Surfaces containing two isotropic circles through each point (Q2243185)

The author investigates surfaces in \(\mathbb{R}^3\) that contain \(2\leq \lambda<\infty\) isotropic circles through almost each point. An \textit{isotropic circle} in \(\mathbb{R}^3\) with Cartesian coordinates \((x,y,z)\) is defined as either a parabola whose axis is vertical, or an ellipse whose orthogonal projection to the \(xy\)-plane is a circle (an axis is called \textit{vertical} if it is parallel to the \(z\)-axis). Theorem 1 states under some additional assumptions such a surface admits a parametrization of the form \[ \Phi(u,v)=\left(\frac{P(u,v)}{R(u,v)},\frac{Q(u,v)}{R(u,v)},\frac{Z(u,v)}{R^2(u,v)}\right) \] for some polynomials~\(P,Q,R,Z\in\mathbb{R}[u,v]\) such that \(P\), \(Q\), \(R\) have degree at most 1 in both \(u\) and \(v\), and \(Z\) has degree at most 2 in both \(u\) and \(v\). The additional assumption requires that the surface contains through almost each point two isotropic circles that are parabolas, and these parabolas should not lie in a plane, a paraboloid of revolution with vertical axis, or a circular cylinder with vertical axis. The parameter lines of \(\Phi(u,v)\) correspond to the parabolas. A corollary is that the orthogonal projection of the parabolas to the \(xy\)-plane are lines that are either tangent to one conic, or lie in a union of two pencils of lines. The author proceeds in Theorem 2 by providing a parametrization, where the isotropic circles contained in the surface are not necessarily parabolas. For this case, interesting examples are provided where the orthogonal projection of the isotropic circles to the \(xy\)-plane are all tangent to one curve that is cyclic. The article concludes with some challenging open problems.