This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Partitions of vertices and facets in trees and stacked simplicial complexes

Gunnar Fløystad Matematisk Institutt
Postboks
5020 Bergen
[email protected]
Abstract.

For stacked simplicial complexes, (special subclasses of such are: trees, triangulations of polygons, stacked polytopes with their triangulations), we give an explicit bijection between partitions of facets (for trees: edges), and partitions of vertices into independent sets. More generally we give bijections between facet partitions whose parts have minimal distance s\geq s and vertex partitions whose parts have minimal distance s+1\geq s+1.

Key words and phrases:
tree, partition, independent vertices, stacked simplicial complex, natural numbers, Stirling numbers, bijection
2020 Mathematics Subject Classification:
Primary: 05C05 05C69, 05E45; Secondary: 05C70

1. Introduction

If G=(V,E)G=(V,E) is a graph with vertices VV and edges EE, a set of vertices AVA\subseteq V is independent if no two vertices in AA are on the same edge. Between any two vertices there is a well-defined distance, and the vertices are independent if their distance is 2\geq 2. Distance may also be defined between edges in a graph.

Assume the graph G=TG=T is a tree and consider partitions of vertices in a tree

V=V0V1VrV=V_{0}\sqcup V_{1}\sqcup\cdots\sqcup V_{r}

into r+1r+1 non-empty parts. We can also partition edges

E=E1E2ErE=E_{1}\sqcup E_{2}\sqcup\cdots\sqcup E_{r}

into rr non-empty parts.

0. We give a bijection between:

  • i)

    Partitions of vertices into r+1r+1 non-empty parts, each part consisting of independent vertices, and

  • ii)

    partitions of edges into rr non-empty parts (and no further requirements).

Example 1.1.

Any tree has a unique partition of the vertices into two independent sets (two colors modulo S2S_{2}). This corresponds to the partition of the edges into one part (one color).

Moreover the above generalizes in two directions.

1. Define a set of vertices AVA\subseteq V to be ss-scattered if the distance between any two vertices is s\geq s. Similarly we have the notion of a set of edges BEB\subseteq E being ss-scattered. We show, Theorem 3.6, for a tree and for r,s1r,s\geq 1 there is a bijection between:

  • i)

    Partitions of vertices into r+1r+1 non-empty parts, each part being s+1s+1-scattered, and

  • ii)

    Partitions of edges into rr non-empty parts, each part being ss-scattered

2. Stacked simplicial complexes is a generalization of trees to higher dimensions. Stacked polytopes [8] with the triangulation coming from a stacking of simplices, induce a well-known subclass of stacked simplicial complexes. The main feature of stacked simplicial complexes for us is that between any two facets (maximal faces) there is a unique path. Similarly between any pair of vertices there is a unique path of facets. We show:

Theorem 3.6 Let XX be a stacked simplicial complex of dimension dd and r,s1r,s\geq 1. There is a bijection between:

  • i)

    Partitions of vertices of XX into r+dr+d non-empty parts, each part being s+1s+1-scattered,

  • ii)

    Partitions of facets of XX into rr non-empty parts, each part being ss-scattered

The results on trees appear quite non-trivial even for the simplest of graphs, the line graph. By considering larger and large line graphs, we get in the limit results for partitions of natural numbers.

Theorem 4.1 There is a bijection between partitions of the natural numbers {\mathbb{N}} into rr non-empty parts, each part being ss-scattered, and partitions of {\mathbb{N}} into r+1r+1 non-empty parts, each part being s+1s+1-scattered.

As a consequence we get for instance:

Corollary 4.3 There is a bijection between partitions of {\mathbb{N}} into two parts A0A1A_{0}\sqcup A_{1} (each part automatically 11-scattered) and partitions of {\mathbb{N}} into d+1d+1 parts B0BdB_{0}\sqcup\cdots\sqcup B_{d}, each part being dd-scattered.

(A trivial special case by repeated use of Theorem 4.1, starting from r=s=1r=s=1 and successively increasing r,sr,s: There is a unique partition of {\mathbb{N}} into rr parts, each being rr-scattered. These parts are of course the congruence classes modulo rr.)

Let us explain in more detail the bijection, first for trees, and secondly in an example of triangulations of polygons. Such triangulations are stacked simplicial complexes of dimension two.

Given a tree and a partition of the vertices VV, make a partition of the edges EE as follows: If vv and ww are vertices consider the unique path in TT linking vv and ww. Let ff, respectively gg, be the edge incident to vv, respectively ww, on this path. If (i) vv and ww are in the same part ViV_{i} of VV and (ii) no other vertex on this path is in the part ViV_{i}, then put ff and gg into the same part of EE, and write fEgf\sim_{E}g. The partition of edges is the equivalence relation generated by E\sim_{E}.

Conversely given a partition of the edges EE, make a partition of the vertices VV as follows: Let vv and ww be distinct vertices, and consider again the path from vv to ww. Let the edges f,gf,g be as above. If (i) the edges ff and gg are distinct (equivalently the vertices v,wv,w are independent), (ii) ff and gg are in the same part EjE_{j}, and (iii) no other edge on this path is in the part EjE_{j}, then put vv and ww in the same part of VV, and write vVwv\sim_{V}w. The partition of vertices is the equivalence relation generated by V\sim_{V}.

Example 1.2.

In Figure 1 we partition the edges into two parts, colored red and black. The vertices are then partitioned into three parts, each consisting of independent vertices. The partition of the vertex set of the first tree is

{1,3,5}{2,6}{4},\{1,3,5\}\cup\{2,6\}\cup\{4\},

and that of the second tree is

{1,3,5,8,10}{2,4,7}{6,9}.\{1,3,5,8,10\}\cup\{2,4,7\}\cup\{6,9\}.
112233445566
1122334455667788991010
Figure 1. Partitions of edges into two parts and corresponding partitions of vertices into three parts
Example 1.3.

Consider now a triangulation of the heptagon, which is a stacked simplicial complex. We partition the facets into two parts: three blue and two red, Figure 2. This corresponds, in a similar way as for trees, to a partition of the vertices, now into four parts of independent vertices: two yellow vertices 3,73,7, three red 1,4,61,4,6, one blue 22, and one green 55. In fact the facet parts here are 22-scattered and so each of the vertex parts will even be 33-scattered.

Note that the colors have no real significance here, they are just added for pedagogical and visualisation purposes. In particular there is no connection between colors of facets and colors of vertices.

11223344556677
Figure 2. Partition of facets into two parts and corresponding partition of vertices into four parts

Our results here came out of work in [7] by M.Orlich and the author on triangulations of polygons and more generally on stacked simplicial complexes.

Results related to the present article are to be found in enumerative combinatorics. A first result in this direction is W.Yang [14] considering trees on n+1n+1 vertices, showing that partitions into independent sets of vertices are counted by Bell numbers BnB_{n}. He also considers generalized dd-trees and show that partitions into independent sets of vertices of such a tree with n+dn+d vertices is counted by the Bell number BnB_{n}. A generalized dd-tree is the same as the edges (or 11-skeleton) of a stacked simplicial complex of dimension dd.

B.Duncan and R.Peele [4] consider enumerative aspects of partitions of vertices of graphs into independent sets. For trees they show that partitions of a tree with n+1n+1 vertices into k+1k+1 independent parts are in bijection with partitions of an nn-set into kk parts. They give, mostly illustrated by a large example, a bijection which is essentially the same as we give. But we gain here the conceptual advantage of expressing this in terms of edges of the tree. Their statement involves choosing an arbitrary vertex rr, a root, and then relating independent vertex partitions of VV to partitions of V\{r}V\backslash\{r\}. Their less conceptual form may also be the reason they do not have the extension to ss-scattered parts. [9] and [11] considers enumerative aspects of graphical Bell numbers further, and the latter also generalized dd-trees.

