Three-coloring triangle-free planar graphs in linear time (SODA 09’)

Download Presentation

Three-coloring triangle-free planar graphs in linear time (SODA 09’)

Loading in 2 Seconds...

- 79 Views
- Uploaded on
- Presentation posted in: General

Three-coloring triangle-free planar graphs in linear time (SODA 09’)

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.

- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Three-coloring triangle-free planar graphs in linear time (SODA 09’)

Zdenek Dvorak, Ken-ichi Kawarabayashi, Robin Thomasz

Presented By Chen Yang

- Planar Graph
- no subgraph that is isomorphic to or is a subdivision of K5 or K3, 3

- Triangle-free graph
- no three vertices forming a triangle of edges

- Every triangle-free planar graph can be colored with only 3 colors.
- Not every non-planar triangle-free graph is 3-colorable.
- Cannot be generalized to all planar K4-free graphs.

- Thomassen presented an quadratic time algorithm.
- J. Gimbel and C. Thomassen, Coloring graphs with fixed genus and girth, Trans. Amer. Math. Soc., 349 (1997), pp. 4555--4564.

- Kowalik and Kurowski achieved an O (nlogn) time by providing a sophisticated structure.
- Lukasz Kowalik: Fast 3-Coloring Triangle-Free Planar Graphs. ESA 2004: 436-447
- Lukasz Kowalik, Maciej Kurowski: Oracles for bounded-length shortest paths in planar graphs. ACM Transactions on Algorithms 2(3): 335-363 (2006)

- This paper presents a linear time algorithm for three coloring triangle-free planar graph.
- It avoids any complex data structure and easy to implement.

- Exhibit five reducible configurations, called “multigram”
- Show that every triangle-free planar graph G contains one of those reducible configurations.
- Look for a reducible configuration in G
- Modify G to a smaller graph G’, that is r-reduction
- State every 3-coloring of G’ can be converted into a 3-coloring of G in constant time.
- Apply algorithm recursively to G’ to get a 3-coloring

- Tetragram
- A sequence (v1, v2, v3, v4) of vertices which form a facial cycle in G in the order listed.

- Hexagram
- A sequence, (v1, v2, v3…, v6) is defined similarly.

- Pentagram
- A sequence (v1, v2, v3, v4, v5) of vertices which form a facial cycle in G in the order listed
- v1, v2, v3, v4 all have degree exactly 3

- Decagram:
- A pentagram (v1,v2..v5) is called a decagram if v5 has degree exactly three.

- Monogram:
- A sequence (v) comprised of a vertex v belonging to V(G) of degree at most two.

- Let G be a connected triangle-free plane graph and let f0 be the unbounded face f0 of G. Assume that the boundary of f0 is a cycle C of length at most six, and that every vertex of G not on C has degree at least three. If G doesn’t equal C, then G has either a tetragram or a pentagram (v1, v2, v3, v4, v5) such that v1, v2, v3, v4 are not in V(C).

- Tetragram: it is safe if every path in G of length at most three with ends v1 and v3 if a subgraph of the cycle v1v2...vk.
- Hexagram: the safety definition is the same with Tetragram.

- For i=1,2,3,4, let xi be the neighbor of vi distinct from vi-1 and vi+1(where v0 = v5 ). Then xi that are not in {v1…v5} because G is triangle-free.
- The vertices x1, x2, x3, v4 are pairwise distinct and pairwise non-adjacent.
- No path in G \ {v1, v2, v3, v4} of length at most three from x2 to x5.
- Every path in G\{v1,v2,v3,v4} of length at most three from x3 to x4 has length exactly two, and its completion via the path x3v3v4x4 results in a facial cycle of length five in G. in particular, there is at most one such path.

- Every triangle-free plane graph G of minimum degree at least three has a safe tetragram, pentagram or hexagram.

- Lemma 2.1 makes sure that the connected triangle-free plane graph has either a tetragram, or a pentagram.
- Define what is safe tetra, penta and hexagram.
- Lemma 2.2 states that every triangle-free plane graph G of minimum degree at least 3 has a safe tetra-, penta- or hexagram.
- Proof if G contains safe tetra-, penta- or hexagram, it will be 3-colorable.
- If k = 4 or k = 6, then apply induction to the graph obtained from G by identifying v1 and v3. It follows from the definition of safety that the new graph has no triangles, and clearly every 3-coloring of the new graph extends to a 3-coloring of G.

