The main result shows that the exact average-case complexity of this fundamental problem is $\Theta(nk)$ for every $k \leq n^{c'}$ and some $c'\in (0, 1)$, and reveals the average sublinear nature of $k$-colorability: the average-case complexity is linear in $n$, and thus sublinear in the size of the input.
Abstract
Graph $k$-coloring is one of the classic NP-complete problems. Previous work has studied its average time complexity, defined to be the average runtime of computing a $k$-coloring over the set of all $k$-colorable graphs on $n$ vertices. A highly influential result of Dyer-Frieze from 1989 gave an algorithm with $O(n^2)$ average runtime for constant $k$. This quadratic runtime appeared natural (and possibly even optimal) since almost all $k$-colorable graphs have $\Theta(n^2)$ edges, so one needs at least this time in order to read the (entire) input. However, this was later improved by Ku\v{c}era in 1995 to average runtime $O(n^2/k)$ for every $k \leq n^{c}$ where $c \in (0, 1)$. Nevertheless, in the most interesting case of $k = O(1)$, the best-known bound remained quadratic in $n$. The true average complexity of the $k$-coloring problem has remained elusive for the last three decades. We break the longstanding quadratic barrier. Our main result in this paper shows that the exact average-case complexity of this fundamental problem is $\Theta(nk)$ for every $k \leq n^{c'}$ and some $c'\in (0, 1)$. For $k = O(1)$, this reveals the average sublinear nature of $k$-colorability: the average-case complexity is linear in $n$, and thus sublinear in the size of the input. We further show that our $\Theta(nk)$ average runtime is optimal, since a simple bound proves that every algorithm that correctly $k$-colors all $k$-colorable graphs requires $\Omega(n k)$ average runtime. Our proofs draw on ideas from sublinear and local algorithms and also yield a local computation algorithm (LCA) for $k$-coloring with average-case probe complexity $\text{poly}(k)$. A key new ingredient in our algorithm is a method for certifying the unique colorability of random subgraphs, using tools from the theory of graph regularity.
This work generalizes and combines tools from the $(k+2)-coloring to $k$-list-coloring reduction of [Zamir, ICALP 2021] and the hypergraph-containers based approach in [Zamir, STOC 2023] and yields an iterable reduction from $(k+1)$-list-coloring to $k$-list-coloring over fixed palettes.
We study the parameterized complexity of $k$-Coloring in $H$-free graphs, when $H$ is a linear forest (i.e., a disjoint union of paths) as an induced subgraph. We show two hardness results: * $k$-Coloring is W[1]-hard in $2P_2$-free graphs when parameterized by $k$. * $3$-Coloring is W[1]-hard in $P_t$-free graphs when parameterized by $t$. Moreover, assuming the ETH, these problems admit no algorithms solving $n$-vertex instances in time $f(k) \cdot n^{o(k)}$ and $f(t) \cdot n^{o(t/\log t)}$, respectively, for any computable function $f$. The first result resolves in a strong form a long-standing open problem, originally posed by Ho\`ang, Kami\'nski, Lozin, Sawada, and Shu [Algorithmica, 2010]. The second result answers a question of Golovach, Johnson, Paulusma, and Song [Journal of Graph Theory, 2017].
We show that for all $k$, there exists $\varepsilon_k>0$ such that graph $k$-coloring can be solved by a randomized algorithm with one-sided error in time $O((2-\varepsilon_k)^n)$. Prior to this work and independent concurrent work of Zamir [arXiv, 2026], exponential improvements over the $2^n \cdot \mathrm{poly}(n)$-time algorithm of Bj\"orklund, Husfeldt, and Koivisto [SIAM Journal on Computing, 2009] were only known for $k \le 6$.
For an integer $k\geq2$, let $\chi_k'(G)$ denote the minimum number of colors in an edge-coloring of a graph $G$ such that every nonzero degree in each color subgraph is congruent to $1\pmod{k}$. A graph is a $0_k$-graph if every vertex degree is divisible by $k$. We disprove a conjecture of Berthe et al.\ (On modular edge colorings of graphs, SIAM J. Discrete Math. 40 (2026) 897--904), which states that $\chi_k'(G)\leq k+o(k)$ for every $0_k$-graph $G$. We prove a lower bound for $0_k$-graphs with degree set $\{k,2k\}$ and a specified vertex partition. With a suitable choice of the part sizes, if the number of edges inside one part is $o(k^2)$, then $\chi_k'(G)\geq(4-2\sqrt2+o(1))k$. This gives connected bipartite and connected nonbipartite counterexamples. In particular, the same examples also disprove the earlier conjecture of Botler, Colucci, and Kohayakawa (The mod $k$ chromatic index of graphs is $O(k)$, J. Graph Theory 102 (2023) 197--200), which states that $\chi_k'(G)\leq k+C$ for some absolute constant $C$.
The first $poly(\Delta,\log n)-round algorithm for $(\Delta + 1)$-edge coloring in the CONGEST model is presented and the $n$-dependency of its runtime, $\tilde{O}(\log^5 n)$, matches the best published dependency in the LOCAL model.
Sebastian Brandt, Ananth Narayanan, Alexandre Nolin· 0 citations
We give a sublinear algorithm for the planted $k$-coloring problem. Given an expander $G$ with a planted coloring, the goal is to efficiently determine the color class of a given vertex. We work in the adversarial planted coloring model of David and Feige [STOC 2016], where an adversary chooses a $d$-regular spectral $\lambda$-expander $G$ on $n$ vertices and plants a balanced $k$-coloring by partitioning the vertices into $k$ equal parts and deleting all edges within each part. This model generalizes the earlier random graph models studied by Blum and Spencer [J. Algorithms 1995] and Alon and Kahale [STOC 1994]. We give the first sublinear-time algorithm for recovering planted colorings in this model. The algorithm has preprocessing time and space $\widetilde O\left(n^{1/2+O(1/\log(d/\lambda))}\right)$, and produces a data structure that answers color queries in time $\widetilde O\left(n^{1/2+O(1/\log(d/\lambda))}\right)$, such that the resulting labeling agrees with the planted coloring on all but an $O(\sqrt{\lambda /d})$ fraction of vertices, up to a permutation of the $k$ colors. The algorithm gives sublinear-time inner product access to the bottom eigenspace of the normalized adjacency matrix, which allows us to adapt the classical spectral approach of Alon and Kahale in sublinear time.