Color-Critical Graphs: Theory & Applications

• Color-critical graphs are defined as those whose chromatic number decreases upon deletion of a specific edge, serving as a cornerstone in graph coloring theory.
• They exhibit tight extremal bounds and structural characterizations, such as Gallai’s theorem and sharp lower bounds on edge counts.
• Their study drives advances in algorithm design, spectral analysis, and extensions to hypergraph, circular, and fractional coloring.

A color-critical graph is a graph whose chromatic number drops upon the deletion of at least one edge; formally, a graph $F$ is color-critical if there exists $e \in E(F)$ with $\chi(F-e) < \chi(F)$ (Pikhurko et al., 2012, Li et al., 19 Nov 2025, Zheng et al., 10 Apr 2025, Chakraborti et al., 2022). These graphs lie at the intersection of extremal graph theory, coloring theory, and algebraic methods, and possess structural, spectral, and algorithmic properties that are central to the understanding of chromatic phenomena.

1. Formal Definitions and Basic Properties

Let $G$ be a finite simple graph.

• $k$-critical graph: $G$ is $k$-critical if $\chi(G)=k$ and every proper subgraph of $G$ is $(k-1)$-colorable (Kostochka et al., 2012, Sebő, 2019).
• Color-critical edge: $e\in E(G)$ is called color-critical if $\chi(G-e) < \chi(G)$ (Simonyi et al., 2019).
• Color-critical graph: $F$ is color-critical if it contains a color-critical edge; equivalently, there exists $e$ with $\chi(F-e) = \chi(F) - 1$.

Special classes include vertex-color-critical graphs (removal of any vertex reduces $\chi$), as exemplified by Schrijver graphs (Gujgiczer et al., 2022). The notion generalizes in fractional and circular coloring settings.

Further, the minimum edge count in a $k$-critical graph with $n$ vertices is denoted $f_k(n)$ (Kostochka et al., 2012).

2. Extremal Bounds and Structure Theorems

Classical results established tight bounds and structural characterizations:

• Gallai’s Theorem: Every connected $k$-color-critical graph has at most $2k-1$ vertices, with equality if and only if it is the odd cycle $C_{2k-1}$ (Sebő, 2019).
• Ore’s Conjecture (almost true): For $k\geq 4$ and $n\geq k+2$,

$$f_k(n+k-1) = f_k(n) + \frac{k-1}{2}\left(k - \frac{2}{k-1}\right)$$

holds for all but $O(k^3)$ values of $n$ (Kostochka et al., 2012). There exists a sharp lower bound:

$$f_k(n) \geq \left\lceil\frac{(k+1)(k-2)n - k(k-3)}{2(k-1)}\right\rceil$$

with tightness for $k=4$ and $n\geq 6$, and all $n\equiv 1 \pmod{k-1}$ for $k\geq 5$.

• Supersaturation: For color-critical $F$ with chromatic number $r+1$, the Turán graph $T_r(n)$ maximizes the edge count in $F$-free graphs. Upon adding $q$ edges to $T_r(n)$, the minimum number $h_F(n,q)$ of $F$-copies grows linearly for $q=O(n)$ (Pikhurko et al., 2012). More generally, threshold phenomena govern when the "add edges to Turán graph" construction is optimal (Ma et al., 2023).

3. Cycle Structure and Local Certificates

A foundational property of color-critical graphs is the forced presence of many cycles of specific length congruence:

• Quantitative Kempe-chain Theorems: In a $k$-critical graph,
  • Every critical edge lies in at least $\prod_{i=1}^{r-1}(k-i)$ cycles of length $1 \pmod r$ for $2 \leq r \leq k$.
  • Upon edge deletion, at least $2\prod_{i=1}^{r-1}(k-i)$ cycles of length divisible by $r$ remain (Moore et al., 2019).

These results generalize Tuza’s theorem and show that color-criticality enforces a rich supply of cycles with prescribed modularity, unifying and sharpening earlier ad-hoc congruence results. For circular coloring, similar cycle-existence bounds hold for $(k,d)$-colorings, linking chromatic obstruction to algebraic cycle parity (Moore et al., 2019).

