Diameter and stationary distribution of random r-out digraphs

被引:7
作者
Addario-Berry, Louigi [1 ]
Balle, Borj A. [2 ]
Perarnau, Guillem [3 ]
机构
[1] McGill Univ, Dept Math & Stat, Montreal, PQ, Canada
[2] McGill Univ, Sch Comp Sci, Montreal, PQ, Canada
[3] Univ Politecn Catalunya UPC, Dept Matemat, Barcelona, Spain
基金
加拿大自然科学与工程研究理事会;
关键词
D O I
10.37236/9485
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Let D(n, r) be a random r-out regular directed multigraph on the set of vertices {1, ..., n}. In this work, we establish that for every r >= 2, there exists eta(r) > 0 such that diam(D(n, r)) = (1 + eta(r)+ o(1)) log(r) n. The constant eta(r) is related to branching processes and also appears in other models of random undirected graphs. Our techniques also allow us to bound some extremal quantities related to the stationary distribution of a simple random walk on D(n, r). In particular, we determine the asymptotic behaviour of pi(m)(ax) and pi(min) the maximum and the minimum values of the stationary distribution. We show that with high probability pi(max) = n(-1+o(1)) and pi(min) = n(-(1+eta r)+o(1)). Our proof shows that the vertices with 7r(v) near to 7i m i n lie at the top of "narrow, slippery towers"; such vertices are also responsible for increasing the diameter from (1 + o(1)) log(r) n to (1 + eta(r) + o(1)) log(r) n.
引用
收藏
页码:1 / 41
页数:41
相关论文
共 38 条
[1]  
Angluin D., 2010, INT C ALG LEARN THEO
[2]  
Angluin D., 2009, INT C ALG LEARN THEO
[3]  
Angluin D, 2015, PROC INT C ALGORITHM, P119
[4]  
[Anonymous], 1998, Markov chains. 2
[5]  
[Anonymous], 2009, Markov chains and mixing times
[6]  
[Anonymous], 1994, An Introduction to Computational Learning Theory, DOI DOI 10.7551/MITPRESS/3897.001.0001
[7]  
Athreya K.B., 1972, Grundlehren der mathemat. Wissenschaften
[8]  
Balle B., 2013, ARXIV13116830
[9]  
Bassino F., 2012, ALGORITHMICA
[10]   The state complexity of random DFAs [J].
Berend, Daniel ;
Kontorovich, Aryeh .
THEORETICAL COMPUTER SCIENCE, 2016, 652 :102-108