Ultracontractivity of Heat semigroups in L 2 ( Ω ) \mathrm{L}^{2}\left(\Omega\right) with non-local Robin boundary conditions using Nash’s inequality
Christoph Schwerdt · 2026
Read Ultracontractivity of Heat semigroups in L 2 ( Ω ) \mathrm{L}^{2}\left(\Omega\right) with non-local Robin boundary conditions using Nash’s inequality by Christoph Schwerdt (2026) 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-2605-13413.txt.
Introduction
Christoph Schwerdt studies heat flow on bounded Lipschitz domains, regions with mildly rough edges, under generalised Robin edge conditions. The heat operator is second-order and uniformly elliptic, so it spreads heat in a regular way. His edge conditions allow an operator that can break the rule that solutions stay non-negative. Using Nash's inequality on the Sobolev space H1, he proves ultracontractivity of the solution semigroup, a strong smoothing property that bounds later values by early totals.
Catalog record
| CREATOR | Christoph Schwerdt |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49962 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.13413
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Ultracontractivity of Heat semigroups in L 2 ( Ω ) \mathrm{L}^{2}\left(\Omega\right) with non-local Robin boundary conditions using Nash’s inequality Christoph Schwerdt Abstract We study heat equations ∂ u ∂ t − div ( A ∇ u ) = 0 \frac{\partial u}{\partial t}-\operatorname{div}\left(A\nabla u\right)=0 on bounded Lipschitz domains Ω \Omega in ℝ d \mathbb{R}^{d} for d > 2 d>2 , where − div ( A ∇ ⋅ ) -\operatorname{div}\left(A\nabla\cdot\right) is a second-order uniformly elliptic operator with generalised Robin boundary conditions. These boundary conditions are formally given by ν ⋅ A ∇ u + B u = 0 \nu\cdot A\nabla u+Bu=0 where ν \nu is the outer unit normal on ∂ Ω \partial\Omega and B ∈ ℒ ( L 2 ( ∂ Ω ) ) B\in\mathcal{L}\left(\mathrm{L}^{2}\left(\partial\Omega\right)\right) is a general operator which is allowed to destroy the positivity preserving property of the solution semigroup. Ultracontractivity of the solution semigroup is shown by using Nash’s inequality on the Sobolev space H 1 ( Ω ) H^{1}(\Omega) . 1 Introduction Let \Omega\subset\mathbb{R}^{d} for d\in\mathbb{N} be a bounded Lipschitz domain and A\colon\Omega\to\mathbb{R}^{d\times d} be a matrix-valued function such that every coefficient a_{ij}\colon\Omega\to\mathbb{R} is a bounded and measurable function on \Omega . Furthermore let A be uniformly elliptic on \Omega , i.e. \exists\alpha>0\ \forall\xi\in\mathbb{C}^{d},x\in\Omega\ :\ \Re\left(\left(A(x)\xi\right)^{T}\overline{\xi}\right)\geq\alpha\left|\xi\right|^{2}. We study solutions u=u(t,x) to the second-order parabolic equation \frac{\partial u}{\partial t}-\operatorname{div}\left(A\nabla u\right)=0\ \text{ in }\ (0,\infty)\times\Omega (1) with a generalised Robin boundary condtion formally given by \nu\cdot A\nabla u+B\gamma(u)=0\ \text{ on }\ (0,\infty)\times\partial\Omega, (2) where \nu is the outer unit normal on \partial\Omega and B is a linear operator on \mathrm{L}^{2}\left(\partial\Omega\right) . In this article