4. Structural Theory: Special Cases and Constructions

Certain graph families admit explicit characterizations:

• Schrijver and Kneser Graphs: The Schrijver graph $\mathrm{SG}(n,k)$ is an induced subgraph of the Kneser graph on $k$-subsets of $[n]$, and is vertex-color-critical (removal of any vertex lowers $\chi$ by one) (Gujgiczer et al., 2022).
  • In $\mathrm{SG}(2k+2,k)$, every interlacing edge (those whose elements alternate around $C_n$) is color-critical. Non-interlacing edges are never color-critical (Simonyi et al., 2019).
  • Special quadrangulating subgraphs of $\mathrm{SG}(2k+2,k)$ afford edge-color-criticality through combinatorial embeddings on nonorientable surfaces (e.g., Klein bottle, projective plane).
• $(P_5,\overline{P}_5)$-free Graphs: The set of $k$-critical graphs free of $P_5$ and its complement is finite for every fixed $k$.
  • Constructed recursively either as joins of smaller critical graphs or by substituting in "buoys" (blow-ups of $C_5$) with tight constraints on bag sizes (Dhaliwal et al., 2014).

5. Spectral and Algebraic Extremal Characterizations

Recent advances establish spectral analogues of classical extremal graph bounds for color-critical graphs.

• Adjacency and Signless Laplacian Spectral Turán Theorems: For color-critical $F$ with $\chi(F)=r+1\ge 4$, every $F$-free graph $G$ with $m$ edges satisfies (Li et al., 19 Nov 2025):

$\lambda(G) \leq \sqrt{(1 - 1/r) 2m}$

and for $n$ vertices,

$q(G) \leq q(T_{n,r})$

(where $q(G)$ is the largest eigenvalue of $D(G)+A(G)$), with equality if and only if $G$ is a regular complete $r$-partite graph (Zheng et al., 10 Apr 2025). Structural stability and eigenvector perturbation bounds drive these results.

• Degree Moment Bounds: For color-critical $F$ and $G$ $F$-free,

$$\sum_{v \in V(G)} d(v)^2 \leq 2\left(1- \frac{1}{r}\right)mn$$

with equality precisely for regular Turán graphs (Zheng et al., 10 Apr 2025).

6. Supersaturation and Rainbow Extremal Problems

Supersaturation theory quantifies the minimal number of forbidden subgraphs forced by surplus edges in near-extremal settings.

• Excess edge regime: For color-critical $F$, adding $q \leq c_1 n$ edges to $T_r(n)$ achieves the minimum $F$-subgraphs ($q \cdot c(n,F)$), where $c(n,F)$ is the single-edge contribution; for larger $q$, unbalanced constructions become optimal (Pikhurko et al., 2012, Ma et al., 2023).
• Rainbow extremal functions: Given $k$ colorings of edges, for color-critical $H$ (chromatic number $r$), the maximum edge count in $k$-colored graphs avoiding rainbow $H$ is determined by two basic constructions (cliques and Turán graphs), with sharp phase transitions and nontrivial behaviour for non-critical and bipartite graphs (Chakraborti et al., 2022).

7. Algorithmic and Hypergraph Generalizations

Algorithmic consequences include certifying algorithms for coloring in hereditary subclasses, and uniform hypergraph frameworks:

• Certifying algorithms: For $(P_5,\overline{P}_5)$-free graphs, all critical obstructions can be exhaustively listed, yielding polynomial-time coloring and certification (Dhaliwal et al., 2014).
• Hypergraph viewpoint: Covering by stable sets translates coloring to minimum hyperedge covers in hereditary hypergraphs. Criticality in coloring generalizes to other combinatorial covering problems (clique cover, dichromatic number, matroid circuits) with min-max theorems and matching-based proofs (Sebő, 2019).

8. Fractional and Circular Color-Criticality

Fractional chromatic number extends classic notions; certain subgraphs (e.g., $Q(n,k)$ in $\mathrm{SG}(n,k)$) become vertex-critical for the fractional chromatic number, linking to circular complete graphs and circular coloring theory (Gujgiczer et al., 2022).

• Deletion of a vertex or a cycle-edge in $Q(n,k)$ strictly reduces $\chi_f$, characterizing the loci of fractional criticality within Schrijver graphs.

---

In sum, color-critical graphs form a cornerstone of structural graph theory, linking algebraic extremal properties, chromatic phenomena, supersaturation, and algorithmic tractability. Their study via extremal, spectral, combinatorial, and hypergraph methods continues to drive advances in graph coloring, forbidden subgraph problems, and combinatorial optimization.