No label defined (Q4338410)

The author considers generalizations of derivatives. One of them has the form \[ D^2_x f(x_0,y_0): =\lim_{t\to 0,(x,y) \to(x_0,y_0)} \bigl[f(x+t,y) +f(x-t,y) -2f(x,y)\bigr] t^{-1}. \] It is known that \(\exists f_{xx} \Rightarrow \exists D^2_xf\), where \(f_{xx}\) is the usual second-order derivative. The inverse implication is no longer true. Let \[ Uf(x,y,z)= {z\over 2\pi} \iint_{\mathbb{R}^2} f(t,\tau) \bigl[(t-x)^2 +(\tau-y)^2 +z^2\bigr]^{-3/2} dt d\tau \] be the Poisson integral. The author proves the following relation: \[ \lim_{(x,y,z) \to(x_0,y_0,0)} \bigl(Uf(x,y,z) \bigr)_{xx} =D^2_x f(x_0,y_0), \] if \(D^2_x f(x_0,y_0)\) exists.