Quantcast
Channel: User Gro-Tsen - MathOverflow
Viewing all articles
Browse latest Browse all 130

Simplicial complexes on $[n] := \{0,\ldots,n\}$ that are identical under any contraction of consecutive vertices

$
0
0

For $n\in\mathbb{N}$, let us denote by $\Omega(n)$ the set of all (possibly empty) “abstract” simplicial complexes on $[n] := \{0,\ldots,n\}$ (“on $[n]$” means “labeled by the elements of $[n]$”). To be completely explicit, $C \in \Omega(n)$ means that $C$ is a subset of $\mathcal{P}([n]) \setminus \{\varnothing\}$ such that $\sigma\in C$ and $\varnothing\neq\sigma'\subseteq\sigma$ implies $\sigma'\in C$.

If $g\colon[n]\to[k]$ is order-preserving and $C \in \Omega(n)$, let me denote $g_!(C)$ the simplicial complex whose elements are the $g(\sigma)$ for $\sigma\in C$.

Let me say that $C \in \Omega(n)$ is a contraction-compatible complex when: for every $k\leq n$, the simplicial complex $g_!(C)$ is the same for every $g\colon[n]\twoheadrightarrow[k]$ order-preserving and surjective. In other words, no matter how you contract the vertices of $[n]$ down to $[k]$ in an order-preserving way, we get the same $g_!(C)$.

Alternatively, we can say that an infinite sequence $(C_n)_{n\in\mathbb{N}}$ of simplicial complexes, with $C_n \in \Omega(n)$ is a contraction-compatible sequence when for any $k\leq n$ and any $g\colon[n]\twoheadrightarrow[k]$ order-preserving and surjective we have $g_!(C_n) = C_k$. (So of course, the $C_n$ must be contraction-compatible in the sense of the previous paragraph, and additionally they are compatible between each other.)

As an example of these notions, if $d\in\mathbb{N}$ is fixed, the set $\mathrm{Skel}^{(d)}_n$ of all subsets having $\leq(d+1)$ elements (i.e. “dimension $\leq d$ simplices”) of $[n]$, is a contraction-compatible complex, and indeed the sequence $(\mathrm{Skel}^{(d)}_n)_{n\in\mathbb{N}}$ is a contraction-compatible sequence. To give further examples, if $C_n = \{\{0\}\}$ (“the first vertex”) then this also defines a contraction-compatible sequence, and the same holds for $C_n = \{\{n\}\}$ (“the last vertex”; but no other value will do); the same is true for $C_n = \{\{0\},\{1\},\{0,1\}\}$ (“the first edge”); or $C_n = \{\{i\} : i\in [n]\} \cup \{\{i,i+1\} : i\in[n-1]\}$.

General questions: Has this notion appeared in the literature before? Have such objects been studied? Can we, for example, compute or estimate the number of contraction-compatible complexes in $\Omega(n)$?

More specific question: If $C \in \Omega(n)$ is a contraction-compatible complex and $m\geq n$, does there always exist a contraction-compatible $C' \in \Omega(m)$ such that $g_!(C') = C$ for some (and hence any) $g\colon[m]\twoheadrightarrow[n]$ order-preserving and surjective? (Equivalently, can we extend any contraction-compatible complex into a contraction-compatible sequence?)


Motivation: contraction-compatible sequences correspond exactly to prenuclei¹ on $\Omega$ in the topos of simplicial sets (i.e., Lawvere-Tierney pretopologies), so they are a fairly natural thing to come up. (See the question “internal logic of the topos of simplicial sets” for a discussion of the subobject classifier $\Omega$ of simplicial sets.) Given a contraction-compatible sequence $(C_n)$ and any simplicial complex $D \in \Omega(n)$, we can define $\widehat{D}$ by “filling in” any simplex when we already have its parts designated by some $C_n$, that is, $\widehat{D}$ is the set of nonempty $\sigma \subseteq [n]$ such that $f^*(D)\supseteq C_n$ where $f\colon[r]\hookrightarrow[n]$ is the unique order-preserving and injective map with image $\sigma$ (here $r+1$ is the number of elements of $\sigma$) and $f^*(D) := \{\tau\subseteq[r] : f(\tau)\in D\}$. For $(C_n)$ to correspond to a nucleus means that (★) $\widehat{\widehat{D}} = \widehat{D}$ for every $D$. (For example, in the case of $(C_n) = (\mathrm{Skel}^{(d)}_n)$ this means that $\widehat{D}$ consists of every simplex whose $d$-dimensional faces are already in $D$, and this is the case for $(\mathrm{Skel}^{(d)}_n)$.)

The above motivation is not really relevant to the question, but it strongly suggests that the structure of these contraction-compatible sequences $(C_n)$, with or without the additional (★) property, should have interesting order-theoretic properties probably worth studying². Please feel free to add the (★) hypothesis if it helps answer the above questions.

  1. A prenucleus on a frame $L$ is a map $j\colon L\to L$ satisfying $j(x)\geq x$ and $j(x\wedge y) = j(x)\wedge j(y)$. (Here $j(D)$ is the $\widehat{D}$ described above.) A prenucleus is said to be a nucleus when additionally it satisfies $j(j(x))=j(x)$.

  2. The set of contraction-compatible sequences $(C_n)$ satisfying the additional (★) property is a Heyting algebra for the order given by reverse inclusion for all $n$. It is to the topos of simplicial sets what is studied in this paper for the effective topos, which certainly suggests that it might be an interesting mathematical object.


Viewing all articles
Browse latest Browse all 130

Trending Articles