Pith. sign in

REVIEW 2 major objections 6 minor 23 references

Network realignment complexes retract onto star trees

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · glm-5.2

2026-07-09 11:44 UTC pith:A5THMPH6

load-bearing objection Solid extension of Kozlov's network realignment complexes to general graphs, with one proof gap worth checking the 2 major comments →

arxiv 2607.07404 v1 pith:A5THMPH6 submitted 2026-07-08 math.CO math.AT

Network Realignment Complexes over General Graphs

Fabienne Klatt , Iason Papadopoulos , Marc Raffelsiefen This is my paper
classification math.CO math.AT MSC 05E4557Q0505C0555U10
keywords network realignment complexleaf slidespanning tree reconfigurationequivariant discrete Morse theorycubical complexautomorphism groupgraph diameterflag complex
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper generalizes Kozlov's network realignment complexes from complete graphs to arbitrary connected base graphs G. The central objects are the network realignment complex X_G (a cubical complex encoding simultaneous independent leaf-slide reconfigurations of spanning trees of G) and, for the complete graph K_n, the network realignment graph G_n (its 1-skeleton, encoding single leaf slides). The paper's three main contributions are: (1) For any connected G, X_G admits an Aut(G)-equivariant strong deformation retraction onto the disjoint union of a complete graph K_{S_G} (whose vertices are the star trees of G) and a discrete Aut(G)-space D_G. This means the equivariant homotopy type is completely determined: the only possibly non-trivial component is a complete graph on the vertices of G that support star trees. (2) For K_n, explicit upper and lower bounds on the diameter of G_n are established, partially answering an open question of Kozlov. The upper bound is n^2/2 - n (even n) or (n^2-1)/2 - n (odd n); the lower bound is floor((3n^2-4n)/8) (even) or floor((3n^2-6n)/8) (odd). The key mechanism is the invariant Ψ(T), which counts edge-by-edge imbalance in the tree and changes by at most one per leaf slide, providing a tight lower bound on distance to the nearest star tree when the target center is a centroid. (3) For n ≥ 5, every automorphism of X_n is induced by a vertex relabeling, so Aut(X_n) ≅ S_n. This is proved by first showing X_n is a cubical flag complex (determined by its 1-skeleton), then identifying a rigid subcomplex H whose automorphism group is S_n, and showing H is closed under 4-cycle completion in G_n, forcing any automorphism of G_n to be determined by its restriction to H.

Core claim

The central discovery is that the topology, geometry, and symmetry of network realignment complexes are governed by star trees. Topologically, the entire complex X_G equivariantly collapses onto a complete graph whose vertices are star trees plus a discrete space of isolated points. Geometrically, the distance from any spanning tree to its nearest star tree in G_n is exactly the tree invariant Ψ(T), achieved when the star center is a centroid of T. This invariant also controls the diameter bounds. And algebraically, the rigid configuration of star trees and their neighboring cells in the subcomplex H is sufficient to pin down all automorphisms of X_n as relabeling symmetries for n ≥ 5.

What carries the argument

The proofs rest on three mechanisms. (1) An extension of Freij's equivariant generalized discrete Morse theory to cubical complexes (Theorem 2.3, proved in Appendix A), which allows label-free Morse matchings where all active vertices are simultaneously (co-)specified. (2) The tree invariant Ψ(T) = Σ_e (σ_T(e) - 1), where σ_T(e) is the smaller side of the cut induced by edge e; this changes by at most one per leaf slide and equals zero only for star trees, making it a sharp distance lower bound. (3) The 4-cycle completion property: any 4-cycle in G_n is uniquely determined by three of its vertices (Lemma 4.2), which propagates rigidity from a small subcomplex H to the entire graph.

Load-bearing premise

The extension of equivariant generalized discrete Morse theory to cubical complexes requires that the critical cells of the generalized Morse matching form a subcomplex. The acyclicity proof of the matching relies on the claim that the set of non-active specified vertices strictly decreases along any putative cycle, which depends on verifying that no cospecification can reintroduce a vertex into the non-active set.

