betaFeedback

WorksMathematics

Strong sl2\mathfrak{sl}_2 Actions and Symmetric Chain Decompositions

Abstract

We introduce a notion of a strong sl2\mathfrak{sl}_2-action on a poset, and conjecture the existence of such an action is equivalent to the existence of a symmetric chain decomposition of that poset. We explain the intuition behind the conjecture, go through a number of explicit examples, and prove some special cases and relaxations of the conjecture. Then we explain a computational experiment where we tested the conjecture on millions of randomly constructed rank-symmetric, rank-unimodal, and strongly Sperner posets.

Introduction

Given a ranked (i.e. graded) poset PP, we can renormalize the ranks so that the maximal and minimal ranks are nn and −n-n respectively for some nn, with steps of size 2. With this convention we have the following definition:

Definition 1.1. A strong sl2\mathfrak{sl}_2-action on PP is the data of linear operators EE, FF, and HH on CP\mathbb{C}P such that:

  1. HH scales elements x∈Px \in P according to the rank: Hx=ω(x)xHx = \omega(x)x, where ω\omega is the (renormalized) rank function.

  2. EE acts upwards along cover relations: Ex=∑y⋗xex,yyEx = \sum_{y \gtrdot x} e_{x,y}y for some complex numbers ex,ye_{x,y}.

  3. FF acts downwards along cover relations: Fx=∑y⋖xfx,yyFx = \sum_{y \lessdot x} f_{x,y}y for some complex numbers fx,yf_{x,y}.

  4. The operators satisfy the sl2\mathfrak{sl}_2 relations: EF−FE=HEF-FE = H, HE−EH=2EHE-EH = 2E, and HF−FH=−2FHF-FH = -2F

The “strong” modifier is to distinguish this from an earlier notion due to Proctor in [10] where we only require the raising operator EE to act along cover relations, and FF just has to lower the rank. A poset admits an sl2\mathfrak{sl}_2-action in the sense of Proctor if and only if it is rank symmetric, unimodal, and satisfies the strong Sperner property – such posets are called Peck. See § 2.1 for more details.

As an example, here is a visual depiction of a strong sl2\mathfrak{sl}_2-action on the Young lattice L(3,3)L(3,3):

Visual depiction of a strong sl2-action on the Young lattice L(3, 3)

Figure 1.

The blue upwards arrows encode the raising operator EE, and the red downwards arrows are represent the lowering operator FF. The operator HH acts according to the rank, which from top to bottom go from 9 to −9-9 in steps of size 2. Famously, the Young lattices L(m,n)L(m,n) are conjectured to always admit a symmetric chain decomposition – a covering of the poset by disjoint saturated chains that begin and end in opposite ranks.

In this paper we won’t actually look at the L(m,n)L(m,n) posets at all. Instead we will focus on what we feel is actually the main underlying question about symmetric chain decompositions:

Open Problem: Give an algebraic characterization of posets which admit a symmetric chain decomposition.

With this broader goal in mind we offer the following conjecture1:

Conjecture 1.2. A ranked poset PP admits a strong sl2\mathfrak{sl}_2-action if and only if it admits a symmetric chain decomposition.

One direction is easy, if PP admits a symmetric chain decomposition then we can define a sl2\mathfrak{sl}_2-action along the chains, see Proposition 2.2. The goal of this paper is to provide computational evidence and partial results toward the other direction. To start we’ll provide some intuition for why this might be true by looking at a simple example:

Hasse diagram of the poset with elements r, s, t, u, v, and w

Figure 4 (PDF p. 2).

Example 1.3.

Claim: The above poset does not admit a symmetric chain decomposition.

Proof: Assume a symmetric chain decomposition exists.

rr has no upwards neighbors, so it must be the start of a chain of length 4.

tt is the unique downwards neighbor of rr, so it must be in a chain of length 4.

The element vv has no downwards neighbors, so it must be the end of a chain of length 2.

tt is the unique upwards neighbor of vv. Therefore, tt must be in a chain of length 2.

This is a contradiction: tt cannot simultaneously be in chains of two different lengths. □

Claim: The above poset does not admit a strong sl2\mathfrak{sl}_2-action.

Proof: Assume a strong sl2\mathfrak{sl}_2-action exists.

rr has no upwards neighbors, so it must be a highest-weight vector in the L(3)L(3)-isotypic component.

tt is the unique downwards neighbor of rr, so t=cF(r)t=cF(r) for a nonzero constant cc, and tt is therefore in the L(3)L(3)-isotypic component.

The element vv has no downwards neighbors, so it must be a lowest-weight vector in the L(1)L(1)-isotypic component.

tt is the unique upwards neighbor of vv, so t=dE(v)t=dE(v) for a nonzero scalar dd, and therefore tt is in the L(1)L(1)-isotypic component.

This is a contradiction: tt cannot simultaneously be in two different isotypic components. □

This example is very simple, and there are a number of different arguments one could come up with for both of these facts. However we hope these parallel arguments demonstrate the following informal observation, which underpins why we believe this could be true:

Obstructions to a symmetric chain decomposition seem to translate, often directly, into obstructions to a strong sl2\mathfrak{sl}_2-action.

We have constructed many examples of Peck posets which do not admit a symmetric chain decomposition. Not only have we not found any admitting a strong sl2\mathfrak{sl}_2-action, but as a general rule we’ve found that the easier it is to prove the poset does not have an SCD the easier it is to prove it does not admit a strong sl2\mathfrak{sl}_2-action.

The remainder of the paper is structured as follows:

  • In Section 2 we set our conventions and cover the necessary background information. This includes a brief overview of the representation theory of sl2\mathfrak{sl}_2, an explanation of the dictionary between matrices and bipartite graphs leading to the Lindström–Gessel–Viennot lemma, several equivalent characterizations of a Peck poset, and a discussion of Donnelly’s support graph framework. We also introduce the notion of a fractional symmetric chain decomposition.

  • In Section 3 we look at how the combinatorics of the poset can constrain a possible strong sl2\mathfrak{sl}_2-action, and vice versa. We give a number of examples of Peck posets which do not admit a symmetric chain decomposition, and explain why these constraints prevent them from having a strong sl2\mathfrak{sl}_2-action. We close the section with a proof that the weak Bruhat order for the Coxeter group of type H3 does not admit a strong sl2\mathfrak{sl}_2-action.

  • In Section 4 we look at special cases and relaxations of the conjecture. We prove the conjecture for several families of rank vectors, and for “diamond-free” posets. We also show that every poset with 4 ranks which admits a strong sl2\mathfrak{sl}_2-action also admits a fractional symmetric chain decomposition.

  • In Section 5 we discuss the computational experiments we ran, testing the conjecture on millions of randomly generated Peck posets, including thousands of randomly generated worst-case-scenario SCD-critical posets.

Acknowledgments. Thanks to Joshua Mundinger for many helpful conversations. This project was started at the 2022 MathILy-EST REU funded by NSF Grant DMS-2149647, and later the first author was supported by NSF grant DMS-2401515.

AI Acknowledgment. We used ChatGPT (primarily version 5.6) to write essentially all of the code used in the experiments discussed in Section 5, to find relevant references, to create all of the diagrams and tables in this paper, and to help edit the paper. All of the proofs, definitions, and conjectures were made by humans.

Background and Definitions

Representations of sl2\mathfrak{sl}_2

For simplicity, we will work over the complex numbers, although much of this paper makes sense over arbitrary fields of characteristic zero. The Lie algebra sl2=sl2(C)\mathfrak{sl}_2 = \mathfrak{sl}_2(\mathbb{C}) is the space of 2×22 \times2 complex matrices of trace zero. It has a standard basis:

e=[0100]f=[0010]h=[100−1]e = \begin{bmatrix} 0 & 1 \\ 0 & 0 \end{bmatrix} \qquad f = \begin{bmatrix} 0 & 0 \\ 1 & 0 \end{bmatrix} \qquad h = \begin{bmatrix} 1 & 0 \\ 0 & -1 \end{bmatrix}

which satisfy the relations [e,f]=h[e,f] = h, [h,e]=2e[h,e] = 2e, and [h,f]=−2f[h,f] = -2f where [a,b]=ab−ba[a,b] = ab - ba denotes the Lie bracket. A representation of sl2\mathfrak{sl}_2 is a complex vector space VV equipped with a Lie algebra action of sl2\mathfrak{sl}_2. Concretely, a representation is the data of three linear maps E,F,H∈End⁡C(V)E,F,H \in\operatorname{End}_{\mathbb{C}}(V) satisfying EF−FE=HEF - FE = H, HE−EH=2EHE - EH = 2E and HF−FH=−2FHF - FH = -2F.

  • A subspace W⊆VW \subseteq V is a subrepresentation if it is preserved by EE, FF and HH. Note that in this case WW is naturally an sl2\mathfrak{sl}_2-representation, as is the quotient V/WV/W.

  • We say VV is irreducible if it has no nonzero proper subrepresentations.

  • The eigenvalues of HH are called weights. A vector v⃗∈V\vec{v} \in V is called a weight vector if it is an eigenvector for HH. If we specify the eigenvalue λ\lambda we’ll call v⃗\vec{v} a λ\lambda-weight vector.

  • A weight basis for VV is a basis consisting of weight vectors.

  • If v⃗∈V\vec{v} \in V is a weight vector satisfying Ev⃗=0E\vec{v}=0 we say v⃗\vec{v} is a highest-weight vector.

  • If v⃗∈V\vec{v} \in V is a weight vector satisfying Fv⃗=0F\vec{v}=0 we say v⃗\vec{v} is a lowest-weight vector.

As an illustrative example, let L(n)L(n) denote the space of homogeneous degree nn polynomials in two variables, xx and yy. This is an (n+1)(n+1)-dimensional space with a natural basis {xn,xn−1y,…,xyn−1,yn}\{x^n, x^{n-1}y, \ldots, xy^{n-1}, y^n\}. If we consider the operators

E=xddyF=yddxH=xddx−yddyE=x\frac{d}{dy} \qquad F=y\frac{d}{dx} \qquad H=x\frac{d}{dx}-y\frac{d}{dy}

acting on L(n)L(n) in the natural way it is a standard exercise to verify

a) These operators satisfy the sl2\mathfrak{sl}_2 relations and therefore define an sl2\mathfrak{sl}_2-representation on L(n)L(n).

b) L(n)L(n) is irreducible.

c) The standard basis {xn,xn−1y,…,xyn−1,yn}\{x^n, x^{n-1}y, \ldots, xy^{n-1}, y^n\} is a weight basis with weights n,n−2,…,−(n−2)n,n-2,\ldots,-(n-2), and −n-n respectively.

d) Up to scaling, xnx^n and yny^n are the unique highest-weight and lowest-weight vectors in L(n)L(n) respectively.

In fact, these spaces L(n)L(n) essentially completely describe the finite dimensional representation theory of sl2\mathfrak{sl}_2, as summarized by the following proposition:

Proposition 2.1. (For a standard reference, see [7] chapters 6 and 7.) Let VV be a finite dimensional sl2\mathfrak{sl}_2-representation.

a) VV can be decomposed as a direct sum of irreducible representations, each isomorphic to an L(k)L(k):

V≅L(k1)⊕L(k2)⊕⋯⊕L(kℓ)V \cong L(k_1) \oplus L(k_2) \oplus\cdots\oplus L(k_\ell)

b) This decomposition is typically not unique, but the coarser decomposition

V≅(L(0)⊕m0)⊕(L(1)⊕m1)⊕⋯⊕(L(n)⊕mn)V \cong(L(0)^{\oplus m_0}) \oplus(L(1)^{\oplus m_1}) \oplus\cdots\oplus(L(n)^{\oplus m_n})

into isotypic components is unique. In particular any two decompositions of VV into irreducibles have the same multiplicities mkm_k.

c) The multiplicity mkm_k is equal to the dimension of the space of weight kk highest-weight vectors in VV (or equivalently, the dimension of the space of weight −k-k lowest-weight vectors in VV)

d) The decomposition into isotypic components V(k)=L(k)⊕mkV^{(k)}=L(k)^{\oplus m_k} coincides with the eigenspace decomposition for the Casimir operator Ω=EF+FE+12H2∈End⁡(V)\Omega=EF+FE+\frac{1}{2}H^2 \in\operatorname{End}(V).

Posets and Hasse Diagrams

A partially ordered set or poset is a set PP equipped with a reflexive, antisymmetric, and transitive relation ⪯\preceq. There are additional relations ≺\prec, ⪰\succeq, and ≻\succ, which are all naturally defined in terms of ⪯\preceq.

Given two elements aa and bb in a poset PP we say aa covers bb, written a⋗ba \gtrdot b if a≻ba \succ b and if cc satisfies a⪰c⪰ba \succeq c \succeq b then either c=ac=a or c=bc=b. In other words, aa covers bb if aa is bigger than bb and there is nothing in between them. The data of which elements cover one another is called the cover relations of PP. If PP is a finite poset, it is a standard exercise to show that PP is determined by its cover relations.

