Viscosity Solutions in Martinet Spaces
Thomas Bieske, Frederic Bowen · 2026
Read Viscosity Solutions in Martinet Spaces by Thomas Bieske, Frederic Bowen (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-2601-19864.txt.
Introduction
Thomas Bieske and Frederic Bowen study a kind of solution to partial differential equations in an unusual geometry. Martinet spaces lack the algebraic group law of Carnot groups. They also lack the triangular vector fields of Grushin-type spaces. That makes standard methods hard to apply.
The authors establish the basic properties of viscosity solutions in these spaces. Viscosity solutions are a weak notion designed for equations with no smooth solutions. They then prove uniqueness for strictly monotone elliptic equations and for the infinite Laplace equation. Uniqueness matters because a model with several right answers predicts nothing. The work extends a workhorse theory to a setting that had resisted it. The proofs are strict and step by step. Each claim rests on the last.
Catalog record
| CREATOR | Thomas Bieske, Frederic Bowen |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-19864
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Viscosity Solutions in Martinet Spaces Thomas Bieske, Frederic Bowen Abstract . In this paper, we establish the properties of viscosity solutions in Martinet spaces, which lack both the algebraic group law of Carnot groups and the triangular vector fields of Grushin-type spaces. We then prove the uniqueness of viscosity solutions to strictly monotone elliptic PDEs and to the infinite Laplace equation. 1991 Mathematics Subject Classification Primary: 53C17, 35D40, 35H20 1. introduction Properties of viscosity solutions in sub-Riemannian spaces have been a topic of recent study. In particular, the existence-uniqueness of viscosity solutions to the p -Laplace equation for 1<p\leq\infty in Carnot groups was established by exploiting the (non-abelian) algebraic group law [ 4 ] and the triangular structure of the vector fields in Grushin-type spaces enabled existence-uniqueness of viscosity solutions to the infinite Laplace and \infty(x) -Laplace equations to be proved. [ 3 , 5 ] . In [ 6 ] , existence of viscosity solutions to the infinite Laplace equation in general sub-Riemannian spaces was proved, leading to the question of uniqueness especially in spaces that lack an algebraic group law or lack the vector field structure of Grushin-type spaces. In this paper, we establish uniqueness of viscosity solutions to the infinite Laplace equation in such a space, namely, Martinet spaces. This paper is part of an undergraduate research project of the second author under the guidance of the first author. The second author wishes to thank the University of South Florida for the opportunities and support. 2. Martinet Space We begin with \mathbb{R}^{3} and coordinates (x_{1},x_{2},x_{3}) . Consider the vector fields \displaystyle X_{1} \displaystyle= \displaystyle\frac{\partial}{\partial x_{1}} \displaystyle\textmd{and\ \ }X_{2} \displaystyle= \displaystyle\frac{\partial}{\partial x_{2}}+f(x_{1})\frac{\partial}{\partial x_{3}} where f:\mathbb{R}\to\mathbb{R} is real-analytic. That is, it has a nontrivial Taylor series over \mathbb{R} . The space \mathbb{R}^{3} with these vector fields, denoted \mathfrak{m} , is the tangent space to the Martinet space \mathbb{M} with coordinates (x_{1},x_{2},x_{3}) . Now, \mathbb{M} need not be a group because [X_{1},X_{2}]=f^{\prime}(x_{1})\frac{\partial}{\partial x_{3}}=\vec{0} when f^{\prime}(x_{1})=0 and [X_{1},X_{2}]=f^{\prime}(x_{1})\frac{\partial}{\partial x_{3}}\neq\vec{0} when f^{\prime}(x_{1})\neq 0 , meaning the Lie algebra grading can depend on the point (x_{1},x_{2},x_{3}) , such as the case when f^{\prime}(x_{1})\not\equiv 0 but f^{\prime}(x_{1}) has zeros. Note that when f(x_{1})=x_{1} , we have a presentation of the Heisenberg algebra. We shall denote the vector [X_{1},X_{2}] by X_{3} . Because we will assume the zeros of f(x_{1}) have finite order, there is an r_{x_{1}}\in\mathbb{N} such that the Lie bracket iteration of minimal length r_{x_{1}} satisfies the condition [X_{1},[X_{1},[...,[X_{1},X_{2}]...]]