What would settle it

Construct a connected graph G where the claimed Aut(G)-equivariant deformation retraction of X_G onto K_{S_G} ⊔ D_G fails because the Morse matching is cyclic (a cycle exists in the matching relation) or because the critical cells do not form a subcomplex.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • The equivariant homotopy type X_G ≃ K_{S_G} ⊔ D_G means the Betti numbers and fundamental group of X_G are determined purely by the number of universal vertices (vertices of degree n-1) in G, making the topology computable from local degree data.
  • The diameter bounds on G_n constrain the computational complexity of shortest-path algorithms for leaf-slide reconfiguration of spanning trees: any algorithm must handle distances that grow quadratically in n.
  • The result Aut(X_n) ≅ S_n for n ≥ 5 means the complex has no hidden symmetries beyond vertex relabeling, which is a rigidity property useful for distinguishing X_n from other cubical complexes arising in reconfiguration.
  • The characterization of connected components of X_G via spine and spine assignment (Proposition 2.37) provides a concrete invariant for classifying reconfiguration orbits under Aut(G).
  • The cubical flag property of X_n means the higher-dimensional structure is fully encoded in the reconfiguration graph, so combinatorial data about single moves suffices to reconstruct the full topology.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The condition n ≥ 5 for Aut(X_n) ≅ S_n likely reflects a threshold where the complex has enough local structure to force rigidity; below this, small-n coincidences (like X_4 ≅ Σ_{K_4} with its larger automorphism group) allow extra symmetries. This suggests a general phenomenon where rigidity of reconfiguration complexes stabilizes above a critical size.
  • The matching distance δ, introduced as a lower-bound tool for the diameter, may be of independent interest as a metric on trees: it compares inter-vertex distances across matchings and could serve as a computationally cheaper proxy for the leaf-slide distance in algorithmic applications.
  • If the upper diameter bound is sharp (as computations for 6 ≤ n ≤ 9 suggest), then the worst-case leaf-slide distance between two spanning trees of K_n is achieved by pairs of path trees with specific labelings, which would mean path trees are the geometric antipodes of the realignment graph.
  • The extension of equivariant Morse theory to cubical complexes (Appendix A) is developed in a self-contained way and could be applied to other cubical reconfiguration complexes beyond network realignments, such as state complexes of reconfigurable systems more generally.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 6 minor

Summary. The paper generalizes Kozlov's network realignment complexes from the complete graph $K_n$ to arbitrary connected base graphs $G$. The main results are: (1) an $Aut(G)$-equivariant strong deformation retraction of $X_G$ onto $K_{S_G} sqcup D_G$ (Theorem 2.1), where $S_G$ is the set of universal vertices and $D_G$ is a discrete $Aut(G)$-space; (2) explicit upper and lower bounds on the diameter of the network realignment graph $mathcal{G}_n$ (Theorem 3.3); (3) a proof that $X_n$ is a cubical flag complex (Lemma 4.3); and (4) a determination of $Aut(X_n) cong S_n$ for $n geq 5$ (Theorem 4.14). The proofs use equivariant generalized discrete Morse theory, combinatorial optimization, and 4-cycle completion arguments.

Significance. The paper makes substantial contributions to the topology and geometry of reconfiguration complexes. The equivariant Morse-theoretic framework (Theorem 2.1, Appendix A) extends Freij's theory to cubical complexes and yields a clean, parameter-free homotopy type for arbitrary base graphs. The diameter bounds (Theorem 3.3) partially answer an open question of Kozlov. The automorphism result (Theorem 4.14) is a strong rigidity statement. The cubical flag complex property (Lemma 4.3) and the 4-cycle completion argument (Proposition 4.11) are well-constructed. The matching distance lower bound (Proposition 3.22) is an elegant auxiliary tool. The paper is essentially self-contained.

