arXiv++ Combinatorics

Poset games form a fundamental class of combinatorial games that includes classical games such as Nim and Chomp. Determining the outcome of a poset game is known to be PSPACE-complete even for posets of height at most three. However, the exact computational complexity of Two-Level Poset Game, played on posets of height at most two, has remained an open problem for over a decade. We resolve this problem by proving that Two-Level Poset Game is PSPACE-complete via a polynomial-time reduction from Avoid True on positive CNF formulas.

Let $(X_1,d_1),\dots,(X_N,d_N)$ be metric spaces with $d_i:X_i\times X_i\to[0,1]$ and let $X=\prod_{i=1}^N X_i$. Let $G$ be a directed graph with vertex set $[N]=\{1,\dots,N\}$ and edge set $E(G)$. Given edge weights $P=\{p_{ij}\in(0,1]:(i,j)\in E(G)\}$, define the closed in-neighborhood $B_j(G):=\{j\}\cup\{i:(i,j)\in E(G)\}$ and put $p_{jj}=1$. Define, for $g,h\in X$, \[ d_{X,G,P}(g,h)\;:=\;1-\frac{1}{N}\sum_{j=1}^N\ \prod_{i\in B_j(G)}\bigl(1-d_i(g_i,h_i)\bigr)^{1/p_{ij}}. \] We prove that $d_{X,G,P}$ is a metric and analyze its behavior under graph operations (disjoint unions, Cartesian products). Two quantitative principles govern this family: (i) \emph{edge densification} (adding edges) increases the distance pointwise and in expectation, with an exact single-edge increment formula and a uniform $1/N$ sensitivity bound; and (ii) \emph{weight monotonicity}: increasing an edge weight $p_{ij}$ decreases the metric. When the coordinate pairs are independent across coordinates, we obtain a closed-form factorization of the mean distance.} In the discrete specialization $X=\mathbb{F}_q^N$ with $d_i\in\{0,1\}$ and $G$ acyclic, after replacing $G$ by the oriented reflexive transitive closure $G^{\preceq}$ of Corollary~1.4, the unnormalized distance $D_{X,G^{\preceq},P}:=N d_{X,G^{\preceq},P}$ coincides with the size of the hereditary closure of the support, yielding a poset/digraph \emph{sphere-packing bound} via the counting of lower ideals. Finally, we give a continuous analog of the construction, yielding a metric on $C([0,1],\mathbb C)$ through an explicit integral formula. These results unify and generalize several graph-based metrics used in coding theory while providing new design levers for mean-distance control.

We study alternating Hamilton paths and cycles in the comparability graphs of finite posets. We prove that if a finite poset $P$ with at least two elements admits an alternating Hamilton path, then $P\times B_r$ admits an alternating Hamilton cycle for every $r\ge2$, where $B_r$ is the Boolean lattice. As a corollary, we settle a conjecture of Katona [Discrete Appl. Math., 2013]. Furthermore, we characterize the divisor lattices admitting alternating Hamilton paths and cycles.

Let $G$ be a split real semisimple group of adjoint type. We prove that the cellular decomposition of the totally nonnegative part of its wonderful compactification is a regular CW complex, confirming a conjecture of Galashin, Karp, and Lam. In particular, the closure of every cell is homeomorphic to a closed ball. The proof combines a Bruhat atlas compatible with total positivity and a CL-labeling of the augmented face poset. The atlas models the strata locally with open Richardson varieties in the flag variety of a larger Kac-Moody group and identifies their totally nonnegative cells.

For a finite poset, we partition the edges of the order polytope and of the chain polytope into classes indexed by the connected convex subsets of the poset, and we establish that corresponding classes have the same cardinality. This yields a closed formula for the common number of edges, and it identifies the bijection of Hibi, Li, Sahara and Shikama, given by them through an explicit formula, as a disjoint union of simple bijections between corresponding classes, which explains why it is a bijection and how it acts on edge directions and lengths. As consequences, the two polytopes have equally many edge directions, with matching multiplicities; the chain polytope has at least as many edges parallel to each coordinate subspace as the order polytope; and the edge lengths of the chain polytope are dominated by those of the order polytope. Strict inequality occurs in the last two comparisons exactly when the poset contains a three-element chain. We discuss implications for linear and convex combinatorial optimization over ideals and antichains. We also express the number of edges in terms of the comparability graph, in a form that extends to stable-set polytopes of arbitrary graphs, give a recursion for series-parallel posets and closed formulas for layered, zigzag and crown posets, characterize the distributive lattices for which the two polytopes are unimodularly equivalent, and demonstrate that, although the two polytopes have the same number of edges, either one can have the larger diameter, by an arbitrary amount; on the other hand, both diameters are bounded by the width of the poset, and they coincide for ordinal sums, series-parallel posets, and zigzag and crown posets.

