Given the regular $n$-dimensional simplex $S\subset\Bbb R^n$ with $n\ge 4$, as well as two orthogonal subspaces $V,W\subset\Bbb R^n$ of dimension $\ge2$ (not necessarily of same dimension, not necessarily $V\oplus W=\Bbb R^n$).

We consider orthogonal projections of $S$ onto these subspaces, denoted by $S_V$ (resp. $S_W$). A face of $S$ is said to be exposed in a projection, if it gets mapped onto the boundary of $S_V$ (resp. $S_W$).

Question: Can it happen that $S_V$ and $S_W$ expose all 0-faces (vertices) of $S$, and expose the exact same 1-faces (edges) of $S$?

Especially interesting are examples in which the projections are not neighborly (i.e. not all edges are exposed).


  • The question what sets of edges of $S$ can be exposed is equivalent to the question what graphs are the 1-skeletons of polytopes $-$ a famously unsolved problem. This question only asks if a set of edges, if exposable at all, can possibly be exposed by different projections to orthogonal subspaces.

  • If $V$ and $W$ are not required to be orthogonal, then this is obviously possible. We can even require the projections $S_V$ and $S_W$ to be not combinatorially equivalent: take two non-equivalent neighborly polytopes with $n$ vertices and represent them as projections of the same $(n-1)$-dimensional simplex.

  • It is not sufficient that the projections are combiantorially equivalent. They have to use the exact same 1-faces of $S$. I am not sure whether this is a restriction since we might be able to apply a symmetry transformation of $S$ to one of the subspaces to make the exposed 1-faces line up. However, it is not clear that this preserves the orthogonality of $V$ and $W$.

  • I am aware of the paper [1], which shows that for $n\to\infty$ the projection of $S\subset\Bbb R^n$ onto a randomly chosen $d$-dimensional subspace with $d=\lfloor \delta n\rfloor,\delta\in(0,1)$ is neighborly with probability converging to one. This seems to make it very likely that such two spaces $V$ and $W$ might exist. However, this is non-obvious and an explicit example would be welcome. It is because of this paper that I consider it more interesting when the projections are not neighborly. In higher dimensions it seems to be unlikely that a certain (non-complete) set of 1-faces is exposed, therefore even more unlikely that two projections onto orthogonal spaces expose the same 1-faces.

[1] Donoho, Tanner: Neighborliness of randomly projected simplices in high dimensions


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.