The Effect of Induced Subgraphs on Quasi-randomness Asaf Shapira & Raphael Yuster.

Slides:



Advertisements
Similar presentations
Long cycles, short cycles, min-degree subgraphs, and feedback arc sets in Eulerian digraphs Raphael Yuster joint work with Asaf Shapira Eilat 2012.
Advertisements

Great Theoretical Ideas in Computer Science
CS 473Lecture 131 CS473-Algorithms I Lecture 13-A Graphs.
1 Decomposing Hypergraphs with Hypertrees Raphael Yuster University of Haifa - Oranim.
Deterministic vs. Non-Deterministic Graph Property Testing Asaf Shapira Tel-Aviv University Joint work with Lior Gishboliner.
On the Density of a Graph and its Blowup Raphael Yuster Joint work with Asaf Shapira.
Great Theoretical Ideas in Computer Science for Some.
Heuristics for the Hidden Clique Problem Robert Krauthgamer (IBM Almaden) Joint work with Uri Feige (Weizmann)
1 NP-completeness Lecture 2: Jan P The class of problems that can be solved in polynomial time. e.g. gcd, shortest path, prime, etc. There are many.
CS138A Single Source Shortest Paths Peter Schröder.
Combinatorial Algorithms
The Probabilistic Method Shmuel Wimer Engineering Faculty Bar-Ilan University.
The number of edge-disjoint transitive triples in a tournament.
© 2006 Pearson Addison-Wesley. All rights reserved14 A-1 Chapter 14 Graphs.
What is the next line of the proof? a). Let G be a graph with k vertices. b). Assume the theorem holds for all graphs with k+1 vertices. c). Let G be a.
Conditional Regularity and Efficient testing of bipartite graph properties Ilan Newman Haifa University Based on work with Eldar Fischer and Noga Alon.
NP-Complete Problems Reading Material: Chapter 10 Sections 1, 2, 3, and 4 only.
A general approximation technique for constrained forest problems Michael X. Goemans & David P. Williamson Presented by: Yonatan Elhanani & Yuval Cohen.
Foundations of Quasirandomness Joshua N. Cooper UCSD / South Carolina.
1 Packing directed cycles efficiently Zeev Nutov Raphael Yuster.
Is the following graph Hamiltonian- connected from vertex v? a). Yes b). No c). I have absolutely no idea v.
1 CSE 417: Algorithms and Computational Complexity Winter 2001 Lecture 23 Instructor: Paul Beame.
1 On the Benefits of Adaptivity in Property Testing of Dense Graphs Joint work with Mira Gonen Dana Ron Tel-Aviv University.
CSE 421 Algorithms Richard Anderson Lecture 27 NP Completeness.
Clique Cover Cook’s Theorem 3SAT and Independent Set
Introduction to Proofs Goals 1.Introduce notion of proof & basic proof methods. 2.Distinguish between correct & incorrect arguments 3.Understand & construct.
Undirected ST-Connectivity In Log Space Omer Reingold Slides by Sharon Bruckner.
Hardness Results for Problems
Fractional decompositions of dense hypergraphs Raphael Yuster University of Haifa.
The Quasi-Randomness of Hypergraph Cut Properties Asaf Shapira & Raphael Yuster.
The Turán number of sparse spanning graphs Raphael Yuster joint work with Noga Alon Banff 2012.
Lecture 22 More NPC problems
1 Edge-bipancyclicity of star graphs under edge-fault tolerant Applied Mathematics and Computation, Volume 183, Issue 2, 15 December 2006, Pages
Approximating the Minimum Degree Spanning Tree to within One from the Optimal Degree R 陳建霖 R 宋彥朋 B 楊鈞羽 R 郭慶徵 R
Approximation Algorithms Department of Mathematics and Computer Science Drexel University.
Edge-disjoint induced subgraphs with given minimum degree Raphael Yuster 2012.
Fan-planar Graphs: Combinatorial Properties and Complexity results Carla Binucci, Emilio Di Giacomo, Walter Didimo, Fabrizio Montecchiani, Maurizio Patrignani,
Week 10Complexity of Algorithms1 Hard Computational Problems Some computational problems are hard Despite a numerous attempts we do not know any efficient.
Complexity 25-1 Complexity Andrei Bulatov Counting Problems.
1 The number of orientations having no fixed tournament Noga Alon Raphael Yuster.
1 Rainbow Decompositions Raphael Yuster University of Haifa Proc. Amer. Math. Soc. (2008), to appear.
Monochromatic Boxes in Colored Grids Joshua Cooper, USC Math Steven Fenner, USC CS Semmy Purewal, College of Charleston Math.
Maximum density of copies of a graph in the n-cube John Goldwasser Ryan Hansen West Virginia University.
Testing the independence number of hypergraphs
1 Asymptotically optimal K k -packings of dense graphs via fractional K k -decompositions Raphael Yuster University of Haifa.
Ramsey Properties of Random Graphs; A Sharp Threshold Proven via A Hypergraph Regularity Lemma. Ehud Friedgut, Vojtech Rödl, Andrzej Rucinski, Prasad.
NP-Complete Problems. Running Time v.s. Input Size Concern with problems whose complexity may be described by exponential functions. Tractable problems.
Regularity partitions and the topology of graphons László Lovász Eötvös Loránd University, Budapest Joint work Balázs Szegedy August
1/19 Minimizing weighted completion time with precedence constraints Nikhil Bansal (IBM) Subhash Khot (NYU)
1 Decomposition into bipartite graphs with minimum degree 1. Raphael Yuster.
The Evolution of a Hard Graph Theory Problem – Secure Sets Ron Dutton Computer Science University of Central Florida 1.
1 Quasi-randomness is determined by the distribution of copies of a graph in equicardinal large sets Raphael Yuster University of Haifa.
Artur Czumaj DIMAP DIMAP (Centre for Discrete Maths and it Applications) Computer Science & Department of Computer Science University of Warwick Testing.
CPS Computational problems, algorithms, runtime, hardness (a ridiculously brief introduction to theoretical computer science) Vincent Conitzer.
Dense graphs with a large triangle cover have a large triangle packing Raphael Yuster SIAM DM’10.
1 Complexities of some interesting problems on spanning trees M Sohel Rahman King’s College, London M Kaykobad KHU, NSU and BUET.
C&O 355 Lecture 19 N. Harvey TexPoint fonts used in EMF. Read the TexPoint manual before you delete this box.: A A A A A A A A A A.
The NP class. NP-completeness Lecture2. The NP-class The NP class is a class that contains all the problems that can be decided by a Non-Deterministic.
Theory of Computational Complexity Probability and Computing Lee Minseon Iwama and Ito lab M1 1.
The NP class. NP-completeness
Markov Chains and Random Walks
Algorithm Analysis Fall 2017 CS 4306/03
Computability and Complexity
Parameterised Complexity
Clustered representations: Clusters, covers, and partitions
Packing directed cycles efficiently
Around the Regularity Lemma
Existence of 3-factors in Star-free Graphs with High Connectivity
Locality In Distributed Graph Algorithms
Integer and fractional packing of graph families
Presentation transcript:

The Effect of Induced Subgraphs on Quasi-randomness Asaf Shapira & Raphael Yuster

Background Chung, Graham, and Wilson ’89, Thomason ‘87: 1. Defined the notion of a p-quasi-random = A graph that “behaves” like a typical graph generated by G(n,p). 2. Proved that several “natural” properties guarantee that a graph is p-quasi-random. Abstract Question: What it means for a graph to be random? “Concrete” problem: Which graph properties “force” a graph to behave like a “truly” random one.

The CGW Theorem Theorem [CGW ‘89]: Fix any 0<p<1, and let G=(V,E) be a graph on n vertices. The following are equivalent: 1.Any set of vertices U  V spans  ½p|U| 2 edges 2.Any set of vertices U  V of size ½n spans  ½p|U| 2 edges 3. 1 (G)  pn and 2 (G) = o(pn) 4.For any graph H on h vertices, G has  n h p e(H) copies of H 5. G contains  ½pn 2 edges and  p -4 n 4 copies of C 4 6.All but o(n 2 ) of the pairs u,v have  p 2 n common neighbors Definition: A graph that satisfies any (and therefore all) the above properties is p-quasi-random, or just quasi-random. “  ” = (1+o(1)) Note: All the above hold whp in G(n,p).

Background Relation to (theoretical) computer science: 1. Conditions of randomness that are verifiable in polynomial time. For example, using number of C 4, or using 2 (G). 2. Algorithmic version of Szemeredi’s regularity-lemma [ADLRY ‘95], uses equivalence between regularity and number of C 4. Relation to Extremal Combinatorics: 1. Central in the strong hypergraph generalizations of Szemeredi’s regularity-lemma.

More on the CGW Theorem Definition: Say that a graph property is quasi-random if it is equivalent to the properties in the CGW theorem. “The” Question: Which graph properties are quasi-random? Any (natural) property that holds in G(n,p) whp? Example: Having  ½pn 2 edges and  p -3 n 3 copies of K 3 is not a quasi-random property. Example: Having  ¼ pn 2 edges crossing all cuts of size (½n,½n) is not a quasi-random property. Recall that if we replace K 3 with C 4 we do get a quasi-random property. No! …but if we consider cuts of size (¼n,¾n) then the property is quasi-random.

Subgraph and Quasi-randomness The effect of subgraphs on quasi-randomness: 1. Having the correct number of edges + correct number of C 4 is a quasi-random property. True also for any C 2k. 2. Having the correct number of edges + correct number of K 3 is not a quasi-random property. True for any non-bipartite H. Question: Is it true that for any single H, the “distribution” of copies of H affects the quasi-randomness of a graph? [Simonovits and Sos ’97]: Yes. If all vertex sets U  V span  |U| h p e(H) copies of H, then G is p-quasi-random. Intutition: “Randomness is a hereditary property”.

