Let $f$ be a positive function over the unit sphere $S^{d-1}$. Minkowski's problem is to find a convex body $K$ in ${\mathbb R}^d$, whose Gauss curvature is prescribed as a function of the normal direction: $\kappa=f(\vec n)$. An "obvious" necessary condition is the vector-valued identity $$\int_{S^{d-1}}f(\omega)\omega ds(\omega)=0.$$ Pogorelov proved that this condition (plus some modest regularity) is actually sufficient. He presented the complete proof in a book published in 1978 by Wiley & Sons.

Is there a close formula for the volume of $K$ in terms of $f$ ? If not, are there nice estimates of this volume in terms of $f$ ?

  • 1
    $\begingroup$ This might help, but I have not penetrated enough to be confident that it is relevant. Alexandrov, Victor, Natalia Kopteva, and S. S. Kutateladze. "Blaschke addition and convex polyhedra." arXiv preprint math/0502345 (2005). $\endgroup$ – Joseph O'Rourke Jan 11 at 14:10
  • 1
    $\begingroup$ The function $f$ is the reciprocal of Gauss curvature, $f = \kappa^{-1}$. I do not know of any closed form formula for the volume of the solution $K$. The volume of $K$ is given by $$ V = \frac{1}{d}\int_{S^{d-1}} h(\omega)f(\omega)\,ds(\omega), $$ where $ds$ is the standard $(d-1)$-volume measure on $S^{d-1}$ and $h$ is the support function of $K$. $\endgroup$ – Deane Yang Jan 11 at 15:08
  • 1
    $\begingroup$ By the way, no regularity assumptions are needed at all. The integrand $f\,ds$ is well defined as a measure $dS$ on $S^{n-1}$ (called the surface area measure) for any convex body. Minkowski solved the Minkowski problem for polytopes, where the measure is a finite discrete measure. Alexandrov and, independently Fenchel-Jessen, extended Minkowski's solution to arbitrary Borel measures on the sphere such that the measure is not supported on a hypersphere and satisfies $$ \int_{S^{d-1}} \omega\,dS(\omega) = 0$$ $\endgroup$ – Deane Yang Jan 11 at 15:13
  • 2
    $\begingroup$ The function $f$ is the determinant of the Hessian of the support function $h$, and the volume can be computed from $h$ by the formula given in the first comment of Deane Yang. So, it is unlikely that there is an explicit formula for the volume in terms of $f$. $\endgroup$ – Ivan Izmestiev Jan 11 at 16:34
  • 1
    $\begingroup$ Also, in the discrete case the function $f$ gives the areas of facets (faces of codimension $1$) with prescribed directions of normals. One could compute the volume if one knew the heights of all facets. But I am quite sure that there is no explicit formula for the heights. $\endgroup$ – Ivan Izmestiev Jan 11 at 16:37

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.