--- title: Generalized Color-Critical Graphs url: https://www.emergentmind.com/topics/generalized-color-critical-graphs type: topic --- # Generalized Color-Critical Graphs Generalized color-critical graphs are minimal-obstruction objects for coloring theories that extend ordinary chromatic criticality. In the classical case, a graph is \(k\)-critical if \(\chi(G)=k\) and every proper subgraph is \((k-1)\)-colorable; modern work extends the same minimality paradigm to hereditary graph classes, list and DP-coloring, property-based colorings, star coloring, hereditary hypergraphs, and several extremal settings in which color-criticality determines the exact extremal configuration [1403.8027][1804.06338][1908.00282][2408.04538][2305.17956]. ## 1. Foundational notions and scope The subject now comprises several distinct but closely related criticality notions. Some are vertex-minimal or subgraph-minimal non-colorability notions; others are edge-criticality notions, where deleting one edge lowers the relevant chromatic parameter; still others are obstruction notions for list, DP, or property-based colorings. The shared theme is that a critical object is a smallest witness that a coloring threshold cannot be lowered. | Notion | Defining condition | Ambient framework | |---|---|---| | \(k\)-critical graph | \(\chi(G)=k\) and every proper subgraph is \((k-1)\)-colorable | Ordinary coloring | | Color-critical graph | \(G\) contains an edge \(e\) such that \(\chi(G-e)<\chi(G)\) | Extremal graph theory | | \((\mathcal P,L)\)-critical hypergraph | \(H\) is not \((\mathcal P,L)\)-colorable, but \(H-v\) is for all \(v\) | Property colorings of hypergraphs | | \((\mathcal P,\mathrm{DP})\)-critical graph | Every proper induced subgraph \(G'\) satisfies \(\chi_{\rm DP}(G':\mathcal P)<\chi_{\rm DP}(G:\mathcal P)\) | Generalized DP-coloring | | Strong \(k\)-chromatic-choosable graph | \(\chi(G)=k\) and every bad \((k-1)\)-assignment is constant | List coloring | | Robustly \(k\)-critical graph | \(G\) is \(k\)-critical and every bad \((k-1)\)-fold cover is canonical | DP-coloring | | \(k\)-critical for star coloring | \(\chi_s(G)=k\) and \(\chi_s(G-e)<\chi_s(G)\) for every \(e\in E(G)\) | Star coloring | In hereditary graph classes, critical graphs are naturally interpreted as minimal forbidden induced subgraphs for a coloring bound. In hereditary hypergraphs, Sebő replaces the chromatic number by the partition parameter \(p(H)\), the minimum number of hyperedges covering the vertex set, and defines criticality by the condition \(p(H-v)=p(H)-1\) for every vertex \(v\) [1910.11302]. In smooth hypergraph-property colorings, Schweser studies \((\mathcal P,L)\)-critical hypergraphs, where color classes are required to induce members of a hereditary property \(\mathcal P\), and \(H-v\) is colorable for every vertex although \(H\) itself is not [1804.06338]. In generalized DP-coloring, Kostochka, Schweser, and Stiebitz define \((\mathcal P,\mathrm{DP})\)-critical graphs and the more local notion of a \(\mathcal P\)-critical cover, which functions as the natural obstruction object for correspondence coloring with property-constrained color classes [1908.00282]. This diversity suggests that “generalized color-critical graphs” are best understood not as a single class but as a family of minimal-obstruction theories sharing a common structural agenda: degree lower bounds, low-vertex decompositions, exact extremal constructions, and finite or recursive obstruction sets. ## 2. Structural principles and decomposition mechanisms Several recurring principles organize generalized color-critical graph theory. In the class of \((P_5,\overline{P}_5)\)-free graphs, critical graphs are connected, have no comparable vertices, and, if they are not cliques, have no clique cutset. Modules inherit criticality, joins preserve criticality exactly, and replacing a module by a clique of the same chromatic number preserves \(k\)-criticality [1403.8027]. These statements are not merely local lemmas: they allow criticality to descend through decomposition and then be rebuilt from smaller critical pieces. For reliable graph properties \(\mathcal P\), generalized DP-criticality has a parallel low-vertex theory. Writing \(r=d(\mathcal P)\), a \(\mathcal P\)-critical cover forces \(d_G(v)\ge r|X_v|\) for every vertex. If equality holds, the vertex is low, and each block of the low-vertex subgraph is highly restricted: it is either a brick, or \(tB'\) with \(B'\in CR(\mathcal P)\) \(r\)-regular, or \(tB'\) with \(B'\in \mathcal P\) and \(\Delta(B')\le r\) [1908.00282]. Schweser proves the analogous hypergraph statement for \((\mathcal P,L)\)-critical hypergraphs: if \(F\) is the low-vertex hypergraph, then each block of \(F\) is a brick, or belongs to \(\mathcal F(\mathcal P)\) and is \(r\)-regular, or already lies in \(\mathcal P\) with maximum degree at most \(r\) [1804.06338]. Sebő’s hereditary-hypergraph reformulation reveals the same Gallai-type mechanism from a different angle. In a connected, hereditary, critical hypergraph, \(p(H)\le (n+1)/2\); if equality holds, minimum covers consist of one singleton and edges only, and the graph of 2-element hyperedges is factor-critical [1910.11302]. This places matching theory at the core of generalized criticality: in the equality regime, larger hyperedges cease to matter, and the obstruction is controlled by a factor-critical graph. A plausible implication is that generalized criticality is structurally rigid precisely when the coloring model admits a controllable “low part,” whether that part is a module decomposition, a brick decomposition, or a factor-critical 2-section. ## 3. Hereditary classes and finite obstruction theories A particularly sharp hereditary example is the class of \((P_5,\overline{P}_5)\)-free graphs. Let \(\mathcal C_k\) denote the \(k\)-critical members of this class. The central structure theorem states that \(G\in \mathcal C_k\) if and only if either \(G\) is the join of graphs in \(\mathcal C_{k_1}\) and \(\mathcal C_{k_2}\) with \(k_1+k_2=k\), or \(G\) is a buoy with bags \(B_1,\dots,B_5\) such that each \(B_i\in \mathcal C_{k_i}\), the cyclic constraints \(k_i+k_{i+1}\le k-1\) hold, and \(\sum_{i=1}^5 k_i=2k-1\) [1403.8027]. The buoy, obtained from a \(C_5\) by substitution, is the class-specific template that plays the role of an irreducible non-perfect obstruction. This recursive grammar yields a finiteness theorem: for every fixed \(k\), \(\mathcal C_k\) is finite. The paper records the initial values \[ \mathcal C_1=\{K_1\},\qquad \mathcal C_2=\{K_2\},\qquad \mathcal C_3=\{K_3,C_5\}, \] and the counts \[ |\mathcal C_4|=3,\qquad |\mathcal C_5|=9,\qquad |\mathcal C_6|=31,\qquad |\mathcal C_7|=185,\qquad |\mathcal C_8|=1487 \] [1403.8027]. The contrast with \(P_5\)-free graphs alone is decisive: there are infinitely many \(k\)-critical \(P_5\)-free graphs for every \(k\ge 5\), while forbidding both \(P_5\) and \(\overline P_5\) restores finiteness for every fixed \(k\). The same finiteness result has an algorithmic consequence. Because every non-\(k\)-colorable graph in the class contains an induced \((k+1)\)-critical subgraph, and because the obstruction family is finite for fixed \(k\), there is a certifying algorithm for fixed-\(k\) coloring of \((P_5,\overline P_5)\)-free graphs [1403.8027]. This is an archetypal finite-obstruction theorem: a hereditary coloring problem becomes a recursively generated obstruction theory with an effective NO-certificate. ## 4. Alternative coloring models and generalized criticality Property-based coloring extends ordinary coloring by replacing “independent color classes” with color classes inducing members of a fixed hereditary property. In Schweser’s hypergraph framework, if \(\mathcal P\) is non-trivial, hereditary, and additive with \(d(\mathcal P)=r\), then \[ \chi^\ell(H:\mathcal P)\le \frac{\Delta(H)}{r}+1, \] with equality only in explicitly listed exceptional cases involving complete graphs, odd cycles in the \(r=1\) regime, or \(r\)-regular members of \(\mathcal F(\mathcal P)\) [1804.06338]. The generalized DP analogue has the same scaling parameter: if \(\mathcal P\) is reliable with \(d(\mathcal P)=r\), then for connected simple graphs \[ \chi_{\rm DP}(G:\mathcal P)\le \left\lceil\frac{\Delta(G)}{r}\right\rceil \] except for complete graphs \(K_{kr+1}\), \(r\)-regular graphs in \(CR(\mathcal P)\), and cycles when \(\mathcal P=\mathcal O\) [1908.00282]. In both settings, \(d(\mathcal P)\) plays the role that \(1\) plays in classical critical graph theory. List and DP variants introduce more refined notions of minimal obstruction. A graph is strong \(k\)-chromatic-choosable if \(\chi(G)=k\) and every bad \((k-1)\)-assignment is constant; such graphs are chromatic-choosable and vertex-critical, and the join \(G\vee K_p\) is strong \((k+p)\)-chromatic-choosable [1805.02147]. The DP-strengthening is robust criticality: \(G\) is robustly \(k\)-critical if \(G\) is \(k\)-critical and every bad \((k-1)\)-fold cover is canonical. If \(G\) is critical with \(m\) edges, then \(G\vee K_t\) is strongly critical for all \(t\ge 3m\), and robustly critical for all \(t\ge 100m^3\) [2408.04538]. This shows that generalized criticality can be forced by a large clique join even in the correspondence-coloring regime. Star coloring gives a different generalization, because the obstruction is no longer merely adjacency but the existence of a 2-colored \(P_4\). Here \(G\) is \(k\)-critical if \(\chi_s(G)=k\) and \(\chi_s(G-e)<\chi_s(G)\) for every edge. The 3-critical graphs are exactly \(K_3\) and \(P_4\). For non-complete graphs on \(n\ge 5\), \((n-1)\)-criticality is characterized by \((I_3,2K_2)\)-freeness together with the property that every edge deletion creates an induced \(I_3\) or \(2K_2\). Under the paper’s stated hypothesis that the graph contains \(I_3\) or \(2K_2\), \((n-2)\)-criticality is characterized by \((I_4,2K_2+K_1,P_3+P_2)\)-freeness together with the property that every edge deletion creates one of those induced subgraphs [2305.17956]. A distinct topological-combinatorial example comes from Schrijver graphs. For the 4-chromatic family, \[ \mathrm{SG}(2k+2,k)\cong D'_{k+1,\,2k+2}, \] and an edge is color-critical if and only if it is interlacing [1912.03724]. This shows that generalized criticality can also be encoded by cyclic order and surface-embedding structure, not only by density or hereditary exclusion. ## 5. Extremal sparsity, degree bounds, and counting The minimum-edge problem remains central. For ordinary critical graphs, if \(k\ge 4\) and \(G\) is \(k\)-critical, then \[ |E(G)|\ge F(k,|V(G)|),\qquad F(k,n)=\left\lceil \frac{(k+1)(k-2)n-k(k-3)}{2(k-1)}\right\rceil. \] This is sharp for every \(n\equiv 1 \pmod{k-1}\), sharp for \(k=4\) and every \(n\ge 6\), yields \[ \phi_k=\lim_{n\to\infty}\frac{f_k(n)}{n}=\frac{k}{2}-\frac{1}{k-1}, \] and shows that Ore’s conjectured recurrence can fail for at most \(\frac{k^3}{12}-\frac{k^2}{8}\) values of \(n\) [1209.1050]. The same result also produces a polynomial-time algorithm for \((k-1)\)-coloring every graph \(G\) such that \(|E(G[W])|kn+k-2. \] Equality can occur only for the classical family \(Dir_k\), and the same consequence holds for list-critical graphs [1609.09122]. This is structurally notable because DP-coloring permits even cycles to behave differently from list coloring, yet the sharp Dirac equality class remains unchanged. The generalized DP and defective DP settings produce comparable quantitative constraints. For reliable \(\mathcal P\) with \(r=d(\mathcal P)\), a simple \((\mathcal P,\mathrm{DP})\)-critical graph with \(\chi_{\rm DP}(G:\mathcal P)=k+1\) satisfies \(\delta(G)\ge rk\) and, unless \(G=K_{kr+1}\), \[ 2|E(G)|\ge \left(kr+\frac{kr-2}{(kr+1)^2-3}\right)|G|+\frac{2kr}{(kr+1)^2-3}. \] For