major comments (2)
  1. Proposition 2.20, case (b): The acyclicity argument claims that a vertex $v notin Act_Y(N_i)$ 'can never be cospecified later,' and hence $|B setminus Act_Y|$ decreases monotonically along any putative cycle. This claim requires more careful justification. A vertex $v$ is in $Act_Y^-(N)$ when $v$ is a leaf of $T$ whose parent is an unoccupied leaf of $Sp(N)$. The fact that $v notin Act_Y(tilde{N}_i)$ at one step does not, by itself, prevent $v$ from entering $Act_Y$ at a later step $tilde{N}_j$: cospecifications of other vertices (via $u_Y$) modify the tree $T$ by removing vertices in $Act_Y^-(N_j)$, which can alter which vertices are leaves of $T$ and which spine leaves are unoccupied. A vertex specified to an interior vertex of the spine could potentially become a leaf of $T$ after such removals, and if its parent then becomes an unoccupied spine leaf, it would enter $Act_Y^-$. The mon
  2. Proposition 2.20, case (b) (continued): monotonicity claim on $|B setminus Act_Y|$ is load-bearing for the acyclicity of the Morse matching, which in turn is the foundation of Theorem 2.1. The authors should either prove that the tree modifications along the cycle cannot reintroduce a vertex into $Act_Y$, or provide a more robust monotone quantity. Note that the analogous argument in Proposition 2.29 (for the matching on $M_G setminus Sigma_G$) uses a different, cleaner monotonicity argument based on $Act_M$ and equation (2.4), which does not appear to suffer from the same issue.
minor comments (6)
  1. Section 3.4, proof of Proposition 3.22: The proof is given only for $n=4k$. The modifications for the other congruence classes are described briefly. It would help the reader to at least state the matching $M$ and the trees $T_1, T_2$ explicitly for one odd case (e.g., $n=4k+1$), since the unmatched vertex introduces a subtlety.
  2. Lemma 3.6, equation (3.1): The equality case for even $n$ is proved by a parity argument. The logic is sound but the exposition is slightly compressed; explicitly stating that $min(a, n-a)$ and $min(a-1, n-a+1)$ have opposite parity when $n$ is even would improve readability.
  3. Figure 2: The projection of $X_G$ for $G = K_5 setminus {e}$ is difficult to parse. The caption mentions violet squares and grey 3-cubes, but the distinction is not immediately clear in the figure. Consider enlarging or adding a schematic.
  4. Section 4.2, proof of Lemma 4.7: The notation $T^{v rightsquigarrow u}$ is defined in Definition 4.6, but the operation $tilde{T}^{v_1 rightsquigarrow i}$ used in the proof involves sliding to a vertex $i$ rather than to a vertex $u$ adjacent to $p_T(v)$. Clarify that $i$ plays the role of $u$ here, or adjust notation.
  5. Corollary 2.47: The formula for $f(n)$ involves a sum with a $k!$ factor inside. It would be helpful to verify this formula for small $n$ (e.g., $n=5,6$) and state the values explicitly, to assist the reader in checking the enumeration.
  6. The paper states that computations suggest the upper bound in Theorem 3.3 is sharp for $6 leq n leq 9$. It would strengthen the paper to include a table of these computed diameters alongside the bounds.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for a careful and constructive report. The referee identifies a genuine gap in the acyclicity argument for Proposition 2.20, case (b), which we acknowledge and will repair. No other major comments were raised.

read point-by-point responses

  1. Referee: Proposition 2.20, case (b): The acyclicity argument claims that a vertex v not in Act_Y(N_i) 'can never be cospecified later,' and hence |B setminus Act_Y| decreases monotonically along any putative cycle. The referee argues this claim requires more careful justification, because cospecifications of other vertices modify the tree T by removing vertices in Act_Y^-(N_j), which can alter which vertices are leaves of T and which spine leaves are unoccupied. A vertex specified to an interior vertex of the spine could potentially become a leaf of T after such removals, and if its parent then becomes an unoccupied spine leaf, it would enter Act_Y^-.

