The sets of quasi-uniform density are not a topology (Q2467957)

For a Lebesgue measurable subset \(A\) of \(\mathbb{R}\) the set \(\varphi(A)\) of density points of \(A\) can be characterized as the set of points \(x\in\mathbb{R}\) such that \(\chi_{n(A-x)\cap[-1,1]}\) converges in measure to \(\chi_{[-1,1]}\). Replacing convergence in measure by quasi-uniform convergence one obtains the author's definition of the set \(\Phi_{qu}(A)\) of quasi-uniform density points of \(A\). The author proves that in the definition of \(\Phi_{qu}(A)\) one can replace quasi-uniform convergence neither by uniform convergence nor by pointwise convergence. It is known that \(\varphi\) has the following properties: (1) \(A\sim\varphi(A)\) (i.e. the symmetric difference of \(A\) and \(\varphi(A)\) is a null set); (2) \(A\sim B\) implies \(\varphi(A) =\varphi(B)\); (3) \(\varphi(A\cap B)=\varphi(A)\cap\varphi(B)\); (4) the measurable subsets \(A\) of \(\mathbb{R}\) with \(A\subset\varphi(A)\) form a topology. The author proves that analogous statements for \(\Phi_{qu}\) are not true.