Cayley Graphs: Structure and Properties
- Cayley graphs are highly symmetric graphs defined from a group and its symmetric generating set, ensuring regularity, connectivity, and vertex-transitivity.
- They bridge group theory with combinatorics and spectral graph theory, serving as canonical models in analyzing expansion, coloring, and spectral phenomena.
- Their well-defined automorphism and spectral properties enable effective algorithmic recognition, enumeration, and characterization in advanced research.
A Cayley graph is a highly symmetric, combinatorially defined graph that encodes the structure of a group through its generating set. Formally, for a finite group and a subset satisfying , the Cayley graph has vertex-set and edges . Cayley graphs provide deep connections between group theory, combinatorics, and spectral graph theory, and serve as canonical models for vertex-transitive graphs and as test cases for fundamental questions in expansion, coloring, connectivity, and spectral phenomena.
1. Fundamental Definition and Basic Properties
A finite (undirected, simple) Cayley graph is specified by a group and a “connection set” , with guaranteeing undirectedness. The vertex set is , and 0 is an edge if 1. This construction ensures several universal properties:
- Regularity: Every vertex has degree 2.
- Vertex-transitivity: The right-regular action of 3 is by automorphisms and acts transitively on 4.
- Connectivity: The graph is connected if and only if 5 generates 6 (Ganesan, 2017).
- Normal Cayley graphs: A graph 7 is called normal if 8 is a union of conjugacy classes. In this case, the full automorphism group is exactly 9, where 0 is the group of right multiplication automorphisms (Ganesan, 2017).
- Integral Cayley graphs: A graph is integral if all eigenvalues of its adjacency matrix are integers (Guo et al., 2018, Árnadóttir et al., 2023).
2. Structural and Characterization Results
Cayley graphs can be characterized among all (possibly directed, labeled) graphs by precise graph-theoretic properties:
- Determinism and simplicity: Cayley graphs are deterministic and simple as labeled digraphs, meaning outgoing edges from each vertex are labeled distinctly, and multiple edges and loops are excluded (Caucal, 2016, Caucal, 2018).
- Vertex-transitivity: For a (possibly labeled) digraph, being simple, deterministic, rooted, and vertex-transitive suffices to ensure it is a Cayley graph (Caucal, 2016).
- Alternative characterizations: Cayley graphs coincide with strongly connected, deterministic, and circular graphs (every vertex has the same cycle language) (Caucal, 2016).
- Combinatorial propagation: Arc-symmetry and determinism are minimal requirements for Cayley graphs of left-cancellative monoids, and adding co-determinism upgrades to group structures (Caucal, 2019).
- Extensions: For generalized algebraic structures (magmas, semigroups, monoids, groups, quasigroups), there are sharp hierarchies of graph-theoretic properties corresponding to these algebraic axioms (Caucal, 2019, Caucal, 2018).
- Automorphism group criterion: A finite graph is a Cayley graph for some group if and only if it admits a subgroup of 1 acting regularly ((Morris et al., 2022), Sabidussi’s theorem).
3. Spectral Theory and Association Schemes
Spectral analysis of Cayley graphs utilizes group representations. The adjacency algebra (Bose–Mesner algebra) of a normal Cayley graph is a commutative semisimple algebra, and its eigenvalues are determined by irreducible characters:
- For normal, Eulerian subsets 2, all eigenvalues are integers (Guo et al., 2018).
- In odd order groups, any non-empty integral normal Cayley graph must possess at least one odd eigenvalue, a property detected via association schemes and frame quotient arguments (Árnadóttir et al., 2023).
- Spectral formulas: For 3, the eigenvalues are 4, each with multiplicity 5.
- Twisted Cayley graphs and other variants preserve eigenvalue sets up to 6 factors, with deep consequences for cospectral/non-isomorphic constructions (Biswas et al., 2021).