We will be representing finite posets via their Hasse diagrams. These are graphs drawn in the plane satisfying:

  • The underlying graph has a vertex for each element in PP, and an edge between aa and bb if a⋗ba \gtrdot b.

  • The graph is drawn in the plane so that if a≻ba \succ b then aa is higher in the plane than bb (i.e. its yy-coordinate is larger)

As an example, if P={1,2,3,4,12}P=\{1,2,3,4,12\} ordered by divisibility then the Hasse diagram is: We often won’t really distinguish between the poset itself and its Hasse diagram. In particular if we use graph theoretic terms like edges, vertices, paths, or connected components when describing posets, we are always referring to a corresponding feature of the Hasse diagram.

Ranked posets

A ranked poset (sometimes called a graded poset) is a poset (P,⪯)(P,\preceq) equipped with a rank function ω:P→Z\omega:P\to\mathbb{Z} such that

If a⪰b then ω(a)≥ω(b).(1)\text{If }a\succeq b\text{ then }\omega(a)\geq\omega(b). \tag*{(1)}
If a⋗b then ω(a)=ω(b)+2.(2)\text{If }a\gtrdot b\text{ then }\omega(a)=\omega(b)+2. \tag*{(2)}
For all a,b∈Pω(a)≡ω(b)(mod2)(3)\text{For all }a,b\in P\quad\omega(a)\equiv\omega(b)\pmod{2} \tag*{(3)}

Notice that we are allowing negative ranks and are taking steps of size two along cover relations (ω(a)=ω(b)+2\omega(a)=\omega(b)+2 instead of ω(a)=ω(b)+1\omega(a)=\omega(b)+1). These are non-standard conventions, but they amount to reindexing from the usual notion. Namely, we scale the rank by a factor of 2 and possibly shift it. Note that if the Hasse diagram is connected then condition 3 follows from the first two. Ultimately we will want our rank function to agree with the weights of an sl2\mathfrak{sl}_{2}-representation, which is why we make this convention.

For ranked posets we impose an additional condition on their Hasse diagrams: the height of a vertex in the diagram is equal to the rank of the corresponding poset element. For example if P=B(3)P=B(3) is the Boolean lattice on 3 elements, there is a natural rank function ω(S)=∣S∣−∣Sc∣\omega(S)=|S|-|S^{c}| and the Hasse diagram looks like:

Hasse diagram of a poset

Figure 5 (PDF p. 5).

For a ranked poset PP we let PnP_n denote the set of rank nn elements, i.e. those p∈Pp\in P with ω(p)=n\omega(p)=n. The rank vector records the sizes ∣Pn∣|P_n| of the nonempty ranks, starting from the highest non-empty rank. We say that a ranked poset is rank symmetric if ∣Pn∣=∣P−n∣|P_n|=|P_{-n}| for all nn. For PP connected and rank symmetric we say PP is unimodal if ∣Pn∣≥∣Pn+2∣|P_n|\geq|P_{n+2}| for all n≥0n\geq0. Ranked posets which are supported in either all even or all odd ranks (which is automatic if they are connected) and are rank symmetric and unimodal are called RSU-posets. In the example above, the Boolean lattice B(3)B(3) is an RSU-poset with rank vector [1,3,3,1][1,3,3,1].

Hasse diagram of the Boolean lattice

Figure 6.

Under our conventions, given a ranked poset PP, a symmetric chain decomposition (SCD) of PP is a collection of chains (totally ordered subsets of PP) satisfying:

  1. Each chain is saturated, meaning it does not skip any ranks: For integers ℓ<m<n\ell< m < n which are congruent modulo 2, if a chain contains elements of rank ℓ\ell and nn, then it must contain an element of rank mm as well.

  1. If the top element of a chain is in rank kk, then the bottom element of that chain must be in rank −k-k.

  1. Each element of the poset is contained in exactly one chain in the collection.

Proposition 2.2. If a poset PP admits an SCD then PP admits a strong sl2\mathfrak{sl}_2 action.

Proof. Given SCD on PP, first set all EE- and FF-coefficients along edges not in the SCD equal to zero. Then for a symmetric chain in the SCD of maximal rank kk, identify its elements with the standard basis for L(k)L(k) to define an action along that chain. □\square

2.4. Matrices and Bipartite Graphs. There is a natural correspondence between n×mn \times m matrices and bipartite graphs on {a1,…,an}∪{b1,…,bm}\{a_1,\ldots,a_n\} \cup\{b_1,\ldots,b_m\} with weighted edges: Given an n×mn \times m matrix AA, if an entry Ai,jA_{i,j} is nonzero we add an edge with that weight from bjb_j to aia_i.

Correspondence between a matrix and a weighted bipartite graph

Figure 1 (PDF p. 6).

2.4.1. Minors and matchings: Given subsets II and JJ of the rows and columns respectively, we can look at the induced subgraph on those corresponding vertices. If ∣I∣=∣J∣|I| = |J| we can interpret the corresponding minor as an alternating sum over matchings on this induced subgraph. In particular if such a minor is nonzero then there must be at least one matching between those vertices in the bipartite graph.

Matrix minor and matchings in a bipartite graph

Figure 2 (PDF p. 6).

2.4.2. Multiplication and graph stacking: Under this identification, matrix multiplication corresponds to the following operation: First we stack the two bipartite graphs into a 3-layer graph. Then we construct a new bipartite graph just on the two outer layers of vertices, with edge weights given by the sum over all length 2 paths in the 3-layer graph of the product of weights along those paths.

Cauchy-Binet and disjoint paths

Combining these, the Cauchy-Binet formula expresses a minor of a product ABAB of matrices as a sum of products of minors of AA and BB. A combinatorial manifestation of this is the Lindström–Gessel–Viennot lemma (see [9] and [5]) which computes a minor of a product of two or more matrices graphically by stacking the bipartite graphs and then taking a signed sum over collections of disjoint paths connecting II to JJ of the product of weights along those paths.

Bipartite graphs illustrating weighted disjoint paths

Figure 1 (PDF p. 7).

In particular, if the I×JI \times J minor of a product is known to be nonzero then there exists at least one collection of disjoint paths connecting II to JJ in the stacked graph. Or more generally, if the I×JI \times J submatrix of the product is known to have rank at least kk, then we can find at least kk disjoint paths connecting II to JJ.

Matrix product and disjoint paths

Figure 2 (PDF p. 7).

Peck posets

We’ll now introduce the main class of posets we will be interested in, so-called Peck posets. If a poset admits an SCD then it is automatically Peck, and hence a failure to satisfy these weaker conditions is an obstruction to admitting an SCD. However there are many examples of Peck posets which do not admit an SCD, and there seems to be very little in the literature towards understanding the failure of Peck posets to admit SCDs.

The Sperner Property

We say a ranked poset has the Sperner property if the every antichain has size at most the largest rank.

Ranked poset diagram with highlighted antichain

Figure 3.

In the above example the highlighted vertices form an antichain of size 4, but the maximum rank size is 3. Therefore this poset does not satisfy the Sperner property. More generally, we say a ranked poset is strongly Sperner if for all k≥1k \ge1 every disjoint union of kk antichains has size at most the union of the kk largest ranks.

Ranked poset with highlighted vertices

Figure 1 (PDF p. 8).

In this example the largest antichain has size 6, equal to the largest rank. However the largest union of two antichains has size 10 which is larger than the union of the two largest ranks. Therefore this poset is Sperner but not strongly Sperner.

Griggs’ property T

We say a rank symmetric poset PP satisfies Griggs’ property T if for each pair of opposite ranks PnP_n, P−nP_{-n} there exists a collection of disjoint, saturated chains beginning and ending at those ranks, containing every element in those ranks.

Families of disjoint chains connecting opposite ranks

Figure 2 (PDF p. 8).

Note that there does not need to be any compatibility between these collections of disjoint chains for different pairs of opposite ranks. In the above example, each color corresponds to a family of disjoint edges connecting opposite ranks. The different families have some overlap but they do not form an SCD.

Proctor’s sl2\mathfrak{sl}_2-actions

Following [10], we say a ranked poset PP admits an sl2\mathfrak{sl}_2-action in the sense of Proctor (a weaker notion than the strong actions defined in the introduction) if there is an action of the Lie algebra sl2\mathfrak{sl}_2 on the vector space CP\mathbb{C}P such that:

(1) HH scales elements x∈Px \in P according to the rank: Hx=ω(x)xHx = \omega(x)x, where ω\omega is the rank function.

(2) EE acts along cover relations: Ex=∑y≻xcx,yyEx = \sum_{y \succ x} c_{x,y}y for some complex numbers cx,yc_{x,y}.

Note that FF is not required to act along cover relations (that would make it a strong action), but the sl2\mathfrak{sl}_2-relations do ensure that if x∈Pnx \in P_n then Fx∈CPn−2Fx \in\mathbb{C}P_{n-2}. Here is an example of an sl2\mathfrak{sl}_2-action on the poset from Example 1.3:

Support graph for $L(1)^{\otimes 3}$

Figure 1 (PDF p. 9).

The FF operator, represented with dotted red lines, does not follow the edges of the poset. Thus, the action depicted is not a strong sl2\mathfrak{sl}_2-action on this poset. Indeed, in Example 1.3 we saw that this poset does not admit any strong action. We’ll note that it can be viewed as a strong action the larger poset obtained by adding the dotted red edges. Indeed any action in the sense of Proctor can be thought of as a strong action on a larger poset, by adding edges corresponding to nonzero FF-coefficients.

Peck poset definition

Definition/Theorem 2.1. (Combining [10] and [6]) Let PP be a rank-symmetric poset. The following are equivalent:

  1. PP is strongly Sperner.

  2. PP satisfies Griggs’ property TT.

  3. PP admits an sl2\mathfrak{sl}_2-action in the sense of Proctor.

Such posets are called Peck posets.

It is easy to see (using any of these conditions) that if a poset PP admits an SCD it is automatically Peck. However we see that Example 1.3 is a Peck poset, which does not admit an SCD. We will see many more such examples in this paper.

SCD-critical posets

We say that a poset PP is an SCD-critical poset if

  1. PP is Peck.

  2. PP does not have an SCD.

  3. Adding any edge to the Hasse diagram creates an SCD.

Note that any Peck poset not admitting an SCD can always be enlarged to an SCD-critical poset. So to prove Conjecture 1.2 it would suffice to prove it for SCD-critical posets. For small enough rank vectors we can sometimes just enumerate all of the SCD-critical posets to verify the conjecture, although this quickly becomes computationally unfeasible.

sl2\mathfrak{sl}_2 support graphs

Let VV be a finite dimensional representation of sl2\mathfrak{sl}_2 with a weight basis {b1,b2,…,bn}\{b_1,b_2,\ldots,b_n\}. We define a ranked poset as follows:

  • The elements of the poset are the basis elements b1,b2,…,bnb_1,b_2,\ldots,b_n.

  • The rank function ω(bi)\omega(b_i) is the weight of bib_i (its HH-eigenvalue).

  • Cover relations bi⋗bjb_i \gtrdot b_j if either bjb_j appears with nonzero multiplicity in the expansion of FbiF b_i, or bib_i appears with nonzero multiplicity in the expansion of EbjE b_j.

By construction we see this poset admits a strong sl2\mathfrak{sl}_2-action. These posets (or the underlying graphs of their Hasse diagrams), along with the numerical data of the EE- and FF-coefficients along each edge are called sl2\mathfrak{sl}_2-support graphs. For example if we take a standard basis {v1,v−1}\{v_1,v_{-1}\} for L(1)L(1), we get a natural tensor product basis {v±1⊗v±1⊗v±1}\{v_{\pm1}\otimes v_{\pm1}\otimes v_{\pm1}\} for L(1)⊗3L(1)^{\otimes3} with a support graph that looks like: These were introduced by Donnelly in [4] (and further developed in [3]) along with analogs for all semisimple Lie algebras g\mathfrak{g}, where instead of a poset we get a directed graph with multiple edge colors, one for each simple root in the root system associated to g\mathfrak{g}. Donnelly asked whether support graphs for a semisimple Lie algebra g\mathfrak{g} always contain a combinatorial g\mathfrak{g}-crystal. For g=sl2\mathfrak{g} = \mathfrak{sl}_2 that simply means every support graph admits an SCD. Since any poset admitting a strong sl2\mathfrak{sl}_2-action contains an sl2\mathfrak{sl}_2-graph in its support (just delete any edges unused by the action), we see that an affirmative answer to the sl2\mathfrak{sl}_2 case of Donnelly’s question is equivalent to our formulation of Conjecture 1.2.

![Directed graph with multiple edge colors] (media/fig-1.png) bbox=0.36,0.12,0.64,0.29

Figure 7.

Multicommodity Flows and Fractional SCDs

The existence of an SCD in a ranked poset can be formulated as a multicommodity flow problem as follows:

  1. Make the Hasse diagram directed by orienting all edges downward.

  2. For each n≥0n \ge0 add an auxiliary vertex xnx_n with edges to every element of PnP_n, and an auxiliary yny_n with edges from every element of P−nP_{-n}.