In 1989, Nešetřil and Pudlák posed the following challenging question: Do planar posets have bounded Boolean dimension? We show that every poset with a planar cover graph and a unique minimal element has Boolean dimension at most $10$. As a consequence, we are able to show that there is a reachability labeling scheme with labels consisting of $\mathcal{O}(\log n)$ bits for planar digraphs with a single source. The best known scheme for general planar digraphs uses labels with $\mathcal{O}(\log^2 n)$ bits [Thorup JACM 2004], and it remains open to determine whether a scheme using labels with $\mathcal{O}(\log n)$ bits exists. The Boolean dimension result is proved in tandem with a second result showing that the dimension of a poset with a planar cover graph and a unique minimal element is bounded by a linear function of its standard example number. However, one of the major challenges in dimension theory is to determine whether dimension is bounded in terms of standard example number for all posets with planar cover graphs.

Richardson varieties provide geometric representatives for products of Schubert classes, with stratifications indexed by the subintervals of Bruhat intervals. We construct analogous geometric models for tilted Bruhat intervals for all split connected reductive groups. Our construction uses Richardson varieties in affine flag varieties and yields abstract tilted Richardson varieties whose isomorphism classes, together with their labeled stratifications, are independent of the choices involved. We obtain an explicit formula for the point-counting polynomials of their open parts in terms of increasing minimal weight paths, extending Dyer's formula for Kazhdan--Lusztig $R$-polynomials. Over the real numbers, their totally nonnegative parts are regular CW complexes homeomorphic to closed balls, which gives a natural geometric realization of the posets of tilted Bruhat subintervals in all Lie types. Over the complex numbers, we prove that their fundamental classes push forward under a continuous loop evaluation map to the Poincaré duals of the minimal-degree coefficients of quantum Schubert products.

The cardinalities of Kiselman's semigroups $K_n$ are known to grow double exponentially in $n$: the sequence $2^{-n/2}\log_2|K_n|$ converges along the even and odd $n$ subsequences to positive limits $c_{\mathrm{even}}$ and $c_{\mathrm{odd}}$ that appear to coincide. In this paper we prove that they are indeed distinct. Since Kiselman's semigroups are $\mathcal{L}$-trivial, the suffix relation defines a partial order on $K_n$. We introduce the \emph{Kiselman transform} $\mathsf{Kis}(P)$ of a poset $P$, a construction recovering the poset associated with $K_{n+1}$ from that associated with $K_n$. Counting strictly increasing chains in $P$ and $\mathsf{Kis}(P)$ yields a linear recurrence governed by a symmetric infinite matrix with explicit binomial entries. This can be used towards efficiently computing $|K_n|$, allowing the values listed in OEIS sequence A125625 to be extended beyond $n=14$. A generating functions analysis gives formulas for the above asymptotic limits, together with rigorous numerical bounds establishing that \[ c_{\mathrm{even}} - c_{\mathrm{odd}} = 1.9945066605\ldots\times10^{-6} \]

The independent set polytope, or stable set polytope, of a graph $G$ is the 0/1-polytope defined by the convex hull of the characteristic vectors of all independent sets of $G$. We present a simple algorithm for computing a Hamilton path on the independent set polytope of a given $n$-vertex graph $G$ with amortized delay $\mathcal{O}(n)$. The independent sets are listed such that two consecutive sets differ either in removing a vertex, or adding a vertex and removing its neighbors from the independent set, i.e., the symmetric difference between two consecutive independent sets induces a star in $G$. As applications of this result, we obtain an algorithm to compute a Hamilton path on the matching polytope of an $m$-edge graph $G$ with worst-case delay $\mathcal{O}(m)$, which lists all matchings of $G$ in such a way that the symmetric difference between two consecutive matchings is a path on at most three edges. Furthermore, we obtain an algorithm to compute a Hamilton path on the chain polytope and order polytope of an $n$-element poset $P$ with amortized delay $\mathcal{O}(n)$, which lists all antichains of $P$ or all ideals of $P$, respectively, by star exchanges. Our algorithms are derived from the generic framework proposed by Merino and Mütze (FOCS'23+SICOMP'24) for computing Hamilton paths on arbitrary 0/1-polytopes, which uses a linear optimization procedure as a black box. Our algorithms bypass solving the computationally intractable maximum weight independent set problem by a simple and purely combinatorial greedy rule.

