Derangement Graph Overview

• Derangement graphs are Cayley graphs formed from fixed-point-free elements in a permutation group, providing a clear framework for studying group actions.
• They translate intersecting family problems into independent set and clique questions, enabling applications such as EKR theorems and spectral analysis via character theory.
• Generalizations including k-derangement graphs, even derangement graphs, and perfect matching schemes extend the classical theory to diverse combinatorial and geometric structures.

A derangement graph most commonly denotes the Cayley graph associated with a permutation action: if a finite permutation group $G$ acts on a finite set $\Omega$, and $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$ is the set of derangements, then the derangement graph is $\Gamma(G,D(G))$, with vertex set $G$ and adjacency $g\sim h$ if and only if $g^{-1}h\in D(G)$ (Meagher et al., 2013). In the literature, however, closely related but distinct notions also occur: the derangement graph on $S_n$ with adjacency “disagree in every position” (Lv et al., 2022), generalized $k$-derangement graphs on $S_n$ (Jackson et al., 2011), derangement action digraphs on an arbitrary set $\Omega$0 generated by fixed-point-free permutations (Iradmusa et al., 2018), and “graph derangements,” which are fixed-point-free adjacency-respecting permutations of the vertex set of a graph and are explicitly not the same object as the permutation-group derangement graph (Clark, 2013). The term therefore denotes a family of constructions unified by fixed-point-free behavior, but differentiated by whether the vertices are group elements, permutations, tuples, perfect matchings, or vertices of an underlying graph.

1. Core definition in permutation group theory

For a finite permutation group $\Omega$1, the standard derangement graph is the Cayley graph

$\Omega$2

where $\Omega$3 is the set of elements of $\Omega$4 with no fixed points on $\Omega$5 (Razafimahatratra, 2021). Equivalently,

$\Omega$6

so adjacency records whether the relative permutation between two group elements is a derangement (Fusari et al., 2023).

This graph is regular of valency $\Omega$7, vertex-transitive by the right-regular action of $\Omega$8, and normal because the derangement set is a union of conjugacy classes (Razafimahatratra et al., 2020). For a normal Cayley graph, the adjacency eigenvalues are given by irreducible characters: $\Omega$9 a formula repeatedly used in spectral analyses of derangement graphs (Razafimahatratra, 2021).

The independent sets of $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$0 are precisely the intersecting families in $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$1: a subset $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$2 is intersecting when every $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$3 satisfy that $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$4 fixes some point of $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$5, equivalently $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$6 (Meagher et al., 2013). This identification makes the derangement graph a standard encoding device for Erdős–Ko–Rado-type problems in permutation groups. Dually, cliques are subsets $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$7 such that all pairwise quotients are derangements (Fusari et al., 2023).

Two recurrent numerical invariants are the independence number $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$8 and the clique number $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$9. The clique–coclique bound

$\Gamma(G,D(G))$0

links clique growth to upper bounds on intersecting families (Fusari et al., 2023). This suggests why both cocliques and cliques are central in the literature: cocliques control EKR-type extremal problems, while cliques force corresponding upper bounds.

2. The symmetric-group derangement graph and its cycle structure

For the classical derangement graph on $\Gamma(G,D(G))$1, the vertex set is $\Gamma(G,D(G))$2, and two permutations $\Gamma(G,D(G))$3 are adjacent exactly when they disagree in every position: $\Gamma(G,D(G))$4 equivalently $\Gamma(G,D(G))$5, where $\Gamma(G,D(G))$6 is the set of derangements in $\Gamma(G,D(G))$7 (Lv et al., 2022). In this formulation the graph is the normal Cayley graph

$\Gamma(G,D(G))$8

with $\Gamma(G,D(G))$9 and degree $G$0 (Cao et al., 18 Aug 2025).

Several basic structural facts are established explicitly. The graph is connected for $G$1; $G$2 is a single edge; and $G$3 is disconnected, consisting of two 3-cycles (Cao et al., 18 Aug 2025). For connected cases, the diameter is $G$4: if $G$5, then for $G$6, one has $G$7 if $G$8, and otherwise $G$9 (Li et al., 2016). The same paper uses the diameter-two identity

$g\sim h$0

for the distance matrix $g\sim h$1 in terms of the adjacency matrix $g\sim h$2, enabling a direct translation from adjacency spectra to distance spectra (Li et al., 2016).