  3. Make xnx_n and yny_n a source and sink respectively for commodity nn, set the demand between them to ∣Pn∣−∣Pn+2∣|P_n| - |P_{n+2}|.

  4. Give all vertices in the original poset total capacity 1.

![Multicommodity flow network with auxiliary source and sink vertices] (media/fig-2.png) bbox=0.36,0.54,0.64,0.80

Figure 8.

The observation is that integer flow solutions to this flow problem exactly correspond to SCDs on the original poset. It’s then natural to ask: What do fractional flow solutions correspond to?

Definition. Let PP be a ranked poset, and let C(P)C(P) denote the set of symmetric saturated chains in PP. A fractional symmetric chain decomposition is a function τ:C(P)→R≥0\tau: C(P) \to\mathbb{R}_{\ge0} such that for all p∈Pp \in P

∑c∈C(P),c∋pτ(c)=1.\sum_{\substack{c \in C(P),\,c \ni p}} \tau(c) = 1.

In other words, a fractional SCD of PP is a collection of weighted symmetric saturated chains in PP such that each vertex is covered by chains of total weight 1.

As far as we can tell this is the first time this notion is being defined explicitly. We will note though that the recent preprint [2] uses coverings by weighted, but not necessarily symmetric, chains extensively. Moreover, in Appendix B they implicitly use fractional SCDs to give a new proof of Sperner’s theorem. Our Proposition 2.4 below is essentially a generalization of that argument.

Note that if all weights are rational we can renormalize and say that a fractional SCD is a collection of symmetric saturated chains covering every vertex of PP the same number of times. We’ll see that in fact if any fractional SCD exists then a rational one does too, so effectively we could have taken that as an alternative definition.

Proposition 2.3. For a ranked poset PP the associated multicommodity flow problem described above has a (not necessarily integral) solution if and only if PP admits a fractional SCD.

Proof. One direction is clear, given a fractional SCD we get a solution to the flow problem by flowing each commodity along the chains of the corresponding length according to the weight of that chain. The fact that each vertex is covered by chains of weight 1 translates to each vertex in the poset having one total unit of flow, satisfying the capacity.

Going the other direction, starting with a flow we will construct a fractional SCD one weighted chain at a time. Note that if there is flow from xix_i to yiy_i we can find a path connecting them with nonzero flow. This path intersects the poset in a symmetric chain with nonzero flow along each edge. Here’s the procedure:

  1. Pick xix_i and yiy_i with a nonzero amount of flow between them, and choose a path connecting them with nonzero flow along it.

  2. Add the corresponding chain to the fractional SCD with weight equal to the smallest flow along any edges involved.

  3. Remove that amount of flow from each edge along the path, creating a new smaller flow.

  4. Repeat this process.

At every step the flow of the chosen commodity becomes zero on at least one additional edge, so this process terminates after finitely many steps. □

Remark: Note that this is not a bijection. Every fractional SCD gives rise to a flow in a unique way, but many fractional SCDs can give rise to the same flow. The procedure above for converting a flow into a fractional SCD depends heavily on the choices of which commodity to look at and which path to choose at every step.

An example of a fractional SCD

Every SCD is automatically a fractional SCD – give the chains in the SCD weight 1 and all others weight 0. Moreover, given two or more SCDs of the same poset, a convex combination of them will be a fractional SCD. However not all fractional SCDs come from ordinary SCDs. Here is an example which cannot be obtained that way:

Flow diagrams illustrating a fractional SCD

Figure 9 (PDF p. 11). On the left we have depicted as a flow: the bold edges at the top carry 1 unit of flow, and the remaining edges carry 12\frac{1}{2} a unit of flow. We have a total of 2 units of purple flow, which begin at rank 3 and end at rank −3-3, and 2 units of green flow going from rank 1 to rank −1-1. On the right we show

it as a collection of 8 symmetric saturated chains, which collectively cover each vertex exactly twice. One can easily verify that this poset does not admit an ordinary SCD. We’ll see in Example 3.4 that this poset does not admit a strong sl2\mathfrak{sl}_2-action either.

Fractional SCD implies Peck

Recall that if a poset PP admits an SCD then it is automatically Peck. We have the following generalization:

Proposition 2.4. If PP is a poset admitting a fractional SCD, then PP is Peck.

Proof. First observe that PP must be rank-symmetric and unimodal, since the rank vector is the weighted sum of the rank vectors for the chains in the fractional SCD. We will check the strong Sperner condition directly. Fix a fractional SCD τ:C(P)→R≥0\tau:C(P)\to\mathbb{R}_{\ge0} and let SS be a disjoint union of kk antichains.

Let DkD_k denote the union of the largest kk ranks. For any symmetric chain c∈C(P)c\in C(P) note that ∣c∩S∣≤min⁡(∣c∣,k)=∣c∩Dk∣|c\cap S|\le\min(|c|,k)=|c\cap D_k|. Therefore:

∣S∣=∑x∈S∑c∋xτ(c)=∑c∈C(P)τ(c)∣S∩c∣≤∑c∈C(P)τ(c)∣c∩Dk∣=∑y∈Dk∑c∋yτ(c)=∣Dk∣|S|=\sum_{x\in S}\sum_{c\ni x}\tau(c)=\sum_{c\in C(P)}\tau(c)|S\cap c|\le\sum_{c\in C(P)}\tau(c)|c\cap D_k|=\sum_{y\in D_k}\sum_{c\ni y}\tau(c)=|D_k|

The flow polytope

One advantage of this perspective is that fractional flows can naturally be identified with points inside a polytope, with SCDs corresponding to integer points. For general references on integer programming and polyhedral realizations of multicommodity flow problems we refer to [1] and [11]. In our context, the flow polytope is defined as follows:

  • For every edge ee in the augmented poset and every commodity cc we have a variable fe,cf_{e,c} representing the amount of that commodity flowing along that edge.

fe,c≥0and∑cfe,c≤1 for all edges ef_{e,c}\ge0 \qquad\text{and}\qquad\sum_c f_{e,c}\le1\text{ for all edges }e
  • For every vertex in the original poset the amount of flow in equals the amount of flow out for each commodity:

∑e→vfe,c=∑v→efe,c for all vertices v and commodities c\sum_{e\to v}f_{e,c}=\sum_{v\to e}f_{e,c}\text{ for all vertices }v\text{ and commodities }c
  • The total flow into each vertex in the original poset is 1.

∑c∑e→vfe,c=1\sum_c\sum_{e\to v}f_{e,c}=1
  • The auxiliary vertices xnx_n and yny_n are sources and sinks for commodity cnc_n with appropriate demands DnD_n.

∑xn→efe,cn=Dnfe,cm=0 if e comes out of xn and m≠n\sum_{x_n\to e}f_{e,c_n}=D_n \qquad f_{e,c_m}=0\text{ if }e\text{ comes out of }x_n\text{ and }m\ne n
∑e→ynfe,cn=Dnfe,cm=0 if e goes in to yn and m≠n\sum_{e\to y_n}f_{e,c_n}=D_n \qquad f_{e,c_m}=0\text{ if }e\text{ goes in to }y_n\text{ and }m\ne n

In particular this means one can use linear programming and integer programming techniques to determine whether a poset admits a fractional or ordinary SCD. For larger posets this was by far the most efficient way we found to test them, see § 7.2 for more discussion.

Some questions about fractional SCDs

As far as we can tell this is the first time that this notion explicitly appears in the literature. While this isn’t the main focus of this paper, we do think fractional SCDs are quite interesting so we’d like to include some questions about them we think are interesting:

Open Problem: Do the Young lattices L(m,n)L(m,n), and more generally quotients Bn/GB_n/G of the Boolean lattice for G⊆SnG \subseteq S_n, always admit fractional SCDs?

The analogous questions for ordinary SCDs are famously open, but fractional SCDs are more flexible so this could be an interesting intermediate result. Likely this could be verified computationally for more values of mm and nn too.

Open Problem: Can the existence of a fractional SCD on a Peck poset be characterized by a max-flow-min-cut type criterion?

In general for multicommodity flow problems we do not expect this to be true, but for this very restricted class of networks it seems conceivable that there is a nice characterization.

Open Problem: For a poset which does admit a fractional SCD, can we bound the minimum denominator needed for a rational solution in terms of the size or number of ranks in the poset?

Since the flow polytope is defined by integer equations, if any real point exists, so does a rational one. In all of the small examples we looked at, if there was a fractional SCD at all then there was one with denominator at most 2, although we expect that may grow with the number of ranks.

Combinatorics of strong sl2\mathfrak{sl}_2-actions

In this section we will explore how the combinatorics of a poset can give us linear algebraic information about any strong sl2\mathfrak{sl}_2-actions on the poset (and vice-versa). Although perhaps a better title for this section might be “how to tell when a Peck poset does not admit a strong sl2\mathfrak{sl}_2-action”.

All examples in this section are Peck posets which do not admit symmetric chain decompositions, and we will be showing that they do not admit strong sl2\mathfrak{sl}_2-actions. Other than the last example in the section, these posets were all part of randomly generated batches, and then we selected ones to highlight particular proof ideas.

By Proposition 2.2, a poset not admitting a strong sl2\mathfrak{sl}_2-action cannot admit a symmetric chain decomposition. Nevertheless, we encourage the reader to try to prove directly that the posets here do not admit symmetric chain decompositions. In particular, all of these proofs were inspired by similar (and usually simpler) arguments to show the posets do not admit a symmetric chain decomposition, and one can often reverse engineer that original argument.

Isotypic vectors

We say that a nonzero vector v⃗\vec{v} in an sl2\mathfrak{sl}_2 representation VV is an L(λ)L(\lambda)-isotypic vector if v⃗\vec{v} lies in the L(λ)L(\lambda)-isotypic component. If v⃗\vec{v} is a highest weight vector, this is equivalent to v⃗\vec{v} generating an irreducible subrepresentation of VV with highest-weight λ\lambda. If we don’t want to specify which component we are talking about we will just say “v⃗\vec{v} is an isotypic vector”.

A first observation is that if a poset PP admits a strong sl2\mathfrak{sl}_2-action then we can tell that certain vectors must be isotypic vectors, even without conferring the whole action. The fact then that a nonzero vector cannot be in two different isotypic components gives us an easy obstruction. Here is a simple example of this type of reasoning:

Example 3.1. The poset

Poset diagram

Figure 10. does not admit a strong sl2\mathfrak{sl}_2-action

Proof. Assume there is a strong sl2\mathfrak{sl}_2-action on this poset. The elements t1t_1 and t2t_2 have no downwards neighbors, so they are both lowest-weight vectors in the L(1)L(1) component. EE is injective between ranks −1-1 and 11, and t1t_1 and t2t_2 have exactly two upwards neighbors altogether. Hence, span⁡(E(t1),E(t2))=span⁡(s1,s2)\operatorname{span}(E(t_1),E(t_2))=\operatorname{span}(s_1,s_2). Therefore span⁡(s1,s2)\operatorname{span}(s_1,s_2) consists entirely of L(1)L(1)-isotypic vectors.

On the other hand, rr has no upwards neighbors so it is a highest-weight L(3)L(3)-isotypic vector. So then F(r)F(r) is a nonzero L(3)L(3)-isotypic vector, which is a contradiction since F(r)∈span⁡(s1,s2)F(r)\in\operatorname{span}(s_1,s_2) so we know it must be a L(1)L(1)-isotypic vector by the previous paragraph. □

Poset elements as isotypic vectors

Suppose PP is a poset which admits a strong sl2\mathfrak{sl}_2-action. Here are some ways we can tell from the combinatorics of PP when certain elements must be isotypic vectors:

  • Highest-weight and lowest-weight vectors: If v⃗\vec{v} is in rank λ\lambda and E(v)=0E(v)=0 then v⃗\vec{v} is highest-weight vector, and therefore L(λ)L(\lambda)-isotypic. In particular if p∈Pp\in P has no upwards neighbors in PP then this must happen. The story is similar for lowest-weight vectors.

  • Unique neighbor propagation: Suppose v⃗\vec{v} is an L(λ)L(\lambda)-isotypic vector in rank k>−λk>-\lambda, then F(v⃗)F(\vec{v}) is nonzero and L(λ)L(\lambda)-isotypic. So if p∈Pkp\in P_k (with k>−λk>-\lambda) is known to be L(λ)L(\lambda)-isotypic and pp has a unique neighbor qq in Pk−2P_{k-2}, then qq is also L(λ)L(\lambda)-isotypic. Similarly we can propagate upwards along using EE, with the opposite restriction on ranks.

