\vert ^{p-1}T_{p}\left( Ω ight) ^{1-p}$, where $|Ω|$ stands for the Lebesgue measure of $Ω$. The main purpose of this paper is to compare the values of $T(p;Ω)$ for bounded convex domains having different inradii. We prove that for any $0

연구 동기 및 목표

• Bounded convex 도메인들 사이에서 내접 반지름에 따른 정규화된 p-비틀림 강성 T(p;Ω)의 변화에 대한 연구 동기 부여.
• p-토션 문제와 정규화된 기능적 T(p;Ω) 정의를 통해 규모 불변 비교를 가능하게 함.
• 내접 반지름 a<b를 주었을 때, 차원 및 p에 의존하는 예리한 임계치(gamma_{D,p})를 규정하여 T(p;Ω_b) ≤ T(p;Ω_a)가 성립하는지 판단.
• 모델 계열(직사각형, 직교정육면체, 타원, 삼각형)을 이용한 p의 점근적 구간(p→1^+ 및 p→∞)의 분석 및 예리성 논의.
• 기하 제약 하에서 거리 기반 함수들과 Saint-Venant 형의 결과로의 확장 가능성 탐구

제안 방법

• Introduce the p-torsion problem -Δ_p u = 1 in Ω, u = 0 on ∂Ω, with T_p(Ω) = ∫_Ω u_p dx and T(p;Ω) = |Ω|^{p-1} T_p(Ω)^{1-p}.
• Use the variational characterization T_p(Ω)^{p-1} = sup_{u∈W_0^{1,p}(Ω) eq0} (∫_Ω |u| dx)^p / ∫_Ω |∇u|^p dx.
• Develop a scale-invariant quantity Q_p(Ω) = (T(p;Ω) R_Ω^p / ((2p-1)/(p-1))^{p-1})^{1/p} and define α, β as its infimum/supremum over convex domains, leading to γ_{D,p} = α/β.
• Prove the main Theorem 2: T(p;Ω_b) ≤ T(p;Ω_a) for all Ω_a∈P^D(a), Ω_b∈P^D(b) iff γ_{D,p} b ≥ a, with equality asymptotically when γ_{D,p} b = a.
• Employ scaling T(p; tΩ) = t^{-p} T(p;Ω) and derive bounds using sharp Hersch-Protter and Buser-type inequalities, plus geometric estimates R_Ω P(Ω)/|Ω|.
• Section 5 computes Q_p on model families (rectangles, orthotopes, ellipses, triangles) to illustrate sharpness and asymptotic behavior.

실험 결과

연구 질문

• RQ1 Does a universal, dimension- and p-dependent criterion exist that orders p-torsional rigidity across convex domains with different inradii?
• RQ2 What is the precise relation between inradius, domain geometry, and the scale-invariant T(p;Ω) that yields T(p;Ω_b) ≤ T(p;Ω_a) when a<b?
• RQ3 How do asymptotic regimes p→1^+ and p→∞ affect the comparison, and are the bounds sharp in model domains?
• RQ4 Can the comparison principle be extended via distance-based functionals (δ(Ω)) and to Saint-Venant type results under geometric constraints?
• RQ5 What is the behavior of Q_p and related constants on key convex families (rectangles, orthotopes, ellipses, triangles) and how do these illustrate sharpness of the bounds?

주요 결과

• There exists γ_{D,p} ∈ [1/D, 1) depending only on D and p such that T(p;Ω_b) ≤ T(p;Ω_a) for all convex Ω_a with inradius a and Ω_b with inradius b if and only if γ_{D,p} b ≥ a.
• If γ_{D,p} b > a, the inequality is strict; if γ_{D,p} b = a, equality is attained asymptotically via degenerating sequences of domains.
• Corollaries extend the result to a single γ_D, showing a threshold relationship across all p>1 when γ_D b ≥ a ensures T(p;Ω_b) ≤ T(p;Ω_a) for every p, with strictness when γ_D b > a.
• Theorem 3 extends the result using an average-distance framework δ(Ω) and yields a similar γ̄_{D,p} with bounds [2/(D(D+1)), 1).
• Model-family analysis shows sharpness of γ_{D,p} for rectangles, orthotopes, ellipses, and triangles; extremal values are approached via degenerating degenerations (e.g., elongated domains or collapsing polytopes).
• As p→∞, the problem aligns with distance-to-boundary behavior, and the average-distance framework provides extended comparison results (Theorem 3 and Corollaries).