W.Chen, E.Deng, and R.Du [2] consider ordered sets and use the terminology mm-regular for mm-scattered. They show that partitions of an ordered n+1n+1-set into k+1k+1 parts which are m+1m+1-regular are in bijection with partitions of an nn-set into kk parts which are mm-regular. This corresponds to the case of line graphs, and for this case the bijection they give is essentially the same as ours. We discuss this in Section 4. They also show that we have bijections in this case when considering non-crossing partitions. Related enumerative results are found in [13] and [3]. The book [12] is a comprehensive account of partitions of ordered sets.

Remark 1.4.

The results of this paper dropped out of investigations in [7], concerning Stanley-Reisner rings of stacked simplicial complexes. Among such simplicial complexes there are certain separated models, and from these models we obtain every stacked simplicial complex by partitioning the vertices into independent classes and then collapsing each class into a single vertex. This is done such that the essential algebraic and homological properties of the associated rings are preserved.

The algebra involved here should generalize to much larger classes of simplicial complexes. Polarizing Artin monomial ideals gives separated models, and in [1] we describe all such for polarizations of any power of a graded maximal ideal in a polynomial ring. The case of stacked simplicial complex is the case of the second power. Polarizing Artin monomial ideals in general still goes much further than [1]. The results in the present article therefore likely have vast generalizations, see Subsection 8.2 of [6]. This should involve partitions of vertices, but it is a challenge what other ingredients and statements should be involved.

Let us also mention that the main result of [6] is a fundamental theorem of combinatorial geometry of monomial ideals. Namely that any polarization of an Artin monomial ideal, via the Stanley-Reisner correspondence, is a simplicial complex whose topological realization is a ball.

We describe the organization of this article. Section 2 gives the notion of stacked simplicial complex of dimension dd. It characterizes these, Proposition 2.9, as the simplicial complexes for which there is a unique path between any pair of facets, and the number of vertices is dd more then that number of facets. Section 3 describes the correspondence between partitions of vertices and partitions of facets, and shows our main Theorem 3.6. In the last Section 4 we specialize these results to bijections between partitions of natural numbers, the parts having lower bound requirements on minimal distance.

Acknowledgements. I thank M.Orlich for making Figure 1.

2. Stacked simplicial complexes

We recall the notion of stacked simplicial complex. Its main feature from our perspective, is that it generalizes the property of trees, that for every two faces, there is a unique path between them.

2.1. Paths

Let VV be a finite set. A simplicial complex XX on VV is a family of subsets of VV closed under taking subsets of each element of the family. So for an element FXF\in X and GFG\subseteq F, then also GXG\in X. The elements of VV are vertices, the elemens of XX are faces, and the maximal elements of XX for the inclusion relation are facets. A simplicial complex has a natural geometric realization. A face with dd vertices is then realized as a simplex of dimension d1d-1.

Given any family YY of subsets of VV, the simplicial complex generated by YY is the family of all subsets GG of VV such that GFG\subseteq F for some FYF\in Y.

Definition 2.1.

A pure simplicial complex (i.e., where all the facets have the same dimension) is stacked if there is an ordering of its facets F0,F1,,FkF_{0},F_{1},\ldots,F_{k} such that if Xp1X_{p-1} is the simplicial complex generated by F0,,Fp1F_{0},\ldots,F_{p-1}, then for p1p\geq 1 the facet FpF_{p} is attached to Xp1X_{p-1} along a single codimension one face of FpF_{p}. So we may write Fp=Gp{vp}F_{p}=G_{p}\cup\{v_{p}\} where GpG_{p} is a codimension one face of Xp1X_{p-1} and vpv_{p} is not a vertex of Xp1X_{p-1}. The vertex vpv_{p} is the free vertex of FpF_{p}, in this stacking order.

Remark 2.2.

This is a special case of shellable simplicial complexes, see [10, Subsection 8.2]. It is not the same as the notion of simplicial complex being a tree as in [5], even if the tree is pure. Rather the notion of stacked simplicial complex is more general. For instance the triangulation of the heptagon given in Figure 2, is not a tree in the sense of [5], since removing the triangles 234234 and 257257 one has no facet which is a leaf.

Definition 2.3.

A (gallery) walk in a pure simplicial complex, is a sequence of facets f1,,fpf_{1},\ldots,f_{p} such that each gi=fifi+1g_{i}=f_{i}\cap f_{i+1} has codimension one in fif_{i} (and hence also in fi+1f_{i+1}). The left end vertex is the single element of f1\f2f_{1}\backslash f_{2} and the right end vertex is the single element of fp\fp1f_{p}\backslash f_{p-1}.

If fifi+1=fjfj+1f_{i}\cap f_{i+1}=f_{j}\cap f_{j+1} for some 1i<j<p1\leq i<j<p, we can make a shorter walk from f1f_{1} to fpf_{p}. If fifj+1f_{i}\neq f_{j+1} then

f1,,fi,fj+1,,fpf_{1},\ldots,f_{i},f_{j+1},\ldots,f_{p}

is a shorter walk. If fi=fj+1f_{i}=f_{j+1}, then

f1,,fi1,fj+1,,fpf_{1},\ldots,f_{i-1},f_{j+1},\ldots,f_{p}

is a shorter walk from f1f_{1} to fpf_{p}.

Definition 2.4.

A path is a walk f1,f2,,fpf_{1},f_{2},\ldots,f_{p} where all the fifi+1f_{i}\cap f_{i+1} are distinct. The length of the path is p1p-1.

By the explanation before this definition, any walk from f1f_{1} to fpf_{p} may be reduced to a path between f1f_{1} and fpf_{p}.

Lemma 2.5.

Let XX be a stacked simplicial complex and f1,f2,,fpf_{1},f_{2},\ldots,f_{p} a path in XX.

  • a.

    Let fif_{i} come last among the facets on the path, for a stacking order of XX, and let vv be its free vertex. Then either fi=f1f_{i}=f_{1} and {v}=f1\f2\{v\}=f_{1}\backslash f_{2} or fi=fpf_{i}=f_{p} and {v}=fp\fp1\{v\}=f_{p}\backslash f_{p-1}.

  • b.

    All the fif_{i} are distinct.

Proof.

a. If 1<i<p1<i<p, then fi=(fifi1)(fifi+1)f_{i}=(f_{i}\cap f_{i-1})\cup(f_{i}\cap f_{i+1}), and so vv would be on two facets. But this is not so. So fif_{i} must be one of the end vertices and {v}\{v\} must be as above.

b. If f1=fpf_{1}=f_{p} then p3p\geq 3 and as vv is on only one facet on the path, f1f2=f1\{v}=fp\{v}=fp1fpf_{1}\cap f_{2}=f_{1}\backslash\{v\}=f_{p}\backslash\{v\}=f_{p-1}\cap f_{p}, contradicting that we have a path.

If 1j<kp1\leq j<k\leq p, then fj,fj+1,,fkf_{j},f_{j+1},\ldots,f_{k} is also a path and so fjf_{j} and fkf_{k} must be distinct. ∎∎

Definition 2.6.

Let XX be a pure simplicial complex, and h,kh,k faces in XX such that hkh\cup k is not contained in any codimension one face. A path between hh and kk, written

h|f1,,fp|kh|f_{1},\ldots,f_{p}|k

is a path f1,,fpf_{1},\ldots,f_{p} such that

hf1,hf1f2,kfp,kfpfp1.h\subseteq f_{1},\,h\not\subseteq f_{1}\cap f_{2},\quad k\subseteq f_{p},\,k\not\subseteq f_{p}\cap f_{p-1}.

The face-distance (or simply distance) between hh and kk is pp. In particular, if h,kh,k are contained in a common facet, and not in a codimension one face, their distance is one. If hkh\cup k is contained in a codimension one face, a path between is an empty path consisting of no facets, written h||kh||k. Their distance is defined to be zero.

We mostly use this definition when hh and kk are single vertex sets {v}\{v\} and {w}\{w\}, in which case we simply write vv instead of {v}\{v\}, and similarly for ww.

Lemma 2.7.

Let XX be a stacked simplicial complex, and h,kh,k faces of XX.

  • a.