  • Hall-tight propagation: More generally suppose p1,p2,…,pm∈Pkp_1,p_2,\ldots,p_m\in P_k are L(λ)L(\lambda)-isotypic with k>−λk>-\lambda. Suppose also that they collectively have exactly mm neighbors q1,q2,…,qmq_1,q_2,\ldots,q_m in Pk−2P_{k-2}. Then span⁡(q1,…,qm)=span⁡(F(p1),…,F(pm))\operatorname{span}(q_1,\ldots,q_m)=\operatorname{span}(F(p_1),\ldots,F(p_m)). Therefore, all of the qiq_i's are L(λ)L(\lambda)-isotypic too. Again, a similar rule applies in the other direction using EE.

  • Reverse propagation: In general F(v⃗)F(\vec{v}) being L(λ)L(\lambda)-isotypic does not imply that v⃗\vec{v} is also L(λ)L(\lambda)-isotypic, because v⃗\vec{v} could differ from the L(λ)L(\lambda)-isotypic component by an element of the kernel of FF. However if FF is injective from rank kk to k−2k-2 and p∈Pkp\in P_k has all of its neighbors in Pk−2P_{k-2} known to be L(λ)L(\lambda)-isotypic, then pp must also be L(λ)L(\lambda)-isotypic. Again, similar rules for EE apply.

  • Multi-step propagation: We can also propagate using FnF^n or EnE^n. For example suppose p∈Pkp\in P_k is a L(λ)L(\lambda)-isotypic vector with k>−λ+2k>-\lambda+2, and that q∈Pk−4q\in P_{k-4} is its unique two-step downward neighbor. Then q=F2(p)q=F^2(p) up to scalar and therefore qq is also L(λ)L(\lambda)-isotypic. This can also be combined with Hall-tight or reverse propagation in suitable circumstances.

  • EFEF-eigenvectors: A weight vector v⃗\vec{v} is an eigenvector for EFEF if and only if vv is an isotypic vector—the eigenvalue will depend on the isotypic component as well as the rank. Suppose p∈Pkp\in P_k has a unique neighbor q∈Pk−2q\in P_{k-2} and pp is also the unique neighbor of qq. Then both pp and qq are L(λ)L(\lambda)-isotypic vectors for some λ\lambda (but we don’t know which λ\lambda a priori).

Partial isotypic information

Typically it is not the case that all of the poset elements themselves are forced to live in isotypic components. More often we are only able to get partial information about the relationship between the poset elements and the isotypic components.

  • Isotypic vectors in spans: If p∈Pkp \in P_k is L(λ)L(\lambda)-isotypic with k>−λk > -\lambda, then F(p)F(p) is also L(λ)L(\lambda)-isotypic. Even if pp does not have a unique downwards neighbor, we can still conclude that there is a L(λ)L(\lambda)-isotypic vector in the span of pp’s downward neighbors.

  • Sums of isotypic components: If p∈Pkp \in P_k has all of its downwards neighbors known to be L(λ)L(\lambda)-isotypic, but FF is not injective from rank kk to k−2k-2, we can’t use the reverse propagation rule but we can still conclude that pp is in the direct sum of the L(λ)L(\lambda) and L(k)L(k) components.

  • Propagating partial information: We can use the same methods for individual poset elements in isotypic components to propagate partial information about isotypic vectors in spans of poset elements and poset elements in sums of isotypic components. For example if p∈Pkp \in P_k is in the sum of two isotypic components, and qq is the unique downward neighbor of pp, then qq must also be in the sum of those components.

  • Intersections and Unions: This partial information can sometimes be combined to get stronger results. For example suppose SS and TT are subsets of PkP_k where we know both span⁡(S)\operatorname{span}(S) and span⁡(T)\operatorname{span}(T) contain a L(λ)L(\lambda)-isotypic vector. Then either span⁡(S∪T)\operatorname{span}(S \cup T) contains two linearly independent L(λ)L(\lambda)-isotypic vectors or span⁡(S∩T)\operatorname{span}(S \cap T) contains a L(λ)L(\lambda)-isotypic vector. If say we additionally knew L(λ)L(\lambda) had multiplicity 1, we could conclude we are in the latter case.

Forbidden Configurations

Once we know where the isotypic vectors live in the representation, certain combinatorial obstructions to strong sl2\mathfrak{sl}_2-actions emerge. Essentially these are contradictions one arrives at when trying to prove a given poset does not admit such an action.

  • Multiple isotypic components: The simplest obstruction is that a poset element is forced to be in two different isotypic components simultaneously. This was what happened in Example 1.3.

  • Overcrowded spans: More generally if S⊆PkS \subseteq P_k is forced to contain more than ∣S∣|S| linearly independent isotypic vectors in its span. In Example 3.1 span⁡(s1,s2)\operatorname{span}(s_1,s_2) was forced to contain two linearly independent L(1)L(1)-isotypic vectors, as well as one L(3)L(3)-isotypic vector.

  • Too many isotypic vectors: If L(λ)L(\lambda) appears with multiplicity mm, then we can only have at most mm linearly independent L(λ)L(\lambda)-isotypic vectors in any given rank. So if say the multiplicity of L(λ)L(\lambda) is one and we know p1∈Pkp_1 \in P_k is L(λ)L(\lambda)-isotypic, and we also know that there is a L(λ)L(\lambda)-isotypic vector in span⁡(p2,p3)\operatorname{span}(p_2,p_3) (also in rank kk) that is a contradiction.

  • Rank conditions and Lindström–Gessel–Viennot: Suppose S⊆PkS \subseteq P_k and T⊆Pk−2nT \subseteq P_{k-2n} are such that the submatrix FS,TF_{S,T} is known to have rank mm. Then the Lindström–Gessel–Viennot Lemma tells us there must be at least mm disjoint downwards paths connecting SS to TT. For example, suppose span⁡(S)\operatorname{span}(S) and span⁡(T)\operatorname{span}(T) are both known to contain all of the L(λ)L(\lambda)-isotypic vectors in their respective ranks. If there aren’t enough disjoint paths connecting them, we obtain a contradiction. We’ll note that often this can be converted into one of the other contradictions listed above.

Shrinking the support

Given a poset PP with a strong sl2\mathfrak{sl}_2-action, for p∈Pp \in P a priori we know that F(p)F(p) is a linear combination of the downwards neighbors of pp (and similarly for E(p)E(p) and upwards neighbors). However it could be that in fact F(p)F(p) is a linear combination of some smaller subset of its downwards neighbors – possibly just a single one or maybe even F(p)=0F(p)=0.

The next observation about the combinatorics of posets with a strong sl2\mathfrak{sl}_2-action is that sometimes you can tell, just from the combinatorics of the poset, that any strong action on the poset cannot actually use certain edges in one direction or the other (or both). Similarly, we can sometimes tell from the combinatorics that certain edge coefficients must be nonzero in any strong sl2\mathfrak{sl}_2-action. Here is an argument that uses this sort of reasoning:

Example 3.2. The poset

The poset

Figure 1 (PDF p. 16).

does not admit a strong sl2\mathfrak{sl}_2-action.

Proof. Assume this admits a strong sl2\mathfrak{sl}_2-action. First note that r1r_1, r2r_2, u2u_2, v1v_1, and v2v_2 are all L(4)L(4)-isotypic vectors, and that s2s_2, s4s_4, t2t_2, t4t_4, and u4u_4 are all L(2)L(2)-isotypic vectors. All of these follow from the highest/lowest-weight and the unique neighbor propagation deduction rules from the previous section.

Now observe that E(u2)E(u_2) is in span⁡(t2,t3)\operatorname{span}(t_2,t_3), but it cannot be a multiple of t2t_2 as they are in different isotypic components. Therefore E(u2)E(u_2) must have a nonzero t3t_3 term. This forces F(t2)F(t_2) not to have a u2u_2 term. Otherwise, EF(t2)EF(t_2) would have a nonzero t3t_3 term but EF(t2)=2t2EF(t_2)=2t_2 since t2t_2 is a L(2)L(2)-isotypic vector in rank 0.

The known coefficients and isotypic vectors in the poset

Figure 2 (PDF p. 16).

The above picture is a depiction of what’s known so far: The purple vertices are L(4)L(4)-isotypic vectors, and the green ones are L(2)L(2)-isotypic. The solid arrow from u2u_2 to t3t_3 indicates the coefficient that we know must be nonzero, and the missing red arrow between t2t_2 and u2u_2 indicates the coefficient that we now know must be zero. The remaining edges are left dotted as a priori they could be either zero or nonzero.

From here we arrive at a contradiction since we see that F(t1)F(t_1) and F(t2)F(t_2) must both be multiples of u1u_1, but that can’t happen since FF is injective between ranks 0 and −2-2 in the representation L(4)2⊕L(2)2L(4)^2 \oplus L(2)^2. □

:::

Support deductions

Support deductions: Suppose PP is a poset admitting a strong sl2\mathfrak{sl}_2-action. Here are some situations where we can make these sorts of deductions:

  • Highest and lowest-weight vectors: Suppose because of one of our isotypic vector propagation rules we know that p∈Pkp \in P_k is a kk-isotypic vector. Well then E(p)=0E(p)=0 so we know that the EE-coefficients along any upwards edge out of pp must all be 0. Similarly for downward edges out of p∈P−kp \in P_{-k}.

  • Unique neighbors: Suppose p∈Pkp \in P_k has a unique neighbor in Pk−2P_{k-2} and that FF is known to be injective between those ranks. Then the FF-coefficient along that edge must be nonzero. This can also apply if pp has multiple neighbors in Pk−2P_{k-2} but we have already determined that the other FF-coefficients along those edges must be zero.

  • Unique non-isotypic neighbors: Suppose p∈Pkp \in P_k is known to be L(λ)L(\lambda)-isotypic, and that all but one of its neighbors in Pk−2P_{k-2} are known to be in other isotypic components. Then the FF-coefficient along the edge to the last vertex must be nonzero.

  • Unique isotypic neighbors: Suppose p∈Pkp \in P_k is known to be L(λ)L(\lambda)-isotypic, it has a neighbor q∈Pk−2q \in P_{k-2} which is also L(λ)L(\lambda)-isotypic. Suppose additionally that the span of all pp’s neighbors in Pk−2P_{k-2} is known to have exactly one L(λ)L(\lambda)-isotypic vector up to scalars. Then F(p)F(p) must be a scalar multiple of qq. Consequently, the FF-coefficient from pp to qq is nonzero and all other FF-coefficients out of pp are zero.

  • Unique matchings in Hall-tight neighborhoods: More generally suppose S⊂PkS \subset P_k has neighborhood T⊂Pk−2T \subset P_{k-2} with ∣S∣=∣T∣|S|=|T| and FF is injective between those ranks. At least one perfect matching between SS and TT must have all nonzero FF-coefficients as the determinant of the submatrix FS,TF_{S,T} is nonzero. If there is only one such matching, or if certain edges are contained in all such matchings, we can conclude those edges have nonzero FF-coefficients.

  • Down-up paths from isotypic vectors: Although EE and FF are not inverses to one another, for an isotypic weight vector v⃗\vec{v} we have EF(v⃗)=cv⃗EF(\vec{v})=c\vec{v} for some explicit constant cc. This constant depends on the weight and isotypic component, and is nonzero provided v⃗\vec{v} is not a lowest-weight vector. In particular if p∈Pkp \in P_k is an isotypic but not lowest-weight vector, and there is a unique two-step down-up path to some other q∈Pkq \in P_k then one of the coefficients along that path must be zero. If one is already known to be nonzero, then the other must be zero. We used this in Example 3.2.

  • Two-step paths from general vectors: For a non-isotypic vector in rank kk we don’t have EF(v⃗)=cv⃗EF(\vec{v})=c\vec{v} but we do have EF(v⃗)−FE(v⃗)=H(v⃗)=kv⃗EF(\vec{v})-FE(\vec{v})=H(\vec{v})=k\vec{v}. In particular if p∈Pkp \in P_k is any element and there is a unique two-step path (both down-up and up-down) to some other q∈Pkq \in P_k then one of the coefficients along that path must be zero. Again, if one of those is already known to be nonzero the other must be zero.

  • Longer paths and higher order relations: Suppose v⃗\vec{v} is an isotypic weight vector and F2(v⃗)≠0F^2(\vec{v}) \ne0, then EF2(v⃗)=cF(v⃗)EF^2(\vec{v})=cF(\vec{v}) for an explicit constant cc depending on the weight and isotypic component. If p∈Pkp \in P_k is known to be L(λ)L(\lambda)-isotypic with k>−λ+2k > -\lambda+2 then for any q∈Pk−2q \in P_{k-2} which is a neighbor of pp but not reachable by a down-down-up path from pp, the FF-coefficient from pp to qq must be zero. Similar arguments can be made for different kinds of paths and EmFnE^mF^n or FnEmF^nE^m.

Forbidden Configurations

Once we have more information about which edge coefficients are forced to be zero or nonzero, there are new ways to see a strong sl2\mathfrak{sl}_2-action cannot exist.

  • Simultaneously zero and nonzero: We may find that for one reason a coefficient is forced to be zero, but for another reason it is forced to be nonzero. Obviously this is a contradiction.

