A sharp threshold for Ramsey properties of random sets of integers (A Socratic dialogue) Ehud Friedgut, Weizmann Institute Joint work with Hiệp Hàn, Yuri.

Slides:



Advertisements
Similar presentations
Completeness and Expressiveness
Advertisements

Following Directions [1] A: Excuse me. Do you speak English
Thomas A. Stewart Literacy Test (OSSLT) Prep Guide 2013
How to make a good power point presentation!
Welcome to plexiom! Thank you for taking a few minutes to see what we’re about. plexiom.com.
A.
DESTINY SHOBE , NA’STYSHIA SULLIVAN , JHANARRI DURHAM
Farewell Argument Paper Weeds & Roses. Why this weeds and roses is extremely important:  You are likely to have concerns, comments, questions, gripes.
By Bryce Hanson. There are many different types of customers here’s a few Angry Irritable/moody Complaining Impatient Leave me Alone Insulting Argumentative.
Section 3.5 – P lyg n Angle-Sum Theorems Created by Leon Tyler Funchess.
Greeting and Introduction By Thidawan Wichanee. Listen and Watch Take notes Please! What are the similarities in these two dialogues?
The basics of ratio. The purpose of this staff tutorial is to introduce the basic ideas of ratio and to show how ratio relates to fractions. Using this.
Inapproximability of Hypergraph Vertex-Cover. A k-uniform hypergraph H= : V – a set of vertices E - a collection of k-element subsets of V Example: k=3.
Bison Management Suppose you take over the management of a certain Bison population. The population dynamics are similar to those of the population we.
A Teenager’s Guide To Asexuality Am I Ace?. Am I Asexual? You’re not into sex the way other people are. You’re not sure you really get what people mean.
Week 10 Generalised functions, or distributions
Introducing your Technical Author... Alison Peck.
Decision Rules. Decision Theory In the final part of the course we’ve been studying decision theory, the science of how to make rational decisions. So.
Teresa Harding US Founding Presidential Diamond
Communication at International Conference
Probability and Induction
8 meeting Action points.
Hello, Pig! Hello, Rabbit! Look at this – I am making a list!
Bugz A Musical Play For Young Voices. 2. Goin’ On a Picnic.
Writing an 1st Year Intermediate
Escape Pythagoreans Dungeon! Alex Marling Geometry Middle school Time to Escape!
On the robustness of dictatorships: spectral methods. Ehud Friedgut, Hebrew University, Jerusalem.
Discussion Skills For the 2nd semester.
Customers Request the Darndest Things* 10 Challenges for VUI Designers Eduardo Olvera User Interface Designer.
MA 1128: Lecture 06 – 2/15/11 Graphs Functions.
Copyright 2009 by Richard P. Walters TURN ON SOUND, THEN CLICK TO BEGIN.
Five Ways to Sabotage Your Business By Nancy Friedman, Telephone Doctor.
What’s wrong with this statement? Yes, I agree that she is to blame for what happened. In my opinion it is clear to see from the reading of the book that.
Deterministic vs. Non-Deterministic Graph Property Testing Asaf Shapira Tel-Aviv University Joint work with Lior Gishboliner.
I Want to Change for the BETTER! Mahragan el Keraza Sunday, May 6 th 2012.
On the Density of a Graph and its Blowup Raphael Yuster Joint work with Asaf Shapira.
Copyright © Cengage Learning. All rights reserved.
Lecture 22: April 18 Probabilistic Method. Why Randomness? Probabilistic method: Proving the existence of an object satisfying certain properties without.
SEQUENCES and INFINITE SERIES
Upper bounds for asymmetric Ramsey properties of random graphs Reto Spöhel, ETH Zürich Joint work with Yoshiharu Kohayakawa, Universidade de São Paulo.
On the relation between probabilistic and deterministic avoidance games Torsten Mütze, ETH Zürich Joint work with Michael Belfrage (ETH Zürich), Thomas.
Offline thresholds for online games Reto Spöhel, ETH Zürich Joint work with Michael Krivelevich and Angelika Steger TexPoint fonts used in EMF. Read the.
1 List Coloring and Euclidean Ramsey Theory TexPoint fonts used in EMF. Read the TexPoint manual before you delete this box.: A A A A Noga Alon, Tel Aviv.
Asymmetric Ramsey Properties of Random Graphs involving Cliques Reto Spöhel Joint work with Martin Marciniszyn, Jozef Skokan, and Angelika Steger TexPoint.
Online Graph Avoidance Games in Random Graphs Reto Spöhel Diploma Thesis Supervisors: Martin Marciniszyn, Angelika Steger.
Ramsey games in random graphs PhD defense 22 nd Nov 2011 Torsten Mütze TexPoint fonts used in EMF. Read the TexPoint manual before you delete this box.:
Hunting for Sharp Thresholds Ehud Friedgut Hebrew University.
Online Vertex Colorings of Random Graphs Without Monochromatic Subgraphs Reto Spöhel, ETH Zurich Joint work with Martin Marciniszyn.
Online Ramsey Games in Random Graphs Reto Spöhel, ETH Zürich Joint work with Martin Marciniszyn and Angelika Steger.
Copyright © Cengage Learning. All rights reserved. CHAPTER 11 ANALYSIS OF ALGORITHM EFFICIENCY ANALYSIS OF ALGORITHM EFFICIENCY.
MATH 224 – Discrete Mathematics
Chapter 6.7 Determinants. In this chapter all matrices are square; for example: 1x1 (what is a 1x1 matrix, in fact?), 2x2, 3x3 Our goal is to introduce.
How conversation works Conversational English Compiled by Victor Nickolz Grand Lyceum 2004 For classes 7-11.
The Effect of Induced Subgraphs on Quasi-randomness Asaf Shapira & Raphael Yuster.
The “Intercepts” Form and graphing linear inequalities Alegbra I, Standard 6.
Tonga Institute of Higher Education Design and Analysis of Algorithms IT 254 Lecture 8: Complexity Theory.
Online Ramsey Games in Random Graphs Reto Spöhel, ETH Zürich Joint work with Martin Marciniszyn and Angelika Steger TexPoint fonts used in EMF. Read the.
Online Vertex-Coloring Games in Random Graphs Reto Spöhel (joint work with Martin Marciniszyn; appeared at SODA ’07)
Extending the Definition of Exponents © Math As A Second Language All Rights Reserved next #10 Taking the Fear out of Math 2 -8.
Edge-disjoint induced subgraphs with given minimum degree Raphael Yuster 2012.
Ramsey Properties of Random Graphs; A Sharp Threshold Proven via A Hypergraph Regularity Lemma. Ehud Friedgut, Vojtech Rödl, Andrzej Rucinski, Prasad.
The Practice of Statistics, 4 th edition – For AP* STARNES, YATES, MOORE Chapter 8: Estimating with Confidence Section 8.2 Estimating a Population Proportion.
Module 2.2.  Learn the importance of using probing questions during the tutoring session.
1 Decomposition into bipartite graphs with minimum degree 1. Raphael Yuster.
Week 6. Statistics etc. GRS LX 865 Topics in Linguistics.
Copyright © Cengage Learning. All rights reserved. CHAPTER 8 RELATIONS.
Copyright © Zeph Grunschlag, Induction Zeph Grunschlag.
The inference and accuracy We learned how to estimate the probability that the percentage of some subjects in the sample would be in a given interval by.
Philosophy of Religion Ontological Argument
On the effect of randomness on planted 3-coloring models
Presentation transcript:

