翻訳と辞書
Words near each other
・ Michael Schopper
・ Michael Schorr
・ Michael Schottenberg
・ Michael Schraa
・ Michael Sackler-Berner
・ Michael Sacks
・ Michael Sadgrove
・ Michael Sadleir
・ Michael Sadler
・ Michael Sadler (educationist)
・ Michael Sadowsky
・ Michael Sagmeister
・ Michael Sahl Hansen
・ Michael Sak
・ Michael Saks
Michael Saks (mathematician)
・ Michael Salafia
・ Michael Salazar
・ Michael Salcman
・ Michael Salgado
・ Michael Salinger
・ Michael Salomon
・ Michael Salter
・ Michael Salu
・ Michael Salvatori
・ Michael Salyer Stone House
・ Michael Salzhauer
・ Michael Sam
・ Michael Sampson
・ Michael Sams


Dictionary Lists
翻訳と辞書 辞書検索 [ 開発暫定版 ]
スポンサード リンク

Michael Saks (mathematician) : ウィキペディア英語版
Michael Saks (mathematician)
Michael Ezra Saks is an American mathematician. He was (2006–2010) director of the Mathematics Graduate Program at Rutgers University. Saks received his Ph.D from the Massachusetts Institute of Technology in 1980 after completing his dissertation entitled ''Duality Properties of Finite Set Systems'' 〔http://en.scientificcommons.org/2859536〕 under his advisor Daniel J. Kleitman.
A list of his publications and collaborations may be found at.〔http://www.informatik.uni-trier.de/~ley/db/indices/a-tree/s/Saks:Michael_E=.html〕
==Research==
Saks research in computational complexity theory, combinatorics, and graph theory has contributed to the study of lower bounds in order theory, randomized computation, and space-time tradeoff.
In Kahn and Saks (1984) it was shown there exist a tight information-theoretical lower bound for sorting under partially ordered information up to a multiplicative constant.
In () the first super-linear lower bound for the noisy broadcast problem was proved. In a noisy broadcast model, n+1 processors P_0,P_1,\ldots,P_n are assigned a local input bit x_i. Each processor may perform a ''noisy broadcast'' to all other processors where the received bits may be independently flipped with a fixed probability. The problem is for processor P_0 to determined f(x_1,x_2,\ldots,x_n) for some function f. Saks et al. showed that an existing protocol by Gallager was indeed optimal by a reduction from a generalized noisy decision tree and produced a \Omega(n\log(n)) lower bound on the depth of the tree that learns the input.
In Beame et al. (2003) the first time-space lower bound trade-off for randomized computation of decision problems was proved.

抄文引用元・出典: フリー百科事典『 ウィキペディア(Wikipedia)
ウィキペディアで「Michael Saks (mathematician)」の詳細全文を読む



スポンサード リンク
翻訳と辞書 : 翻訳のためのインターネットリソース

Copyright(C) kotoba.ne.jp 1997-2016. All Rights Reserved.