a(B) always is a sesqulinear form in \mathrm{L}^{2}\left(\Omega\right) formally given by a(B)(u,v)=\int_{\Omega}A\nabla u\cdot\overline{\nabla v}\,\mathrm{d}x+\int_{\partial\Omega}B\gamma(u)\ \overline{\gamma(v)}\,\mathrm{d}\sigma(x) (3) for u,v\in D\left(a(B)\right)=H^{1}(\Omega) where \gamma is a trace operator mapping H^{1}(\Omega) into \mathrm{L}^{2}\left(\partial\Omega\right) . Let L(B) be the associated operator to a(B) such that -L(B) generates a C_{0} -semigroup \left(\mathrm{e}^{-tL(B)}\right)_{t\geq 0} in \mathrm{L}^{2}\left(\Omega\right) . Then u(t)=\mathrm{e}^{-tL(B)}u_{0} is a solution to ( 1 ) taking ( 2 ) into account in the sense of Subsection 2.4 in the article [ 1 ] by Jochen Glück and Jonathan Mui. For details about previous articles on the topic, more information how to model heat equations with non-local boundary conditions as well as an implication of an eventual positivity of the solution semigroup please see Section 1 of [ 1 ] . Here we focus on one of their main results in form of Theorem 3.2 on page 10 which reads: Theorem 1.1 . Let \Omega and A be as described above. Furthermore let B\in\mathcal{L}\left(\mathrm{L}^{2}\left(\partial\Omega\right)\right) act boundedly on \mathrm{L}^{1}\left(\partial\Omega\right) and on \mathrm{L}^{\infty}\left(\partial\Omega\right) . Then the semigroup \left(\mathrm{e}^{-tL(B)}\right)_{t\geq 0} is ultracontractive in \mathrm{L}^{2}\left(\Omega\right) , i.e. \exists\ C,\mu>0\ \forall u\in\mathrm{L}^{2}\left(\Omega\right)\ :\ \left\|\mathrm{e}^{-tL(B)}u\right\|_{\mathrm{L}^{\infty}\left(\Omega\right)}\leq Ct^{-\frac{\mu}{4}}\left\|u\right\|_{\mathrm{L}^{2}\left(\Omega\right)} is true for any time t\in(0,1] Remark 1.2 . i) Note that B is defined on \mathrm{L}^{\infty}\left(\partial\Omega\right) since \mathrm{L}^{\infty}\left(\partial\Omega\right)\subset\mathrm{L}^{2}\left(\partial\Omega\right) is implied by \Omega being a bounded Lipschitz domain. Demanding B to act boundedly on \mathrm{L}^{\infty}\left(\partial\Omega\right) means that Bu is contained in the subspace \mathrm{L}^{\infty}\left(\partial\Omega\right)\subset\mathrm{L}^{2}\left(\partial\Omega\right) for every u\in\mathrm{L}^{\infty}\left(\partial\Omega\right) and furthermore that there exists a constant C>0 such that \left\|Bu\right\|_{\mathrm{L}^{\infty}\left(\partial\Omega\right)}\ \leq\ C\left\|u\right\|_{\mathrm{L}^{\infty}\left(\partial\Omega\right)} is true for every u\in\mathrm{L}^{\infty}\left(\partial\Omega\right) . ii) Furthermore \mathrm{L}^{2}\left(\partial\Omega\right) is contained in \mathrm{L}^{1}\left(\partial\Omega\right) since \Omega is a bounded Lipschitz domain. Hence Bu is contained in \mathrm{L}^{1}\left(\partial\Omega\right) for every u\in\mathrm{L}^{2}\left(\partial\Omega\right) . Demanding B to act boundedly on \mathrm{L}^{1}\left(\partial\Omega\right) means that there exists a constant \tilde{C}>0 such that \left\|Bu\right\|_{\mathrm{L}^{1}\left(\partial\Omega\right)}\ \leq\ \tilde{C}\left\|u\right\|_{\mathrm{L}^{1}\left(\partial\Omega\right)} is true for every u\in\mathrm{L}^{2}\left(\partial\Omega\right) . Then there exists a unique extension of B