\neq 0 . Then by [ 1 , Theorem 7.34] , d(p,q)\approx|x-x_{0}|+|y-y_{0}|+|z-z_{0}|^{(r_{x_{1}}+1)^{-1}}. (2.1) Even though \mathbb{M} need not be a group, it is a metric space whose natural metric is the Carnot-Carathéodory distance, which is defined for the points p and q as follows: d(p,q)=\inf_{\Gamma}\int_{0}^{1}\|\gamma^{\prime}(t)\|dt, where \|\cdot\| is a norm that makes \{X_{1},X_{2}\} an orthonormal set and where \Gamma is the set of all curves \gamma joining p and q with \gamma(0)=p , \gamma(1)=q , and \gamma^{\prime}(t)\in\operatorname{span}\{X_{1},X_{2}\} . By the Chow-Rashevskii theorem (see e.g. [ 1 ] ), any two points can be connected by such a curve, which means d(p,q) is an honest metric. Using this metric, we can define a Carnot-Carathéodory ball of radius r centered at a point p_{0} by B=B(p_{0},r)=\{p\in\mathbb{M}:d(p,p_{0})<r\}; similarly, we shall denote a bounded domain in \mathbb{M} by \Omega . 3. Calculus Given a smooth function u on \mathbb{M} , we define the horizontal gradient of u , denoted \nabla_{0}u by \nabla_{0}u(p)=(X_{1}u(p),X_{2}u(p)), the semi-horizontal gradient of u , denoted \nabla_{1}u by \nabla_{1}u(p)=(X_{1}u(p),X_{2}u(p),X_{3}u(p)), and the symmetrized second-order (horizontal) derivative matrix, denoted (D^{2}u(p))^{\star} by ((D^{2}u(p))^{\star})_{ij}=\frac{1}{2}(X_{i}X_{j}u(p)+X_{j}X_{i}u(p)) for i,j=1,2 . Definition 1 . The function u:\mathbb{M}\to\mathbb{R} is said to be C_{sub}^{1} if X_{1}u and X_{2}u are continuous. Similarly, the function u is C_{sub}^{2} if X_{3}u(p) is continuous and X_{i}X_{j}u(p) is continuous for all i,j=1,2 . It should also be noted that, for any open set \mathcal{O}\subset\mathbb{M} , the function u is in the horizontal Sobolev space W^{1,q}(\mathcal{O}) if u,X_{1}u,X_{2}u are in L^{q}(\mathcal{O}) . Replacing L^{q}(\mathcal{O}) by L^{q}_{loc}(\mathcal{O}) , the space W^{1,q}_{loc}(\mathcal{O}) is defined similarly. The space W^{1,q}_{0}(\mathcal{O}) is the closure in W^{1,q}(\mathcal{O}) of smooth functions with compact support. Locally Lipschitz functions are those functions u such that \|\nabla_{0}u\|_{L^{\infty}_{loc}}<\infty. 4. Martinet Jets and Viscosity Solutions Using the derivatives of the previous section, the class of equations we consider are given by F(p,u(p),\nabla_{1}u(p),(D^{2}u(p))^{\star})=0 (4.1) where the continuous function F:\mathbb{M}\times\mathbb{R}\times\mathfrak{m}\times S^{2}\to\mathbb{R} satisfies F(p,r,\eta,X)\leq F(p,s,\eta,Y). when r\leq s and Y\leq X . (That is, F is proper [ 7 ] ). Recall that S^{2} is the set of 2\times 2 real symmetric matrices. An example of this type of equation is the quasilinear horizontal q-Laplacian \operatorname{div}(\|\nabla_{0}u\|^{q-2}\nabla_{0}u)=X_{1}(\|\nabla_{0}u\|^{q-2}X_{1}u)+X_{2}(\|\nabla_{0}u\|^{q-2}X_{2}u) for 2\leq q<\infty . Formally taking the limit as q\to\infty yields the horizontal infinite Laplacian \Delta_{\infty}f=\sum_{i,j=1}^{2}X_{i}fX_{j}fX_{i}X_{j}f=\langle(D^{2}f)^{\star}\nabla_{0}f,\nabla_{0}f\rangle. We now wish to define solutions to the equation F(p,u(p),\nabla_{1}u(p),(D^{2}u(p))^{\star})=0 in the viscosity sense. In order to do so, we must define the subelliptic jets. (For a thorough discussion of jets, the interested reader is directed to [ 7 ] .) Given an open set \mathcal{O}\subset\mathbb{M} , a point p_{0}\in\mathcal{O} , and a function u:\mathcal{O}\to\mathbb{R} , we consider classes of test functions which “touch” the function u at the point