    In a path f1,,fpf_{1},\ldots,f_{p} between hh and kk we have hfifi+1h\not\subseteq f_{i}\cap f_{i+1} and kfifi+1k\not\subseteq f_{i}\cap f_{i+1} for each 1i<p1\leq i<p.

  • b.

    Let ff be the facet on the path which is last for a stacking order on XX. The free vertex of ff is in hh or in kk.

Proof.

a. Suppose for hh there was an rr such that hfrfr+1h\subseteq f_{r}\cap f_{r+1}, and let rr be minimal such. Then r2r\geq 2 and consider the path

f1,,frf_{1},\ldots,f_{r}

Let ff be the last among these facets in the stacking order of XX, and vv its free vertex. Then f=f1f=f_{1} or f=frf=f_{r}. If f=f1f=f_{1}, since hf1h\subseteq f_{1} and hf1f2h\not\subseteq f_{1}\cap f_{2}, hh must contain the single vertex in f1\f1f2f_{1}\backslash f_{1}\cap f_{2}, and this vertex is vv. But then vv is also in frf_{r}. Since vv is only on a single facet, then fr=f1f_{r}=f_{1}, and this contradicts all facets on a path being distinct.

b. The last facet is one of the end facets, say f1f_{1}. If p=1p=1 then f1=hkf_{1}=h\cup k and the statement holds. If p2p\geq 2, the f1=h(f1f2)f_{1}=h\cup(f_{1}\cap f_{2}) and since vv is not on two facets in the path, we get vhv\in h. ∎∎

2.2. Existence and uniqueness of paths

Proposition 2.8.

Let h,kh,k be faces with union hkh\cup k not contained in any codimension one face. Then there is a unique path between them.

Proof.

Existence: If hkh\cup k is a facet ff, then h|f|kh|f|k is a path between them. Assume then hkh\cup k is not contained in a facet.

Let ff be a facet containing hh and ff^{\prime} a facet containing kk. Let

f=f1,,fp=ff=f_{1},\ldots,f_{p}=f^{\prime}

be the path between them. Let ii be maximal such that hfih\subseteq f_{i}. Let jij\geq i be minimal such that kfjk\subseteq f_{j}. Then we have a path from hh to kk:

h|fi,,fj|k.h|f_{i},\ldots,f_{j}|k.

Uniqueness: Let

h|f1,,fp|k,h|f1,,fq|kh|f_{1},\ldots,f_{p}|k,\quad h|f_{1}^{\prime},\ldots,f_{q}^{\prime}|k

be paths with p,q1p,q\geq 1.

Let ff be the last of the facets in these paths, in the stacking order, and with free vertex vv. By Lemma 2.7, vv is in say hh. As vv is in a single facet on these paths, we get f=f1=f1f=f_{1}^{\prime}=f_{1}. If p=q=1p=q=1 we are done.

i) Suppose exactly one of p,qp,q is 2\geq 2, say p=1p=1 (and q2q\geq 2). Then f1=f1=hkf_{1}^{\prime}=f_{1}=h\cup k so kf1k\subseteq f_{1}^{\prime}. Since kf1f2k\not\subseteq f_{1}^{\prime}\cap f_{2}^{\prime} the free vertex vv is also in kk, and so in fqf_{q}^{\prime}. Then fq=f=f1f_{q}^{\prime}=f=f_{1}^{\prime}, which is not so for a path.

ii) Suppose the paths have p,q2p,q\geq 2. Let

g:=f\{v}=f1f2=f1f2.g:=f\backslash\{v\}=f_{1}\cap f_{2}=f_{1}^{\prime}\cap f_{2}^{\prime}.

Since kf1f2=gk\not\subseteq f_{1}\cap f_{2}=g, we have gkg\cup k not included in a codimension one face. So we get paths

g|f2,,fp|k,g|f2,,fq|k.g|f_{2},\ldots,f_{p}|k,\quad g|f_{2}^{\prime},\ldots,f_{q}^{\prime}|k.

By induction on length, these paths are equal. ∎∎

The following generalizes the well-known situation for trees.

Proposition 2.9.

Let XX be a pure simplicial complex of dimension dd with nn facets and vv vertices. Then XX is stacked iff v=n+dv=n+d and between every pair of facets there is a unique path.

Proof.

When XX is stacked it is clear from construction that v=n+dv=n+d. By Proposition 2.8 there is a unique path between any two facets.

Conversely, assume v=n+dv=n+d and that between every pair of facets there is a unique path. Choose a facet ff, order the facets of XX:

(1) f=f1,f2,,fnf=f_{1},f_{2},\ldots,f_{n}

such that the distance dist(f,fi)dist(f,fj)\text{dist}(f,f_{i})\leq\text{dist}(f,f_{j}) for iji\leq j. Let YpY_{p} be the simplicial complex generated by f1,,fpf_{1},\ldots,f_{p}. Consider the path from ff to fp+1f_{p+1}:

f=f1,f2,,fr=fp+1.f=f^{1},f^{2},\ldots,f^{r}=f_{p+1}.

Here all facets except the last frf^{r} are in YpY_{p} as they must be listed before fp+1f_{p+1} in (1) due to distance. Since fr1frf^{r-1}\cap f^{r} has codimension one in fp+1f_{p+1}, when passing from YpY_{p} to Yp+1Y_{p+1} we have added at most one new vertex. Since Yn=XY_{n}=X has d+nd+n vertices, we must have added exactly one new vertex each time, and so fnf_{n} has a single free vertex, and Yn1Y_{n-1} has d+n1d+n-1 vertices.

We show that between any two facets of Yn1Y_{n-1} there is a unique path. By induction f1,,fn1f_{1},\ldots,f_{n-1} is a stacking order for Yn1Y_{n-1}, and so (1) gives that YnY_{n} is also stacked.

Let fr,fsf_{r},f_{s} be two facets in Yn1Y_{n-1}. Consider the path fr=f1,,fp=fsf_{r}=f_{1}^{\prime},\ldots,f_{p}^{\prime}=f_{s} in X=YnX=Y_{n}. If the last facet fnf_{n} is on this path, say fn=ftf_{n}=f_{t}^{\prime}, then 1<t<p1<t<p and

fn=ft=(ftft1)(ftt+1)Yn1.f_{n}=f_{t}^{\prime}=(f_{t}^{\prime}\cap f_{t-1}^{\prime})\cup(f_{t}^{\prime}\cap_{t+1})\subseteq Y_{n-1}.

This is not so, and so the path from frf_{r} to fsf_{s} is entirely in Yn1Y_{n-1}. ∎∎

The facet-distance between two facets f,gf,g is defined to be the length of the path f|f1,,fp|gf|f_{1},\ldots,f_{p}|g between them (note that f=f1f=f_{1} and g=fpg=f_{p}), and this length is p1p-1.

Remark 2.10.

Note that for two facets f,gf,g their face-distance is one more than their facet-distance. It may seem awkward to have two notions of distance. But by looking at the graph:

vvffggww

it is natural. The facet-distance between ff and gg is one, and the face-distance between vv and ww is two. (And the face-distance between ff and gg is also two.)

2.3. Distance neighborhoods

Let XX be a pure simplicial complex. Choose a codimension one face gg in XX. For m1m\geq 1, let XmX_{m} be the simplicial complex generated by those facets of XX whose face-distance to gg is m\leq m. In particular X0=X_{0}=\emptyset and the facets of X1X_{1} are the facets of XX containing gg. Let V0V_{0} be the vertices of gg and VmV_{m} the vertices of XmX_{m} for m1m\geq 1.

Lemma 2.11.

Assume XX is a stacked simplicial complex. For m1m\geq 1, if vVm\Vm1v\in V_{m}\backslash V_{m-1}, there is a unique facet fvf_{v} in XmX_{m} containing vv, and fv\{v}f_{v}\backslash\{v\} is a subset of Vm1V_{m-1}.

Proof.