Recently, Merino and Mütze (FOCS'23+SICOMP'24) presented an algorithm for computing a Hamilton path on the skeleton of any 0/1-polytope ${\rm conv}(X)$, where $X\subseteq\{0,1\}^n$. The algorithm uses as a black box an algorithm for solving the classical linear optimization problem $\min\{w\cdot x\mid x\in X\}$ for some weight vector $w\in\mathbb{R}^n$. The resulting delay per visited vertex on the Hamilton path is only by a $\log n$ factor larger than the time to solve one instance of the optimization algorithm. In this paper, we make the Hamilton path algorithm simpler and faster. Namely, we obtain an amortized delay that is only by a constant factor larger than the running time of the optimization algorithm, thus removing the $\log n$ factor. As concrete results, this yields improved algorithms for generating bases and independent sets in a matroid, spanning trees, forests, matchings and maximum matchings in a graph, vertex covers, minimum vertex covers, independent sets and maximum independent sets in a bipartite graph, and antichains, maximum antichains and ideals in a poset. All of these listings correspond to Hamilton paths on the corresponding polytopes. Furthermore, we obtain an $\mathcal{O}(t_{\rm LP})$ amortized delay algorithm for the vertex enumeration problem on 0/1-polytopes $\{x\in\mathbb{R}^n\mid Ax\leq b\}$, where $A\in \mathbb{R}^{m\times n}$ and $b\in\mathbb{R}^m$, and $t_{\rm LP}$ is the time needed to solve the linear program $\min\{w\cdot x\mid Ax\leq b\}$. This improves upon the $\mathcal{O}(t_{\rm LP} \log n)$ delay algorithm of Merino and Mütze, and the previous $\mathcal{O}(t_{\rm LP}\,n)$ delay algorithm of Bussieck and Lübbecke from 1998.

We prove that every finite ordered set of width at most three and cardinality at least four is reconstructible from its deck. This settles the width-three part of an open problem posed by Schröder in Ordered Sets; the separate question concerning ordered sets of height one is not addressed here. The new ingredient is a criterion for recovering the marked maximal deck of an ordered set with exactly three maximal elements from its unmarked maximal deck and the isomorphism type of the suborder obtained by deleting all maximal elements. We express the criterion as an orbital balance condition on the attachment ideals in this common base. Necessity follows by double counting; sufficiency uses the structure of automorphism actions in width at most three. The full-deck theorem then follows from reconstruction results of Schröder concerning maximal cards and marked maximal cards. A local reattachment argument correcting one intermediate step in the two-maximal-element case is included in an appendix.

The Ungar game, introduced by Defant et al., is a combinatorial game played on a finite lattice, which originates from a geometric transformation introduced by Ungar to resolve Scott's slope problem. When played on a distributive lattice $L$, the Ungar game reduces to repeatedly choosing and removing a non-empty set of maximal elements from a finite poset $P$, where $L$ is represented as the lattice $J(P)$ of the order ideals of $P$ by Birkhoff's representation theorem. Defant et al. raised the open question of whether determining the outcome of the Ungar game on $J(P)$ is PSPACE-complete with respect to $|P|$. In this paper, we settle this open question in the affirmative. More precisely, we prove that the problem is PSPACE-complete even when restricted to posets $P$ with maximum chain length $l(P) \le 3$ such that the underlying graph of the Hasse diagram of $P$ is bipartite, where the length of a chain is defined as the number of elements in the chain minus one. Furthermore, we show that the problem is NP-complete for $l(P) \le 2$, whereas the case $l(P) \le 1$ is known to belong to LOGSPACE.