- Remove an edge.
- Add an edge, assuming that the edges preceding and following it in the facial walks are specified.
- Remove an isolated vertex.
- Determine the degree of a vertex v if deg(v) ≤ D or prove that deg(v) > D
- Check whether two vertices u and v such that min (deg(u), dev(v)) ≤ D are adjacent.
- Check whether the distance between two vertices u and v such that max(deg(u), deg(v)) ≤ D is at most two.
- Given an edge e of a face f, either output all vertices incident with the component of the boundary of f that contains e, or establish that the length of f is strictly larger than six.
- Output the subgraph consisting of vertices which is reachable from a vertex v0 through a path v0, v1…. vt of length t ≤ D, such that deg (vi) ≤ D for 0≤i≤t.

- Two essential concepts were introduced, C-admissible and C-forbidden under these assumptions:
- G: plane graph
- k ∈ {1, 4, 5, 6}
- r = (v1, v2... vk) be a mono, tera, penta or hexgram in G.
- C is a subgraph of G.
- A vertex of G is big if it has degree at least 48, and small otherwise.

- A vertex v belonging to V(G) is C-admissible if it is small and does not belong to C; otherwise it is C-forbidden.

- Suppose G is a triangle-free plane graph, r is a safe multigram in G, and let G’ be the r-reduction of G. Then G’ is triangle-free, and every 3-coloring of G’ can be converted to a 3-coloring of G in constant time.

- If r is a monogram, it is always safe
- it is C-secure if v1 doesn’t belong V(C).

- If r is a tetragram,
- itis safe if the only paths in G of length at most three with ends v1 and v3 are subgraphs of facial cycle v1v2v3v4.
- if r is safe, v1 is C-admissible and has degree exactly three, r is C-secure. Letting x denote the neighbor of v1 other than v2 and v4, the vertex x is C-admissible and either v3 is C-admissible, or every neighbor of x is C-admissible.

- If r is a decagram
- For i = 1, 2, 3, 4 let xi be the neighbor of vi other than vi-1 or vi+1, v0 means v5.
- r is safe if x1, x3 are distinct, non-adjacent and there is no path of length two between them. Also r is C-secure if it is safe and the vertices v1, v2 … v5, x1, x3 are all C-admissible.

- If r is a pentagram,
- r is C-secure if it is safe, the vertices v1, v2 … v5, x1, x2, x3, x4 are all C-admissible, either v5 or x2 has not C-forbidden neighbor and either x3 or x4 has no C-forbidden neighbor.

- If r is a hexagram
- r is safe if every path of length at most three in G between v1 and v3 is the path v1v2v3.
- r is C-secure if v1, v3, v6 are C-admissible, v1 has degree exactly three, and the neighbor of v1 other than v2 or v6 is C-admissible.

- G’ is obtained from G be identifying the vertices v1 and v3 and deleting one of the parallel edges

- Algorithm 4.1. There is an algorithm with following specifications:
- Input: A triangle-free planar graph.
- Output: A proper 3-coloring of G.
- Running time: O(|V(G)|).

- Assume that G is a plane graph. The algorithm is recursive.
- maintain a list L that will include the pivots of all secure multigrams in G.
- Initializing the list L to consist of all vertices of G of degree at most three.
- Iteration starts.
- Remove a vertex from L.
- If v∉V(G), go to the next iteration
- Otherwise, check if G has a secure multigram with pivot v. (this can be performed in constant time)
- If no such multigram exist, go to the next iteration.
- Otherwise, let r e one such multigram, and let G’ be the r-reduction of G. G’ is triangle-free (lemma 4.1) and can be constructed in constant time by adding and deleting bounded number of edges.
- Apply the algorithm recursively to G’ and converts the resulting 3-coloring of G’ to one of G.

- Given a triangle-free plane graph G and a vertex v∈V(G), it can be decided in constant time whether G has a secure multigram with pivot v.
- The vertex v1 will be called the pivot of the multigram (v1; v2; ….; vk).

- Input: A triangle-free plane graph G, a facial cycle C in G of length at most five and a proper 3-coloring ∅ of C.
- Output; A proper 3-coloring of G whose restriction to V(C) is equal to ∅.
- Running time: O(|V(G)|).

- Same as Algorithm 1, but use “C-secure” rather than “secure”

- This lemma supports the correctness of Algo 2.
- Let G be a connected triangle-free plane graph and let f0 be its outer face. Assume that f0 is bounded by a cycle of length at most six and that |V(G) – V(C)| ≥ 2. Then either G has a C-secure multigram, or C has length exactly six and includes at least two distinct non-adjacent C-appendices.

- Zdenek Dvorak and Robin Thomas
- Three-coloring triangle-free graphs on surfaces I. Extending a coloring to a disk with one triangle
- CoRR abs/1010.2472

- It proved that a 3-coloring of C does not extend to a 3-coloring of G if and only if C has length exactly six.

- Three-coloring triangle-free graphs on surfaces I. Extending a coloring to a disk with one triangle