Fukuoka Institute of Technology REPOSITORY SYSTEM

HOME    repository info    mypage        Japanese    library    university    Feedback

Fukuoka Institute of Technology Repository >
02.University Bulletins & Report >
null >
Vol.1 >

 
Title :NIST乱数検定ツールの検定項目間対角化手法
Authors :岩崎 淳
Issue Date :Dec-2018
Abstract :Randomness tests included in NIST test tool are not independent each other, i.e., the distributions followed by p-value of each test item are not independent. Even if each test item has no problem, the fact makes it difficult to derive the significance level of the test tool as whole items or to introduce a rational criterion for results of the tool. In this paper, we propose a method to solve this problem. The method transforms the distribution followed by each test item’s p-value to the standard normal distribution and assumes that the joint distribution after transformation is a multidimensional normal distribution. Diagonalizing to the joint distribution, the distribution of each test item's p-vale becomes independent each other. In addition, we evaluate this method numerically.
Type Local :紀要論文
ISSN :24345725
Publisher :福岡工業大学総合研究機構
Comment :情報科学研究所(Computer Science Laboratory)
URI :http://hdl.handle.net/11478/1218
citation :福岡工業大学総合研究機構研究所所報
1
39
42
Citation :福岡工業大学総合研究機構研究所所報 Vol.1 p.39 -42
Appears in Collections:Vol.1

Files in This Item:

File Description SizeFormat
11478-1218_p39.pdf473KbAdobe PDFView/Open