Manu Basavaraju, L. Sunil Chandran, Mathew C. Francis, Karthik Murali
{"title":"Variants of the Gyárfás–Sumner conjecture: Oriented trees and rainbow paths","authors":"Manu Basavaraju, L. Sunil Chandran, Mathew C. Francis, Karthik Murali","doi":"10.1002/jgt.23171","DOIUrl":null,"url":null,"abstract":"<p>Given a finite family <span></span><math>\n \n <mrow>\n <mi>ℱ</mi>\n </mrow></math> of graphs, we say that a graph <span></span><math>\n \n <mrow>\n <mi>G</mi>\n </mrow></math> is “<span></span><math>\n \n <mrow>\n <mi>ℱ</mi>\n </mrow></math>-free” if <span></span><math>\n \n <mrow>\n <mi>G</mi>\n </mrow></math> does not contain any graph in <span></span><math>\n \n <mrow>\n <mi>ℱ</mi>\n </mrow></math> as a subgraph. We abbreviate <span></span><math>\n \n <mrow>\n <mi>ℱ</mi>\n </mrow></math>-free to just “<span></span><math>\n \n <mrow>\n <mi>F</mi>\n </mrow></math>-free” when <span></span><math>\n \n <mrow>\n <mi>ℱ</mi>\n \n <mo>=</mo>\n \n <mrow>\n <mo>{</mo>\n \n <mi>F</mi>\n \n <mo>}</mo>\n </mrow>\n </mrow></math>. A vertex-colored graph <span></span><math>\n \n <mrow>\n <mi>H</mi>\n </mrow></math> is called “rainbow” if no two vertices of <span></span><math>\n \n <mrow>\n <mi>H</mi>\n </mrow></math> have the same color. Given an integer <span></span><math>\n \n <mrow>\n <mi>s</mi>\n </mrow></math> and a finite family of graphs <span></span><math>\n \n <mrow>\n <mi>ℱ</mi>\n </mrow></math>, let <span></span><math>\n \n <mrow>\n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>,</mo>\n \n <mi>ℱ</mi>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n </mrow></math> denote the smallest integer such that any properly vertex-colored <span></span><math>\n \n <mrow>\n <mi>ℱ</mi>\n </mrow></math>-free graph <span></span><math>\n \n <mrow>\n <mi>G</mi>\n </mrow></math> having <span></span><math>\n \n <mrow>\n <mi>χ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mi>G</mi>\n \n <mo>)</mo>\n </mrow>\n \n <mo>≥</mo>\n \n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>,</mo>\n \n <mi>ℱ</mi>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n </mrow></math> contains an induced rainbow path on <span></span><math>\n \n <mrow>\n <mi>s</mi>\n </mrow></math> vertices. Scott and Seymour showed that <span></span><math>\n \n <mrow>\n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>,</mo>\n \n <mi>K</mi>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n </mrow></math> exists for every complete graph <span></span><math>\n \n <mrow>\n <mi>K</mi>\n </mrow></math>. A conjecture of N. R. Aravind states that <span></span><math>\n \n <mrow>\n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>,</mo>\n \n <msub>\n <mi>C</mi>\n \n <mn>3</mn>\n </msub>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mo>=</mo>\n \n <mi>s</mi>\n </mrow></math>. The upper bound on <span></span><math>\n \n <mrow>\n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>,</mo>\n \n <msub>\n <mi>C</mi>\n \n <mn>3</mn>\n </msub>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n </mrow></math> that can be obtained using the methods of Scott and Seymour setting <span></span><math>\n \n <mrow>\n <mi>K</mi>\n \n <mo>=</mo>\n \n <msub>\n <mi>C</mi>\n \n <mn>3</mn>\n </msub>\n </mrow></math> are, however, super-exponential. Gyárfás and Sárközy showed that <span></span><math>\n \n <mrow>\n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>,</mo>\n \n <mrow>\n <mo>{</mo>\n \n <mrow>\n <msub>\n <mi>C</mi>\n \n <mn>3</mn>\n </msub>\n \n <mo>,</mo>\n \n <msub>\n <mi>C</mi>\n \n <mn>4</mn>\n </msub>\n </mrow>\n \n <mo>}</mo>\n </mrow>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mo>=</mo>\n \n <mi>O</mi>\n \n <mrow>\n <mo>(</mo>\n \n <msup>\n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mn>2</mn>\n \n <mi>s</mi>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mrow>\n <mn>2</mn>\n \n <mi>s</mi>\n </mrow>\n </msup>\n \n <mo>)</mo>\n </mrow>\n </mrow></math>. For <span></span><math>\n \n <mrow>\n <mi>r</mi>\n \n <mo>≥</mo>\n \n <mn>2</mn>\n </mrow></math>, we show that <span></span><math>\n \n <mrow>\n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>,</mo>\n \n <msub>\n <mi>K</mi>\n \n <mrow>\n <mn>2</mn>\n \n <mo>,</mo>\n \n <mi>r</mi>\n </mrow>\n </msub>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mo>≤</mo>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>r</mi>\n \n <mo>−</mo>\n \n <mn>1</mn>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>−</mo>\n \n <mn>1</mn>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>−</mo>\n \n <mn>2</mn>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mo>∕</mo>\n \n <mn>2</mn>\n \n <mo>+</mo>\n \n <mi>s</mi>\n </mrow></math> and therefore, <span></span><math>\n \n <mrow>\n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mi>s</mi>\n \n <mo>,</mo>\n \n <msub>\n <mi>C</mi>\n \n <mn>4</mn>\n </msub>\n \n <mo>)</mo>\n </mrow>\n <mspace></mspace>\n \n <mo>≤</mo>\n <mspace></mspace>\n \n <mfrac>\n <mrow>\n <msup>\n <mi>s</mi>\n \n <mn>2</mn>\n </msup>\n \n <mo>−</mo>\n \n <mi>s</mi>\n \n <mo>+</mo>\n \n <mn>2</mn>\n </mrow>\n \n <mn>2</mn>\n </mfrac>\n </mrow></math>. This significantly improves Gyárfás and Sárközy's bound and also covers a bigger class of graphs. We adapt our proof to achieve much stronger upper bounds for graphs of higher girth: we prove that <span></span><math>\n \n <mrow>\n <mi>ℓ</mi>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>,</mo>\n \n <mrow>\n <mo>{</mo>\n \n <mrow>\n <msub>\n <mi>C</mi>\n \n <mn>3</mn>\n </msub>\n \n <mo>,</mo>\n \n <msub>\n <mi>C</mi>\n \n <mn>4</mn>\n </msub>\n \n <mo>,</mo>\n \n <mo>…</mo>\n \n <mo>,</mo>\n \n <msub>\n <mi>C</mi>\n \n <mrow>\n <mi>g</mi>\n \n <mo>−</mo>\n \n <mn>1</mn>\n </mrow>\n </msub>\n </mrow>\n \n <mo>}</mo>\n </mrow>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mo>≤</mo>\n \n <msup>\n <mi>s</mi>\n \n <mrow>\n <mn>1</mn>\n \n <mo>+</mo>\n \n <mfrac>\n <mn>4</mn>\n \n <mrow>\n <mi>g</mi>\n \n <mo>−</mo>\n \n <mn>4</mn>\n </mrow>\n </mfrac>\n </mrow>\n </msup>\n </mrow></math>, where <span></span><math>\n \n <mrow>\n <mi>g</mi>\n \n <mo>≥</mo>\n \n <mn>5</mn>\n </mrow></math>. Moreover, in each case, our results imply the existence of at least <span></span><math>\n \n <mrow>\n <mi>s</mi>\n \n <mo>!</mo>\n \n <mo>∕</mo>\n \n <mn>2</mn>\n </mrow></math> distinct induced rainbow paths on <span></span><math>\n \n <mrow>\n <mi>s</mi>\n </mrow></math> vertices. Along the way, we obtain some new results on an oriented variant of the Gyárfás–Sumner conjecture. For <span></span><math>\n \n <mrow>\n <mi>r</mi>\n \n <mo>≥</mo>\n \n <mn>2</mn>\n </mrow></math>, let <span></span><math>\n \n <mrow>\n <msub>\n <mi>ℬ</mi>\n \n <mi>r</mi>\n </msub>\n </mrow></math> denote the orientations of <span></span><math>\n \n <mrow>\n <msub>\n <mi>K</mi>\n \n <mrow>\n <mn>2</mn>\n \n <mo>,</mo>\n \n <mi>r</mi>\n </mrow>\n </msub>\n </mrow></math> in which one vertex has out-degree or in-degree <span></span><math>\n \n <mrow>\n <mi>r</mi>\n </mrow></math>. We show that every <span></span><math>\n \n <mrow>\n <msub>\n <mi>ℬ</mi>\n \n <mi>r</mi>\n </msub>\n </mrow></math>-free oriented graph having a chromatic number at least <span></span><math>\n \n <mrow>\n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>r</mi>\n \n <mo>−</mo>\n \n <mn>1</mn>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>−</mo>\n \n <mn>1</mn>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mrow>\n <mo>(</mo>\n \n <mrow>\n <mi>s</mi>\n \n <mo>−</mo>\n \n <mn>2</mn>\n </mrow>\n \n <mo>)</mo>\n </mrow>\n \n <mo>+</mo>\n \n <mn>2</mn>\n \n <mi>s</mi>\n \n <mo>+</mo>\n \n <mn>1</mn>\n </mrow></math> and every bikernel-perfect oriented graph with girth <span></span><math>\n \n <mrow>\n <mi>g</mi>\n \n <mo>≥</mo>\n \n <mn>5</mn>\n </mrow></math> having a chromatic number at least <span></span><math>\n \n <mrow>\n <mn>2</mn>\n \n <msup>\n <mi>s</mi>\n \n <mrow>\n <mn>1</mn>\n \n <mo>+</mo>\n \n <mfrac>\n <mn>4</mn>\n \n <mrow>\n <mi>g</mi>\n \n <mo>−</mo>\n \n <mn>4</mn>\n </mrow>\n </mfrac>\n </mrow>\n </msup>\n </mrow></math> contains every oriented tree on at most <span></span><math>\n \n <mrow>\n <mi>s</mi>\n </mrow></math> vertices as an induced subgraph.</p>","PeriodicalId":16014,"journal":{"name":"Journal of Graph Theory","volume":"108 1","pages":"136-161"},"PeriodicalIF":0.9000,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Graph Theory","FirstCategoryId":"100","ListUrlMain":"https://onlinelibrary.wiley.com/doi/10.1002/jgt.23171","RegionNum":3,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"MATHEMATICS","Score":null,"Total":0}
引用次数: 0
Abstract
Given a finite family of graphs, we say that a graph is “-free” if does not contain any graph in as a subgraph. We abbreviate -free to just “-free” when . A vertex-colored graph is called “rainbow” if no two vertices of have the same color. Given an integer and a finite family of graphs , let denote the smallest integer such that any properly vertex-colored -free graph having contains an induced rainbow path on vertices. Scott and Seymour showed that exists for every complete graph . A conjecture of N. R. Aravind states that . The upper bound on that can be obtained using the methods of Scott and Seymour setting are, however, super-exponential. Gyárfás and Sárközy showed that . For , we show that and therefore, . This significantly improves Gyárfás and Sárközy's bound and also covers a bigger class of graphs. We adapt our proof to achieve much stronger upper bounds for graphs of higher girth: we prove that , where . Moreover, in each case, our results imply the existence of at least distinct induced rainbow paths on vertices. Along the way, we obtain some new results on an oriented variant of the Gyárfás–Sumner conjecture. For , let denote the orientations of in which one vertex has out-degree or in-degree . We show that every -free oriented graph having a chromatic number at least and every bikernel-perfect oriented graph with girth having a chromatic number at least contains every oriented tree on at most vertices as an induced subgraph.
期刊介绍:
The Journal of Graph Theory is devoted to a variety of topics in graph theory, such as structural results about graphs, graph algorithms with theoretical emphasis, and discrete optimization on graphs. The scope of the journal also includes related areas in combinatorics and the interaction of graph theory with other mathematical sciences.
A subscription to the Journal of Graph Theory includes a subscription to the Journal of Combinatorial Designs .