The graph also has a rich cycle structure. It was already known to be Hamiltonian and Hamilton-connected, and the stronger result now recorded is that $g\sim h$3 is edge pancyclic for $g\sim h$4: every edge lies on cycles of all lengths $g\sim h$5 (Lv et al., 2022). The proof combines a partition of $g\sim h$6 into $g\sim h$7-cliques $g\sim h$8, a fixed-point counting identity

$g\sim h$9

for $g^{-1}h\in D(G)$0, and the edge pancyclicity of the complement of $g^{-1}h\in D(G)$1 (Lv et al., 2022).

The adjacency spectrum is controlled by the representation theory of $g^{-1}h\in D(G)$2. If $g^{-1}h\in D(G)$3, then the adjacency eigenvalue indexed by $g^{-1}h\in D(G)$4 is

$g^{-1}h\in D(G)$5

with multiplicity $g^{-1}h\in D(G)$6 (Ku et al., 2012). A central recurrence is

$g^{-1}h\in D(G)$7

where $g^{-1}h\in D(G)$8 deletes the last row and $g^{-1}h\in D(G)$9 removes the first column (Ku et al., 2012). The same work proves the Alternating Sign Property

$S_n$0

and strict monotonicity of $S_n$1 under dominance order, settling the Ku–Wales conjecture on extremal eigenvalue magnitudes (Ku et al., 2012).

These spectral facts feed directly into extremal combinatorics. In particular, the smallest adjacency eigenvalue is attained at $S_n$2,

$S_n$3

and Hoffman's bound yields $S_n$4, the classical EKR value for permutations (Ku et al., 2012).

3. Intersecting families, EKR theorems, and geometric classification

The derangement graph formalism is particularly effective for classification theorems on maximum intersecting sets. A definitive example is $S_n$5 acting on the points of the projective plane $S_n$6. In this setting,

$S_n$7

with

$S_n$8

(Meagher et al., 2013).

The main theorem in that case classifies all maximum independent sets. If $S_n$9 is intersecting of maximum size, then $k$0 is a coset of the stabilizer of a point or a coset of the stabilizer of a line, and

$k$1

(Meagher et al., 2013). The abstract of the paper contains the phrase “points of the projective line,” but the paper’s theorem concerns the action on $k$2, and this is explicitly identified as a typo (Meagher et al., 2013).

The proof combines two layers. First, a ratio-bound argument based on character theory shows that the independence number is exactly

$k$3

using the minimum eigenvalue

$k$4

and the largest eigenvalue $k$5 (Meagher et al., 2013). Second, a linear-algebraic classification analyzes a matrix $k$6 indexed by group elements and ordered pairs of points, together with a block $k$7 consisting of the derangement rows. The characteristic vector of a maximum independent set lies in the column space of $k$8, and the right kernel of $k$9 is described explicitly in terms of points and lines in projective geometry (Meagher et al., 2013).

This geometric outcome is noteworthy because the maximum cocliques are not of a single type: both point stabilizers and line stabilizers occur. The paper presents this as the first confirmed case of a higher-dimensional projective conjecture in which two genuinely different geometric families of maximum intersecting sets appear (Meagher et al., 2013). A plausible implication is that the derangement graph does not merely encode fixed-point-free behavior abstractly; in geometric actions it can detect dual incidence structures with full rigidity at the extremal level.

More broadly, the connection between cocliques and intersecting families underlies a sequence of results for transitive groups. The existence of large cliques forces upper bounds on the intersection density $S_n$0, and this mechanism motivates the study of cliques in derangement graphs for innately transitive and other structured permutation groups (Fusari et al., 2023).

4. Cliques, multipartite structure, and Latin-square correspondences

Clique structure in derangement graphs has developed into an independent theme. One universal result is that if $S_n$1 is transitive of degree at least $S_n$2, then the derangement graph $S_n$3 contains a triangle (Razafimahatratra et al., 2020). A stronger recent theorem states that if a transitive permutation group has degree exceeding $S_n$4, then its derangement graph contains a $S_n$5; the only exceptions listed are degrees $S_n$6 and $S_n$7 with specific groups (Cazzola et al., 3 Feb 2025). This theorem is used to deduce a bound $S_n$8 in an index-$S_n$9 covering problem related to Kronecker classes (Cazzola et al., 3 Feb 2025).

A different line of work studies when the whole derangement graph is complete multipartite. For a transitive group $\Omega$00, define

$\Omega$01

Then $\Omega$02 is complete multipartite if and only if $\Omega$03 is intersecting; in that case the parts are exactly the left cosets of $\Omega$04 (Razafimahatratra, 2021). Two infinite families are constructed explicitly. One uses a subgroup $\Omega$05 acting transitively on the affine lines of $\Omega$06, producing a complete $\Omega$07-partite derangement graph of degree $\Omega$08 (Razafimahatratra, 2021). The other gives, for odd $\Omega$09, a transitive group of degree $\Omega$10 whose derangement graph is complete $\Omega$11-partite (Razafimahatratra, 2021).