4. Variants and Generalizations
Several generalizations and refinements of Cayley graphs have been studied:
- Two-sided Cayley graphs: Defined on a group 7 by two (possibly distinct) non-empty subsets 8, 9, with edges from 0 to 1, exhibiting subtler symmetry and connectedness properties and leading to graphs not always covered by standard Cayley territory (Iradmusa et al., 2014).
- Relative Cayley graphs and Cayley-type graphs for group–subgroup pairs: Extension to graphs 2 where 3 and 4, with adjacency conditioned on coset positions. These constructions interpolate between Cayley graphs and other highly structured families (e.g., star graphs, bipartite block structures) (Ghouchan et al., 2015, Reyes-Bustos, 2014).
- Expander families: Specific algebraic families, e.g., those over finite fields via character sum estimates, provide new explicit constructions of connected, expanding, sparse Cayley graphs (Lu et al., 2013, Biswas et al., 2021).
- Graphs on rings and gyrogroups: Involutory Cayley graphs and Cayley graphs on gyrogroups extend the combinatorial framework beyond groups (Keshavarzi et al., 2 Aug 2025, S, 2023).
5. Automorphisms, Symmetry, and Colourings
Cayley graphs are prototypes of vertex-transitive graphs; their automorphism group structure is rich and heavily influenced by the group 5 and generating set 6. Notable points include:
- For normal Cayley graphs, the automorphism group is the semidirect product 7. Exceptions arise for certain generating sets, leading to non-normal Cayley graphs with "unexpected" automorphisms (Ganesan, 2017).
- Most cubic and quartic Cayley graphs are graphical regular representations (GRRs): 8 for large orders (Evans et al., 5 Dec 2025).
- Colouring invariants (vertex, edge, total chromatic numbers) of Cayley graphs receive attention, with most families conjectured to satisfy the Total Colouring Conjecture and, in many cases, to attain the bounds 9 (S, 2023).
6. Enumerative, Asymptotic, and Algorithmic Aspects
Large-scale enumeration and census of Cayley graphs have become feasible via algorithmic advances:
- Algorithms systematically enumerate all non-isomorphic 0-valent Cayley graphs for given 1, using sophisticated orbit and automorphism filtering (Evans et al., 5 Dec 2025).
- Asymptotic phenomena: The proportion of GRRs tends to 1 for fixed valency, and typical Cayley graphs become bipartite for large orders (Evans et al., 5 Dec 2025).
- Structural recognition algorithms—based on characterizations via determinism, simplicity, and vertex-symmetry—grant the ability to algorithmically recover the underlying group or magma from the graph structure (Caucal, 2016, Caucal, 2019).
- For end-regular graphs of finite degree, Cayley graph recognition and algebraic recovery are effective and decidable (Caucal, 2018).
7. Open Problems and Research Directions
Several fundamental problems in Cayley graph theory remain open:
- Complete characterization of possible spectra (especially odd/even eigenvalue multiplicities) for integral/normal Cayley graphs of various orders (Árnadóttir et al., 2023).
- Refining conditions under which two-sided or relative Cayley graphs are genuinely Cayley graphs, and the precise relationship between vertex-transitivity, regular subgroups, and the automorphism group (Iradmusa et al., 2014, Ghouchan et al., 2015, Morris et al., 2022).
- Asymptotic density of orders of Cayley graphs with prescribed girth and the limiting frequency of bipartiteness and GRR property (Evans et al., 5 Dec 2025).
- Classifying all non-abelian twins: pairs of non-isomorphic groups on which the same Cayley graph can be defined (Morris et al., 2022).
- Understanding property-polynomial and testability phenomena in local convergence of Cayley graphs versus diagrams (Timar, 2011).
Cayley graphs thus serve as a central object at the crossroad of group theory, spectral graph theory, algebraic combinatorics, structural graph theory, and algorithmic group theory, continuing to motivate a spectrum of deep structural, enumerative, and algorithmic investigations.