random number [0, 99]
in [C++]
|
From: Pete Becker on 10 Feb 2006 05:57 kevin.hall(a)motioneng.com wrote: > > Ok, fine. But again, the C library implementation of rand() is > notorious for this behavior Which C library is "the C library"? -- and that is what people in this thread > are recommending. Perhaps not all rand() implementations are bad -- > but some certainly are. Maybe. And maybe some implementations of the mersenne twister are bad. > > Anyway, it seems that many people are fighting to defend their practice > of using modulo. Maybe. I'm defending thinking instead of basing engineering decisions on rumors. > >>No, it addresses a different, real, problem. > > > I should have been more clear. I understand that some slots in general > will get more hits than others. I was trying to address the problem > from a different perspective. In my example, I ended up with 8 valid > slots (just like Ulrich's example) and that is what I meant by "This is > the same as above". > > Does this make sense? > No, because it's not the same as above. It's a different problem. Properly reducing the number of values eliminates the imbalance that results from a mismatched number of slots. That's a completely different problem from lack of randomness in low-order bits, assuming it's a problem in the first place. -- Pete Becker Dinkumware, Ltd. (http://www.dinkumware.com) [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ]
From: Alberto Ganesh Barbati on 10 Feb 2006 09:16 quickcur(a)yahoo.com ha scritto: > Hi, I am using the int version of rand() with MS Visual Studio. I would > like to generate random number in the range of [0, 99]. Right now I am > using > > rand() * 100 / (RAN_MAX + 1) > > I found that I got too much 0. 0 appears more than any other numbers. > How can I fix it? I've seen a lot of discussion about rand(), modulo and division in this thread... I'm not going to delve into that, but rather I suggest you throw away rand() as soon as possible and consider using Boost.Random (http://boost.org/libs/random/) as a replacement. It's a solid peer-reviewed random generator library, which means all gory details have probably been settled, so that you don't need to bother about them. Boost.Random has been considered as a reference implementation for the upcoming TR1, so I bet it's going to become quite popular (at least I hope so). Just my opinion, Ganesh [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ]
From: Francis Glassborow on 10 Feb 2006 09:18 In article <1139504596.550672.184530(a)z14g2000cwz.googlegroups.com>, kevin.hall(a)motioneng.com writes >Anyway, it seems that many people are fighting to defend their practice >of using modulo. Division uses higher order bits which should be the >most random in all good PRNGs. Modulo uses the lowest order bits which >is known to NOT be very random in some cases. So which is better >practice for general programming? You must qualify your assertion to modulo a power of 2. Those are the only ones where the high order bits have no contribution. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit For project ideas and contributions: http://www.spellen.org/youcandoit/projects [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ]
From: Francis Glassborow on 10 Feb 2006 09:20 In article <mvednW2NC-NxiHHeRVn-pg(a)giganews.com>, Pete Becker <petebecker(a)acm.org> writes >Carlos Moreno wrote: > >> >> rand() tends to be one of those implementations in which the >> above tends to be true. >> > >More accurately, it is frequently said that some implementations of >rand() have that problem. I think some of the early ones did (and a common one used in Basic was terrible) but I think that most modern implementations have fixed that (yes Pete, I am on your side :-) -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit For project ideas and contributions: http://www.spellen.org/youcandoit/projects [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ]
From: Francis Glassborow on 10 Feb 2006 09:32
In article <1139503966.627601.168480(a)o13g2000cwo.googlegroups.com>, kevin.hall(a)motioneng.com writes >What denotes a good random number generator? No PSEUDO-random number >generator is good compared to a TRUE random number generator. Sorry but a PRNG and a RNG are aimed at entirely different things. And define what you mean by a true RNG. There are some pretty lousy RNG's around (ever seen the care that has to be taken when a really random choice must be maid with all elements being equally likely? Many engineering problems require a sequence of 'mathematically random values' and that sequence must be repeatable so that tests can be repeated. Other cases such picking lottery numbers need a real method for picking numbers at random (an ever noticed how much testing goes into the mechanisms for lotteries to avoid bias?) TR1 (IIRC) provides the user with a range of PRNG tools. And on a modern fast desktop with a raw keyboard read I can provide a pretty good RNG for what ever range of values you want. -- Francis Glassborow ACCU Author of 'You Can Do It!' see http://www.spellen.org/youcandoit For project ideas and contributions: http://www.spellen.org/youcandoit/projects [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] |