Bugs item #538118, was opened at 2002-04-01 17:05
You can respond by visiting:
http://sourceforge.net/tracker/?func=detail&atid=112740&aid=538118&group_id=12740Category: None
Group: None
Status: Closed
Resolution: None
Priority: 5
Submitted By: Nobody/Anonymous (nobody)
Assigned to: Nobody/Anonymous (nobody)
Summary: bug in Knuth uniform rnd gen
Initial Comment:
In function ranf_arr_cycle should be generated
1009 "random" numbers, but used only 100 of them.
Knuth puts -1 on 100th place in array to bound output.
----------------------------------------------------------------------
Comment By: Nobody/Anonymous (nobody)
Date: 2002-04-04 14:11
Message:
Logged In: NO
New 2002 thing is related to ranf_start, he improved
initialization algorithm.
QUALITY thing is discussed in 3rd edition of Seminumerical
Algorithms. He says there that it is good to skip chunk of
numbers for programs, which require many random numbers
simultaneously (Especially MonteCarlo simulation).
----------------------------------------------------------------------
Comment By: Luigi Ballabio (lballabio)
Date: 2002-04-03 02:00
Message:
Logged In: YES
user_id=75450
Maybe the extract 1009, use 100 thing was added by Knuth afterwards? He
speaks of it as a 2002 thing.
However, it is fixed in CVS now.
----------------------------------------------------------------------
You can respond by visiting:
http://sourceforge.net/tracker/?func=detail&atid=112740&aid=538118&group_id=12740
Locked
