https://arxiv.org/api/YDqBbKl49dccNiCTLp50lwgYZ3E 2026-06-19T01:22:52Z 15689 510 15 http://arxiv.org/abs/2603.25219v1 2026-03-26T09:18:10Z

It was once conjectured that two graph states are local unitary (LU) equivalent if and only if they are local Clifford (LC) equivalent. This so-called LU-LC conjecture was disproved in 2007, as a pair of 27-qubit graph states that are LU-equivalent, but not LC-equivalent, was discovered. We prove that this counterexample to the LU-LC conjecture is minimal. In other words, for graph states on up to 26 qubits, the notions of LU-equivalence and LC-equivalence coincide. This result is obtained by studying the structure of 2-local complementation, a special case of the recently introduced r-local complementation, and a generalization of the well-known local complementation. We make use of a connection with triorthogonal codes and Reed-Muller codes.

2026-03-26T09:18:10Z Nathan Claudet http://arxiv.org/abs/2603.24708v1 2026-03-25T18:31:10Z

We prove that the directed 3-torus D_3(m), or equivalently the Cartesian product of three directed m-cycles, admits a decomposition into three arc-disjoint directed Hamilton cycles for every integer m >= 3. The proof reduces Hamiltonicity to the m-step return maps on the layer section S=i+j+k=0. For odd m, five Kempe swaps of the canonical coloring produce return maps that are explicitly affine-conjugate to the standard 2-dimensional odometer. For even m, a sign-product invariant rules out Kempe-from-canonical constructions, and a different low-layer witness reduces after one further first-return map to a finite-defect clock-and-carry system. The remaining closure is a finite splice analysis, and the case m=4 is handled separately by a finite witness. A Lean 4 formalization accompanies the construction.

2026-03-25T18:31:10Z 48 pages, 1 figure. Ancillary verification files included. Lean 4 formalization available at https://github.com/aria1th/Torus-Hamilton-Decomposition/ SangHyun Park http://arxiv.org/abs/2504.12764v5 2026-03-25T11:51:34Z

This paper introduces GraphOmni, a comprehensive benchmark designed to evaluate the reasoning capabilities of LLMs on graph-theoretic tasks articulated in natural language. GraphOmni encompasses diverse graph types, serialization formats, and prompting schemes, significantly exceeding prior efforts in both scope and depth. Through extensive systematic evaluation, we identify critical interactions among these dimensions, demonstrating their substantial impact on model performance. Our experiments reveal that state-of-the-art models like Claude-3.5 and o4-mini consistently outperform other models, yet even these leading models exhibit substantial room for improvement. Performance variability is evident depending on the specific combinations of factors we considered, underscoring the necessity of comprehensive evaluations across these interconnected dimensions. Additionally, we observe distinct impacts of serialization and prompting strategies between open-source and closed-source models, encouraging the development of tailored approaches. Motivated by the findings, we also propose a reinforcement learning-inspired framework that adaptively selects the optimal factors influencing LLM reasoning capabilities. This flexible and extendable benchmark not only deepens our understanding of LLM performance on structured tasks but also provides a robust foundation for advancing research in LLM-based graph reasoning. The code and datasets are available at https://github.com/GAI-Community/GraphOmni.

2025-04-17T09:01:16Z Published at ICLR 2026. Project Page: https://gai-community.github.io/Graph-Omni/ Hao Xu Xiangru Jian Xinjian Zhao Wei Pang Chao Zhang Suyuchen Wang Qixin Zhang Zhengyuan Dong Joao Monteiro Bang Liu Qiuzhuang Sun Tianshu Yu http://arxiv.org/abs/2603.06930v2 2026-03-25T04:06:06Z

The general position problem in graphs seeks the largest set of vertices such that no three vertices lie on a common geodesic. Its counting refinement, the general position polynomial $ψ(G)$, asks for all such possible sets. In this paper, We describe general position sets for several classes of graphs and provide explicit formulas for the general position polynomials of complete multipartite graphs. We specialize to balanced complete multipartite graphs and show that for part size $r\le 4$, the polynomial $ψ(K_{r,\dots,r})$ is log-concave and unimodal for all numbers of parts, while for larger $r$, counterexamples show that these properties fail. Finally, we analyze the corona $G\circ K_1$ and prove that unimodality of $ψ(G)$ is retained for numerous natural classes (paths, edgeless graphs, combs). This contributes to an open problem, but the general case remains unknown. Our findings support the parallel between general position polynomials and classical position-type parameters, and identify balanced multipartite graphs and coronas as promising testbeds for additional research.

