Deterministic Distributed Vertex Coloring: Simpler, Faster, and without Network Decomposition

@article{Ghaffari2022DeterministicDV,
  title={Deterministic Distributed Vertex Coloring: Simpler, Faster, and without Network Decomposition},
  author={Mohsen Ghaffari and Fabian Kuhn},
  journal={2021 IEEE 62nd Annual Symposium on Foundations of Computer Science (FOCS)},
  year={2022},
  pages={1009-1020}
}
  • M. Ghaffari, F. Kuhn
  • Published 9 November 2020
  • Computer Science
  • 2021 IEEE 62nd Annual Symposium on Foundations of Computer Science (FOCS)
We present a simple deterministic distributed algorithm that computes a ($\Delta+1$)-vertex coloring in $O(\text{log}^{2}\Delta. \text{log}\ n)$ rounds. The algorithm can be implemented with $O(\text{log}\ n)$-bit messages. The algorithm can also be extended to the more general ($degree+1$)-list coloring problem. Obtaining a polylogarithmic-time deterministic algorithm for ($\Delta +1$)-vertex coloring had remained a central open question in the area of distributed graph algorithms since the… 

Distributed Graph Coloring Made Easy

TLDR
A deterministic CONGEST algorithm to compute an O(kΔ)-vertex coloring in O( Δ/k)+łog^* n rounds, where Δ is the maximum degree of the network graph and 1łeq kłeq O(Δ) can be freely chosen.

Locally-iterative $(\Delta+1)$-Coloring in Sublinear (in $\Delta$) Rounds

TLDR
This paper gives the first locally-iterative (∆ + 1) -coloring algorithm with sublinear-in- ∆ running time, and answers the main open question raised in a recent breakthrough.

Distributed $\Delta$-Coloring Plays Hide-and-Seek

TLDR
Lower bounds as a function of ∆ are proved for a large class of distributed symmetry breaking problems, which can all be solved by a simple sequential greedy algorithm.

Near-optimal distributed degree+1 coloring

TLDR
A randomized distributed algorithm for D1LC that is optimal under plausible assumptions about the deterministic complexity of the problem is given, matching the best bound known for (Δ+1)-coloring.

Distributed Edge Coloring in Time Polylogarithmic in $\Delta$

TLDR
It is shown that a (2∆ − 1)-edge coloring can be computed in time poly log ∆+ O (log ∗ n ) in the LOCAL model, which improves a result of Balliu, Kuhn, and Olivetti [PODC ’20], who gave an algorithm with a quasi-polylogarithmic dependency on ∆.

Distributed Edge Coloring in Time Polylogarithmic in Δ

TLDR
It is shown that a (2Δ - 1)-edge coloring can be computed in time poly log Δ + O(log* n) in the LOCAL model, which improves a result of Balliu, Kuhn, and Olivetti [PODC '20], who gave an algorithm with a quasi-polylogarithmic dependency on Δ.

Deterministic graph coloring in the streaming model

TLDR
It is proved that there is no deterministic single-pass semi-streaming algorithm that given a graph G with maximum degree Δ, can output a proper coloring of G using any number of colors which is sub-exponential in Δ.

Ultrafast Distributed Coloring of High Degree Graphs

TLDR
A new randomized distributed algorithm that can color all n-node graphs of maximum degree ∆ ≥ log n in O(log∗ n) rounds and shows that the randomized complexity of ∆ + 1-list coloring in Congest depends inherently on the deterministic complexity of related coloring problems.

Fast Distributed Vertex Splitting with Applications

TLDR
A randomized poly log log n -round CONGEST algorithm for (1 + ε )∆ -edge coloring n -node graphs of sufficiently large constant maximum degree ∆, for any ε > 0 .

Improved Deterministic (Δ+1) Coloring in Low-Space MPC

TLDR
The Chang-Li-Pettie algorithm runs in T_local =poly(loglog n) rounds, which sets the state-of-the-art randomized round complexity for the problem in the local model, and employs a combination of tools, notably pseudorandom generators (PRG) and bounded-independence hash functions.

References

SHOWING 1-10 OF 40 REFERENCES

Faster Deterministic Distributed Coloring Through Recursive List Coloring

  • F. Kuhn
  • Computer Science, Mathematics
    SODA
  • 2020
TLDR
An improved deterministic $2^{O(\sqrt{\log\Delta})}\cdot\log^3 n$-round algorithm for $\Delta$-coloring non-complete graphs with maximum degree $\Delta\geq 3$.

Improved Deterministic Network Decomposition

TLDR
A modified version of the CONGEST network decomposition algorithm is presented, constructing a decomposition whose quality does not depend on the identifiers, and thus improves the randomized round complexity for various problems.

Local Conflict Coloring Revisited: Linial for Lists

TLDR
The state-of-the-art truly local $(deg+1)$-list coloring algorithm from Barenboim et al. [PODC'18] is improved by slightly reducing the runtime to $O(\sqrt{\Delta\log\Delta})+\log^* n$ and significantly reducing the message size (from huge to roughly $\Delta$).

Efficient Deterministic Distributed Coloring with Small Bandwidth

We show that the (degree + 1)-list coloring problem can be solved deterministically in O(D · log n · log2 Δ) rounds in the CONGEST model, where D is the diameter of the graph, n the number of nodes,

Efficient randomized distributed coloring in CONGEST

TLDR
This work presents a new randomized distributed vertex coloring algorithm for the standard CONGEST model, where the network is modeled as an n-node graph G, and where the nodes of G operate in synchronous communication rounds in which they can exchange O(logn)-bit messages over all the edges of G.

Deterministic Distributed Edge-Coloring via Hypergraph Maximal Matching

TLDR
A polylogarithmic-time deterministic distributed maximal independent set (MIS) algorithm for graphs with bounded neighborhood independence is obtained, hence answering Open Problem 5 of Barenboim and Elkins book.

Improved Distributed Approximation to Maximum Independent Set

TLDR
Improved results for approximating Maximum Independent Set ($\MaxIS$) in the standard LOCAL and CONGEST models of distributed computing and implies that finding an $O(\Delta)-approximation to $\MaxIS$.

Improved distributed algorithms for coloring and network decomposition problems

TLDR
It is shown that A-coloring G is reducible in 0(log3 n/log A) time to (A+ I)-vertex coloring G in a distributed model, which leads to fast distributed algorithms, and a linear–processor NC algorithm, for Acoloring.

Deterministic distributed edge-coloring with fewer colors

TLDR
This work presents a deterministic distributed algorithm, in the LOCAL model, that computes a (1+o(1))Δ-edge-coloring in polylogarithmic-time, so long as the maximum degree Δ=Ω(logn), which are the first deterministic algorithms to go below the natural barrier of 2Δ−1 colors.

Network decomposition and locality in distributed computation

The authors introduce a concept of network decomposition, a partitioning of an arbitrary graph into small-diameter connected components, such that the graph created by contracting each component into