Optimal function spaces for continuity of the Hessian determinant as a distribution (Q2355444)

The authors show that the Hessian determinant is a continuous mapping from the Besov space \(B(2-\frac{2}{N},N)\) of scalar-valued functions on \({\mathbb R}^N\), \(N\geq 3\), into the space of distributions. Consequently, the Hessian determinant is continuous on every Besov space \(B(s,p)\) satisfying \(B(s,p)\subset B_{\text{loc}}(2-\frac{2}{N},N)\). The result is inspired by a recent result due to \textit{H. Brezis} and \textit{H.-M. Nguyen} [Invent. Math. 185, No. 1, 17--54 (2011; Zbl 1230.46029)] on the Jacobian determinant. Moreover, the continuity fails for any Besov space for which the previous inclusion does not hold. This is the most cumbersome part of the paper and it covers the Sections 3, 4 and 5. For instance, the entire Section 5 is dedicated to construct a counterexample for continuity in \(B(2-\frac{2}{N},p)\) with \(p > N\).