Let f1f_{1} be a facet in XmX_{m} containing vv, and

f1|f1,,fr|gf_{1}|f_{1},\ldots,f_{r}|g

the unique path from f1f_{1} to gg. We have rmr\leq m. Let ss be maximal with vfsv\in f_{s}. Then in XmX_{m}:

v|fs,,fr|gv|f_{s},\ldots,f_{r}|g

is the unique path from vv to gg. Since vVm\Vm1v\in V_{m}\backslash V_{m-1} we must have rs+1=mr-s+1=m, and so r=mr=m and s=1s=1. Then f1f_{1} is the first facet in the unique path from vv to gg and f1\{v}f2Vm1f_{1}\backslash\{v\}\subseteq f_{2}\subseteq V_{m-1}. ∎∎

Corollary 2.12.

XmX_{m} is a stacked simplicial complex on VmV_{m}.

Proof.

Let f1,,frf_{1},\ldots,f_{r} be any ordering of the facets in XmX_{m} such that the face-distance between gg and fif_{i} is weakly increasing with ii. Choose 1kr1\leq k\leq r and let dd be the distance from gg to fkf_{k}. The facets f1,,fkf_{1},\ldots,f_{k} are then all in XdX_{d}. Let f1,,fd=fkf_{1}^{\prime},\ldots,f_{d}^{\prime}=f_{k} be the unique path from gg to fkf_{k}. Then:

  • If d2d\geq 2 then fd1=fif_{d-1}^{\prime}=f_{i} for some i<ki<k,

  • fk=fvf_{k}=f_{v} for some vVd\Vd1v\in V_{d}\backslash V_{d-1},

  • fifk=fk\{v}f_{i}\cap f_{k}=f_{k}\backslash\{v\},

  • By Lemma 2.11, vv is in none of the fif_{i} for i<ki<k.

Thus f1,,frf_{1},\ldots,f_{r} is a stacking order for XmX_{m}. ∎∎

3. Bijections between partitions of facets and of vertices

We show how partitions of facets and partitions of vertices into independent sets correspond, and we show that this correspondence is really a bijection. Our arguments are by induction on the distance neighbourhood XmX_{m}. We develop some lemmata before the proof of the main theorem.

3.1. Bijections between partitions

Definition 3.1.

Let XX be a be a stacked simplicial complex with vertex set VV, and ss an integer 1\geq 1. A subset AVA\subseteq V is ss-scattered if the face-distance between any two distinct vertices in AA is s\geq s. The vertex set is independent if it is 22-scattered, i.e. no two vertices in AA are on the same facet.

Similarly a subset BB of the facets is ss-scattered if the facet-distance between any two facets in BB is s\geq s.

We consider partitions of the vertices

(2) V=V1V2VrV=V_{1}\sqcup V_{2}\sqcup\cdots\sqcup V_{r}

into non-empty disjoint sets. Note that the order here is not relevant, so if we switch ViV_{i} and VjV_{j} we have the same partition.

Remark 3.2.

If the ViV_{i} are independent, this is almost the same as a graph coloring of vertices, but not quite. A coloring of the vertices VV is a map f:VCf:V\rightarrow C where C={c1,,cr}C=\{c_{1},\ldots,c_{r}\} is a set of colors, such that each inverse image f1(ci)f^{-1}(c_{i}) is a set of independent vertices. The symmetric group SrS_{r} acts on colorings by permuting the colors c1,,crc_{1},\ldots,c_{r}. So a partition as above (2) is an orbit for the action of SrS_{r}. The class of such orbits, or equivalently of partitions (2) are also called non-equivalent vertex colorings, see [9].

We also consider partitions of the facets

F=F1F2Fs.F=F_{1}\sqcup F_{2}\sqcup\cdots\sqcup F_{s}.

3.1.1. From vertex partitions to facet partitions.

Now we make a correspondence as follows. Given a partition of VV into non-empty independent sets, given by an equivalence relation V\sim_{V}. For ease of following the arguments, we will think of each part as having a specific color. Make a partition of FF as follows. Let ff and ff^{\prime} be distinct facets. Consider the unique path in XX between them:

f=f1,,fp=f,f=f_{1},\ldots,f_{p}=f^{\prime},

and let vv and ww be respectively the left and right end vertices. If vv and ww have the same color, say blue, and none of the facets f2,,fp1f_{2},\ldots,f_{p-1} has any blue vertex, then write fFff\sim^{\prime}_{F}f^{\prime}. This means that ff and ff^{\prime} will be in the same part of facets, these facets get the same color. The colors of vertices and edges are however unrelated, so the facets ff and ff^{\prime} get some color unrelated to blue. The relation F\sim_{F} on FF is the equivalence relation generated by F\sim_{F}^{\prime}.

3.1.2. From facet partitions to vertex partitions

Conversely given a partition of the facets FF, given by an equivalence relation F\sim_{F}. Make a partition of the vertices VV as follows. Let vv and ww be independent vertices, and consider the path from vv to ww:

v|f1,,fp|w.v|f_{1},\ldots,f_{p}|w.

If f1f_{1} and fpf_{p} have the same color, say green, and none of the facets f2,,fp1f_{2},\ldots,f_{p-1} are green, then let vVwv\sim^{\prime}_{V}w. The relation V\sim_{V} is the equivalence relation generated by V\sim^{\prime}_{V}. See Example 1.2.

We want to show that these correspondences are inverse to each other. To do this we show:

A. From an equivalence relation F\sim_{F} on FF, we have constructed the equivalence relation V\sim_{V} on VV. We show that the equivalence relation V\sim_{V} in turn induces the equivalence relation F\sim_{F} by showing:

  • 1.

    If vVwv\sim_{V}w are distinct and, say blue, and

    v|f1,,fp|wv|f_{1},\ldots,f_{p}|w

    where none of f2,,fp1f_{2},\ldots,f_{p-1} have a blue vertex, then p2p\geq 2 and f1Ffpf_{1}\sim_{F}f_{p} (in the original equivalence relation for FF)

  • 2.

    If fFgf\sim_{F}g in the original relation with f,gf,g distinct, there is a sequence f=f0,f1,,fp=gf=f_{0},f_{1},\ldots,f_{p}=g such that

    (3) f0Ff1FFfpf_{0}\sim_{F}^{\prime}f_{1}\sim_{F}^{\prime}\cdots\sim_{F}^{\prime}f_{p}

    where F\sim^{\prime}_{F} is the relation constructed from V\sim_{V}.

  • 3.

    We also show that if the original F\sim_{F} is ss-scattered, then V\sim_{V} is s+1s+1-scattered.

B. From an equivalence relation V\sim_{V} on VV, we have constructed the equivalence relation F\sim_{F} on FF. We show that this in turn induces the equivalence relation V\sim_{V}, by showing:

  • 1.

    If we have a path with p2p\geq 2

    v|f1,,fp|wv|f_{1},\ldots,f_{p}|w

    where f1f_{1} and fpf_{p} are, say green, and none of f2,,fp1f_{2},\ldots,f_{p-1} are green, then vVwv\sim_{V}w (in the original equivalence relation for VV).

  • 2.

    If vVwv\sim_{V}w in the original relation, there is a sequence v=v0,v1,,vp=wv=v_{0},v_{1},\ldots,v_{p}=w such that

    (4) v0Vv1VVvp,v_{0}\sim_{V}^{\prime}v_{1}\sim_{V}^{\prime}\cdots\sim_{V}^{\prime}v_{p},

    where V\sim_{V}^{\prime} is the relation constructed from F\sim_{F}.

  • 3.

    We also show that if the original V\sim_{V} is s+1s+1-scattered, then F\sim_{F} is ss-scattered.

3.2. Induction arguments on XmX_{m}

We show A, B for the XmX_{m} by induction on mm. For this we need some lemmata.

Lemma 3.3.

Given a partition of the facets FF, and consider the relation V\sim_{V}^{\prime} on vertices, constructed above in Subsection 3.1.2. Let vVm\Vm1v\in V_{m}\backslash V_{m-1}, and u,wu,w distinct in VmV_{m}. If vVuv\sim_{V}^{\prime}u and vVwv\sim_{V}^{\prime}w, then uVwu\sim_{V}^{\prime}w.