  • Breaking injectivity or surjectivity: For sl2\mathfrak{sl}_2 representations we know exactly when the EE and FF operators are injective or surjective between adjacent ranks. Since PP is Peck, the necessary Hall-matching conditions on the edges will all initially be satisfied automatically. However setting additional coefficients to zero might break these conditions, and making it so EE or FF cannot be injective/surjective in certain ranks.

Subrepresentations and quotients

Let PP be a poset with a strong sl2\mathfrak{sl}_2-action. Suppose S⊆PS \subseteq P is such that all of the EE- and FF-edge coefficients from elements of SS to elements of ScS^c are zero (the coefficients from ScS^c to SS may still be nonzero). It follows that that span⁡(S)\operatorname{span}(S) forms a submodule of CP\mathbb{C}P. Thus SS itself inherits a strong sl2\mathfrak{sl}_2-action. Moreover though, the image of ScS^c forms a basis for the quotient representation CP/span⁡(S)\mathbb{C}P/\operatorname{span}(S) with the same edge coefficients between them as in PP.

It turns out that sometimes you can tell from the combinatorics of PP that this must happen for any strong sl2\mathfrak{sl}_2-action on PP. Here is an example of how we might make such an argument:

Example 3.3. The poset

Poset diagramPoset diagrams showing the highlighted subrepresentation and quotient

Figure 11 (PDF p. 18).

does not admit a strong sl2\mathfrak{sl}_2-action.

Proof. Suppose we have a strong sl2\mathfrak{sl}_2-action on this poset. The element u1u_1 is a lowest-weight L(3)L(3)-isotypic vector. Unique neighbor propagation tells us t5t_5, s5s_5, and r3r_3 are all also L(3)L(3)-isotypic vectors. We can also then conclude using the up-down path arguments that the FF-edge coefficients from r3r_3 to s1s_1, from r3r_3 to s4s_4, and from t5t_5 to u3u_3 are all zero. In particular we see that r3r_3, s5s_5, t5t_5, u1u_1 must span a subrepresentation.

Therefore if we remove these vertices the images of the remaining vectors in the quotient representation naturally define a strong sl2\mathfrak{sl}_2-action on the complement. However in this new smaller poset we can easily spot a familiar contradiction: t~4\tilde{t}_4 is forced to be both L(1)L(1)-isotypic and L(3)L(3)-isotypic. □

Note that we can also go the other direction. Given strong sl2\mathfrak{sl}_2-actions on SS and ScS^c we can define one on PP by setting all coefficients between SS and ScS^c, in both directions, to zero. This gives us a recursive, divide and conquer approach to determining when strong sl2\mathfrak{sl}_2-actions exist: Assuming you can show that all EE- and FF-coefficients from SS to ScS^c must be zero then a strong sl2\mathfrak{sl}_2-action on PP exists if and only if such actions exist on SS and ScS^c.

Branching arguments

When proving posets do not admit symmetric chain decompositions, inevitably one needs to start branching arguments saying things like “The chain through vertex AA must connect to either BB or CC, if it connects to BB...” or “Either DD is the bottom of its chain, or it connects to EE...”.

Even though strong sl2\mathfrak{sl}_2-actions are more flexible than symmetric chain decompositions in that EE and FF can be supported along multiple edges, there are still times where these sorts of binary choices occur and we need to use similar branching arguments. Here is an example of such an argument:

Example 3.4. The poset

The poset

Figure 12 (PDF p. 19).

does not admit a strong sl2\mathfrak{sl}_2-action.

Proof. Assume we have a strong sl2\mathfrak{sl}_2-action. We can see immediately that r1r_1, r2r_2, s2s_2, s3s_3, u1u_1, and u2u_2 are L(3)L(3)-isotypic, and that s1s_1 and s4s_4 are L(1)L(1)-isotypic. We then know by injectivity that at least one of the FF-coefficients along the edges down from s1s_1 must be nonzero so let’s branch based on that.

Case 1: F(s1)F(s_1) has a nonzero t1t_1 term.

Chain reaction diagrams

Figure 13 (PDF p. 19).

This causes a chain reaction in the middle layer:

  • First we see that since EF(s1)=s1EF(s_1)=s_1 we must have Et1,s1≠0E_{t_1,s_1}\ne0 and that Et1,s2=0E_{t_1,s_2}=0, and therefore we see that t1t_1 is 11 isotypic as E(t1)=cs1E(t_1)=cs_1.

  • Next since EF(s2)=4s2EF(s_2)=4s_2 we must have Fs2,t1=0F_{s_2,t_1}=0 as well as Fs2,t3≠0F_{s_2,t_3}\ne0 and Et3,s2≠0E_{t_3,s_2}\ne0. We see F(s2)=ct3F(s_2)=ct_3 so t3t_3 is L(3)L(3)-isotypic.

  • Applying the same reasoning one more time, EF(s4)=s4EF(s_4)=s_4 we must have Fs4,t3=0F_{s_4,t_3}=0 as well as Fs4,t4≠0F_{s_4,t_4}\ne0 and Et4,s4≠0E_{t_4,s_4}\ne0. We see F(s4)=ct4F(s_4)=ct_4 so t4t_4 is L(1)L(1)-isotypic. We arrive at a contradiction, since span⁡(t1,t4)\operatorname{span}(t_1,t_4) consists entirely of L(1)L(1)-isotypic vectors but E(u1)∈span⁡(t1,t4)E(u_1)\in\operatorname{span}(t_1,t_4) is L(3)L(3)-isotypic.

Case 2: F(s1)F(s_1) has a nonzero t2t_2 term.

A completely analogous chain reaction happens, this time supported on the complementary set of edges on the middle row, and an analogous contradiction is arrived at, this time involving t2t_2, t3t_3, and u2u_2. Since both cases lead to a contradiction we therefore conclude there cannot be a strong sl2\mathfrak{sl}_2-action. □

Things to branch on

Here are a few situations where branching arguments have proven useful:

  • Two neighbors: In Example 3.4 the element s1s_1 had two downwards neighbors, and we knew at least one had to appear with nonzero coefficient in F(s1)F(s_1). This seems to be particularly useful when the source vertex is isotypic, and especially if there are other nearby elements with two neighbors to cause a chain reaction like in that example.

  • Two isotypic components: Suppose we know an element p∈Pp\in P is in the direct sum of the kk-isotypic component and the jj-isotypic component. It is sometimes useful to condition on whether or not pp actually has a jj-isotypic component or not. This is particularly useful if pp is in rank kk (or similar for −k-k), so then either pp is a highest-weight vector or E(p)E(p) is nonzero and jj-isotypic, either of which can be useful information.

  • Isotypic vectors in a span: Suppose S⊆PkS\subseteq P_k is known to contain both F(p)F(p) and E(q)E(q) where pp and qq are L(λ)L(\lambda)-isotypic vectors in ranks k+2k+2 and k−2k-2 respectively. Well, either F(p)=E(q)F(p)=E(q) up to a scalar, or they are linearly independent. Either of these options provides new and potentially useful information. This is particularly useful if SS is small or if it is known to contain other isotypic vectors.

Without loss of generality arguments

Sometimes one can show “If any strong sl2\mathfrak{sl}_2-action exists on PP, then there must exist an action such that [insert additional property] holds”. So without loss of generality we may assume the additional condition holds. This can in some circumstances simplify things greatly. Here is an example of such an argument:

Example 3.5. The poset

The poset

Figure 3.5. The poset does not admit a strong sl2\mathfrak{sl}_2-action.

Proof. First observe that r1r_1, r2r_2, and s1s_1 are all highest-weight vectors. Moreover, span⁡(F(r1),F(r2))=span⁡(s2,s3)\operatorname{span}(F(r_1), F(r_2)) = \operatorname{span}(s_2, s_3) so s2s_2 and s3s_3 are L(5)L(5)-isotypic. This poset has 180∘180^\circ rotational symmetry, so analogous claims hold on the bottom ranks as well.

There is actually much more symmetry here than just the rotational symmetry. There are a number of pairs of vertices that have the exact same neighbors. Here is a new picture with the isotypic data and symmetry highlighted:

Poset with isotypic data and symmetry highlighted

Figure 14 (PDF p. 21).

Here is the key observation: Suppose we have a basis for L(5)2⊕L(3)⊕L(1)2L(5)^2 \oplus L(3) \oplus L(1)^2 with support graph contained in this poset. For each of these groups of “twin” vertices (connected in orange) we may replace those basis vectors with any other basis for the same span and the support graph will still be contained in the poset.

So for example since span⁡(t3,t4,t5)\operatorname{span}(t_3,t_4,t_5) contains F(s2)F(s_2) and F(s3)F(s_3), we may assume without loss of generality that t3t_3 and t4t_4 are L(5)L(5)-isotypic. Using the 180∘180^\circ rotational symmetry again we can make a similar argument to assume u2u_2 and u3u_3 are L(5)L(5)-isotypic. Once we make those assumptions we have found the entire L(5)L(5)-isotypic component and can quotient it out:

Quotient of the poset after removing the isotypic component

Figure 15 (PDF p. 21).

In this quotient we can see that F(t~1)=cF(t~2)F(\widetilde{t}_1) = cF(\widetilde{t}_2), which is a contradiction as FF needs to be injective between rank 1 and −1-1. □

Remark: The 180∘180^\circ rotational symmetry in the above example is fairly rare. It was just convenient as an example to simplify the exposition. However we’ll note that having some combinatorial symmetry coming from “twin” vertices empirically seems fairly common among SCD-critical Peck posets.

The weak Bruhat order for H3H_3

Finally, we’ll close this section out with an example of a Peck poset which does not admit an SCD which occurs “in the wild”. Leclerc proved in [8] that the weak Bruhat order for the Coxeter group of type H3H_3 is Peck but does not admit an SCD. We have the following strengthening:

Theorem 3.6. The weak Bruhat order for the Coxeter group of type H3H_3 does not admit a strong sl2\mathfrak{sl}_2-action.

Proof. Suppose a strong sl2\mathfrak{sl}_2-action exists. Mimicking Leclerc’s argument (and reproducing his diagram), here are the middle four ranks poset plus a few relevant elements from the ranks above and below them:

Middle four ranks poset and relevant elements from the ranks above and below

Figure 16 (PDF p. 22).

Let SS denote the set of 5 vertices in rank 3 highlighted in purple. Collectively the elements of SS only have 4 neighbors in rank 5. Therefore, span⁡(S)\operatorname{span}(S) must contain a highest-weight vector v⃗\vec{v} of weight 3. Similarly let TT denote the set of 5 vertices in rank −3-3 highlighted in green. Collectively the elements of TT only have 4 neighbors in rank −5-5. Therefore, span⁡(T)\operatorname{span}(T) must contain a lowest-weight vector w⃗\vec{w}.

In this poset ∣P3∣−∣P5∣=1|P_3|-|P_5|=1. Thus, v⃗\vec{v} and w⃗\vec{w} are the unique L(3)L(3)-isotypic vectors in their respective ranks, up to scaling. Consequently, E3(w⃗)=λv⃗E^3(\vec{w})=\lambda\vec{v} for some nonzero scalar λ\lambda. Therefore ET,S3E^3_{T,S} has rank ≥1\ge1 so by the Lindström–Gessel–Viennot Lemma (the combinatorial version of the Cauchy–Binet Theorem, see Section 2.4) there must be at least one upward path connecting TT to SS. However by inspection we see no such path exists, a contradiction. □

Special cases and partial results

In this section we’ll prove some special cases and partial results toward Conjecture 1.2.

A single isotypic component

We’ll prove Conjecture 1.2 in the case where all of the ranks of PP have the same size. Such posets correspond to sl2\mathfrak{sl}_2-representations L(λ)kL(\lambda)^k with a single isotypic component. In fact we will only need an sl2\mathfrak{sl}_2-action in the sense of Proctor for this.

Theorem 4.1. Let PP be a ranked poset where all of the non-empty ranks have the same size. If PP admits an sl2\mathfrak{sl}_2-action in the sense of Proctor then PP admits an SCD.

Proof. In this case, the EE operator is injective between every pair of adjacent ranks, so on each layer we can pick a perfect matching. Those matchings combine to form an SCD. (Alternatively, we can observe that the outermost family of chains guaranteed by Griggs’ property TT form an SCD in this case) □

Indeed in this case we can get away only using edges corresponding to nonzero EE coefficients. We’ll note that if we have a strong sl2\mathfrak{sl}_2-action, we could also use all edges corresponding to nonzero FF-coefficients instead. Here are a couple more easy variations of this theorem:

Corollary 4.2. Let PP be an RSU poset where all of the non-empty ranks have the same size except for P0P_0. If PP admits an sl2\mathfrak{sl}_2-action in the sense of Proctor then PP admits an SCD.

Proof. The same argument works: The chains connecting the outer ranks will miss some vertices on the middle rank. Taking those missed vertices to be one vertex chains gives an SCD. □

Corollary 4.3. Let VV be a finite dimensional representation of sl2\mathfrak{sl}_2 with either all even or all odd highest-weights. If {b1,b2,…bn}\{b_1,b_2,\ldots b_n\} is a weight basis for VV consisting entirely of isotypic vectors, then the corresponding support graph admits an SCD.