The complete multipartite case is spectrally rigid. For a complete $\Omega$12-partite graph with equal part size $\Omega$13,

$\Omega$14

and the spectrum is

$\Omega$15

(Razafimahatratra, 2021). In the affine-line family, this yields

$\Omega$16

while the maximum cocliques are exactly the cosets of $\Omega$17 (Razafimahatratra, 2021).

A separate but related clique theory emerges for the symmetric-group derangement graph $\Omega$18, where every maximal clique has size exactly $\Omega$19 (Anderson et al., 2024). Here a Latin square $\Omega$20 of order $\Omega$21 gives a maximal clique

$\Omega$22

and two orthogonal Latin squares produce two disconnected maximal cliques (Anderson et al., 2024). The paper proves a bijection between pairs of orthogonal Latin squares and pairs of disconnected ordered maximal cliques in $\Omega$23 (Anderson et al., 2024).

This bijection is then combined with modular obstructions derived from the natural representation. For $\Omega$24,

$\Omega$25

and in characteristic dividing $\Omega$26 the dimension of the image drops from $\Omega$27 to $\Omega$28 (Anderson et al., 2024). These modular dependencies are used to analyze small $\Omega$29, culminating in a short proof that there do not exist two disconnected maximal cliques in $\Omega$30, hence no pair of orthogonal Latin squares of order $\Omega$31 (Anderson et al., 2024).

5. Generalizations on permutation spaces

The ordinary derangement graph on $\Omega$32 is only one member of a broader family of Cayley and association-scheme constructions.

A first extension is the generalized $\Omega$33-derangement graph $\Omega$34 on $\Omega$35, where a permutation is a $\Omega$36-derangement if it fixes no $\Omega$37-subset setwise (Jackson et al., 2011). Two permutations $\Omega$38 are adjacent when $\Omega$39 is a $\Omega$40-derangement (Jackson et al., 2011). The cycle criterion states that $\Omega$41 is a $\Omega$42-derangement if and only if its cycle decomposition contains no submultiset of cycle lengths summing to $\Omega$43 (Jackson et al., 2011). For $\Omega$44 and $\Omega$45, $\Omega$46 is connected, and it is Eulerian if and only if either $\Omega$47 is even or both $\Omega$48 and $\Omega$49 are odd (Jackson et al., 2011). When $\Omega$50 is an odd prime power, the $\Omega$51-derangement graph satisfies

$\Omega$52

(Jackson et al., 2011).

Another extension is the even derangement graph $\Omega$53, where $\Omega$54 is the set of even derangements in $\Omega$55 (Deng et al., 2011). Its tensor powers $\Omega$56 remain connected and non-bipartite for $\Omega$57, have diameter $\Omega$58, and satisfy

$\Omega$59

(Deng et al., 2011). Moreover, every maximum independent set is a coordinate fiber

$\Omega$60

and the full automorphism group is determined explicitly (Deng et al., 2011).

A third construction is the perfect matching derangement graph $\Omega$61, whose vertices are perfect matchings of $\Omega$62, with adjacency defined by disjointness of edges (Zhang et al., 2023). Its eigenvalues are indexed by partitions $\Omega$63, and a new recurrence is derived by passing to the sign-normalized quantities

$\Omega$64

(Zhang et al., 2023). The resulting monotonicity theorem parallels Ku–Wong’s theorem for $\Omega$65: if $\Omega$66 and $\Omega$67, then $\Omega$68, with equality if and only if $\Omega$69 and all remaining parts are at most $\Omega$70 (Zhang et al., 2023).

Finally, the phrase “derangement graph” also appears in graph representation theory. A graph $\Omega$71 has a derangement $\Omega$72-representation if there is an injective map $\Omega$73 such that

$\Omega$74

equivalently $\Omega$75 is an induced subgraph of the classical derangement graph $\Omega$76 (Ashofteh et al., 2024). The paper proves that every finite graph has such a representation for some $\Omega$77, and defines the derangement representation number $\Omega$78 as the least such $\Omega$79 (Ashofteh et al., 2024). This suggests a converse viewpoint: instead of studying the internal structure of derangement graphs, one can use the family $\Omega$80 as a universal host family for arbitrary graphs.