Proof.

We show first that u,wu,w are independent. Recall, Lemma 2.11, that fvf_{v} is the unique facet on XmX_{m} containing vv.

i) Suppose u,wu,w were on the same face ff. Let

fv=f1,,fp=ff_{v}=f_{1},\ldots,f_{p}=f

be the path from fvf_{v} to ff. If u,wu,w are both on fp1f_{p-1} we may make a shorter path. So we may assume u,wfpu,w\in f_{p} and not both in fpfp1f_{p}\cap f_{p-1}. Assume then that uu is the right end vertex of fpf_{p}. Then the above must be the unique path from vv to uu. Since vVuv\sim_{V}^{\prime}u, f1f_{1} and fpf_{p} have the same color, say green. We have wfp1w\in f_{p-1} and wf1w\not\in f_{1} (since v,wv,w are independent). Let r2r\geq 2 be minimal such that wfrw\in f_{r}. We then have a path

v|f1,,fr|wv|f_{1},\ldots,f_{r}|w

and this is the unique path from vv to ww. By definition of V\sim_{V}^{\prime}, f1f_{1} and frf_{r} also have the same color, which must be green. But by definition of vVuv\sim_{V}^{\prime}u there should not have been any green color between the faces f1f_{1} and fpf_{p}. Hence u,wu,w must be independent.

ii) Let

(5) v|f1,,fp|u,v|f1,,fq|w,v|f_{1},\ldots,f_{p}|u,\quad v|f^{\prime}_{1},\ldots,f^{\prime}_{q}|w,

be the unique paths where f1=f1f_{1}=f_{1}^{\prime} by Lemma 2.11. Here f1f_{1} and fpf_{p} have the same color, say green, and f1(=f1)f_{1}^{\prime}(=f_{1}) and fqf_{q}^{\prime} have the same color, also green, and none of the facets in between have color green. None of the two paths is then a subpath of the other. Hence there is an rr such that frfrf_{r}\neq f_{r}^{\prime}, and let rr be minimal such, so r2r\geq 2. If r3r\geq 3 there is a walk

(6) u|fp,,fr,fr1=fr1,fr,,fq|wu|f_{p},\ldots,f_{r},f_{r-1}=f_{r-1}^{\prime},f_{r}^{\prime},\ldots,f_{q}^{\prime}|w

where fpf_{p} and fqf_{q}^{\prime} are the only green facets. If r=2r=2 then f2,f2f1\{v}f_{2},f_{2}^{\prime}\supseteq f_{1}\backslash\{v\} and so f2f2f_{2}\cap f_{2}^{\prime} has codimension one. There is then a walk

(7) u|fp,,f2,f2,,fq|wu|f_{p},\ldots,f_{2},f_{2}^{\prime},\ldots,f_{q}^{\prime}|w

where only the end facets are green. By reducing these walks like before Definition 2.4, we get a path giving uVwu\sim_{V}^{\prime}w. ∎∎

Note. The process of suitably cutting the sequences (5), then splicing them, (6) or (7), and lastly reducing to a path, will be used a couple of times and we call it cut-splice-reduction.

Lemma 3.4.

Given a partition of VV into independent sets, and the relation F\sim_{F}^{\prime} constructed as in Subsection 3.1.1. Let ff be a facet in XmX_{m} which is not in Xm1X_{m-1}. If fFgf\sim^{\prime}_{F}g and fFhf\sim^{\prime}_{F}h, then gFhg\sim^{\prime}_{F}h.

Proof.

Note first that ff is fvf_{v} for a unique vv. By Lemma 2.11 any path in XmX_{m} starting from ff must have left end vertex vv. So we have the following paths

v|f=f1,,fp=g|u,v|f=f1,,fq=h|wv|f=f_{1},\ldots,f_{p}=g|u,\quad v|f=f_{1}^{\prime},\ldots,f_{q}^{\prime}=h|w

where v,uv,u have the same color blue and none of f2,,fp1f_{2},\ldots,f_{p-1} have blue vertices. Similarly v,wv,w have the same color blue, and none of f2,,fq1f^{\prime}_{2},\ldots,f^{\prime}_{q-1} have blue vertices. None of these paths is then a subpath of the other, and hence there is r2r\geq 2 such that frfrf_{r}\neq f_{r}^{\prime} and let rr be minimal such. If r3r\geq 3, as above (6), we get a walk

w|h=fq,,fr,fr1=fr1,fr,,fp=g|u,w|h=f_{q}^{\prime},\ldots,f_{r}^{\prime},f^{\prime}_{r-1}=f_{r-1},f_{r},\ldots,f_{p}=g|u,

where none of the intermediate facets have blue vertices. If r2r\geq 2, as above (7), get a walk

(8) u|fp,,f2,f2,,fq|w.u|f_{p},\ldots,f_{2},f_{2}^{\prime},\ldots,f_{q}^{\prime}|w.

Aagain as above these walks may be reduced to paths (cut-splice-reduction) and so gFhg\sim^{\prime}_{F}h. ∎∎

Lemma 3.5.

Suppose the relation F\sim_{F}, induces the relation V\sim_{V} on VV. Consider the subcomplex XmX_{m}. i) The restricted relation F|Xm\sim_{F}|_{X_{m}} then induces the restricted relation V|Xm\sim_{V}|_{X_{m}}. Similarly, ii) if V\sim_{V} induces F\sim_{F} the restricted relation V|Xm\sim_{V}|_{X_{m}} induces the restricted relation F|Xm\sim_{F}|_{X_{m}}.

Proof.

Note that if v,wVmv,w\in V_{m} and vVwv\sim_{V}^{\prime}w and v|f1,,fp|wv|f_{1},\ldots,f_{p}|w is the path in XX between them, then since XmX_{m} is stacked, this path is entirely in XmX_{m}.

i) Suppose given F\sim_{F}. Let v,wVmv,w\in V_{m} such that vVwv\sim_{V}w, so we have

v=v0Vv1VVvt=w.v=v_{0}\sim^{\prime}_{V}v_{1}\sim^{\prime}_{V}\cdots\sim^{\prime}_{V}v_{t}=w.

Let \ell minimal such that all viVv_{i}\in V_{\ell}. If >m\ell>m then some viV\V1v_{i}\in V_{\ell}\backslash V_{\ell-1} where 0<i<t0<i<t. By the above Lemma 3.3 we may replace vi1VviVvi+1v_{i-1}\sim^{\prime}_{V}v_{i}\sim^{\prime}_{V}v_{i+1} by vi1Vvi+1v_{i-1}\sim^{\prime}_{V}v_{i+1}. In this way we may reduce the above so all viVmv_{i}\in V_{m}, and thus F|Xm\sim_{F}|_{X_{m}} induces V|Xm\sim_{V}|_{X_{m}}.

ii) Suppose given V\sim_{V}. Let f,gXmf,g\in X_{m} and fFgf\sim_{F}g, so we have

f=f1Ff2FF=g.f=f_{1}\sim^{\prime}_{F}f_{2}\sim^{\prime}_{F}\cdots\sim^{\prime}_{F}=g.

Again using Lemma 3.4 above we may reduce to all fiXmf_{i}\in X_{m}. ∎∎

3.3. The main theorem

Theorem 3.6.

Let XX be a stacked simplicial complex of dimension dd, with vertex set VV and facet set FF, and ss an integer 1\geq 1. The correspondences in Subsections 3.1.1 and 3.1.2 give a one-to-one correspondence between partitions of the vertices VV into r+dr+d non-empty sets, each s+1s+1-scattered, and partitions of the facets FF into rr non-empty sets, each ss-scattered.

Proof.

A. Assume we have started from a partition F\sim_{F} of facets FF, and have constructed the equivalence relation V\sim_{V} corresponding to a partition of the vertices VV. We show properties A1, A2, A3 for XmX_{m} by induction on mm, so that V\sim_{V} in turn induces the original partition F\sim_{F}.