    Authors: The referee is correct that the monotonicity argument in case (b) of Proposition 2.20 is not adequately justified as written. The claim that a vertex v not in Act_Y(N_i) 'can never be cospecified later' does not follow from v not being active at a single step, because the tree modifications induced by cospecifications via u_Y can change which vertices are leaves of T and which spine leaves are unoccupied. We have carefully re-examined the argument and confirmed that the gap is real: the quantity |B setminus Act_Y| is not, by itself, obviously monotone along a putative cycle, for precisely the reason the referee identifies. We will revise the proof to use a more robust monotone quantity. Specifically, we observe that along any step N_i > tilde{N}_{i+1} in the putative cycle, a vertex v in B setminus Act_Y(N_i) is specified to an interior vertex of the spine (since specifying to a leaf of the spine would keep v in Act_Y). Once specified to an interior spine vertex, v enters A and becomes part of the spine Sp(N) for all subsequent cells in the cycle (since all cells in the cycle share the same spine by Lemma 2.18). A vertex that lies in the interior of the spine cannot become a leaf of T whose parent is an unoccupied leaf of Sp(N), because interior spine vertices are, by definition, not leaves of the spine. The key observation is that the spine is invariant along the entire cycle (Lemma 2.18), so the set of interior spine vertices is fixed. A vertex specified to an interior spine vertex remains in the interior of the spine for all subsequent steps, and hence can never enter Act_Y^- (which requires being a leaf of T whose parent is an unoccupied leaf of Sp(N)). This gives a genuine monotone quantity: the number of vertices in B that are not active strictly decreases at each revision: no

  2. Referee: Proposition 2.20, case (b) (continued): The monotonicity claim on |B setminus Act_Y| is load-bearing for the acyclicity of the Morse matching, which in turn is the foundation of Theorem 2.1. The referee notes that the analogous argument in Proposition 2.29 uses a different, cleaner monotonicity argument based on Act_M and equation (2.4), which does not suffer from the same issue.

    Authors: We agree with the referee's observation that the argument in Proposition 2.29 is cleaner and does not suffer from the same issue. The monotonicity in Proposition 2.29 relies on equation (2.4), which gives an explicit description of the non-active vertices as the neighbourhood of the occupied vertex u_N, and the inclusion N_{T_i}(u_{N_i}) subseteq N_{tilde{T}_{i+1}}(u_{tilde{N}_{i+1}}) provides a clean, strictly monotone quantity. We will revise the proof of Proposition 2.20, case (b), to provide a similarly rigorous argument. As described in our response to the first comment, the spine invariance along the cycle (Lemma 2.18) provides the needed monotonicity: once a non-active vertex in B is specified to an interior spine vertex, it remains in the interior of the invariant spine and can never re-enter Act_Y. This makes |B setminus Act_Y| genuinely strictly decreasing along the cycle, yielding the desired contradiction. We will rewrite the proof to make this argument explicit and self-contained, rather than relying on the brief and insufficient justification currently in the manuscript. revision: no

Circularity Check

0 steps flagged

No circularity found

full rationale

The paper is essentially self-contained in its derivations. The three main results (Theorem 2.1, Theorem 3.3, Theorem 4.14) are derived from definitions and first principles. The diameter bounds use quantities Φ_c and Ψ that are defined directly from tree structure (Definition 3.4), not fitted to data; the lower bound uses explicit tree constructions (path trees with specific labelings in Proposition 3.22). The matching distance δ (Definition 3.19) is defined independently of the graph distance it bounds. Citations to Kozlov [16] are to a different author's published work, not self-citation; where arguments are adapted (e.g., Lemma 2.9 reducing spine containment to the K_n case, Lemma 2.14 adapting [16, Lemma 4.18]), the adaptations are explicitly described and the new content (occupied leaves, general base graph G) is re-proven. The extension of Freij's equivariant Morse theory to cubical complexes (Theorem 2.3) is proved in full in Appendix A. The automorphism group determination (Theorem 4.14) proceeds through a chain of independent lemmas (cubical flag complex, rigid subcomplex, 4-cycle completion closure) without any step reducing to its own inputs. No fitted parameters are renamed as predictions, no uniqueness theorem is invoked from the authors' own prior work, and no ansatz is smuggled in via citation.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 3 invented entities