in \mathcal{L}\left(\mathrm{L}^{1}\left(\partial\Omega\right)\right) since \mathrm{L}^{2}\left(\partial\Omega\right) is a dense subset in \mathrm{L}^{1}\left(\partial\Omega\right) . The main argument in [ 1 ] is to gradually construct an operator B_{2}\in\mathcal{L}\left(\mathrm{L}^{2}\left(\partial\Omega\right)\right) such that -B_{2} is positive and satisfies \left|Bu\right|\ \leq\ -B_{2}|u| pointwise almost everywhere on \partial\Omega for every u\in\mathrm{L}^{2}\left(\partial\Omega\right) . Furthermore -L(B_{2}) is the generator of another C_{0} -semigroup \left(\mathrm{e}^{-tL(B_{2})}\right)_{t\geq 0} such that \mathrm{e}^{-tL(B_{2})} dominates \mathrm{e}^{-tL(B)} at every time t>0 , i.e. \left|\mathrm{e}^{-tL(B)}u\right|\ \leq\ \mathrm{e}^{-tL(B_{2})}|u| (4) is true almost everywhere in \Omega for every u\in\mathrm{L}^{2}\left(\Omega\right) . Please compare with Step 4 on page 13 in [ 1 ] . The construction of B_{2} allows to utilize the results of Proposition 3.4 and Lemma 3.5 both on page 11 to conclude that \mathrm{e}^{-tL(B_{2})} restricts to a bounded operator on \mathrm{L}^{\infty}\left(\Omega\right) which is crucial to prove the ultracontractivity of \mathrm{e}^{-tL(B)} in \mathrm{L}^{2}\left(\Omega\right) using ( 4 ). In this article we put more focus on the ellipticity of A on \Omega . Note that for B being a bounded in \mathrm{L}^{2}\left(\partial\Omega\right) the domain D\left(a(B)\right) is defined as the Sobolev space H^{1}\left(\Omega\right) in \mathrm{L}^{2}\left(\Omega\right) . Furthermore notice that if the adjoint operators \left(\mathrm{e}^{-tL(B)}\right)^{\ast}=\mathrm{e}^{-tL(B)^{\ast}} in \mathrm{L}^{2}\left(\Omega\right) are continuous from \mathrm{L}^{1}\left(\Omega\right) to \mathrm{L}^{2}\left(\Omega\right) for every time t>0 , then an ultracontractivity of the original semigroup \left(\mathrm{e}^{-tL(B)}\right)_{t\geq 0} in \mathrm{L}^{2}\left(\Omega\right) is implied. Nash’s inequality beautifully connects all 3 of the involved function spaces \mathrm{L}^{1}\left(\Omega\right) , \mathrm{L}^{2}\left(\Omega\right) and H^{1}\left(\Omega\right) . Using Nash’s inequality as an alternative argumentation offers the same result as Theorem 1.1 . It reads as follows: Theorem 1.3 . Again let \Omega and A be as above. Furthermore let B\in\mathcal{L}\left(\mathrm{L}^{2}\left(\partial\Omega\right)\right) . Suppose there exists a positive operator \overline{B}\in\mathcal{L}\left(\mathrm{L}^{2}\left(\partial\Omega\right)\right) that is a continuous operator in \mathrm{L}^{\infty}\left(\partial\Omega\right) and satisfies: i) \left|B\gamma(u)\right|\ \leq\ \overline{B}|\gamma(u)| pointwise almost everywhere in \partial\Omega for any u\in H^{1}\left(\Omega\right) and ii) 1+\left(\|\overline{B}\|_{\infty}+\|\overline{B}\|_{2}\right)\|\gamma\|_{H^{1}\to\mathrm{L}^{2}}^{2}\leq\alpha for \alpha being the ellipticity constant of A . Then \left(\mathrm{e}^{-tL(B)}\right)_{t\geq 0} is ultracontractive in \mathrm{L}^{2}\left(\Omega\right) . In particular there is a constant C>0 satisfying \left\|\mathrm{e}^{-tL(B)}u\right\|_{\mathrm{L}^{\infty}\left(\Omega\right)}\leq