Property A1: Suppose vVwv\sim_{V}w are distinct and, say blue, and let mm be minimal such that v,wXmv,w\in X_{m}. We may assume vVm\Vm1v\in V_{m}\backslash V_{m-1}. Suppose we have a path from vv to ww:

v|f1,,fp|wv|f_{1},\ldots,f_{p}|w

where f2,,fp1f_{2},\ldots,f_{p-1} have no blue vertices. We want to show f1Ffpf_{1}\sim_{F}f_{p} (in the original equivalence relation for FF), that is, they have the same color, say green. Let f1f_{1} have color green, and let

(9) v=v0Vv1VVvt=w.v=v_{0}\sim^{\prime}_{V}v_{1}\sim^{\prime}_{V}\cdots\sim^{\prime}_{V}v_{t}=w.

By Lemma 3.5 we may assume all viXmv_{i}\in X_{m}. Also, if viVm\Vm1v_{i}\in V_{m}\backslash V_{m-1} for some 0<i<t0<i<t, we may by Lemma 3.3 reduce to a shorter such sequence. We may therefore assume the viv_{i} for 0<i<t0<i<t are in Vm1V_{m-1}. If t=1t=1, then f1Ffpf_{1}\sim_{F}f_{p} by definition of V\sim_{V}^{\prime}, and we are done. So assume t2t\geq 2 and consider the path in XmX_{m}:

(10) v=v0|f1,,fq|v1.v=v_{0}|f_{1}^{\prime},\cdots,f_{q}^{\prime}|v_{1}.

Then f1=f1f_{1}^{\prime}=f_{1} and fqf_{q}^{\prime} have the same color green by definition of V\sim_{V}^{\prime} and none of f2,,fq1f_{2}^{\prime},\ldots,f^{\prime}_{q-1} are green. Both v=v0v=v_{0} and v1v_{1} are blue. We claim that none of f2,,fq1f_{2}^{\prime},\ldots,f_{q-1}^{\prime} has any blue vertex. Otherwise, let 2rq12\leq r\leq q-1 be maximal such that frf^{\prime}_{r} has a blue vertex vv^{\prime}. We get a sequence v|fr,,fq|v1v^{\prime}|f^{\prime}_{r},\ldots,f^{\prime}_{q}|v_{1}. Then vVm1v^{\prime}\in V_{m-1}, since g=f1\{v}Vm1g=f_{1}^{\prime}\backslash\{v\}\subseteq V_{m-1} by Lemma 2.11, and the path from gg to v1v_{1} is in Vm1V_{m-1}). By induction (on mm) the facets frf_{r}^{\prime} and fqf_{q}^{\prime} have the same color, a contradiction. Thus none of f2,,fq1f_{2}^{\prime},\ldots,f_{q-1}^{\prime} has a blue vertex.

We claim that vv and ww are independent, which is now equivalent to show that ww is not in f1f_{1}. This will give p2p\geq 2. Recall by Lemma 2.11 that f1=fvf_{1}=f_{v} is the only facet in XmX_{m} containing vv. If ww was in f1f_{1} then wf1\{v}=f1\{v}w\in f_{1}\backslash\{v\}=f_{1}^{\prime}\backslash\{v\}, which is f1f2=f1f2f_{1}\cap f_{2}=f_{1}^{\prime}\cap f_{2}^{\prime}. In particular wVm1w\in V_{m-1}. By induction, since w,v1w,v_{1} are in Xm1X_{m-1} and are related by (9), they are either equal or independent. If equal, vv and ww are independent since vVwv\sim_{V}^{\prime}w, and so not both in f1f_{1}. If ww and v1v_{1} are independent, ww is not in fqf_{q}^{\prime}. Then q3q\geq 3, and f2f_{2}^{\prime} has a blue vertex ww, contradicting that no intermediate facet in (10) has a blue vertex. The upshot is that ww is not in f1f_{1}, and so p2p\geq 2.

By cut-splice-reducing the sequences,

(11) v|f1,,fp|w,v|f1,fq|v1,( where f1=f1),v|f_{1},\ldots,f_{p}|w,\quad v|f_{1}^{\prime},\ldots f_{q}^{\prime}|v_{1},(\text{ where }f_{1}=f_{1}^{\prime}),

as in Lemma 3.4 we reduce to a path

(12) v1|fq,,fp|w,v_{1}|f_{q}^{\prime},\ldots,f_{p}|w,

where no intermediate facet has blue vertices.

Case 1: wVm1w\in V_{m-1}. Then by induction on mm, since v1v_{1} and ww are in Xm1X_{m-1}, fqf_{q}^{\prime} and fpf_{p} have the same color green. As fqf_{q}^{\prime} and f1=f1f_{1}^{\prime}=f_{1} has the same color, green, we get that f1f_{1} and fpf_{p} are both green.

Case 2: wVm\Vm1.w\in V_{m}\backslash V_{m-1}. We have wVv1w\sim_{V}v_{1}, and only one of w,v1w,v_{1} (i.e. ww) is in Vm\Vm1V_{m}\backslash V_{m-1}. Then we can start the argument of Property A1 over again and reduce to Case 1. So we conclude again that fqf_{q}^{\prime} and fpf_{p} have the same color. Again as fqf_{q}^{\prime} and f1=f1f_{1}^{\prime}=f_{1} have the same color, green, we get that f1f_{1} and fpf_{p} are green.

Property A2: Suppose f,gf,g are distinct and green. Let

f=f1,f2,,fp=gf=f^{1},f^{2},\ldots,f^{p}=g

be the unique path from ff to gg. Let q>1q>1 be minimal such that fqf^{q} is green, and consider the path

v|f1,,fq|w.v|f^{1},\ldots,f^{q}|w.

Then vVwv\sim_{V}w and so are, say blue. If one frf^{r} where r[2,q1]r\in[2,q-1] has a blue vertex, let rr be minimal such, and let vv^{\prime} be this vertex, so we have a path

v|f1,,fr|v.v|f^{1},\ldots,f^{r}|v^{\prime}.

By part A1, f1f^{1} and frf^{r} have the same color, which must be green. This is a contradiction and so none of f2,,fq1f^{2},\ldots,f^{q-1} has a blue vertex. Thus f1Ffqf^{1}\sim_{F}^{\prime}f^{q}. Let f0=ff_{0}=f and f1=fqf_{1}=f^{q}. Considering now the shorter path fq,fq+1,,fpf^{q},f^{q+1},\ldots,f^{p}. By induction on length of path, there are

fq=f1FFfr=fp,f^{q}=f_{1}\sim_{F}^{\prime}\cdots\sim_{F}^{\prime}f_{r}=f^{p},

and so we get part A2.

Property A3: Suppose F\sim_{F} is ss-scattered. Let v,wv,w be distinct blue vertices whose face-distance pp is as small as possible. We showed in the argument of Property A1 that vv and ww are independent, and so we have a path

v|f1,f2,,fp|wv|f^{1},f^{2},\ldots,f^{p}|w

with p2p\geq 2. None of f2,,fp1f^{2},\ldots,f^{p-1} can then have blue vertices. Thus f1Ffpf^{1}\sim_{F}f^{p} by what we showed in A1, and their facet-distance is p1sp-1\geq s. Whence ps+1p\geq s+1 and the blue vertices are s+1s+1-scattered.

B. We have started from a partition V\sim_{V} of the vertices VV into independent sets. We have constructed from this an equivalence relation F\sim_{F} and corresponding partition of the facets. We show that properties B1, B2, B3 holds for XmX_{m} by induction on mm, so that F\sim_{F} induces the original partition V\sim_{V}.

Property B1: Suppose fFgf\sim_{F}g, and the path from ff to gg is

(13) v|f=f1,,fp=g|wv|f=f_{1},\ldots,f_{p}=g|w

where f1,fpf_{1},f_{p} are green and none of f2,,fp1f_{2},\ldots,f_{p-1} are green. We show that vVwv\sim_{V}w, they have the same color, say blue.