We study the diagonal Specht ideals $I_λ\subseteq\mathcal{R}_{m,n}:=\mathbb{C}[\mathbf{x}_1,\dots,\mathbf{x}_m]$, $\mathbf{x}_r=(x_{r,1},\dots,x_{r,n})$, generated by the $λ$-isotypic component of $\mathcal{R}_{m,n}$ for the diagonal action of $S_n$. We characterize the set of zeros of $I_λ$ as $V_λ=\bigcup_{μ\not\trianglelefteqλ}H_μ$ and prove that $λ\mapsto V_λ$ is an isomorphism of posets from $(\mathcal{P}_n,\trianglelefteq)$ to $(\{V_λ:λ\in \mathcal{P}_n\},\supseteq)$ for all $m$, while $λ\mapsto I_λ$ is an isomorphism of posets from $(\mathcal{P}_n,\trianglelefteq)$ to $(\{I_λ:λ\in \mathcal{P}_n\},\subseteq)$ if and only if $m=1$ or $n\leq3$. Unlike the case $m=1$, where Specht ideals are always radical, radicality in the diagonal setting depends on $λ$ and $m$. We prove radicality for hook partitions of length at most three by providing explicit Gröbner bases, and we develop three criteria for non-radicality via content, multidegree, and total degree, showing in particular that $I_λ$ is not radical for any non-hook partition $λ$ when $m\geq\operatorname{len}(λ)$. Using the $\operatorname{GL}_m(\mathbb{C})$-action on $\mathcal{R}_{m,n}$, we show that $I_λ$ is radical for all $m$ if and only if it is radical for $m=n$. Finally, for $m\geq2$, we prove that $\mathcal{R}_{m,n}/I_λ$ is Cohen-Macaulay if and only if $λ=(n)$ or $λ=(n-1,1)$, and the same is true for $\mathcal{R}_{m,n}/\operatorname{rad}(I_λ)$.

A graph is beautiful if each of its induced subgraphs is the intersection graph of all maximal nonempty 1-rectangles of a binary matrix, with adjacency defined by a common cell. Beautiful graphs were introduced as a hereditary class of Berge graphs, but a complete forbidden-induced-subgraph characterization was not obtained. We prove that a finite graph is beautiful if and only if it has no induced $C_4$, gem, net, watch, or odd hole. More precisely, these graphs are exactly the $C_4$-free comparability graphs admitting a partial order in which every interval is a chain. The proof constructs such an order from inclusion-maximal closed neighbourhoods: their representatives induce a bipartite graph whose domination regions admit compatible orientations. One order matrix then represents the graph and, through its principal submatrices, every induced subgraph. The representation step is formulated using classical double-bound graphs and the established correspondence between maximal bicliques and interval-intersection-closed posets. Consequences include polynomial-time recognition, the exact minimal obstruction families, and a corrected characterization in the $K_4$-free case.

Given a finite poset $\mathcal P$, its induced saturation number, $\text{sat}^*(n,\mathcal P)$ is the smallest size a family of subsets of $[n]$ can have such that it does not contain an induced copy of $\mathcal P$, but adding any other set to it creates one. The saturation numbers already show a sharp dichotomy -- Freschi, Piga, Sharifzadeh and Treglown showed that for any given poset its saturation number is either bounded or at least $2\sqrt n$. The dominant conjecture is that in fact, the saturation numbers are either bounded, or exactly linear. In this paper we show that this dichotomy is true for all posets on at most 4 points. Most of these posets, most notably the butterfly and the diamond, had individually long resisted analysis, and the missing piece which this paper tackles is the poset comprised of a $\mathcal V$ and an isolated point, denoted by $\widehat{\mathcal V}$, for which we show linear saturation. The architecture of the proof uncovers the structure of an arbitrary $\widehat{\mathcal V}$- saturated family, namely an antichain with at most two chains above it, and an upside-down forest below it. Moreover, we anticipate that, at least partially, this structure is relevant to the wider class of posets that contain an isolated point.

