Many triangulated 3-spheres (Q707439): Difference between revisions

All two-dimensional spheres are polytopal, i.e. they arise as boundary complexes of three-dimensional polytopes. On the other hand, there exist examples of simplicial spheres \(S^d\), \(d>2\) which are not the boundary complexes of \((d+1)\)-dimensional polytopes (Bruckner sphere, Barnette sphere, etc). Moreover, it follows from Kalai's lower bound \[ \log s(d,n) \geq c n^{[d/2]}, \] for the number \(s(d,n)\) of distinct combinatorial types of simplicial \textbf{PL} \(d\)-spheres on \(n\) vertices and from Goodman and Pollack's upper bound \[ \log p(d,n) \leq d(d+1)n \log n \] for the number \(p(d,n)\) of combinatorial types of simplicial \(d\)-polytopes on \(n\) vertices, that for \(d>3\) there are more simplicial \(d\)-spheres than simplicial \((d+1)\)-polytopes. So, most of simplicial \(d\)-spheres are not polytopal, when \(d>3\). As for three-dimensional case, only few families of non-polytopal 3-spheres were known. The author proves that a stronger lower bound \[ \log s(3,n)\geq c n^{5/4}, \] holds for some positive constant \(c\) and for all sufficiently large \(n\). As consequence, for \(n\) large enough there are more simplicial 3-spheres than 4-polytopes on \(n\) vertices. The proof is based on two constructions proposed in a recent paper by Eppsteine, Kuperberg and Ziegler.