p_{0} . Namely, we have the so-called “touching above” functions \mathcal{TA}(u,p_{0})=\{\varphi\in C_{sub}^{2}(\Omega):0=\varphi(p_{0})-u(p_{0})<\varphi(p)-u(p)\>\text{near}\>p_{0}\}; we have also the “touching below” functions at p_{0} defined by \mathcal{TB}(u,p_{0})=\{\varphi\in C_{sub}^{2}(\Omega):0=u(p_{0})-\psi(p_{0})<u(p)-\psi(p)\>\text{near}\>p_{0}\}. We then use these test functions to define Martinet jets. Definition 2 . Given u:\Omega\subset\mathbb{M}\to\mathbb{R} , we define the second-order superjet for u by J^{2,+}u(p_{0})=\{(\nabla_{1}\varphi(p_{0}),(D^{2}\varphi)^{\star}(p_{0}))\in\mathfrak{m}\times S^{2}:\varphi\in\mathcal{TA}(u,p_{0})\} and the second-order subjet for u by J^{2,-}u(p_{0})=\{(\nabla_{1}\psi(p_{0}),(D^{2}\psi)^{\star}(p_{0}))\in\mathfrak{m}\times S^{2}:\psi\in\mathcal{TB}(u,p_{0})\} It is easy to see from the definition that J^{2,-}u(p_{0})=-J^{2,+}(-u)(p_{0}) . In addition, we say that the ordered pair (\eta,X)\in\mathfrak{m}\times S^{2} belongs to the closure of the superjet, written (\eta,X)\in\overline{J}^{2,+}u(p_{0}) , if there exists a sequence of points \{p_{k}\}\subset\Omega and jet entries (\eta_{k},X_{k})\in J^{2,+}u(p_{k}) so that (p_{k},u(p_{k}),\eta_{k},X_{k})\to(p_{0},u(p_{0}),\eta,X). Definition 3 . Let \mathcal{O} be a open set in \mathbb{M} and let u:\mathcal{O}\to\mathbb{R} . If u is upper semicontinuous and F(p,u(p),\eta,X)\leq 0\quad\text{for all}\quad p\in\mathcal{O},\quad\text{for all}\quad(\eta,X)\in\overline{J}^{2,+}_{\mathcal{O}}u(p) then u is a viscosity subsolution of F(p,u(p),\nabla_{1}u(p),(D^{2}u(p))^{\star})=0. If u is lower semicontinuous and F(p,u(p),\eta,X)\geq 0\quad\text{for all}\quad p\in\mathcal{O},\quad\text{for all}\quad(\eta,X)\in\overline{J}^{2,-}_{\mathcal{O}}u(p) then u is a viscosity supersolution of F(p,u(p),\nabla_{1}u(p),(D^{2}u(p))^{\star})=0. The function u is a viscosity solution if it is both a viscosity subsolution and a viscosity supersolution. In order to use the machinery of [ 7 ] to prove comparison principles, a relationship between Euclidean and Martinet jets must be established. This is accomplished through the following lemma, whose proof is a special case of [ 4 , Corollary 3.2] and omitted. Lemma 4.1 (Twisting Lemma) . Let O\subset\mathbb{M} be open, let u:O\to\mathbb{R} , and let p_{0}=(x_{1}^{0},x_{2}^{0},x_{3}^{0})\in O . Suppose that (\eta,X)\in J^{2,+}_{\textmd{eucl}}u(p_{0}) with \eta=(\eta_{1},\eta_{2},\eta_{3}) and X=\{X_{ij}\} a 3\times 3 symmetric matrix. Then (B(p_{0})\cdot\eta,A(p_{0})\cdot X\cdot A^{T}(p_{0})+\mathcal{T}(\eta,p_{0}))\in J^{2,+}u(p_{0}) (4.2) where A(p_{0}) is the 2\times 3 matrix of Martinet vector coefficients defined by A(p_{0})=\begin{pmatrix}1&0&0\\ 0&1&f(x_{1}^{0})\end{pmatrix}_{,} the 3\times 3 matrix B(p_{0}) is given by B(p_{0})=A(p_{0})\oplus(0,0,f^{\prime}(x_{1}^{0})=\begin{pmatrix}1&0&0\\ 0&1&f(x_{1}^{0})\\ 0&0&f^{\prime}(x_{1}^{0})\end{pmatrix}_{,} and the 2\times 2 twisting matrix \mathcal{T}(\eta,p_{0}) is given by \mathcal{T}(\eta,p_{0})=\frac{\eta_{3}}{2}\begin{pmatrix}0&f^{\prime}(x_{1}^{0})\\ f^{\prime}(x_{1}^{0})&0\end{pmatrix}_{.} More specifically, B(p_{0})\cdot\eta=\begin{pmatrix}\eta_{1}\\ \eta_{2}+f(x_{1}^{0})\eta_{3}\\ f^{\prime}(x_{1}^{0})\eta_{3}\end{pmatrix} and \displaystyle A(p_{0})\cdot X\cdot A^{T}(p_{0})+\mathcal{T}(\eta,p_{0})= \displaystyle\begin{pmatrix}x_{11}&x_{12}+f(x_{1}^{0})x_{13}+\frac{1}{2}f^{\prime}(x_{1}^{0})\eta_{3}\\ \mbox{}&\mbox{}\\ x_{12}+f(x_{1}^{0})x_{13}+\frac{1}{2}f^{\prime}(x_{1}^{0})\eta_{3}&x_{22}+2f(x_{1}^{0})x_{23}+f(x_{1}^{0})^{2}x_{33}\end{pmatrix}_{.} 5. Maximum Principle We begin by stating a lemma analogous to [ 7 , Lemma 3.1] . The proof is similar and thus is omitted. Remark 5.1 . We wish to emphasize that in Lemma 5.2 below, the penalty function \varphi(p,q) not only differs from