C\mathrm{e}^{t\alpha}t^{-\frac{d}{4}}\left\|u\right\|_{\mathrm{L}^{2}\left(\Omega\right)} for every time t>0 and any u\in\mathrm{L}^{2}\left(\Omega\right) . Remark 1.4 . i) Note that \|B\|_{2}\|\gamma\|_{H^{1}\to\mathrm{L}^{2}}^{2}\leq\|\overline{B}\|_{2}\|\gamma\|_{H^{1}\to\mathrm{L}^{2}}^{2} implies 1+\|B\|_{2}\|\gamma\|_{H^{1}\to\mathrm{L}^{2}}^{2}\leq\alpha . Then the form a(B) is accretive, continuous and closed. Hence a(B) meets all requirements of [ 2 ] which establishes the existence of an associated C_{0} -semigroup \left(\mathrm{e}^{-tL(B)}\right)_{t\geq 0} in \mathrm{L}^{2}\left(\Omega\right) as shown in Subsection 2.1 . Please also see Appendix A . ii) Similar to above 1+\left(\|\overline{B}\|_{\infty}+\|\overline{B}\|_{2}\right)\|\gamma\|_{H^{1}\to\mathrm{L}^{2}}^{2}\leq\alpha establishes a second, auxiliary C_{0} -semigroup \left(\mathrm{e}^{-tL(\|\overline{B}\|_{\infty}-\overline{B})}\right)_{t\geq 0} in \mathrm{L}^{2}\left(\Omega\right) which dominates \left(\mathrm{e}^{-tL(B)}\right)_{t\geq 0} , i.e. \left|\mathrm{e}^{-tL(B)}u\right|\ \leq\ \mathrm{e}^{-tL(\|\overline{B}\|_{\infty}-\overline{B})}\left|u\right| for every u\in\mathrm{L}^{2}\left(\Omega\right) . Furthermore every operator of this auxiliary operator semigroup is continuous in \mathrm{L}^{\infty}\left(\Omega\right) which implies \mathrm{e}^{-tL(B)^{\ast}} to be continuous in \mathrm{L}^{1}\left(\Omega\right) . This last property is crucial for our argumentation using Nash’s inequality on \left(\mathrm{e}^{-tL(B)}\right)_{t\geq 0} . iii) Note that for B\in\mathcal{L}\left(\mathrm{L}^{2}\left(\partial\Omega\right)\right) being positive and continuous in \mathrm{L}^{\infty}\left(\partial\Omega\right) with 1+\left(\|B\|_{\infty}+\|B\|_{2}\right)\|\gamma\|_{H^{1}\to\mathrm{L}^{2}}^{2}\leq\alpha we can directly choose \overline{B}=B . iv) If B\in\mathcal{L}\left(\mathrm{L}^{2}\left(\partial\Omega\right)\right) acts boundedly on \mathrm{L}^{1}\left(\partial\Omega\right) and on \mathrm{L}^{\infty}\left(\partial\Omega\right) , then the existence of such an operator \overline{B} with \left|B\gamma(u)\right|\ \leq\ \overline{B}|\gamma(u)| for every u\in\mathrm{L}^{2}\left(\Omega\right) is implied by Step 1 on page 12 in [ 1 ] . However the sesquilinear forms in [ 1 ] might not meet all the requirements of [ 2 ] . Finally let us give some motivation by refering to the second main result in [ 1 ] in form of Theorem 4.4 on page 18 which reads: Theorem 1.5 . Let \Omega and A be as above. Furthermore let B\in\mathcal{L}\left(\mathrm{L}^{2}\left(\partial\Omega\right)\right) act boundedly on \mathrm{L}^{1}\left(\partial\Omega\right) and on \mathrm{L}^{\infty}\left(\partial\Omega\right) . Furthermore let B+B^{\ast} be positive semi-definte and B1_{\partial\Omega}=0 . Then there exists t_{0}\geq 0 and a constant \delta>0 such that \mathrm{e}^{-tL_{B}}u\ \geq\ \delta\left(\int_{\Omega}u\,\mathrm{d}x\right) is true almost everywhere in \Omega for every t\geq t_{0} and every 0\leq u\in\mathrm{L}^{2}\left(\Omega\right) . In particular, \left(\mathrm{e}^{-tL_{B}}\right)_{t\geq 0} is uniformly eventually positive. We see on page 18 in [ 1 ] that