There is a sequence

f=f0Ff1FFft=g.f=f^{0}\sim^{\prime}_{F}f^{1}\sim^{\prime}_{F}\ldots\sim^{\prime}_{F}f^{t}=g.

Let mm be smallest such that f,gf,g is in XmX_{m}. By Lemma 3.5 we may assume all fif^{i} are in XmX_{m}. If some fif^{i} for 0<i<t0<i<t is in Xm\Xm1X_{m}\backslash X_{m-1}, by Lemma 3.4 we may reduce to a shorter sequence. So we may assume fiXm1f^{i}\in X_{m-1} for 0<i<t0<i<t.

If t=1t=1 then vVwv\sim_{V}w and we are done. So assume t2t\geq 2. Since vVm\Vm1v\in V_{m}\backslash V_{m-1}, by Lemma 2.11 we have f0=fvf^{0}=f_{v}. Now look at at the path from f0f^{0} to f1f^{1}

v=v0|f0=f1,,fq=f1|v1,v=v^{0}|f^{0}=f_{1}^{\prime},\ldots,f_{q}^{\prime}=f^{1}|v^{1},

where v=v0v=v^{0} and v1v^{1} are blue, and f2,,fq1f_{2}^{\prime},\ldots,f_{q-1}^{\prime} do not have any blue vertex. The facets f1=f0=ff_{1}^{\prime}=f^{0}=f and fqf_{q}^{\prime} have the same color, which is green. Are there any green facets in between? Suppose 2rq12\leq r\leq q-1 is maximal such that frf_{r}^{\prime} is green. We have a path

v2|fr,,fq|v1v^{2}|f_{r}^{\prime},\cdots,f_{q}^{\prime}|v^{1}

where all fr,,fqf_{r}^{\prime},\ldots,f_{q}^{\prime} are in Xm1X_{m-1}. By induction on mm, v2v^{2} and v1v^{1} have the same color, blue. This is a contradiction, as frf_{r}^{\prime} has no blue vertex. So f1f_{1}^{\prime} and fqf_{q}^{\prime} are green, while no facets in between are green.

Look at the two paths:

v|f=f1,,fp=g|w,v=v0|f1=f1,,fq|v1v|f=f_{1},\ldots,f_{p}=g|w,\quad v=v^{0}|f_{1}=f_{1}^{\prime},\ldots,f_{q}^{\prime}|v^{1}

where v1Vm1v^{1}\in V_{m-1}. None of these is a sub-sequence of the other, as fpf_{p} and fqf_{q}^{\prime} are green, and no intermediate facet is green. As in Lemma 3.4 we may cut-splice-reduce these together to get a path

v1|fq,,fp|w,v^{1}|f_{q}^{\prime},\ldots,f_{p}|w,

where only the end facets are green.

Case 1: wVm1w\in V_{m-1}. Then in the path from v1v^{1} to ww, the end facets are green, and no intermediate facet is green. By induction on mm (since both v1v^{1} and ww are in Xm1X_{m-1}), we get that v1v^{1} and ww have the same color. Furthermore vv and v1v^{1} have the same color, blue, so both vv and ww are blue.

Case 2: wVm\Vm1w\in V_{m}\backslash V_{m-1}. We have exactly one of w,v1w,v^{1} (i.e. ww) in Vm\Vm1V_{m}\backslash V_{m-1}. The path from v1v^{1} to ww has end facets green and no intermediate facet green. But then we can start the argument of Property B1 over again, and reduce to Case 1. So we conclude that ww and v1v^{1} have the same color. Since vv and v1v^{1} are both blue, we get that vv and ww are both blue.

Property B2: Suppose v,wv,w are distinct and blue. Let

v|f1,,fp|wv|f^{1},\ldots,f^{p}|w

be the unique path from vv to ww. Since vv and ww are independent, p2p\geq 2. Let q2q\geq 2 be minimal such that fqf^{q} contains a blue vertex v=v1v^{\prime}=v_{1}. Then f1Ffqf^{1}\sim_{F}f^{q} by construction, say they are green. Consider the path

v|f1,,fq|v.v|f^{1},\ldots,f^{q}|v^{\prime}.

If one frf^{r} for r[2,q1]r\in[2,q-1] is green, let rr minimal such. Then we have a path

v|f1,,fr|v′′v|f^{1},\ldots,f^{r}|v^{\prime\prime}

and by B1 we have vVv′′v\sim_{V}v^{\prime\prime}, both blue. This contradicts the choice of qq. Thus vVvv\sim_{V}^{\prime}v^{\prime}. Now vfqv^{\prime}\in f^{q}. If q=pq=p we have v=wv^{\prime}=w and we are done. If q<pq<p then vwv^{\prime}\neq w. Let rr be maximal with qr<pq\leq r<p such that vfrv^{\prime}\in f^{r}. We then get

v|fr,,fp|wv^{\prime}|f^{r},\ldots,f^{p}|w

where both vv^{\prime} and ww are blue.

By induction on path length there are

v=v1Vv2VVvs=w,v^{\prime}=v_{1}\sim_{V}^{\prime}v_{2}\sim_{V}^{\prime}\cdots\sim_{V}^{\prime}v_{s}=w,

and so we get part B2.

Property B3: Suppose V\sim_{V} is s+1s+1-scattered. Let fFgf\sim_{F}g be distinct green facets whose facet-distance p1p-1 is as small as possible with path

v|f=f1,f2,,fp=g|w.v|f=f^{1},f^{2},\ldots,f^{p}=g|w.

None of the intermediate facets f2,,fp1f^{2},\ldots,f^{p-1} are green. Then we have just shown in B1 that vVwv\sim_{V}w and so their face-distance ps+1p\geq s+1. Then p1sp-1\geq s and the green facets are ss-scattered.

Final part: We show that if there are rr facet parts, there are r+dr+d vertex parts. This is by induction on the number of facets. Clearly this is true if we have one facet, a simplex. For a stacked simplicial complex XX let mm be minimal such that Xm=XX_{m}=X, and let X=Xm1X^{\prime}=X_{m-1}. Let ff be a facet in Xm\Xm1X_{m}\backslash X_{m-1}, and vv the free vertex of ff. By induction, if there are rr facet parts in XX^{\prime}, there are r+dr+d vertex parts.

If the facet ff makes a part of its own in XX, the free vertex vv becomes a part of its own, by construction of vertex classes. Then we have r+1r+1 facet parts and r+1+dr+1+d vertex parts in XX.

If the facet ff is put into an existing part, say the green part, look at paths v|f=f1,,fp|wv|f=f_{1},\ldots,f_{p}|w where f1,fpf_{1},f_{p} are green and the intermediate facets are not green. Then vv will be given the color of ww, say blue. If v|f=f1,,fq|wv|f=f_{1}^{\prime},\ldots,f_{q}^{\prime}|w^{\prime} is another path with f=f1f=f_{1}^{\prime} and fqf_{q}^{\prime} green, and not intermediate facet is green, by Lemma 3.4 we may cut-splice-reduce and get in XX^{\prime} a path w|fp,,fq|ww|f_{p},\ldots,f_{q}^{\prime}|w^{\prime} with fp,fqf_{p},f_{q}^{\prime} green and with no intermediate green facets. Both ww and ww^{\prime} have the same color. Then vv is uniquely in the blue color class. So XX has rr facet parts and r+dr+d vertex parts. ∎∎

Corollary 3.7.

Let XX be a tree, and s1s\geq 1. There is a bijection between partitions of the vertices into r+1r+1 non-empty parts, each part s+1s+1-scattered, and partitions of the edges into rr non-empty parts, each ss-scattered.

Corollary 3.8.

Let XX be a triangulation of a polygon and s1s\geq 1. There is a bijection between partitions of the vertices into r+2r+2 non-empty parts, each part s+1s+1-scattered, and partitions of the triangles into rr non-empty parts, each ss-scattered.

4. Partitions of natural numbers

The main theorem here appears quite surprising and non-trivial even for the simplest of trees, the line graph. Then it corresponds to studying ordered set partitions, which has a comprehensive theory [12].