Induced Subgraph and Quasi-Rand Related Question: Can we expect to be able to deduce from the distribution of a single H that a graph is p-quasi-random. [Simonovits and Sos ’97]: For any H, if all vertex sets U  V span  |U| h p e(H) copies of H, then G is p-quasi-random. Question: Is it true that for any single H, the “distribution” of induced copies of H affects the quasi-randomness of a graph? Answer: No. For any p and H, there is a q=q(p,H) for which G(n,p) and G(n,q) behave identically w.r.t. induced copies of H.

Induced Subgraph and Quasi-Rand Question: Is it true that for any single H, the “distribution” of induced copies of H affects the quasi-randomness of a graph? [Simonovits and Sos ’03]: No. There are non quasi-random graphs, where all U  V span  |U| 3 p 2 (1-p) induced copies of P 3. [Simonovits and Sos ’97]: For any H, if all vertex sets U  V span  |U| h p e(H) copies of H, then G is p-quasi-random. Question: Is it true that for any H, if all vertex sets U  V span  |U| h  H (p) induced copies of H, then G is quasi-random? Definition:

A New Formulation Lemma: Fix any H on h vertices. The following are equivalent: 1. G is such that all U  V span  |U| h  H (p) induced copies of H. 2. G is such that all h-tuple U 1,…,U h of (arbitrary) size m span  h!m h  H (p) induced copies of H with one vertex in each U i. Observation: Consider any ordered h-tuple U 1,…,U h of (arbitrary) size m in G(n,p). We actually expect U 1,…,U h to span  m h  H (p) induced copies of H with the vertex in U i playing the role of vertex v i of H. Question: Is it true that for any H, if all vertex sets U  V span  |U| h  H (p) induced copies of H, then G is quasi-random? NO Definition: In that case, we say that U 1,…,U h have the correct number of induced embedded copies of H.

Main Result Question: Is it true that for any single H, the “distribution” of induced copies of H affects the quasi-randomness of a graph? Theorem [S-Yuster ‘07]: Yes! 1.Assume all ordered h-tuples U 1,…,U h of (arbitrary) size m span the correct number of induced embedded copies of H. Then G is quasi-random. 2. In fact, G is either p-quasi-random or q-quasi-random. Note: 1. We can’t expect to show that G is p-quasi-random. 2. We can’t consider only the number of induced copies of H in U 1,…,U h.

Proof Overview Lemma [SS 91]: If G is composed of quasi-random graphs with the same density, then G itself is quasi-random. That is, Suppose G is a k-partite graph on vertex sets V 1,…,V k, and most of the bipartite graphs on (V i,V j ) are p-quasi-random. Then G is also p-quasi-random. Overall strategy: Show that G has a k-partition, where most of the quasi-random bipartite graphs have density p or most have density q.

Proof Overview Fact: If G is an h-partite graph on V 1,…,V h and all the bipartite graphs (V i,V j ) are quasi-random, then the number of induced copies of H is determined by the densities between (V i,V j ). denstiy = x 1,3 density = x 2,4  x 1,2 · x 1,3 · x 1,4 · x 2,3 · x 2,4 · (1-x 2,3 ) V1V1 V3V3 V2V2 V4V4 The density of is In fact even the number of induced embedded copies of H.

Proof Overview We will show that in such a partition most densities are p or q. [Szemeredi’s regularity lemma ‘79]: Any graph has a k-partition into V 1,…,V k s.t. most graphs on (V i,V j ) are quasi- random. But not necessarily with the same density… Assumption on G: We “know” the number of induced embedded copies of H in each h-tuple of vertex sets V 1 …,V h. Fact: Given a partition where most bipartite graphs (V i,V j ) are quasi-random, we “know” the number of induced copies of H. We get (k) h polynomial equations relating the densities (V i,V j ). We will show that the only solution is p or q.

Proof Overview Let W be a weighted complete graph on k vertices, with weights 0  w(i,j)  1. For every mapping  :[h]  [k] define Key Lemma: Suppose that for all  :[h]  [k] we have W(  )  (p). Then either most w(i,j)  p or most w(i,j)  q. Notes: We cannot expect to show that most w(i,j)  p. Also, it is NOT true that either all w(i,j)  p or all w(i,j)  q. 1. w(i,j) stands for the density between V i and V j. 2. W(  )   (p) due to our assumption on G.

Proof Overview Key Lemma: Suppose that for all  :[h]  [k] we have Proof idea: Introduce unknowns x ij for each w(i,j). Consider the system of polynomial equations: Then either most w(i,j)  p or most w(i,j)  q. First step: Show that the only solution is x ij  {p,q}.

Proof Overview Proof idea: Suppose that for all  :[h]  [k] we have First step: Show that the only solution is x ij  {p,q}. [Gottlieb ‘66]: If r  k+h, then rank(A(r,h,k)) =. Definition: For integers k  h  r, let A(r,h,k) be the inclusion matrix of the k-element subsets of [r] and its h-element subsets. h-element sets k-element sets A S,T = 1 iif S  T

Proof Overview 3)Use Regularity-lemma + Packing results ( Rodl or Wilson ) to conclude that G is composed of quasi-random graphs with the same densities. 2)Some more (non-trivial) arguments needed to show that in fact most are p or most are q. 1)After (appropriate) manipulations this gives that the unique solution uses p and q.

Thank You