We provide in this text a new framework for a family of combinatorial Hopf algebras. We adopt for this the language of species and twisted bialgebras (also known as Hopf monoids). We define the category of combinatorial twisted bialgebras (briefly, CTBs), a class of twisted bialgebras satisfying strong combinatorial conditions of compatibilities of the product and the coproduct with the underlying combinatorial structure. The considered morphisms also have to respect the combinatorial structure, in some sense. Many classical examples fit into this framework, including Hopf algebras of graphs, rooted trees, posets, hypergraphs, symmetric functions, boolean maps, etc. Most of known combinatorial Hopf algebras can be obtained from a CTB by application of a Fock functor, or by duality. A central result is the introduction of a universal CTB denoted by $\mathbf{Exch}$, built from collections of compositions satisfying a certain exchange condition. It has a special character $ε$, such that the pair $(\mathbf{Exch},ε)$ plays a role analogous to that of quasi-symmetric functions for graded and connected Hopf algebras: it is the terminal object in the category of CTBs equipped with combinatorial characters. This leads to universal morphisms that encode important information about the combinatorial structure of coproducts. The natural order on the set of compositions on a fixed set is used to define three important sub-objects $\mathbf{Exch}\_-$, $\mathbf{Exch}\_+$ and $\mathbf{Exch}\_{+\hspace{-2mm}+}$ of $\mathbf{Exch}$. We also studies polynomial invariants associated with CTBs. It establishes a correspondence between characters and polynomial invariants. Well-known invariants such as the chromatic polynomial of graphs and the Ehrhart polynomials of posets arise naturally within this framework. We show that, for a given degree, only finitely many polynomial invariants can occur. We show that these universal polynomial invariants can be used to give cancellation-free formulas for the antipode or the Eulerian idempotent, under conditions on the CTB, and use this for an example based on rooted trees, coming from a Hopf algebra used in linguistics. Results are also obtained for families of polynomial invariants with the alternating sign condition, or satisfying a reciprocity principle. We also extend the theory to combinatorial twisted double bialgebras (briefly, CTDBs), which involve two interacting coproduct structures. These generalize previously known examples arising from graphs, posets, hypergraphs, and rooted trees. The twisted approach preserves the combinatorial richness of these structures while providing a unified algebraic framework. We show that not any CTB can be made a CTDB, leading to the introduction of the notion of quasi-cocommutativity. In particular, we show that $\mathbf{Exch}\_+$ is not a CTDB, and construct a sub-object which is, in a unique way, and maximal for the inclusion. In the cocommutative case, simplifications naturally occur, and we show that in this case, a CTB of hypergraphs plays a central role, especially two sub-objects based on connectedness systems and antichain covers. We also show that, in the cocommutative case, the universal polynomial invariants are always chromatic polynomials of hypergraphs, no matter which CTB is considered. Finally, forgetting the combinatorial underlying structure, we study the algebraic structure of $\mathrm{Exch}$, with the help of a twisted version of Loday and Ronco's rigidity theorem for infinitesimal bialgebras.

We study the average sizes of ideals and antichains in several families of posets arising from rooted unlabeled trees. For ideals, we encode order ideals by coloring vertices red, with the condition that every descendant of a red vertex is also red. Using generating functions and singularity analysis, we show that for every family governed by a quadratic root decomposition, the average number of red vertices in a tree of size $n$ is asymptotic to $n/3$. This includes binary plane trees, plane 1-2 trees, plane 2-trees, and 0-1-trees, while unrestricted rooted plane trees instead have average ideal size asymptotic to $2n/5$. For antichains, represented by pairwise incomparable blue vertices, the corresponding asymptotic constants depend on the particular branching function; for example, binary plane trees have average antichain size asymptotic to $n/6$. We also consider plane trees in which every vertex has at most $k$ children and prove that the asymptotic proportion of red vertices increases strictly with $k$, from $1/3$ when $k=2$ to the limiting value $2/5$ for unrestricted rooted plane trees. The asymptotic proportion of blue vertices also increases strictly with $k$, tending to $3/10$. The proofs are based on bivariate generating functions together with the analytic implicit-function and smooth implicit-function methods of analytic combinatorics.

Jackson and Tanigawa conjectured that the rigidity matroid $\mathcal{R}^d_n$ and the hyperconnectivity matroid $\mathcal{H}^d_n$ are the unique maximal matroids in the posets of $\{K_{d+2}, K_{d+2,d+2}\}$-matroids and $\{K_{d+2},K_{d+1,d+1}\}$-matroids, respectively. We disprove these conjectures by showing the existence of maximal matroids that are distinct from the proposed candidates.

Transfer systems on finite posets have recently been gaining traction as a key ingredient in equivariant homotopy theory. Additionally, they also naturally occur in the data of a model structure. We give a complete characterization of all model category structures on a finite lattice, using transfer systems as our main tool, resulting in new connections between abstract homotopy theory and equivariant methods.

Regular semisimple Hessenberg varieties are subvarieties of flag varieties defined for any reductive group. Their cohomology carries a representation of the associated Weyl group, known as the dot action representation. In type $A$, these representations can be computed using the chromatic quasisymmetric functions of naturally labeled unit interval graphs. We prove that restricting the dot action representation of any Weyl group to a type $A$ parabolic subgroup yields a representation isomorphic to a direct sum of type $A$ dot action representations. This work is motivated by the conjecture that an analogous property holds for all parabolic subgroups. In particular, for classical Weyl groups of types $B/C$ and $D$, we use chromatic symmetric functions to provide a streamlined, combinatorial formula for the restriction to a specific maximal type $A$ parabolic subgroup. As an application, we prove a conjecture of Lesnevich regarding the representations associated with ideals of types $B$ and $C$ identified under the standard poset isomorphism. Specifically, we show that the corresponding dot action representations are isomorphic if and only if the ideals determine the same reflection sets in their respective root systems. Furthermore, we show that a Catalan number counts the total number of such ideals.