No free parameters are introduced. The axioms are standard mathematical results (Freij's Morse theory, Jordan's centroid theorem) plus one paper-specific proved result (acyclicity of the matching). The invented entities (Y_G, spine assignment, matching distance) are structural tools with independent derivations, not postulated objects.

axioms (4)
  • standard math Freij's equivariant generalised discrete Morse theory (Theorem 2.3, proved in Appendix A)
    The paper adapts Freij's theory to cubical complexes and proves the adaptation in Appendix A. This is the foundational tool for Theorem 2.1.
  • standard math Jordan's theorem on centroids of trees (Lemma 2.42)
    Used in the geometric analysis (Section 3) and the automorphism proof (Section 4) to identify canonical fixed points.
  • domain assumption The lower ideals of the poset P_G are barycentric subdivisions of cubes (from Kozlov [16])
    Stated in Definition 1.6; provides the cubical structure on |Δ(P_G)|. This is a background result from the predecessor paper.
  • ad hoc to paper The acyclicity of the generalised Morse matching ∼ on X_G ∖ Y_G (Proposition 2.20)
    This is proved in the paper but is a load-bearing structural assumption for the deformation retraction. The proof in case (b) relies on a monotonicity argument that is the most fragile part of the main topological result.
invented entities (3)
  • Minimal spine complex Y_G independent evidence
    purpose: Subcomplex of X_G consisting of network realignments with minimal spines; serves as the equivariant deformation retract.
    Defined structurally via occupied leaves (Definition 2.7); its properties are derived from the combinatorics of the base graph, not postulated.
  • Spine assignment c_N independent evidence
    purpose: Classifies connected components of the residual complex R_G by assigning each vertex to a connected component of the spine minus barriers.
    Defined in Definition 2.35; used to prove the product structure of residual components (Proposition 2.40).
  • Matching distance δ independent evidence
    purpose: Auxiliary metric on G_n that lower-bounds the graph distance; used to derive the diameter lower bound.
    Defined in Definition 3.19; its diameter is computed exactly (Proposition 3.22) via explicit tree constructions.

pith-pipeline@v1.1.0-glm · 45000 in / 2798 out tokens · 240914 ms · 2026-07-09T11:44:44.085619+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of Network Realignment Complexes over General Graphs." pith.science (2026). https://pith.science/paper/A5THMPH6

@misc{pith2026260707404,
  author       = {Pith},
  title        = {Pith review of: Network Realignment Complexes over General Graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A5THMPH6}},
  note         = {Machine review of arXiv:2607.07404}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Network realignment complexes were introduced by Kozlov. We generalise their definition to arbitrary connected base graphs. For a connected graph $G$, we characterise the connected components of the associated network realignment complex $X_G$ and show that $X_G$ admits an $\operatorname{Aut}(G)$-equivariant strong deformation retraction onto the disjoint union of a complete graph and a discrete $\operatorname{Aut}(G)$-space. For the complete base graph $K_n$, we study the metric structure of the network realignment graph $\mathcal{G}_n$ and obtain explicit upper and lower bounds for its diameter. Finally, we prove that $X_n$ is a cubical flag complex and that every automorphism of $X_n$ is induced by a relabelling of the underlying vertex set. In particular, $\operatorname{Aut}(X_n)\cong S_n$ for all $n\geq 5$.

Figures

Figures reproduced from arXiv: 2607.07404 by Fabienne Klatt, Iason Papadopoulos, Marc Raffelsiefen.