Proof. The support graph decomposes as a disjoint union of support graphs for its isotypic components. Each of those components has an SCD, so the whole support graph does too. □

Diamond-free Posets

Four elements x,y,z,wx,y,z,w in a ranked poset form a diamond if z⋗x⋗wz \gtrdot x \gtrdot w and z⋗y⋗wz \gtrdot y \gtrdot w. In the Hasse diagram this looks like:

Hasse diagram of a diamond

Figure 17 (PDF p. 23).

We say a poset is diamond-free if no four elements form a diamond. Example 1.3 is an example of a diamond free Peck poset, which we saw does not admit a strong sl2\mathfrak{sl}_2-action. The following theorem generalizes that result:

Theorem 4.4. If PP is a diamond-free ranked poset admitting a strong sl2\mathfrak{sl}_2-action, then PP admits an SCD.

Proof. Suppose a strong sl2\mathfrak{sl}_2-action exists, and consider the action EFEF on a rank PnP_n. Suppose the matrix of EFEF has a nonzero non-diagonal entry (EF)x,y(EF)_{x,y}. Since H=EF−FEH=EF-FE acts diagonally on PnP_n, the corresponding entry (FE)x,y(FE)_{x,y} of FEFE must also be nonzero.

A nonzero entry (EF)x,y(EF)_{x,y} must come from at least one pair of nonzero entries Ex,wE_{x,w} and Fw,yF_{w,y}, and similarly nonzero entry (FE)x,y(FE)_{x,y} must come from at least one pair of nonzero entries Fx,zF_{x,z} and Ez,yE_{z,y}. However then x,y,zx,y,z and ww would form a diamond. Therefore we conclude that EFEF must act diagonally on PnP_n.

As discussed in § 5.2, the eigenvectors EFEF in a weight space are exactly the isotypic vectors. Therefore every element of the poset must be an isotypic vector for this action, which puts us into the situation of Corollary 4.3 and therefore we know an SCD exists. □

Width two posets

Next up we’ll look at posets of width 2, meaning the maximum rank size is 2. These correspond to representations L(λ1)⊕L(λ2)L(\lambda_1)\oplus L(\lambda_2).

Theorem 4.5. Let PP be a ranked poset with maximal rank size 2. PP admits a strong sl2\mathfrak{sl}_2-action if and only if PP admits an SCD.

Proof. First note that if any pair of adjacent ranks PnP_n and Pn−2P_{n-2}, at least one of which has size 2, is complete (meaning all possible edges between them exist) then PP admits an SCD. To see this pick a matching on every layer. If these matchings form an SCD we are done. If not, we can swap the matching at the complete layer to produce an SCD.

Suppose PP is a width 2 poset with a strong action of sl2\mathfrak{sl}_2 and no complete layers (besides between ranks of size 1). Every layer looks like one of the following:

Possible layers in a width two poset

Figure 18 (PDF p. 23).

All of these configurations have the property that if the top rank is known to consist of isotypic vectors then the bottom rank must also consist of isotypic vectors. In particular since the maximal rank in the poset always consists of isotypic vectors, this means the entire poset consists of isotypic vectors, so we know an SCD exists by Corollary 4.3. □

Remark: A variation of this argument can also be used to prove Conjecture 1.2 for posets which are width 2 except for a wide middle layer. These correspond to representations L(λ1)⊕L(λ2)⊕L(0)kL(\lambda_1) \oplus L(\lambda_2) \oplus L(0)^k.

Rank 4 posets

The rank 2 and 3 cases of Conjecture 1.2 are covered by Theorem 4.1 and Corollary 4.2. Indeed for those cases all Peck posets admit an SCD. So in a sense rank 4 is the first real case where the distinction between Proctor’s sl2\mathfrak{sl}_2-actions and our strong actions matters, and we have already seen several examples of rank 4 posets which are Peck but do not admit an SCD (Examples 1.3, 3.1, 3.3, and 3.4). For the remainder of this section we will prove some special cases and partial results in the rank 4 case.

Saturated and cosaturated sets

Let PP be a poset equipped with a strong sl2\mathfrak{sl}_2-action. A set S⊆PkS \subseteq P_k is L(λ)L(\lambda)-saturated if the sl2\mathfrak{sl}_2-equivariant projection πλ\pi_\lambda onto the L(λ)L(\lambda)-isotypic component restricts to a bijection from span⁡(S)\operatorname{span}(S) to the degree kk part of the L(λ)L(\lambda) component. In other words, S⊆PkS \subseteq P_k is L(λ)L(\lambda)-saturated if ∣S∣|S| is equal to the multiplicity of L(λ)L(\lambda) and span⁡(S)\operatorname{span}(S) intersects the sum of the non-L(λ)L(\lambda)-isotypic components trivially.

Lemma 4.6. Let S⊆PkS \subseteq P_k be a L(λ)L(\lambda)-saturated set.

(a) If k<λk < \lambda then we can find a L(λ)L(\lambda)-saturated set S′⊆Pk+2S' \subseteq P_{k+2} such that there is a matching between SS and S′S' consisting of edges with nonzero EE-coefficients.

(b) If k>−λk > -\lambda then we can find a L(λ)L(\lambda)-saturated set S′′⊆Pk−2S'' \subseteq P_{k-2} such that there is a matching between SS and S′′S'' consisting of edges with nonzero FF-coefficients.

Proof. We’ll just prove (a), as the proof for (b) is similar. Since SS is L(λ)L(\lambda)-saturated, πλ\pi_\lambda is full rank from span⁡(S)\operatorname{span}(S) to the degree kk part of the L(λ)L(\lambda) component. Moreover, since k<λk < \lambda we know EE has full rank between ranks kk and k+2k + 2 on the L(λ)L(\lambda) component. Therefore E∘πλE \circ\pi_\lambda is a full rank map from span⁡(S)\operatorname{span}(S) to the degree k+2k + 2 part of the L(λ)L(\lambda) component. However since πλ\pi_\lambda is an equivariant map we have E∘πλ=πλ∘EE \circ\pi_\lambda= \pi_\lambda\circ E. So Cauchy-Binet tells us that if πλ∘E\pi_\lambda\circ E has full rank we can find S′⊂Pk+2S' \subset P_{k+2} such that both ES′,SE_{S',S} and (πλ)L(λ),S′(\pi_\lambda)_{L(\lambda),S'} have full rank. Those conditions correspond to the existence of a matching along edges with nonzero EE-coefficients and S′S' being L(λ)L(\lambda)-saturated respectively. □

Dually we say that S⊆PkS \subseteq P_k is L(λ)L(\lambda)-cosaturated if the coordinate projection πS\pi_S restricts to a bijection from the weight kk part of the L(λ)L(\lambda)-isotypic component to span⁡(S)\operatorname{span}(S). In other words, S⊆PkS \subseteq P_k is L(λ)L(\lambda)-cosaturated if ∣S∣|S| is equal to the multiplicity of L(λ)L(\lambda) and the span of its complement span⁡(Sc)\operatorname{span}(S^c) intersects the L(λ)L(\lambda)-isotypic component trivially. We get a similar result:

Lemma 4.7. Let S⊆PkS \subseteq P_k be a L(λ)L(\lambda)-cosaturated set.

(a) If k<λk < \lambda then we can find a L(λ)L(\lambda)-cosaturated set S′⊆Pk+2S' \subseteq P_{k+2} such that there is a matching between SS and S′S' consisting of edges with nonzero FF-coefficients.

(b) If k>−λk > -\lambda then we can find a L(λ)L(\lambda)-cosaturated set S′′⊆Pk−2S'' \subseteq P_{k-2} such that there is a matching between SS and S′′S'' consisting of edges with nonzero EE-coefficients.

Proof. Nearly identical to the proof of 4.6, just using the inclusion map ιλ\iota_\lambda from the L(λ)L(\lambda) component into the representation instead of the projection map πλ\pi_\lambda. □

The point is that these are linear algebraic conditions which ensure that SS could be the intersection of a collection of L(λ)L(\lambda)-chains with PkP_k.

Corollary 4.8. If S⊆PkS \subseteq P_k is either L(λ)L(\lambda)-saturated or L(λ)L(\lambda)-cosaturated, then we can find mλm_\lambda disjoint saturated chains connecting PλP_\lambda to P−λP_{-\lambda} which intersect PkP_k at SS, where mλm_\lambda is the multiplicity of L(λ)L(\lambda).

Proof. Use Lemma 4.6 or 4.7 to grow the paths outward from SS. □

Now let’s turn our attention back to rank 4. We only have two isotypic components in this case, so we get a tighter relationship between saturated and cosaturated sets:

Lemma 4.9. Let PP be a rank 4 poset with a strong sl2\mathfrak{sl}_2-action.

(1) S⊆P±1S \subseteq P_{\pm1} is L(3)L(3)-saturated if and only if Sc=P±1−SS^{c} = P_{\pm1} - S is L(1)L(1)-cosaturated.

(2) S⊆P±1S \subseteq P_{\pm1} is L(3)L(3)-cosaturated if and only if Sc=P±1−SS^{c} = P_{\pm1} - S is L(1)L(1)-saturated.

Proof. In general a set T⊆PkT \subseteq P_k is L(λ)L(\lambda)-saturated if span⁡(Tc)\operatorname{span}(T^{c}) contains no nonzero L(λ)L(\lambda)-isotypic vectors, and is L(λ)L(\lambda)-cosaturated if span⁡(T)\operatorname{span}(T) intersects the sum of the other isotypic components trivially. In particular if there are only two isotypic components then the cosaturated condition just becomes “TT is L(λ)L(\lambda)-cosaturated if span⁡(T)\operatorname{span}(T) intersects the other component trivially” – but that exactly means TcT^{c} is saturated for the other component. □

Rank vectors [1,a+1,a+1,1][1, a + 1, a + 1, 1] and [a,a+1,a+1,a][a, a + 1, a + 1, a]

We will now prove Conjecture 1.2 in the case where the rank vector of the poset is either [1,a+1,a+1,1][1, a + 1, a + 1, 1] or [a,a+1,a+1,a][a, a + 1, a + 1, a]. In other words, when the underlying representation is either L(3)⊕L(1)aL(3) \oplus L(1)^a or L(3)a⊕L(1)L(3)^a \oplus L(1).

As noted in Section 2.4, if a square matrix has nonzero determinant then the corresponding bipartite graph has a perfect matching. The following lemma gives us a mild strengthening of this when we have a block decomposition satisfying certain rank conditions.

Lemma 4.10. Suppose n=k+ℓ=p+qn = k + \ell= p + q, and MM is an invertible n×nn \times n matrix with a (k+ℓ)×(p+q)(k + \ell) \times(p + q) block decomposition

M=[ABCD]M = \begin{bmatrix} A & B \\ C & D \end{bmatrix}

such that rank⁡(C)<p\operatorname{rank}(C) < p. Then the corresponding graph has a perfect matching using at least one edge corresponding to an entry in AA.

Proof. Expand the determinant along the first pp columns. Because det⁡(M)≠0\det(M) \ne0, some p×pp \times p minor using those columns and its complementary minor must both be nonzero. Since rank⁡(C)<p\operatorname{rank}(C) < p the first minor must use at least one row from AA. The two complementary minors therefore give a matching using an edge corresponding to an entry of AA. □

Theorem 4.11. Let PP be a poset with rank vector [1,a+1,a+1,1][1, a + 1, a + 1, 1]. PP admits a strong sl2\mathfrak{sl}_2-action if and only if PP admits a symmetric chain decomposition.

Proof. Fix a strong sl2\mathfrak{sl}_2-action on PP. Let P1=S⊔TP_1 = S \sqcup T where SS is the set of vertices with an edge to P3P_3, and TT is the set of vertices which do not. Similarly let P−1=S′⊔T′P_{-1} = S' \sqcup T' where S′S' is the set of vertices with an edge to P−3P_{-3}, and T′T' is the set of vertices which do not.

Bipartite graph showing the sets S, T, S′, and T′

Figure 1 (PDF p. 25).

It suffices to show that we can find a perfect matching on the middle layer such that at least one element of SS gets matched with an element of S′S'. Consider the block decomposition of EP1,P−1E_{P_1,P_{-1}} corresponding to the decompositions P1=S⊔TP_1 = S \sqcup T and P−1=S′⊔T′P_{-1} = S' \sqcup T':

EP−1,P1=[ES,S′ES,T′ET,S′ET,T′]E_{P_{-1},P_1}=\begin{bmatrix} E_{S,S'} & E_{S,T'} \\ E_{T,S'} & E_{T,T'} \end{bmatrix}

In particular if we can show the conditions of Lemma 4.10 hold that would prove the theorem. We already know EE is invertible between those ranks, so we just need to check that rank⁡(ET,S′)<∣S′∣\operatorname{rank}(E_{T,S'})<|S'|.

