Existence of minimizers for Newtons's problem of the body of minimal resistance under a single impact assumption (Q5941782): Difference between revisions

In the paper the Newton's problem for the profile of minimal resistance \[ \min\Biggl\{ \int_\Omega{1\over 1+|\nabla u|^2} dx: 0\leq u\leq M,\;u\in X(\Omega)\Biggr\}\tag{1} \] is considered, \(\Omega\) being a strictly convex domain of \(\mathbb{R}^2\) and \(X(\Omega)\) a suitable class of locally Lipschitz functions on \(\Omega\). Here the attention is focused on the class of profiles verifying the so-called ``single impact assumption'' which turns out to require, on the admissible functions \(u\), the inequality \[ {u(x- \tau\nabla u(x))- u(x)\over \tau}\leq {1-|\nabla u(x)|^2\over 2} \] for every point \(x\) of differentiability of \(u\), and for every \(\tau> 0\) such that \(x-\tau\nabla u(x)\in \overline\Omega\). The authors show that for every \(\Omega\) and \(M\) the infimum of problem (1) is strictly positive; moreover, when \(\Omega\) is a ball and the competing functions \(u\) are supposed radially symmetric, the existence of an optimal solution is proved. Passing to general domains \(\Omega\) and dropping the radiality assumption, it is conjectured that the existence for problem (1) fails; this conjecture is supported by a result which proves that no minimizers for (1) can be ``regular on the boundary''.