2026-03-06T23:03:04Z 32 pages; Updated with Theorem 5.3, Remark 5.4, Example 5.5, Theorem 5.6, Corollary 5.7, and Example 5.8 Bilal Ahmad Rather http://arxiv.org/abs/2603.18582v2 2026-03-24T20:18:58Z

DRESS is a deterministic, parameter-free framework for structural graph refinement that iteratively refines the structural similarity of edges in a graph to produce a canonical fingerprint: a real-valued edge vector, obtained by converging a nonlinear dynamical system to its unique fixed point. $Δ$-DRESS is a member of the DRESS family of graph fingerprints that applies a single level of vertex deletion. We test it on a benchmark of 51,813 distinct graphs across 34 hard families, including the complete Spence collection of strongly regular graphs (43,703 SRGs, 12 families), four additional SRG families (8,015 graphs), and 18 classical hard constructions (102 family entries corresponding to 99 distinct graphs). $Δ$-DRESS produces unique fingerprints in 33 of 34 benchmark families at $k=1$, resolving all but one within-family collision among over 576 million non-isomorphic pairs. One genuine collision exists at deletion depth $k=1$, between two vertex-transitive SRGs in SRG(40,12,2,4), which is resolved by a single-step fallback to $Δ^2$-DRESS. For every family with pairwise-comparable full sorted-multiset fingerprints, the minimum observed separation margin remains at least $137 \times ε$, confirming that the reported separations are numerically robust and not artifacts of the convergence threshold. We also show that $Δ$-DRESS separates the Rook $L_2(4)$/Shrikhande pair, proving it escapes the theoretical boundary of 3-WL. The method runs in $\mathcal{O}(n \cdot I \cdot m \cdot d_{\max})$ time per graph.

2026-03-19T07:41:16Z Eduar Castrillo Velilla http://arxiv.org/abs/2602.00657v2 2026-03-24T11:37:25Z

Kirkpatrick et al. [ALT 2019] and Fallat et al. [JMLR 2023] introduced non-clashing teaching and proved that it is the most efficient batch machine teaching model satisfying the collusion-avoidance benchmark established in the seminal work of Goldman and Mathias [COLT 1993]. Recently, (positive) non-clashing teaching was thoroughly studied for balls in graphs, yielding numerous algorithmic and combinatorial results. In particular, Chalopin et al. [COLT 2024] and Ganian et al. [ICLR 2025] gave an almost complete picture of the complexity landscape of the positive variant, showing that it is tractable only for restricted graph classes due to the non-trivial nature of the problem and concept class. In this work, we consider (positive) non-clashing teaching for closed neighborhoods in graphs. This concept class is not only extensively studied in various related contexts, but it also exhibits broad generality, as any finite binary concept class can be equivalently represented by a set of closed neighborhoods in a graph. In comparison to the works on balls in graphs, we provide improved algorithmic results, notably including FPT algorithms for more general classes of parameters, and we complement these results by deriving stronger lower bounds. Lastly, we obtain combinatorial upper bounds for wider classes of graphs.

2026-01-31T11:07:30Z An extended abstract of this paper will appear in the proceedings of ICLR 2026 Sujoy Bhore Liana Khazaliya Fionn Mc Inerney http://arxiv.org/abs/2603.23070v1 2026-03-24T11:07:05Z

The House of Graphs is an online database of graphs which can be accessed at https://houseofgraphs.org/. It serves as a central repository for complete lists of graphs for various graph classes. However, its main feature is a searchable database of so-called "interesting" graphs. The development of the original House of Graphs started in 2010 and it was completely rebuilt in 2021-2022. Each graph in the database is accompanied by a significant amount of meta-data such as a name, drawings, precomputed graph invariants, and comments. Given this volume of information and the importance of reliability in the scientific world, robust data management is essential to ensure accuracy and consistency across the database. In this article, we therefore focus on knowledge management in the House of Graphs and describe the inner workings of the House of Graphs and how we ensure that its data is coherent, qualitative and stable.

2026-03-24T11:07:05Z 17 pages Gauvain Devillez Sven D'hondt Jan Goedgebeur http://arxiv.org/abs/2603.22993v1 2026-03-24T09:35:33Z

In a digraph $D=(V,A)$, an oriented path is a sequence $P=x_1x_2\dots x_p$ of distinct vertices such that either $x_ix_{i+1}\in A$ or $x_{i+1}x_{i}\in A$ or both for every $i\in [p-1]$. If $x_ix_{i+1}\in A$ in $P$, then $x_ix_{i+1}$ is a forward arc of $P$; otherwise, $x_{i+1}x_{i}$ is a backward arc. The independence number $α(D)$ is the maximum integer $p$ such that $D$ has a set of $p$ vertices where there is no arc between any pair of vertices. A digraph is $k$-connected if its underlying undirected graph is $k$-connected. Freschi and Lo (JCT-B 2024) proved that every $n$-vertex oriented graph with minimum degree $δ\ge n/2$ has a Hamilton oriented cycle with at most $n-δ$ backward arcs. We prove that every 2-connected digraph $D$ with $α(D)\le 2$ has a Hamilton oriented cycle with at most five backward arcs, and every 1-connected digraph $D$ with $α(D)\le 2$ has a Hamilton oriented path with at most two backward arcs.

