Upper tails and independence polynomials in random graphs

Bhaswar B. Bhattacharya, Shirshendu Ganguly, Eyal Lubetzky, Yufei Zhao

Research output: Contribution to journalArticle

Abstract

The upper tail problem in the Erdős–Rényi random graph G∼Gn,p asks to estimate the probability that the number of copies of a graph H in G exceeds its expectation by a factor 1+δ. Chatterjee and Dembo showed that in the sparse regime of p→0 as n→∞ with p≥n−α for an explicit α=αH>0, this problem reduces to a natural variational problem on weighted graphs, which was thereafter asymptotically solved by two of the authors in the case where H is a clique. Here we extend the latter work to any fixed graph H and determine a function cH(δ) such that, for p as above and any fixed δ>0, the upper tail probability is exp⁡[−(cH(δ)+o(1))n2pΔlog⁡(1/p)], where Δ is the maximum degree of H. As it turns out, the leading order constant in the large deviation rate function, cH(δ), is governed by the independence polynomial of H, defined as PH(x)=∑iH(k)xk where iH(k) is the number of independent sets of size k in H. For instance, if H is a regular graph on m vertices, then cH(δ) is the minimum between [formula omitted] and the unique positive solution of PH(x)=1+δ.

Original languageEnglish (US)
Pages (from-to)313-347
Number of pages35
JournalAdvances in Mathematics
Volume319
DOIs
StatePublished - Oct 15 2017

Fingerprint

Random Graphs
Tail
Polynomial
Tail Probability
Rate Function
Weighted Graph
Graph in graph theory
Independent Set
Regular Graph
Clique
Maximum Degree
Large Deviations
Variational Problem
Positive Solution
Exceed
Estimate
Independence

Keywords

  • Large deviations
  • Sparse random graphs
  • Subgraph counts
  • Upper tails
  • Variational problems

ASJC Scopus subject areas

  • Mathematics(all)

Cite this

Upper tails and independence polynomials in random graphs. / Bhattacharya, Bhaswar B.; Ganguly, Shirshendu; Lubetzky, Eyal; Zhao, Yufei.

In: Advances in Mathematics, Vol. 319, 15.10.2017, p. 313-347.

Research output: Contribution to journalArticle

Bhattacharya, Bhaswar B. ; Ganguly, Shirshendu ; Lubetzky, Eyal ; Zhao, Yufei. / Upper tails and independence polynomials in random graphs. In: Advances in Mathematics. 2017 ; Vol. 319. pp. 313-347.
@article{0c2171366a5b4d79b52c16f57dbe8cfa,
title = "Upper tails and independence polynomials in random graphs",
abstract = "The upper tail problem in the Erdős–R{\'e}nyi random graph G∼Gn,p asks to estimate the probability that the number of copies of a graph H in G exceeds its expectation by a factor 1+δ. Chatterjee and Dembo showed that in the sparse regime of p→0 as n→∞ with p≥n−α for an explicit α=αH>0, this problem reduces to a natural variational problem on weighted graphs, which was thereafter asymptotically solved by two of the authors in the case where H is a clique. Here we extend the latter work to any fixed graph H and determine a function cH(δ) such that, for p as above and any fixed δ>0, the upper tail probability is exp⁡[−(cH(δ)+o(1))n2pΔlog⁡(1/p)], where Δ is the maximum degree of H. As it turns out, the leading order constant in the large deviation rate function, cH(δ), is governed by the independence polynomial of H, defined as PH(x)=∑iH(k)xk where iH(k) is the number of independent sets of size k in H. For instance, if H is a regular graph on m vertices, then cH(δ) is the minimum between [formula omitted] and the unique positive solution of PH(x)=1+δ.",
keywords = "Large deviations, Sparse random graphs, Subgraph counts, Upper tails, Variational problems",
author = "Bhattacharya, {Bhaswar B.} and Shirshendu Ganguly and Eyal Lubetzky and Yufei Zhao",
year = "2017",
month = "10",
day = "15",
doi = "10.1016/j.aim.2017.08.003",
language = "English (US)",
volume = "319",
pages = "313--347",
journal = "Advances in Mathematics",
issn = "0001-8708",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Upper tails and independence polynomials in random graphs

AU - Bhattacharya, Bhaswar B.

AU - Ganguly, Shirshendu

AU - Lubetzky, Eyal

AU - Zhao, Yufei

PY - 2017/10/15

Y1 - 2017/10/15

N2 - The upper tail problem in the Erdős–Rényi random graph G∼Gn,p asks to estimate the probability that the number of copies of a graph H in G exceeds its expectation by a factor 1+δ. Chatterjee and Dembo showed that in the sparse regime of p→0 as n→∞ with p≥n−α for an explicit α=αH>0, this problem reduces to a natural variational problem on weighted graphs, which was thereafter asymptotically solved by two of the authors in the case where H is a clique. Here we extend the latter work to any fixed graph H and determine a function cH(δ) such that, for p as above and any fixed δ>0, the upper tail probability is exp⁡[−(cH(δ)+o(1))n2pΔlog⁡(1/p)], where Δ is the maximum degree of H. As it turns out, the leading order constant in the large deviation rate function, cH(δ), is governed by the independence polynomial of H, defined as PH(x)=∑iH(k)xk where iH(k) is the number of independent sets of size k in H. For instance, if H is a regular graph on m vertices, then cH(δ) is the minimum between [formula omitted] and the unique positive solution of PH(x)=1+δ.

AB - The upper tail problem in the Erdős–Rényi random graph G∼Gn,p asks to estimate the probability that the number of copies of a graph H in G exceeds its expectation by a factor 1+δ. Chatterjee and Dembo showed that in the sparse regime of p→0 as n→∞ with p≥n−α for an explicit α=αH>0, this problem reduces to a natural variational problem on weighted graphs, which was thereafter asymptotically solved by two of the authors in the case where H is a clique. Here we extend the latter work to any fixed graph H and determine a function cH(δ) such that, for p as above and any fixed δ>0, the upper tail probability is exp⁡[−(cH(δ)+o(1))n2pΔlog⁡(1/p)], where Δ is the maximum degree of H. As it turns out, the leading order constant in the large deviation rate function, cH(δ), is governed by the independence polynomial of H, defined as PH(x)=∑iH(k)xk where iH(k) is the number of independent sets of size k in H. For instance, if H is a regular graph on m vertices, then cH(δ) is the minimum between [formula omitted] and the unique positive solution of PH(x)=1+δ.

KW - Large deviations

KW - Sparse random graphs

KW - Subgraph counts

KW - Upper tails

KW - Variational problems

UR - http://www.scopus.com/inward/record.url?scp=85028024910&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85028024910&partnerID=8YFLogxK

U2 - 10.1016/j.aim.2017.08.003

DO - 10.1016/j.aim.2017.08.003

M3 - Article

VL - 319

SP - 313

EP - 347

JO - Advances in Mathematics

JF - Advances in Mathematics

SN - 0001-8708

ER -