Figure 1
Figure 1. Figure 1: (a) A network realignment N1 over G, (b) a realignment specifica￾tion N3→5 1 , (c) A non-example of a network realignment over G ii) r : B → E(T) is a map satisfying (b, v),(b, w) ∈ E(G) for all b ∈ B with r(b) = (v, w). We call the vertices of T the specified vertices of N, and the elements of B the unspecified vertices of N. As an example, consider the complete graph on eight vertices with three edges er… view at source ↗
Figure 2
Figure 2. Figure 2: Projection of the network realignment complex XG with G = K5 \ {e}. Theorem 2.1. Let G be a connected graph on n ≥ 3 vertices, and let SG = {v ∈ V (G) | degG(v) = n − 1}. The action of Aut(G) on G restricts to an action on SG, and hence induces an action of Aut(G) on the complete graph KSG with vertex set SG. There exists an Aut(G)-equivariant embedding KSG ,→ XG and an Aut(G)-equivariant strong deformatio… view at source ↗
Figure 3
Figure 3. Figure 3: Consider the base graph G = K8 \{(1, 2),(2, 3),(4, 5)}. The missing edges of G are drawn in red. (a) A network realignment over G, (b) a non￾example of a network realignment over G, since 2 is a barrier of 1, and in (c), the vertices 6 and 8 are occupied by 1 and 4, respectively. Hence, the spine, drawn in green, cannot be reduced. realignment over G. As pT (1) = 7, the vertex 1 cannot be realigned to the … view at source ↗
Figure 4
Figure 4. Figure 4: (a) A network realignment in the minimal spine complex YG, where G = K8 \ {(1, 2),(2, 3),(4, 5)}, (b) a network realignment not contained in YG. Lemma 2.10. YG is a subcomplex of XG. Proof. If all leaves of the spine are occupied, then Sp(N) = Int(T). Indeed, in this situation every leaf u of Sp(N) lies in Int(T), because there exists a vertex v ∈ V (T) \ V (Sp(N)) with pT (v) = u, implying degT (u) ≥ 2. S… view at source ↗
Figure 5
Figure 5. Figure 5: Consider the base graph G = K8 \ {(1, 2),(2, 3),(4, 5)}. (a) A network realignment N ∈ MG \ ΣG with occupied vertex 7, (b) the network realignment dM(N), and (c) the network realignment uM(N). Consequently, Nv→wN ∈ MG \ ΣG. If v ∈ Act− M(N), then again Sp(Nv→(uN ,wN ) ) = Sp(N) and Nv→(uN ,wN ) ∈ MG \ ΣG. This motivates the definition of the realignments dM(N) and uM(N), in which all active vertices are sp… view at source ↗
Figure 6
Figure 6. Figure 6: Two network realignments in the minimal spine complex YG, where G = K8 \ {(1, 2),(2, 3),(4, 5)} with the spine assignment of 3 highlighted in blue. Proposition 2.33. KSG is an Aut(G)-equivariant strong deformation retract of MG. Proof. This follows from Proposition 2.32 and Proposition 2.30. □ 2.4. The Residual Components of YG. The preceding subsection gives a complete descrip￾tion of the topology of the … view at source ↗
Figure 7
Figure 7. Figure 7: The spines of the components in each orbit type of Corollary 2.47 together with the occupying vertices. Corollary 2.47. Assume n ≥ 5. Let P be a path of length three, and let G = Kn \ E(P). Then we have the following Aut(G)-equivariant strong deformation retraction of XG: XG ≃(Sn−4×C2) Kn−4 ∪ f a (n) i=1 {∗}, where f(n) = 4(n − 4)(n − 5) nX−6 k=0  n − 6 k  k! ! + (n − 4)(n − 3) + 1, and the union of poin… view at source ↗
Figure 8
Figure 8. Figure 8: The network realignments used in the proof of Lemma 4.7. While there are two paths of length 2 between T and T2, there is a single path of length 2 between f(T) and f(T2). This path passes through the star tree s1. Without loss of generality, assume T ∈ C1,2. Since Aut(Σn) = S ( n 2) n−2 ⋊ Sn, it follows that f|C1,2 is induced by a permutation σ of the vertices 3, . . . , n. Thus, T can be chosen to satisf… view at source ↗
Figure 9
Figure 9. Figure 9: If the tree T minimises the quantity Ψ, then there is no 4-cycle containing T whose other three vertices take strictly lower value in Ψ. In this situation, we can construct a sequence T, T1, . . . , Tk to find a different tree Ti and a desired 4-cycle. Along this sequence, Ψ − Fcent never increases. Case 1: T has a centroid c and two leaves v, w with pT (v) ̸= c ̸= pT (w). Let uv and uw denote the vertices… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

