Building a polyhedron from areas of its faces - MathOverflow most recent 30 from http://mathoverflow.net2013-06-18T21:39:36Zhttp://mathoverflow.net/feeds/question/82333http://www.creativecommons.org/licenses/by-nc/2.5/rdfhttp://mathoverflow.net/questions/82333/building-a-polyhedron-from-areas-of-its-facesBuilding a polyhedron from areas of its facesVladimir Reshetnikov2011-12-01T00:25:07Z2011-12-01T02:14:36Z
<p>Is there a known algorithm which, given a finite multiset (unordered list) of integers $A$, returns a yes/no answer for <code>"Is there a polyhedron such that the multiset of areas of all its faces is exactly $A$?"</code>? </p>
<p>Is there a known general algorithm for $n$-dimensional polytopes?</p>
http://mathoverflow.net/questions/82333/building-a-polyhedron-from-areas-of-its-faces/82338#82338Answer by Joseph O'Rourke for Building a polyhedron from areas of its facesJoseph O'Rourke2011-12-01T01:36:49Z2011-12-01T01:36:49Z<p>I can answer your question with the specialization to <em>convex</em> polyhedra and polytopes.
Specializing further to $\mathbb{R}^3$, the result is that</p>
<blockquote>
<p>$n \ge 4$ positive real numbers are the face areas of a convex polyhedron
if and only if the largest number is not more than the sum of the others.</p>
</blockquote>
<p>I wrote up a short note
establishing this: "Convex Polyhedra Realizing Given Face Areas," <a href="http://arxiv.org/abs/1101.0823" rel="nofollow">arXiv:1101.0823</a>.
The result relies on Minkowski's 1911 theorem, which perhaps you know:</p>
<blockquote>
<p><b>Theorem (Minkowski)</b>. Let $A_i$ be positive faces areas and $n_i$ distinct,
noncoplanar unit face normals,
$i=1,\ldots,n$.
Then if $\sum_i A_i n_i = 0$, there is a closed convex polyhedron
whose faces areas uniquely realize those areas and normals.</p>
</blockquote>
<p>This theorem reduces the problem to finding orientations $n_i$ so that vectors of
length $A_i$ at those orientations sum to zero. And this is not difficult.
Here is Figure 3 from my note from which you can almost infer the construction:
<br />
<img src="http://cs.smith.edu/~orourke/MathOverflow/CylinderCylinder.jpg" alt="Cylinder-Cylinder"><br /></p>
<p>Minkowski's theorem generalizes to $\mathbb{R}^d$ and so does an analog of the above claim
(but I did not work that out in detail in the arXiv note).
In terms of an algorithm, the decision question is linear in the number $n$ of facet areas,
and even constructing the polyhedron is linear in $\mathbb{R}^3$,
and likely $O(dn)$ in $\mathbb{R}^d$ (but again, I didn't work that out).</p>
<p>But you don't mention the word "convex" in your post, so perhaps you are interested
in nonconvex polyhedra and polytopal complexes? </p>