To see this, notice that span⁡(S′)\operatorname{span}(S') contains the unique (up to scaling) L(3)L(3)-isotypic vector in rank −1-1. This vector gets mapped by EE to the unique L(3)L(3)-isotypic vector in rank 11, which we similarly know is contained in span⁡(S)\operatorname{span}(S). The matrix ET,S′E_{T,S'} corresponds to applying EE to span⁡(S′)\operatorname{span}(S') and then projecting onto span⁡(T)\operatorname{span}(T), killing span⁡(S)\operatorname{span}(S). In particular the unique L(3)L(3)-isotypic vector gets killed in the second step, so rank⁡(ET,S′)<∣S′∣\operatorname{rank}(E_{T,S'})<|S'|.

Theorem 4.12

Theorem 4.12. Let PP is a poset with rank vector [a,a+1,a+1,a][a,a+1,a+1,a]. PP admits a strong sl2\mathfrak{sl}_2-action if and only if PP admits a symmetric chain decomposition.

Proof. The proof will proceed similarly to the [1,a+1,a+1,1][1,a+1,a+1,1] case. We define the decompositions P1=S⊔TP_1=S\sqcup T and P−1=S′⊔T′P_{-1}=S'\sqcup T' slightly differently. Let SS be the set of those x∈P1x\in P_1 that either have nonzero projection onto the L(1)L(1)-isotypic component, or those such that the L(1)L(1)-isotypic component is not killed by the coordinate projection onto xx. In other words those xx such that {x}\{x\} is either L(1)L(1)-saturated or L(1)L(1)-cosaturated. Similarly define S′⊆P−1S'\subseteq P_{-1} and let TT and T′T' be their respective complements.

Diagram showing the sets S, S', T, and T' and auxiliary purple edges

Figure 19 (PDF p. 26).

In the above picture the purple edges are auxiliary, representing the nonzero matrix coefficients for the inclusion and projection maps between the L(1)L(1) component and the full representation. An element in the poset having such an edge exactly corresponds to it being L(1)L(1)-saturated or L(1)L(1)-cosaturated.

Now we proceed exactly as in Theorem 4.11, using Lemma 4.10 to show that we can find a matching where at least one a∈Sa\in S is matched with b∈S′b\in S'. Then by Lemma 4.9, if {a}\{a\} is L(1)L(1)-saturated, then P1−{a}P_1-\{a\} is L(3)L(3)-cosaturated. Similarly, if {a}\{a\} is L(1)L(1)-cosaturated then P1−{a}P_1-\{a\} is L(3)L(3)-saturated. In either case P1−{a}P_1-\{a\} has a matching with P3P_3, and P−1−{b}P_{-1}-\{b\} has a matching with P−3P_{-3}. Therefore the matching on the middle layer from Lemma 4.9 can be extended to an SCD.

Fractional SCDs for Rank 4 Posets

Now we will turn our attention to rank 4 posets with arbitrary rank vectors, but relax things to consider fractional symmetric chain decompositions. We have the following theorem:

Theorem 4.13. If PP is a rank 4 poset admitting a strong sl2\mathfrak{sl}_2-action then PP admits a fractional symmetric chain decomposition.

Proof. Let PP be a rank 4 poset with a strong sl2\mathfrak{sl}_2-action. For this argument we will think of a fractional symmetric chain decomposition as a collection of symmetric saturated chains covering every vertex of PP the same number of times.

Choose a L(3)L(3)-saturated set S0⊆P1S_0 \subseteq P_1. By Lemma 4.6, we can find a L(3)L(3)-saturated set S0′⊆P−1S'_0 \subseteq P_{-1} it matches with and we can extend that matching to a collection of disjoint length 4 chains. Set T0=S0′c⊆P−1T_0=S_0^{\prime c}\subseteq P_{-1}. By Lemma 4.9 T0T_0 is L(1)L(1)-cosaturated, so therefore we can match it with a L(1)L(1)-cosaturated set T0′⊆P1T'_0\subseteq P_1 by Lemma 4.7.

Diagram of the matching between saturated and cosaturated sets

Figure 20 (PDF p. 27).

Notice that if T0′=S0cT'_0=S_0^c then we have found a symmetric chain decomposition. If T0′≠S0cT'_0\ne S_0^c we let S1=T0′cS_1=T_0^{\prime c}, which is L(3)L(3)-saturated by Lemma 4.9, and we repeat the process. For each n≥0n\geq0, we get Sn⊆P1S_n\subseteq P_1 and Sn′⊆P−1S'_n\subseteq P_{-1} which are L(3)L(3)-saturated sets with a matching between them, and Tn=Sn′c⊆P−1T_n=S_n^{\prime c}\subseteq P_{-1} and Tn′⊆P1T'_n\subseteq P_1 are L(1)L(1)-cosaturated sets with a matching between them, with Sn+1=Tn′cS_{n+1}=T_n^{\prime c}.

Four diagrams showing the successive matching steps

Figure 21 (PDF p. 27).

There are only finitely many options for the SnS_n’s so at some point this process must repeat. Suppose Tn′c=SmT'_n{}^c=S_m for some m<nm<n. We claim that all of the chains for steps mm through nn form a fractional symmetric chain decomposition. It is clear from the construction that each vertex in ranks 33, −1-1 and −3-3 is covered once in each step. In rank 11 for aa between mm and n−1n-1 Ta′T'_a and Sa+1S_{a+1} together cover all of the vertices once, and then Tn′T'_n and SmS_m cover them one final time.

In the example above we stopped after 4 steps since T3′c=S1T'_3{}^c=S_1. One can easily verify that the chains in the last 3 pictures collectively cover each vertex exactly 3 times.

More questions and special cases

Of course a full solution to Conjecture 1.2 would be great, but there are many interesting special cases and variations one could consider beyond those we just did. Moreover we think many of those directions would still be interesting even if the full conjecture ends up being false. Here are a few we’d like to highlight:

Open Problem: For which rank vectors does Conjecture 1.2 hold?

Of course we think the answer is all of them, but there is lots of room for partial progress. We gave a few infinite families in this section, and have checked it computationally for rank vectors with fewer than 14 elements. All rank 4 posets or width 3 posets might be natural candidates for the next families to look at.

Open Problem: Do rank 5 posets always admit fractional SCDs?

The rank 4 argument breaks down in rank 5. It still produces a family of chains covering every element outside of the middle rank the same number of times. However its possible that one of the middle elements gets used too many times to to be able turn that collection into a fractional SCD. Perhaps one can prevent this from happening though with a more careful argument.

Open Problem: Do posets with positive sl2\mathfrak{sl}_2-actions always admit (fractional) SCDs?

The Young lattices L(m,n)L(m,n) have the additional property that all of the EE and FF coefficients can be taken to be non-negative real numbers. We call these positive actions. These greatly restrict the combinatorics of the poset by preventing certain terms from cancelling in the sl2\mathfrak{sl}_2 relations, and seem to be quite rare.

Computational Experiments

Our main computational experiment can be summarized as follows:

  1. Generate either a random Peck poset or a random SCD-critical Peck poset.

  2. Test whether it admits an SCD or a fractional SCD.

  3. If it does not admit an SCD try to prove it does not admit a strong sl2\mathfrak{sl}_2-action.

  4. If it does not admit an SCD but we fail to prove it cannot admit a strong sl2\mathfrak{sl}_2-action, search for a strong action numerically.

  5. Repeat.

We’ll briefly explain how we did these steps, and explain certain choices we made in the process, and then we will report the outcome of these experiments.

Generating random Peck posets

Recall the characterization of Peck posets due to Griggs (Definition/Theorem 2.1): An RSU poset is Peck if and only if between any pair of opposite ranks there exists a family of disjoint saturated chains connecting them. This allows one to construct random Peck posets via the following procedure:

  1. Fix a symmetric and unimodal rank vector, and put the appropriate number of elements in each rank.

Diagram of elements arranged by rank

Figure 22 (PDF p. 28).

(2) For each pair of opposite ranks randomly select a collection of disjoint saturated chains connecting them, and add those cover relations to the poset.

Diagram of disjoint saturated chains connecting opposite ranks

Figure 23 (PDF p. 29).

(3) Between each pair of adjacent ranks add some additional cover relations with some probability.

Diagram of a poset with additional cover relations between adjacent ranks

Figure 24 (PDF p. 29).

We’ll note that this procedure does not sample from the set of Peck posets with given edge densities uniformly. A Peck poset occurs with probability roughly proportional to the number of families of symmetric saturated chains satisfying the Griggs condition (the number of ways we could have gotten it from this process).

We also tried a procedure that samples uniformly, and it was far less computationally efficient. Moreover, Peck posets with very few Griggs witness families tend to not admit SCDs, fractional SCDs, or strong sl2\mathfrak{sl}_2-actions very often, and the bottlenecks/obstructions tend to be very simple. The procedure described above seems to produce many more interesting examples which can fail to admit these structures for more subtle reasons.

Testing for (fractional) SCDs

Once we have a Peck poset the first thing we want to do is check whether or not it admits an SCD. We experimented with 3 ways to do this:

Method 1: Enumerate all symmetric saturated chains and perform a brute force search.

Method 2: First run a program to find certain easy deductions like “xx must be the start of its chain” or “the chain through yy must also go through zz”, then use these to trim the poset before performing a brute force search as in method 1.

Method 3: Use built-in integer programming tools in Python to search for integer points in the flow polytope (see § 4.10).

Method 1 was the most efficient for relatively small rank vectors or low edge densities. Method 2 was best for certain middle-complexity posets. Once the posets were sufficiently large and complicated, method 3 was the most efficient. Another advantage of method 3 is that it also tells us whether the poset admits a fractional SCD essentially for free since the integer program solvers typically solve the linear program along the way.

Generating random SCD-critical posets

Starting with a Peck poset which does not admit an SCD one can add in edges at random one at a time, verifying each time that the added edge does not create an SCD. Once you cannot add any more edges, the resulting poset is an SCD-critical Peck poset.

For small enough rank vectors this is reasonably computationally tractable (in fact for very small rank vectors there were few enough SCD-critical Peck posets that we could just enumerate them all). However for larger rank vectors this becomes computationally intensive because one needs to keep checking whether the poset admits an SCD, so we were able to test far fewer of these cases for larger rank vectors.

The strong sl2\mathfrak{sl}_2-action autodisprover

Given a Peck poset which we have already verified does not admit an SCD, we then want to prove it also does not admit a strong sl2\mathfrak{sl}_2-action. To do this we wrote a Python program to automatically generate proofs that a given poset does not admit a sl2\mathfrak{sl}_2-action.

This is in some sense the most interesting thing we built for the experiment, but we’ve essentially already explained how it works: Section 3 lists a series of deductions, implications, reductions, and contradictions one can make when trying to prove a given poset does not have a strong sl2\mathfrak{sl}_2-action. The program just iteratively applies these rules until it either finds a contradiction or gets stuck.

Importantly though we’ll note that if the autodisprover does not find a proof, that does not mean the poset admits a strong sl2\mathfrak{sl}_2-action. Indeed, we analyzed many posets that earlier versions of the autodisprover failed to catch. Developing proofs for those examples is how we came up with many of the techniques described in Section 3. The failure rate of the autodisprover is mostly a reflection of where we decided to stop this process rather than some inherent limit on these methods.

Searching for numerical solutions

Whenever our disprover fails we have a candidate counterexample to the conjecture. We then want to just see if we can construct a strong sl2\mathfrak{sl}_2 action, by directly solving the equations involved. For small enough posets we could use Gröbner bases to get a definitive answer one way or another, however for larger posets that became computationally infeasible. Instead here’s what we did:

If we fix the raising operator EE, finding the corresponding FF operator satisfying the sl2\mathfrak{sl}_2-relations is just a linear system of equations. However the solution will typically not be supported along the edges of the poset, so instead we force it to be supported on the poset and find a least squares solution to the sl2\mathfrak{sl}_2-relations. Once we have that FF we flip things around and find a new EE that is supported on the edges we want and is a least squares solution to the sl2\mathfrak{sl}_2-relations. We repeat this process, alternating solving for EE and FF, getting operators supported on our edges which are closer to satisfying the sl2\mathfrak{sl}_2-relations with every step.

This is by no means guaranteed to converge to a solution, and indeed often it gets stuck at local minima even when we know solutions exist. However in practice, with enough starting seeds and a bit of random noise added in, it seems to do pretty well at finding solutions when they exist.

Choosing rank sizes and densities

We wanted to test the conjecture on a variety of posets of different shapes and edge densities. Here’s a bit more on how we decided what to test:

  • The number of samples tested is a function of time. For each rank (rank 4 to rank 10) we ran one ~8 hour sweep, ~6 hours constructing and testing random posets, and ~2 hours constructing and testing SCD-critical posets. In each rank we cycled through the rank vectors being tested so they each got roughly the same number of samples. This did mean that the larger rank vectors tended to use up most of the time. Had we instead given the rank vectors equal time we would have tested far more small cases, but fewer large ones. Since larger rank vectors tended to produce harder more interesting examples, we prioritized those.