6. Graph derangements and derangement action digraphs

A terminological distinction is essential. In graph theory, a “graph derangement” means a fixed-point-free permutation $\Omega$81 of the vertex set of a graph $\Omega$82 such that $\Omega$83 for all $\Omega$84 (Clark, 2013). This is not the permutation-group derangement graph. The paper introducing graph derangements explicitly states that the notion is different from the “derangement graph” used in permutation group theory (Clark, 2013).

Graph derangements interpolate between perfect matchings and Hamiltonian cycles. If every cycle of $\Omega$85 has length $\Omega$86, one gets a perfect matching; if $\Omega$87 is a single $\Omega$88-cycle, one gets a Hamiltonian cycle (Clark, 2013). For locally finite graphs, existence is governed by a Hall-type condition: $\Omega$89 admits a surjective graph derangement if and only if for every finite independent set $\Omega$90,

$\Omega$91

(Clark, 2013). In bipartite graphs this is equivalent to the existence of a perfect matching, hence to a dyadic graph derangement (Clark, 2013).

A game-theoretic reformulation appears in the Territorial Raider Game. For a simple, finite, undirected, connected graph $\Omega$92 and parameter $\Omega$93, the graph admits a derangement if and only if the Territorial Raider Game on $\Omega$94 has a strict Nash equilibrium (Galanter et al., 2015). The implication from a derangement to a strict equilibrium is direct: if every player raids the neighbor prescribed by the derangement, each player receives payoff $\Omega$95, and any unilateral deviation yields strictly smaller payoff (Galanter et al., 2015). The reverse implication proves that any strict equilibrium must be injective and fixed-point-free along edges, hence a graph derangement (Galanter et al., 2015).

A broader dynamical framework is provided by derangement action digraphs. If $\Omega$96 is a non-empty set and $\Omega$97 is finite, the derangement action digraph $\Omega$98 has vertex set $\Omega$99 and arcs $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$00 for $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$01 and $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$02 (Iradmusa et al., 2018). When $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$03 is closed—meaning $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$04 for all $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$05 and $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$06—the resulting undirected graph is regular of valency $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$07 (Iradmusa et al., 2018). This generalizes Cayley graphs: every Cayley digraph is a derangement action digraph, and the class of finite derangement action graphs contains every finite vertex-transitive simple graph and every finite regular simple graph of even valency (Iradmusa et al., 2018).

The infinite case admits a finitary characterization. An infinite simple loopless digraph $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$08 is generated by at most $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$09 derangements if and only if every vertex has in-degree and out-degree at most $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$10 and every finite subset satisfies two explicit neighborhood inequalities (Horsley et al., 2019). The proof passes through the bipartite double $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$11 and the equivalence between derangement generation and the existence of a 1-factor cover of $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$12 by at most $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$13 perfect matchings (Horsley et al., 2019). This suggests that the derangement paradigm extends naturally from group actions to highly non-group-theoretic network constructions.

7. Conceptual synthesis

Across its variants, the derangement graph paradigm organizes fixed-point-free phenomena into graph-theoretic form. In the permutation-group setting, it converts intersection problems into coclique problems and nonintersection problems into clique problems, enabling the use of character theory, Hoffman bounds, and Cayley-graph symmetry (Meagher et al., 2013). In the symmetric-group case it yields a graph that is connected for $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$14, diameter $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$15, Hamilton-connected, and edge pancyclic (Li et al., 2016). In geometric actions such as $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$16 on $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$17, it supports complete classifications of maximum intersecting families in terms of point and line stabilizers (Meagher et al., 2013).

At the same time, the literature shows that “derangement graph” is not a single rigid notion. It can refer to generalized $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$18-subset avoidance on $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$19 (Jackson et al., 2011), even derangements on $D(G)=\{g\in G: g\text{ fixes no point of }\Omega\}$20 (Deng et al., 2011), perfect-matchings schemes (Zhang et al., 2023), disconnected clique geometry tied to orthogonal Latin squares (Anderson et al., 2024), or action digraphs generated directly by fixed-point-free permutations of an arbitrary set (Iradmusa et al., 2018). In contrast, graph derangements concern adjacency-respecting permutations of vertices and belong to a different branch of the subject (Clark, 2013).

This suggests a unifying description: a derangement graph is a graph built from a fixed-point-free relation, with adjacency usually determined by whether a relative move is fixed-point-free. The precise ambient object—group elements, permutations, tuples, matchings, or graph vertices—determines the corresponding algebraic, spectral, and extremal theory.