defective DP-coloring, if \(i=1,2\) and \(j\ge 2i\), then every DP-\((i,j)\)-critical simple graph on \(n\) vertices satisfies \[ g_{DP}(i,j,n)\ge \frac{(2i+1)n+j-i+1}{i+1}, \] and this bound is sharp for infinitely many \(n\) [1908.00282][2306.14295]. Star-coloring criticality has its own extremal profile. For \((n-1)\)-critical graphs, \[ \frac{n^2-n-n\sqrt n}{2}< m\le \frac{(n-1)(n-2)}{2}, \] and the upper bound is attained by the horn graph \(H_n\). For \((n-2)\)-critical graphs, \[ \frac{n(n-3)}{6}\le m\le \frac{n(n-3)}{2} \] [2305.17956]. These inequalities are qualitatively different from ordinary critical graph bounds, reflecting the fact that star coloring penalizes 2-colored \(P_4\)’s rather than only monochromatic edges. ## 6. Rainbow, Ramsey, and spectral manifestations Color-criticality also governs exact extremal problems in nonstandard host models. For generalized books \[ B_{r,k}=K_r\vee \overline{K_k}, \] with \(r\ge 3\), \(k\ge 1\), and sufficiently large \(n=qr+p\), the non-\(r\)-partite extremal number is \[ \ex_{r+1}(n,B_{r,k}) =\left(1-\frac1r\right)\frac{n^2}{2}-\frac nr+\frac{p(p+2)}{2r}-\frac p2+1, \] and the extremal family is exactly \(\mathcal G_1[n],\mathcal G_2[n],\mathcal G_3[n]\) according to \(p=n\bmod r\). The value is the same as the known value of \(\ex_{r+1}(n,K_{r+1})\) for large \(n\) [2508.07533]. In the rainbow Turán problem, if \(H\) is \(4\)-color-critical with \(h=e(H)\), then for large \(n\), \[ \ex_k(n,H)= (h-1)\binom{n}{2}\quad \text{for } h\le k<\frac32(h-1), \] and \[ \ex_k(n,H)=k\,t_3(n)\quad \text{for } k\ge \frac32(h-1), \] while for almost all \(r\)-color-critical graphs with \(r\ge 5\) the analogous threshold is \(\frac{r-1}{r-2}(h-1)\) [2204.02575]. The multicolor Turán problem exhibits the same template. If \(r\ge 5\), \(H\) is \(r\)-color-critical with \(h=e(H)\), \(n\) is sufficiently large, and \[ k\ge 2^{\,r-1}(h-1), \] then \[ ex_k(n,H)=k\,t_{r-1}(n), \] uniquely attained by \(k\) identical copies of \(T_{r-1}(n)\) [2407.14905]. In Ramsey theory, if \(G\) is edge-critical with \(\chi(G)=k+1\), \(t\ge 2\), and \(n\) is sufficiently large, then \[ R(G,K_1+nK_t)=knt+1,\qquad r_{*}(G,K_1+nK_t)=(k-1)nt+t \] [2308.10546]. This extends the exact formulas previously known for complete graphs and odd cycles to arbitrary edge-critical red graphs. Spectral extremal theory reveals the same color-critical boundary. If \(F\) is color-critical with \(\chi(F)=r+1\ge 4\), then for sufficiently large \(m\), every \(m\)-edge \(F\)-free graph satisfies \[ \lambda(G)\le \sqrt{\left(1-\frac1r\right)2m}, \] with equality if and only if \(G\) is a regular complete \(r\)-partite graph [2511.15431]. Under Nikiforov’s condition \[ \rho(G)\ge \sqrt{\left(1-\frac1r\right)2m}, \] a sufficiently large \(m\)-edge graph contains at least \((\gamma_F-o(1))m^{(|F|-2)/2}\) copies of any color-critical \(F\) with \(\chi(F)=r+1\ge 4\), and the leading constant \(\gamma_F\) is optimal [2603.14964]. In the signless Laplacian setting, if \(F\) is color-critical with \(\chi(F)=r+1>4\), then for sufficiently large \(n\), \[ q(G)\le q(T_{n,r}), \] with equality if and only if \(G=T_{n,r}\); as a consequence, \[ \sum_{v\in V(G)} d(v)^2 \le 2\left(1-\frac1r\right)mn, \] with equality if and only if \(G\) is a regular Turán graph [2504.07852]. Taken together, these results suggest that generalized color-critical graphs mark the exact point at which a forbidden configuration begins to enforce Turán-type multipartite structure, sharp density lower bounds, recursive obstruction theories, or canonical covers. Across ordinary coloring, hereditary restrictions, DP-coloring, star coloring, hypergraph property colorings, Ramsey theory, and spectral extremal theory, the same theme persists: criticality identifies the smallest structure whose presence rigidly changes the coloring or extremal behavior. Source: https://www.emergentmind.com/topics/generalized-color-critical-graphs