23 extracted references · 23 canonical work pages

  1. [1]

    Group actions on posets

    E. Babson and D. Kozlov. “Group actions on posets”. In:Journal of Algebra285 (2005), pp. 439–450.doi:https://doi.org/10.1016/j.jalgebra.2001.07.002

    work page doi:10.1016/j.jalgebra.2001.07.002 2005

  2. [2]

    Proof of the Lovasz Conjecture

    E. Babson and D. Kozlov. “Proof of the Lovasz Conjecture”. In:Ann. of Math.165 (2007), pp. 965–1007.doi:10.4007/annals.2007.165.965

    work page doi:10.4007/annals.2007.165.965 2007

  3. [3]

    The connectivity of the leaf-exchange spanning tree graph of a graph

    H. Broersma and X. Li. “The connectivity of the leaf-exchange spanning tree graph of a graph”. In:Ars Combinatoria43 (1996), pp. 225–231.url:https://api.semanticscholar. org/CorpusID:35042375

    work page 1996

  4. [4]

    Hamilton Circuits in Tree Graphs

    R. Cummins. “Hamilton Circuits in Tree Graphs”. In:IEEE Transactions on Circuit Theory13 (1966), pp. 82–90.doi:10.1109/TCT.1966.1082546

    work page doi:10.1109/tct.1966.1082546 1966

  5. [5]

    Diestel.Graph Theory

    R. Diestel.Graph Theory. Graduate Texts in Mathematics. Springer Berlin, Heidelberg, 1997.doi:https://doi.org/10.1007/978-3-662-70107-2

    work page doi:10.1007/978-3-662-70107-2 1997

  6. [6]

    On the chromatic number of tree graphs

    V. Estivill-Castro, M. Noy, and J. Urrutia. “On the chromatic number of tree graphs”. In: Discrete Mathematics223 (2000), pp. 363–366.doi:https://doi.org/10.1016/S0012- 365X(00)00092-3

    work page doi:10.1016/s0012- 2000

  7. [7]

    Morse Theory for Cell Complexes

    R. Forman. “Morse Theory for Cell Complexes”. In:Advances in Mathematics134 (1998), pp. 90–145.doi:https://doi.org/10.1006/aima.1997.1650

    work page doi:10.1006/aima.1997.1650 1998

  8. [8]

    Equivariant discrete Morse theory

    R. Freij. “Equivariant discrete Morse theory”. In:Discrete Mathematics309 (2009), pp. 3821– 3829.doi:https://doi.org/10.1016/j.disc.2008.10.029

    work page doi:10.1016/j.disc.2008.10.029 2009

  9. [9]

    The geometry and topology of reconfiguration

    R. Ghrist and V. Peterson. “The geometry and topology of reconfiguration”. In:Advances in Applied Mathematics38 (2007), pp. 302–323.doi:https://doi.org/10.1016/j.aam. 2005.08.009

    work page doi:10.1016/j.aam 2007

  10. [10]

    Distances between graphs under edge operations

    W. Goddard and H. C. Swart. “Distances between graphs under edge operations”. In: Discrete Mathematics161.1 (1996), pp. 121–132.issn: 0012-365X.doi:https://doi. org/10.1016/0012-365X(95)00073-6

    work page doi:10.1016/0012-365x(95)00073-6 1996

  11. [11]

    On the complexity of reconfiguration problems

    T. Ito et al. “On the complexity of reconfiguration problems”. In:Theoretical Computer Science412 (2011), pp. 1054–1065.doi:https://doi.org/10.1016/j.tcs.2010.12. 005

    work page doi:10.1016/j.tcs.2010.12 2011

  12. [12]

