NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c-part1/2633

Internet Message Format | 1996-08-05 | 2.0 KB

Path: Rezonet.net!news
From: [e-mail] (Ray Dunn)
Newsgroups: comp.lang.c
Subject: Re: swap high and low 16 bits?
Date: 20 Jan 1996 02:35:26 GMT
Organization: Ultimate Technographics Inc.
Message-ID: <4dpkde$[e-mail]>
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
NNTP-Posting-Host: 204.19.230.7
Mime-Version: 1.0
Content-Type: Text/Plain; charset=US-ASCII
X-Newsreader: WinVN 0.99.7

In referenced article, Lawrence Kirby says...
>In article <[e-mail]>
>           [e-mail] "Bill Simpson" writes:
>
>>> The way I understand you is that you are using the result of time()
>>> as your random number.
>>No, as the seed to the RNG.
>>
>>If my program runs and terminates quickly, it will get almost exactly
>>the same seed on successive runs (differing only by low bits).
>>
>>But not if I swap high and low 16 bits.
>>
>>I have done this now and it works nicely.  The seeds are far apart,
>>even for runs <1 minute apart.
>
>Just changing high order bits may not be a good idea since with some
>RNGs this won't filter down to the lower order bits. I suggest you use
>one of the expressions I posted earlier (which are incidentally
>simpler than a full swap).

Now that I see the original poster was in fact using time to seed the
random number generator, I'm not quite sure what the problem is.  The
closesness of one seed to another seed in no way implies that the two
random sequences so produced will be "close to" each other.

A difference of only one in the value given to the seeding function
causes a totally different sequences of numbers to be produced.

--
Ray Dunn (opinions are my own) | Phone: [phone]
Montreal                       | Phax : [phone]
[e-mail]          | Home : [phone]