Viscosity solutions on Grushin-type planes (Q1856427)

Improving his results \textit{T. Bieske} [Commun. Partial Differ. Equ. 27, 727-761 (2002; Zbl 1090.35063)] and employing the superjets and subjets mechanisms \textit{M. Crandall}, \textit{H. Ishii} and \textit{P. L. Lions} [Bull. Am. Math. Soc., New Ser. 27, 1-67 (1992; Zbl 0755.35015)], the author prove a maximum principle for viscosity solutions of a class of nonlinear differential equations under subelliptic environment on Grushin-type planes that is not a group. The paper is rather technical but all fundamental ideas are clearly explained. In more detail, the Grushin plane is determined by vector fields \(X_1= \partial/ \partial x\), \(X_2=\rho(x)\partial/ \partial y\), \(X_3=\rho'(x)\partial/ \partial y\), \(X_4=\rho''(x) \partial/ \partial y,\dots\), where \(\rho:\mathbb{R}\to\mathbb{R}\) is anonconstant polynomial. The plane is moreover equipped with inner product where \(X_1\), \(X_2\) are orthonormal and the relevant Carnot-Carathéodory metric \[ d(P,Q)= \inf_\gamma \int\bigl\|\gamma'(t) \bigr\|dt\bigl(\gamma'(t) \in\text{span}\{X_1,X_2\}\bigr). \] The class of equations considered is \(F(P,u(P),\nabla_0 u(P),(D^2u(p))^*)=0\) with continuous function \(F\) satisfying \(F(P,r,\eta,X)\leq F(P,s,\eta,Y)\) whenever \(r\leq s\) and \(Y\leq X\). Here \(\nabla_0f(P)=(X_1f(P),X_2f(P))\), \[ \bigl(D^2 f(P)\bigr)^* = \left(\begin{matrix} & X_1X_2f(P) & \tfrac 12(X_1X_2+X_2X_1)f(P)\\ & \tfrac 12(X_1X_2+ X_2X_1)f(P) & X_2X_2f(P) \end{matrix}\right). \] The definition of viscosity solution (in terms of subjets and superjets) cannot be stated here. The main result ensures the inequality \(u(P)\leq v(P)\) \((P\in\Omega)\) between upper semicontinuous subsolution \(u\) and lower semicontinuous supersolution \(v\) if \(\lim\sup u(Q)\leq\lim\inf v(Q)\) as \(Q\to R\in\partial\Omega\).