  • The rank vectors tested were chosen by what seemed computationally feasible in test runs, as well as a desire to test posets of a variety of shapes (e.g. thin at the extremes and wide in the middle, or relatively boxy with almost equal rank sizes).

  • We skipped any cases covered by results in Section 4.

  • We skipped all cases with fewer than 14 elements. Bonus Theorem: Conjecture 1.2 holds for ranked posets with fewer than 14 elements. During the process of building and testing the autodisprover we completely settled these cases by enumerating all SCD-critical posets with fewer than 14 elements which are not covered by the results in Section 4 and checking they don’t admit a fractional SCD. At 14 elements the enumeration process became too computationally intensive for some rank vectors, which is why we stopped there.

  • For each rank vector we ran a quick sweep of different edge densities to get an estimate of how often they produce posets with SCDs vs without. We then chose three density profiles. The sparse profile was not so sparse that SCDs never appeared. The medium profile produced a mixture of SCD and non-SCD examples. The dense profile was not so dense that every example admitted an SCD. These different regimes tended to produce examples that feel qualitatively different, but the totals reported below are these 3 combined.

  • For each chosen rank vector we also tested SCD-critical posets, which tend to be much more dense than the “dense” random regime but extremely unlikely to show up in random searches. These are much more computationally intensive to both construct and test, so even though we dedicated a quarter of the total run time to these, they make up a far smaller portion of the total.

Our main interest was in providing evidence for, or finding a counterexample to, Conjecture 1.2, although we did also decide to track how often the posets had fractional SCDs but not ordinary SCDs. There are many other interesting questions one could ask about the distributions of Peck posets produced this way – including a more detailed analysis of how the density affects the probability the poset will have an SCD.

Experimental Results

Below are tables summarizing the results of this computational experiment. The totals in the random poset columns are combined across the sparse, medium, and dense bands. The “Fractional” columns record how many posets we found which admit a fractional SCD but not an ordinary SCD. The “Open” columns record the number of posets where the autodisprover was not able to certify that a strong sl2\mathfrak{sl}_2-action does not exist. For every open case, we also conducted a numerical search for a solution. None of these searches found one.

Rank 4
Random posetsSCD-critical posets
Rank vectorTotalNon-SCDFractionalOpenTestedFractionalOpen
(2, 5, 5, 2)457,722186,592204900
(2, 9, 9, 2)457,722177,8611204900
(3, 5, 5, 3)457,722182,887204900
(3, 8, 8, 3)457,722180,9701604900
(3, 10, 10, 3)457,722177,5401204900
(4, 7, 7, 4)457,722176,066304900
(4, 11, 11, 4)457,721187,2262104911
(6, 8, 8, 6)457,719177,697104900
(6, 11, 11, 6)457,719179,8121804900
(6, 13, 13, 6)457,719174,5282404911
(8, 10, 10, 8)457,719175,549304800
(10, 15, 15, 10)457,719186,703504800
Total5,492,6482,163,431119058622

Table 1.

Rank vectorTotalNon-SCDFractionalOpenTestedFractionalOpen
(1, 4, 4, 4, 1)556,761240,0160041800
(1, 5, 7, 5, 1)556,761263,0127041800
(2, 4, 7, 4, 2)556,761255,38910041800
(2, 5, 5, 5, 2)556,761253,32317041820
(2, 6, 8, 6, 2)556,761247,867370418017
(3, 4, 7, 4, 3)556,758230,1417041800
(3, 5, 6, 5, 3)556,758277,83318041801
(3, 6, 8, 6, 3)556,758242,923320418310
(3, 7, 7, 7, 3)556,758290,867380418319
(4, 5, 5, 5, 4)556,758249,2110041800
(4, 6, 8, 6, 4)556,758269,68538041713
(4, 7, 8, 7, 4)556,758263,773480417318
Total6,681,1113,084,04025205,0141268

Table 2. Rank 5. Random posets: Total, Non-SCD, Fractional, Open. SCD-critical posets: Tested, Fractional, Open.

Rank vectorTotalNon-SCDFractionalOpenTestedFractionalOpen
(1, 2, 4, 4, 2, 1)310,740146,2250017210
(1, 3, 6, 6, 3, 1)310,740158,2459017211
(1, 4, 5, 5, 4, 1)310,740146,34814017200
(1, 5, 7, 7, 5, 1)310,740141,90113017203
(2, 3, 4, 4, 3, 2)310,740142,8963017200
(2, 4, 6, 6, 4, 2)310,740142,22723017211
(2, 5, 5, 5, 5, 2)310,740146,45329017220
(3, 4, 5, 5, 4, 3)310,740143,6403017201
(3, 5, 5, 5, 5, 3)310,740150,87119017202
(3, 5, 7, 7, 5, 3)310,740139,17755017203
(3, 6, 9, 9, 6, 3)310,739124,519870172325
(4, 6, 9, 9, 6, 4)310,737145,87262017129
Total3,728,8761,728,37431702,0631045

Table 3. Rank 6. Random posets: Total, Non-SCD, Fractional, Open. SCD-critical posets: Tested, Fractional, Open.

Rank vectorTotalNon-SCDFractionalOpenTestedFractionalOpen
(1, 2, 3, 4, 3, 2, 1)51,35721,779201800
(1, 2, 5, 6, 5, 2, 1)51,35722,847001801
(1, 4, 4, 7, 4, 4, 1)51,35717,054401800
(1, 5, 7, 9, 7, 5, 1)51,35722,415601802
(2, 3, 4, 5, 4, 3, 2)51,35722,681101800
(2, 4, 6, 6, 6, 4, 2)51,35725,9421001800
(2, 6, 8, 9, 8, 6, 2)51,35723,9713601816
(3, 4, 6, 8, 6, 4, 3)51,35723,2221001801
(3, 6, 6, 9, 6, 6, 3)51,35722,7661601802
(3, 7, 10, 11, 10, 7, 3)51,35720,5687301816
(4, 5, 7, 7, 7, 5, 4)51,35722,107201800
(4, 7, 9, 11, 9, 7, 4)51,35622,6073901827
Total616,283267,9591990216425

Table 4. Rank 7. Random posets: Total, Non-SCD, Fractional, Open. SCD-critical posets: Tested, Fractional, Open.

Rank vectorTotalNon-SCDFractionalOpenTestedFractionalOpen
(1, 2, 3, 4, 4, 3, 2, 1)29,49611,73110800
(1, 2, 5, 6, 6, 5, 2, 1)29,49615,07730801
(1, 4, 4, 7, 7, 4, 4, 1)29,49612,665110800
(1, 5, 7, 9, 9, 7, 5, 1)29,49615,019100800
(2, 3, 4, 5, 5, 4, 3, 2)29,49611,26720800
(2, 4, 6, 6, 6, 6, 4, 2)29,49614,549110801
(2, 6, 8, 9, 9, 8, 6, 2)29,49514,892420813
(3, 4, 6, 8, 8, 6, 4, 3)29,49313,420100801
(3, 6, 6, 9, 9, 6, 6, 3)29,49310,827500813
(4, 5, 7, 7, 7, 7, 5, 4)29,4939,53920800
(4, 7, 9, 10, 10, 9, 7, 4)29,49313,118490824
(4, 8, 10, 10, 10, 10, 8, 4)29,49313,924790803
Total353,936156,028270096416

Table 5.

Takeaways and Caveats

Across 16,872,854 randomly generated Peck posets, every non-SCD example was certified not to admit a strong action. Among 7,975 SCD-critical posets, the autodisprover certified 7,819, leaving 156 unresolved. Numerical searches found no strong action in any unresolved case. While this did produce a small number of unresolved open cases, where we could neither prove nor disprove the existence of a strong sl2\mathfrak{sl}_2-action, overall we think these results are compelling computational evidence for the conjecture. Here are a few additional remarks on these results:

  • The autodisprover has essentially completely saturated testing posets which appear in the random searches. This is likely due in part to the fact that we essentially trained it on this data. We developed the autodisprover iteratively, adding new arguments whenever it failed on an example. By the end, we were training it on examples much rarer and more complicated than those arising in ordinary random searches. It is possible that potential counterexamples have to be far more structured than random posets, and we are just missing them entirely.

  • On SCD-critical posets (the worst case scenario for showing an sl2\mathfrak{sl}_2-action does not exist) the success rate drops. In total we had 156 open cases out of 7975 SCD-critical posets tested – about a 98% success rate overall. This is still a high overall success rate. For some larger shapes, however, the success rate dropped significantly. The worst was (4, 7, 9, 10, 10, 9, 7, 4) with a 50% success rate.

  • Not shown in these tables is the intermediate edge-density regime in between where random searches start producing posets which have SCDs nearly 100% of the time, and the maximally dense SCD-critical posets. One can test these by starting with a random non-SCD poset and iteratively adding edges while making sure it doesn’t create an SCD, but stopping short of making it SCD-critical. We decided to only test the purely random, and maximally difficult regimes for this experiment, but it’s possible that some interesting behavior occurs in this intermediate regime that is not captured by this data.

  • The numerical solver is not guaranteed to find a solution if one exists, so all of the open cases are potential counterexamples to the conjecture. That being said, we ran versions of this experiment many times and tested many more open cases than just the 156 in this official run, and we have yet to find an actual counterexample.

  • This is just a snapshot of where the autodisprover is right now. As we worked on this we watched the success rate on both random and SCD-critical posets creep up with every iteration. We have no reason to think that it couldn’t be improved further to cover more of those remaining open cases.

Gallery of Failures

To close things out, here are a few of the “open” cases in which our autodisprover failed to disprove the existence of a strong sl2\mathfrak{sl}_2-action. Our searches for a numerical solution on these all failed, but they are all potential counterexamples to Conjecture 1.2.

Three graphs of potential counterexamples

Figure 1 (PDF p. 34).

Network diagram with vertices labelled $r_i$, $s_i$, $t_i$, $u_i$, $v_i$, and $w_i$Network diagram with vertices labelled $r_i$, $s_i$, $t_i$, $u_i$, $v_i$, $w_i$, $x_i$, and $y_i$

Figure 25 (PDF p. 35). Do any of these admit a strong sl2\mathfrak{sl}_2-action?

References

  1. [1]Ravindra K. Ahuja, Thomas L. Magnanti, and James B. Orlin. Network Flows: Theory, Algorithms, and Applications. Englewood Cliffs, NJ: Prentice Hall, 1993.
  2. [2]Yaël Dillies, Matthew Johnson, and Aleksandra Kowalska. A Generalisation of Sperner’s Theorem Using Weighted Chains. 2025. arXiv: 2509.26493 [math.CO]. URL: https://arxiv.org/abs/2509.26493.
  3. [3]Robert G. Donnelly. “Extremal Properties of Bases for Representations of Semisimple Lie Algebras”. In: Journal of Algebraic Combinatorics 17.3 (2003), pp. 255–282. DOI: 10.1023/A:1025096704917.DOI
  4. [4]Robert G. Donnelly. “Solitary Bases for Irreducible Representations of Semisimple Lie Algebras”. In: FPSAC ’98. Extended abstract, 12 pages. 1998. URL: https://campus.murraystate.edu/academic/faculty/rdonnelly/Research/FPSAC98-Abstract.pdf.
  5. [5]Ira Gessel and Gérard Viennot. “Binomial determinants, paths, and hook length formulae”. In: Advances in Mathematics 58.3 (1985), pp. 300–321. DOI: 10.1016/0001-8708(85)90121–5.DOI
  6. [6]Jerrold R. Griggs. “On chains and Sperner k-families in ranked posets”. In: Journal of Combinatorial Theory, Series A 28.2 (1980), pp. 156–168. DOI: 10.1016/0097–3165(80)90082–5.DOI
  7. [7]James E. Humphreys. Introduction to Lie algebras and representation theory. Springer-Verlag, New York-Berlin,, 1972, pp. xii+169.DOI
  8. [8]Bruno Leclerc. “A Finite Coxeter Group the Weak Bruhat Order of which is not Symmetric Chain”. In: European Journal of Combinatorics 15.2 (1994), pp. 181–185. ISSN: 0195-6698. DOI: https://doi.org/10.1006/eujc.1994.1020.
  9. [9]Bernt Lindström. “On the vector representations of induced matroids”. In: Bulletin of the London Mathematical Society 5.1 (1973), pp. 85–90. DOI: 10.1112/blms/5.1.85.DOI
  10. [10]Robert A. Proctor. “Representations of sl(2, C) on Posets and the Sperner Property”. In: SIAM Journal on Algebraic Discrete Methods 3.2 (1982), pp. 275–280. DOI: 10.1137/0603026.DOI
  11. [11]Alexander Schrijver. Combinatorial Optimization: Polyhedra and Efficiency. Vol. 24. Algorithms and Combinatorics. Berlin: Springer, 2003.

Paper details

vlnc:W_qhrscd8m
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

ChatGPT

View statement
Looking for related papers…
Contents