2026-03-24T09:35:33Z S. Gerke Q. Guo G. Gutin Y. Hao W. Veeranonchai A. Yeo http://arxiv.org/abs/2603.22688v1 2026-03-24T01:28:53Z

It was proved in (Levit and Mandrescu, 2022) that both $(V(G), Crown(G))$ and $(V(G), CritIndep(G))$ are augmentoids, established partial augmentation phenomena for the family $Ψ(G)$ of local maximum independent sets, and asked in Problem~5.5 to characterize the graphs whose family $Ψ(G)$ is an augmentoid. We prove that the answer is positive in full generality: for every finite simple graph $G$, the set system $(V(G),Ψ(G))$ is an augmentoid. The proof is constructive. If $S,T\inΨ(G)$, then the explicit choice \[ A=S \setminus N[T],\qquad B=T \setminus N[S] \] satisfies \[ T\cup A\inΨ(G),\qquad S\cup B\inΨ(G),\qquad |T\cup A|=|S\cup B|. \] As a structural consequence, for every fixed $S\inΨ(G)$ the map $T\mapsto S\cup T$ induces a canonical bijection from $Ψ(G-N[S])$ onto the members of $Ψ(G)$ containing $S$, and \[ α(G)=|S|+α(G-N[S]). \] This decomposition also yields explicit formulas for the intersection and the union of all the maximum independent sets extending $S$, together with counting formulas for the local maximum and maximum independent sets containing $S$. We also add a short visual guide to the framework $CritIndep(G) \subseteq Crown(G)\subseteq Psi(G)$ and end with several natural follow-up problems suggested by the theorem.

2026-03-24T01:28:53Z 10 pages, 3 figures Vadim E. Levit Eugen Mandrescu http://arxiv.org/abs/2601.02920v2 2026-03-23T22:19:17Z

A theorem of Matoušek asserts that for any $k \ge 2$, any set system whose shatter function is $o(n^k)$ enjoys a fractional Helly theorem of order $k$: in the $k$-wise intersection hypergraph, positive density implies a linear-size clique. Kalai and Meshulam conjectured a generalization of that phenomenon to homological shatter functions. It was verified for set systems with bounded homological shatter functions and ground set with a forbidden homological minor (which includes $\mathbb{R}^d$ by a homological analogue of the van Kampen-Flores theorem). We present two contributions to this line of research: - We study homological minors in certain manifolds (possibly with boundary), for which we prove analogues of the van Kampen-Flores theorem and of the Hanani-Tutte theorem. - We introduce graded analogues of the Radon and Helly numbers of set systems and relate their growth rate to the original parameters. This allows to extend the verification of the Kalai-Meshulam conjecture for sufficiently slowly growing homological shatter functions.

2026-01-06T11:01:57Z Sergey Avvakumov Marguerite Bin Xavier Goaoc http://arxiv.org/abs/2603.22432v1 2026-03-23T18:05:50Z

Spin-glasses are natural Gibbs distributions that have been studied in Theoretical CS for many decades. Recently, they have been gaining attention from the community as they emerge naturally in neural computation and learning, network inference, optimisation and other areas. We study the problem of efficiently sampling from spin-glass distributions when the underlying graph is a typical instance of $G(n,d/n)$, i.e., the random graph on $n$ vertices such that each edge appears independently with probability $d/n$, and $d=Θ(1)$. Our focus is on the 2-spin model at inverse temperature $β$. We consider this distribution to be one of the most interesting case of spin-glasses, and one of the most challenging to analyse, since its Gaussian couplings give rise to unbounded interaction. We employ the well-known Glauber dynamics to sample from the aforementioned distribution. We show that for the typical instances of the 2-spin model on $G(n,d/n)$, the mixing time of Glauber dynamics is $O\left(n^{1+\frac{25}{\sqrt{\log d}}}\right)$, for any $β<\frac{1}{4\sqrt{d}}$. Our results can also be adapted for the case of spin-glass distributions with bounded interactions. In that respect, we obtain rapid mixing of Glauber dynamics for the Viana-Bray model on $G(n,d/n)$ when $β<\frac{1}{4\sqrt{d}}$. This improves on the current best bound which is $β<\frac{0.18}{\sqrt{d}}$. We utilise stochastic localisation, and in particular, we build and improve on the scheme introduced in [Liu, Mohanty, Rajaraman and Wu: FOCS 2024]. This is the first time that stochastic localisation is used for diluted spin-glasses, where both degrees and interactions can be unbounded.