    Edge Rotation and Edge Slide Distance Graphs

    E. Jarrett. “Edge Rotation and Edge Slide Distance Graphs”. In:Computers and Math- ematics with Applications34.11 (1997), pp. 81–87.doi:https://doi.org/10.1016/ S0898-1221(97)00221-6

    work page 1997

  13. [13]

    Jonsson.Simplicial Complexes of Graphs

    J. Jonsson.Simplicial Complexes of Graphs. Lecture Notes in Mathematics. Springer Berlin, Heidelberg, 2007.doi:https://doi.org/10.1007/978-3-540-75859-4

    work page doi:10.1007/978-3-540-75859-4 2007

  14. [14]

    Sur les assemblages de lignes

    C. Jordan. “Sur les assemblages de lignes.” In:Journal f¨ ur die reine und angewandte Math- ematik (Crelles Journal)1869 (), pp. 185–190.url:https://api.semanticscholar.org/ CorpusID:119829832

    work page

  15. [15]

    The Connectivities of Leaf Graphs of 2-Connected Graphs

    A. Kaneko and K. Yoshimoto. “The Connectivities of Leaf Graphs of 2-Connected Graphs”. In:Journal of Combinatorial Theory, Series B76 (1999), pp. 155–169.doi:https://doi. org/10.1006/jctb.1998.1895

    work page doi:10.1006/jctb.1998.1895 1999

  16. [16]

    Network realignment complexes

    D. Kozlov. “Network realignment complexes”. In:J Appl. and Comput. Topology9 (2025). doi:https://doi.org/10.1007/s41468-025-00204-0

    work page doi:10.1007/s41468-025-00204-0 2025

  17. [17]

    The Connectivities of Adjacent Tree Graphs

    G. Liu. “The Connectivities of Adjacent Tree Graphs”. In:Acta Mathematicae Applicatae Sinica3 (1987), pp. 313–317.doi:10.1007/BF02008369

    work page doi:10.1007/bf02008369 1987

  18. [18]

    On connectivities of tree graphs

    G. Liu. “On connectivities of tree graphs”. In:Journal of Graph Theory12 (1988), pp. 453– 459. 46 REFERENCES

    work page 1988

  19. [19]

    Kneser’s conjecture, chromatic number, and homotopy

    L. Lov´ asz. “Kneser’s conjecture, chromatic number, and homotopy”. In:Journal of Combi- natorial Theory, Series A25 (1978), pp. 319–324.doi:https://doi.org/10.1016/0097- 3165(78)90022-5

    work page doi:10.1016/0097- 1978

  20. [20]

    Introduction to Reconfiguration

    N. Nishimura. “Introduction to Reconfiguration”. In:Algorithms11 (2018).doi:10.3390/ a11040052

    work page 2018

  21. [21]

    Reasons to Fall (More) in Love with Combinatorial Reconfiguration

    N. Nishimura. “Reasons to Fall (More) in Love with Combinatorial Reconfiguration”. In: WALCOM: Algorithms and Computation. Ed. by R. Uehara, K. Yamanaka, and H.-C. Yen. Springer Nature Singapore, 2024, pp. 9–14.doi:https://doi.org/10.1007/978- 981-97-0566-5_2

    work page doi:10.1007/978- 2024

  22. [22]

    Spanning Trees: A Survey

    K. Ozeki and T. Yamashita. “Spanning Trees: A Survey”. In:Graphs and Combinatorics 27.1 (2011), pp. 1–26.doi:10.1007/s00373-010-0973-2

    work page doi:10.1007/s00373-010-0973-2 2011

  23. [23]

    State Complexes and Special Cube Complexes

    V. J. Peterson. “State Complexes and Special Cube Complexes”. In:Topology Proceedings 45 (2015), pp. 73–109. REFERENCES 47 AppendixA.Equivariant Generalised Morse Theory Now, we prove the main theorem of generalised Morse theory. The key step is to show that an interval [σ, τ] collapses onto its complement in the cubeτ. By ordering the nontrivial interval...

    work page 2015