Taking the colimit of longer and longer line graphs, we get results for the natural numbers. These are simple consequences of known results for ordered set partitions [2, Thm.2.2] or in a more enumerative form [3, Thm.1], but do not seem to have been stated in this form for natural numbers. In a similar vein, [13] considers partitions of intervals [n][n], but with requirements on the parts which are different from ours here.

Consider the infinite line graph

The set of edges EE may be identified with the natural numbers {\mathbb{N}}, and also the set of vertices VV may be identified with {\mathbb{N}}. Let LnL_{n} be the line graph with nn edges. The bijection between partitions edges into rr parts, each dd-scattered, and partitions of vertices into r+1r+1 parts, each dd-scattered, is compatible with extending the line graph Ln=(Vn,En)L_{n}=(V_{n},E_{n}) with one edge and vertex to the line graph Ln+1L_{n+1}: If (Ei)(E^{i}) is a partition of edges in Ln+1L_{n+1} corresponding to a partition (Vj)(V^{j}) of vertices. Then the partition (EiEn)(E^{i}\cap E_{n}) corresponds to (VjVn)(V^{j}\cap V_{n}).

Thus taking the colimit, we get for the infinite line graph (4) a bijection between partitions of edges (Ei)(E^{i}) into rr parts, each dd-scattered, and partitions of vertices (Vj)(V^{j}) into r+1r+1 parts, each d+1d+1-scattered.

Recall that a subset AA\subseteq{\mathbb{N}} of natural numbers is dd-scattered if for every p<qp<q in AA we have qpdq-p\geq d.

Theorem 4.1.

There is a bijection between partitions of {\mathbb{N}} into rr sets, each dd-scattered, and partitions of {\mathbb{N}} into r+1r+1 sets, each d+1d+1-scattered.

By successively going from rr to r+1r+1 to r+2r+2 and so on we get:

Corollary 4.2.

There is a bijection between partitions of {\mathbb{N}} into r+1r+1 parts A0A1ArA_{0}\sqcup A_{1}\sqcup\cdots\sqcup A_{r} (each set by default being 11-scattered) and partitions of {\mathbb{N}} into r+dr+d parts B0Br+d1B_{0}\sqcup\cdots\sqcup B_{r+d-1}, each dd-scattered.

Specializing to r=0r=0 we get the trivial fact that there is a unique partition of {\mathbb{N}} into dd parts, each dd-scattered. Clearly this partition is the congruence classes modulo dd. However specializing to r=1r=1, we get the quite non-trivial:

Corollary 4.3.

For each d1d\geq 1 there is a bijection between partitions of {\mathbb{N}} into two parts A0A1A_{0}\sqcup A_{1} and partitions of {\mathbb{N}} into d+1d+1 parts B0BdB_{0}\sqcup\cdots\sqcup B_{d}, each being dd-scattered.

Example 4.4.

For r=1r=1 let the partition be {p}\{p}\{p\}\sqcup{\mathbb{N}}\backslash\{p\}, the one part consisting of a single element pp. This corresponds to a partition of {\mathbb{N}} into three parts, each being 22-scattered. We use \ldots to indicate arithmetic progression of step size 22. These parts are:

,p4,p2,\displaystyle\ldots,p-4,p-2, p\displaystyle\,\,p
p+1,p+3,p+5,\displaystyle\,\,p+1,p+3,p+5,\ldots
,p3,p1,\displaystyle\ldots,p-3,p-1, p+2,p+4,\displaystyle\,\,p+2,p+4,\ldots

It also corresponds to a partition of {\mathbb{N}} into four parts, each being 33-scattered. (Here \ldots denotes progression with step size 33.) These parts are:

,p6,p3,\displaystyle\ldots,p-6,p-3, p\displaystyle\,\,p
,p5,p2,\displaystyle\ldots,p-5,p-2, p+1,p+4,p+7,\displaystyle\,\,p+1,p+4,p+7,\ldots
p+2,p+5,\displaystyle\,\,p+2,p+5,\ldots
,p4,p1,\displaystyle\ldots,p-4,p-1, p+3,p+6,.\displaystyle\,\,p+3,p+6,\ldots.
Example 4.5.

Let again r=1r=1. Consider the partition {p,q}\{p,q}\{p,q\}\sqcup{\mathbb{N}}\backslash\{p,q\} where p<qp<q. It corresponds to the following three parts, each 22-scattered. We get two cases, according to whether qpq-p is even or odd. When qpq-p is even:

,p3,\displaystyle\ldots,p-3, p1,p+2,\displaystyle\,\,p-1,p+2,\quad\ldots q2,\displaystyle q-2, q\displaystyle\,\,q
,p4,p2,\displaystyle\ldots,p-4,p-2, p,\displaystyle\,\,p, q+1,q+3,\displaystyle\,\,q+1,q+3,\ldots
p+1,p+3,\displaystyle\,\,p+1,p+3,\quad\ldots q3,q1,\displaystyle q-3,q-1, q+2,q+4,.\displaystyle\,\,q+2,q+4,\ldots.

Note that in the middle part there is quite a long gap from pp to q+1q+1. When qpq-p is odd:

p+1,p+3,\displaystyle\,\,p+1,p+3,\quad\ldots q2,\displaystyle q-2, q\displaystyle\,\,q
,p4,p2,\displaystyle\ldots,p-4,p-2, p,\displaystyle\,\,p, q+1,q+3,\displaystyle\,\,q+1,q+3,\ldots
,p3,\displaystyle\ldots,p-3, p1,p+2,\displaystyle\,\,p-1,p+2,\quad\ldots q1,\displaystyle q-1, q+2,q+4,\displaystyle\,\,q+2,q+4,\ldots

References

  • [1] Ayah Almousa, Gunnar Fløystad, and Henning Lohne, Polarizations of powers of graded maximal ideals, Journal of Pure and Applied Algebra 226 (2022), no. 5, 106924.
  • [2] William YC Chen, Eva YP Deng, and Rosena RX Du, Reduction of m-regular noncrossing partitions, European Journal of Combinatorics 26 (2005), no. 2, 237–243.
  • [3] Wenchang Chu and Chuanan Wei, Set partitions with restrictions, Discrete mathematics 308 (2008), no. 15, 3163–3168.
  • [4] Bryce Duncan and Rhodes Peele, Bell and Stirling numbers for graphs, J. Integer Seq 12 (2009), no. 09.7.1, 1–13.
  • [5] Sara Faridi, The facet ideal of a simplicial complex, Manuscripta Mathematica 109 (2002), no. 2, 159–174.
  • [6] Gunnar Fløystad and Amir Mafi, Polarizations of artin monomial ideals define triangulated balls, https://arxiv.org/abs/2212.09528 (2022).
  • [7] Gunnar Fløystad and Milo Orlich, Triangulations of polygons and stacked simplicial complexes: separating their Stanley–Reisner ideals, Journal of Algebraic Combinatorics (2022), 1–28.
  • [8] Branko Grünbaum, Convex polytopes, second ed., Graduate Texts in Mathematics, vol. 221, Springer-Verlag, New York, 2003, Prepared and with a preface by Volker Kaibel, Victor Klee and Günter M. Ziegler.
  • [9] A Hertz and H Mélot, Counting the Number of Non-Equivalent Vertex Colorings of a Graph, Les Cahiers du GERAD ISSN G-2013-82 (2013), 1–16.
  • [10] Jürgen Herzog and Takayuki Hibi, Monomial ideals, Springer, 2011.
  • [11] Zsófia Kereskényi-Balogh and Gábor Nyul, Stirling numbers of the second kind and Bell numbers for graphs, Australas. J Comb. 58 (2014), 264–274.
  • [12] Toufik Mansour, Combinatorics of set partitions, CRC Press Boca Raton, 2013.
  • [13] Augustine O Munagi, Extended set partitions with successions, European Journal of Combinatorics 29 (2008), no. 5, 1298–1308.
  • [14] Winston Yang, Bell numbers and k-trees, Discrete Mathematics 156 (1996), no. 1-3, 247–252.