2026-03-23T18:05:50Z Charilaos Efthymiou Kostas Zampetakis http://arxiv.org/abs/2602.22980v2 2026-03-23T16:58:19Z

This paper introduces the notion of an $(ι,q)$-critical graph. The isolation number of a graph $G$, denoted by $ι(G)$ and also known as the vertex-edge domination number of $G$, is the size of a smallest subset $D$ of the vertex set of $G$ such that the subgraph induced by the set of vertices that are not in the closed neighbourhood of $D$ has no edges. A graph $G$ is $(ι,q)$-critical if every subdivision of $q$ edges of $G$ gives a graph whose isolation number is greater than $ι(G)$, and $G$ has $q-1$ edges such that subdividing them gives a graph whose isolation number is $ι(G)$. We show that an $(ι,q)$-critical graph exists for every integer $q \ge 1$. We prove that if $G$ is a connected $m$-edge non-star graph, then $G$ is $(ι,q)$-critical for some $q \le m - 1$. We show that this bound is best possible. We provide a general characterization of $(ι,1)$-critical graphs as well as a constructive characterization of $(ι,1)$-critical trees, demonstrating that $(ι,1)$-criticality can be checked in linear time for trees.

2026-02-26T13:22:00Z 15 pages, minor presentation improvements made Karl Bartolo Peter Borg Magda Dettlaff Magdalena Lemańska Paweł Żyliński http://arxiv.org/abs/2312.12919v5 2026-03-23T15:10:48Z

The star chromatic number on a graph is the minimum number of colors in a proper vertex coloring forbidding any $P_4$ with two colors (bicolored). This problem was introduced by Grünbaum (1973) together with the acyclic coloring of graphs, where bicolored cycles are avoided. In this paper, we study a generalization of this problem, by considering proper vertex coloring on graphs forbidding bicolored paths of a fixed length, which was initially discussed by Alon, McDiarmid, and Reed (1991). Here, we study this problem on products of two paths. We show that at least 4 colors are needed to properly color the product of paths, $P_m\square P_n$, avoiding a bicolored $P_k,$ unless $n<k-2$ or $m<k-2.$ With this result, the above question is settled for all $k$ on 2-dimensional grids.

2023-12-20T10:54:34Z Derman Keskinkilic Lale Ozkahya http://arxiv.org/abs/2504.18338v2 2026-03-23T07:44:34Z

An elimination tree of a connected graph $G$ is a rooted tree on the vertices of $G$ obtained by choosing a root $v$ and recursing on the connected components of $G-v$ to obtain the subtrees of $v$. The graph associahedron of $G$ is a polytope whose vertices correspond to elimination trees of $G$ and whose edges correspond to tree rotations, a natural operation between elimination trees. These objects generalize associahedra, which correspond to the case where $G$ is a path. Ito et al. [ICALP 2023] recently proved that the problem of computing distances on graph associahedra is NP-hard. In this paper we prove that the problem, for a general graph $G$, is fixed-parameter tractable parameterized by the distance $k$. Prior to our work, only the case where $G$ is a path was known to be fixed-parameter tractable. To prove our result, we use a novel approach based on a marking scheme that restricts the search to a set of vertices whose size is bounded by a (large) function of $k$. On the negative side, we show that it is unlikely that FPT algorithms exist on a natural generalization of graph associahedra, namely hypergraphic polytopes, by proving that computing distances on them is W[2]-hard parameterized by the distance. We also prove that, on hypergraphic polytopes, the distance cannot be approximated in polynomial time within a factor $c \cdot \log(|V|+|\mathcal{E}|)$ for some constant $c > 0$ unless P = NP, where $H=(V, \mathcal{E})$ is the input hypergraph. This result strengthens the hardness result of Cardinal and Steiner [Combin. Theory 2025], who proved that the problem cannot be approximated within a factor $(1 + \varepsilon)$ for some absolute constant $\varepsilon > 0$ unless P = NP. Finally, we rule out the existence of polynomial kernels parameterized by the number of vertices of the input hypergraph, a parameter for which the problem is easily seen to be FPT.

2025-04-25T13:26:46Z 35 pages, 10 figures Luís Felipe I. Cunha Ignasi Sau Uéverton S. Souza Mario Valencia-Pabon http://arxiv.org/abs/2603.21645v1 2026-03-23T07:26:13Z

We construct automata with input(s) in Fibonacci representation (also known as Zeckendorf representation) recognizing some basic arithmetic relations and study their number of states. We also consider some basic operations on Fibonacci-automatic sequences and discuss their state complexity. Furthermore, as a consequence of our results, we improve a bound in a recent paper of Bosma and Don. We also discuss the state complexity and runtime complexity of using a reasonable interpretation of Büchi arithmetic to actually construct some of the studied automata recognizing relations.

2026-03-23T07:26:13Z Delaram Moradi Narad Rampersad Jeffrey Shallit