A sharp threshold for Ramsey properties of random sets of integers (A Socratic dialogue) Ehud Friedgut, Weizmann Institute Joint work with Hiệp Hàn, Yuri Person, and Mathias Schacht

: Hello, I’m Socrates, I’m 2483 years old : And I thought I was old...Pleased to meet you, Paul Erdos, 100 years old. : I was looking at the title of this talk. Would you oblige me by explaining what a sharp threshold is? : With pleasure.

: So, consider the random graph G(n,p). It becomes connected when p is approximately log(n)/n. Yet the increase from, say, 1% probability to 99%, happens for p in an interval with length of order only 1/n. : Sharp, indeed. Can you give me an example of a non-sharp threshold?

: Sure. Consider the property of containing a triangle. There the critical probability is of order 1/n, and the length of the threshold interval is also of order 1/n. : Hmmm..., any other interesting examples? : Actually, not really.

: ?? : You see, by a result of Friedgut from ’99... : 99 B.C. or A.D. ? : That’s 1999 A.D. Anyhow, he shows that whenever a graph property has a coarse threshold, it’s because of a “local reason”

: I see, so triangle=local=coarse, and connectivity=global=sharp? : Correct! : Let’s see. So, having chromatic number, say, greater than 3... Wait, is that global or local? : In our context, it’s global.

: But, if there’s a K 4 in the graph, that’s local, no? : So true, my scholarly friend. Yet in the threshold setting, where p turns out to be of order 1/n, you’re not going to see any copies of K 4. : But how do you know there isn’t some other “local reason” for non-3-colorability?

: Well, it actually takes some work. Achlioptas and Friedgut showed this in : Achlioptas? Sounds like a co-patriot of mine. Anyhow, I understand it can be tricky proving that a property is “non-local”. : Indeed. For example, for the property of having choice number > 3, it’s not been proven.

: I noticed that you said that Friedgut’s result is about properties of random graphs. But the title of the talk mentioned random sets of integers. : That’s an astute observation. There’s an analogous theorem of Bourgain which takes care of more general settings. : Can you please explain about the case at hand?

: With pleasure. The question is when a random subset of {1,2,...,n} has the property that any coloring of its elements by two colors, say, blue and red, produces a monochromatic 3-term arithmetic progression. : Wait, what density is critical for this property?

: Good question. Rödl and Ruci ń ski proved that the critical density is which is when the expected number of elements and the expected number of 3- AP ’s is approximately the same. : Excuse me, Uncle Paul, can I interrupt for a moment? : Please.

