談話会・数理科学講演会

過去の記録 ~02/05次回の予定今後の予定 02/06~

担当者 加藤晃史、北山貴裕、辻雄(委員長)、三竹大寿
セミナーURL https://www.ms.u-tokyo.ac.jp/seminar/colloquium/index.html

2012年07月06日(金)

16:30-17:30   数理科学研究科棟(駒場) 002号室
旧記録は、上記セミナーURLにあります。
お茶&Coffee&お菓子: 16:00~16:30 (コモンルーム)。

S.R.Srinivasa Varadhan 氏 (Courant Institute of Mathematical Sciences, New York University)
Large Deviations of Random Graphs and Random Matrices (ENGLISH)
[ 講演概要 ]
A random graph with $n$ vertices is a random symmetric matrix of $0$'s and $1$'s and they share some common aspects in their large deviation behavior. For random matrices it is the question of having large eigenvalues. For random graphs it is having too many or too few subgraph counts, like the number of triangles etc. The question that we will try to answer is what would a random matrix or a random graph conditioned to exhibit such a large deviation look like. Since the randomness is of size $n^2$ large deviation rates of order $n^2$ are possible.