Remarks on the continuity of functions of two variables (Q418216)

Beginning calculus students often naively assume that continuity of a function \(f\) of several variables is equivalent to the separate continuity in each variable, i.e., to the continuity of \(f\) along lines parallel to the coordinate axes. Of course, even continuity of \(f\) along all lines through a point does not suffice to deduce continuity of \(f\) at that point, as illustrated by the well-known example \(f(x,y) = x^2y/(x^4+y^2)\), \((x,y) \neq (0,0)\), \(f(0,0)=0\). It is natural to ask: on how rich a collection of curves through a given point \((x_0,y_0)\) must we impose the continuity of \(f\) in order to ensure that \(f\) is continuous at \((x_0,y_0)\)? Lebesgue observed that even the class of analytic curves was too small to draw this conclusion. The following striking results were established by \textit{A. Rosenthal} [Math. Z. 63, 31--38 (1955; Zbl 0064.30003)]: (i) a real-valued function \(f\) defined on a domain \(\Omega\) in \({\mathbb R}^2\) is continuous if and only if its restriction to every convex differentiable curve is continuous, however, (ii) there exist discontinuous functions \(f\) which are continuous when restricted to every twice differentiable curve. In the current paper, the authors give an example of a function \(f: \Omega \to {\mathbb R}\) which is discontinuous at a point \(P\) yet continuous when restricted to every differentiable curve with bounded derivative passing through \(P\). The curves in question are understood to be given as graphs of real-valued functions with respect to a suitable coordinate system in the plane. The authors also give a necessary and sufficient condition for continuity of a function \(f\) at a point \(P\) in terms of continuity along all parametric curves through \(P\).