: I see. So now you want to decide whether the property we’re discussing could possibly be correlated with something “local”. How about containing an interval of twenty consecutive integers? It seems to me that would suffice to force a monochromatic 3- AP. : True, but Rödl and Rucinski showed that all such obvious examples are like a K 4 in G(n,c/n): highly unlikely. The problem is whether some more innocuous examples could, magically, be correlated with the Ramsey property.

: You know, I haven’t yet developed an intuition whether this is a difficult task. : Well, for a similar problem in graphs, having a monochromatic triangle in every bi-coloring, it took Friedgut, Rödl, Ruci ń ski and Tetali more than 100 pages. : Great Zeus! Why so many?

: Well, 65 of these pages were devoted to developing and applying a sparse- hypergraph-regularity-lemma they needed. : Ouch. : Indeed, to quote a good friend of mine: : “When possible, regularity must be avoided at all cost.”

: So, in the current work, was regularity side-stepped? : Yes! Thanks to works of BMS and/or ST. : ? : Balogh-Morris-Samotij, and Saxton- Thomason.

: So, you claim that you can prove that this Ramsey property is not correlated with, say, containing three consecutive integers? : That’s right. : But how? :O.k., I’ll try to sketch the argument.

: O.k. Consider a random subset of {1,2,...,n}, with density approximately. The property we’re discussing is whether every bi-coloring of the set contains a monochromatic 3-term AP. Now, we’re assuming by way of contradiction... : What’s that? : Reductio ad absurdum. : Oh, so why not use simple Latin?

: Sorry, I’ll try to stick to plain terminology. So we assume by way of reductio ad absurdum that the Ramsey property has a non-sharp threshold. By Bourgain’s theorem, there’s a “local culprit”, say an interval of length 3, such that containing it is significantly correlated with the Ramsey property.

:Wait, do you mean a specific interval of length three, or any interval of length three? : Excellent question!! To avoid this delicate point we will work in the cyclical group Z n, rather than in {1,2,...n}, so all intervals of length three are the same.

: O.k., I think I follow so far. But how will this reduce to absurdum? : We’ll see that the fact that for so many x’s, adding {x,x+1,x+2} induces Ramsiness, means that the set of colorings with no monochromatic 3- AP is rather restricted. : and?... That will imply that adding more points at random kills all possible colorings with probability close to 100%, and not 50% as we assumed.

: But it’s not even clear to me how adding {x,x+1,x+2} to Z can ruin all possible colorings. Is it because it forces x-1 to be colored blue? : No, typically x-1 doesn’t even belong to Z. It’s more complicated. Here I think a picture might help.

: What this picture means is that there are elements a,b,c,a’,b’,c’ in Z such that in every proper coloring either [a and a’ are red] or [b and b’ are red] or [c and c’ are blue] : But couldn’t the interaction of {x,x+1,x+2} with Z be of a different nature? a....b...c...x,x+1,x+2,..a’...b’...c’

: Yes, it could, but using probabilistic arguments we can focus on one such prototype. : How can an argument be “probabilistic”? : Oh, that’s an idea that I introduced a mere 65 years ago.

: So all these points whose color is forced to be red, do they contain very many 3- AP ’s? : How did you know?! It follows from a strong version of Szemeredi’s (or Roth’s) theorem. : So if the random addition of more points contains one of these 3- AP s you won’t be able to extend the 3- AP -free coloring?

: Right! You’ve got the drift of it. : But consider the case where these additional points don’t contain any of these 3- AP ’s? : Aha! By Janson’s inequality that’s exponentially unlikely. : I see. So every proper coloring of Z is exponentially unlikely to be extendable to a smallish random set of points. And that contradicts the fact that the random set induces Ramsiness with only 50% probability?

: Not so fast... Each coloring is exponentially unlikely to be extendable, but we have an exponential number of colorings. : ανάθεμα ! So how do we know who wins this race of exponents? : We don’t. Here’s where BMS and/or ST come to the rescue.

: The number of hitting sets (that are exponentially unlikely to be extended) is exponential, but BMS/ST allows us to bunch them into a much smaller number of bunches, where hitting sets in the same bunch induce the same restriction. : So then you can use a union bound! : Ah, you learn so quickly. Such a sage!

:...and that yields the contradiction? With so many supposed local restrictions, it’s extremely unlikely (much less than 50%) that any coloring of Z can be extended to additional random points ? So we have our contradiction! : Quod Erat Demonstrandum. Or as you say, ὅ περ ἔ δει δε ῖ ξαι.

: Wow, that’s very nice. Does this proof extend to more general questions? : Well, the same proof works for arithmetic progressions of length greater than 3, but unfortunately, not for more than two colors. Also, if you replace “arithmetic progression” by some other Rado-type equation, complications arise.

: O.k., you’re getting me giddy here. : How can I repay your infinite patience and kindness in explaining all this to me? : How about a lesson in philosophy? : With pleasure. But be forewarned, philosophy can be very difficult, and our reasoning may be quite different from what you’re used to. : Have no fear. My brain is open...

My brain is open...