random number question.
in [General Programming]
|
Prev: How to write an server application with high performance and high availability.
Next: How to write an server application with high performance andhigh availability.
From: Chad on 13 Apr 2008 14:54 The question stems from the following url http://groups.google.com/group/comp.lang.c/browse_thread/thread/dbb7f5c53d288b0a/4ec31a37499e24f0?hl=en#4ec31a37499e24f0 I had asked what was the difference between what' the difference between random numbers and pseudo-random numbers. I'm not that sure I'm going to get a response since some of the posters are more concerned about using colorful metaphors to describe each other. Usually when the vegititation in comp.lang.c starts to act like a bunch of hyperactive 6th graders, they tend to ignore everyone else. Maybe I will get a response on comp.lang.c.
From: Jens Thoms Toerring on 13 Apr 2008 15:40 Chad <cdalten(a)gmail.com> wrote: > The question stems from the following url > http://groups.google.com/group/comp.lang.c/browse_thread/thread/dbb7f5c53d288b0a/4ec31a37499e24f0?hl=en#4ec31a37499e24f0 > I had asked what was the difference between what' the difference > between random numbers and pseudo-random numbers. I'm not that sure > I'm going to get a response since some of the posters are more > concerned about using colorful metaphors to describe each other. > Usually when the vegititation in comp.lang.c starts to act like a > bunch of hyperactive 6th graders, they tend to ignore everyone else. > Maybe I will get a response on comp.lang.c. Sorry, but perhaps you should try to understand a bit more what different news groups are meant for before you call someone a "hyperactive 6th grader" - they have different topics or we wouldn't need them at all. And the difference between random numbers and pseudo-random numbers isn't anything related in any way to the topic of comp.lang.c, the C programming lan- guage. On the other hand comp.programming has a much broader topic, so it's on topic here, so here we go: Random numbers are numbers that are truely random - there's no way you could ever predict the next one when you already have gotten some from the generator and even if you got all the specifications for the generator. Think e.g. of a gene- rator that gets it's numbers from a source that's known to be strictly random (throwing a coin, radioactive decay etc.). In contrast, pseudo-random numbers are generated via some algorithm. One example of such a generator often used to create pseudo-random numbers employs a rather simple for- mula like x[i+1] = trunc(a * x[i] + b) where trunc() returns the digits after the decimal point. Obviously, the numbers generated this way aren't really random, and once you know the formula and 'a' and 'b' you can easily predict with absolute certainty what x[i+1] will be from x[i] (and if you only have a lot of consecutive number you still might be able to work out what the algo- rithm and its parameters are). Of course, you can use some more complicated formulas, but that doesn't change anything about the basic non- randomness of the numbers created (and, if I remember the chapter of Knuth's "The Art of Computer Programming" about pseudo-random generators correctly, trying to get too clever can actually make things worse;-) Creating really random numbers is difficult since you need some physical process that's truely random to produce them, so you need some extra hardware (some CPUs nowadays come with something like that, probably taking advantage of the randomness of Schottky noise). Pseudo-random numbers, on the other hand, are simple, cheap and fast to produce and for many purposes they're completely sufficient and are thus used in quite a number of applications. Regards, Jens -- \ Jens Thoms Toerring ___ jt(a)toerring.de \__________________________ http://toerring.de
From: Chad on 13 Apr 2008 16:06 On Apr 13, 12:40 pm, j...(a)toerring.de (Jens Thoms Toerring) wrote: > Chad <cdal...(a)gmail.com> wrote: > > The question stems from the following url > >http://groups.google.com/group/comp.lang.c/browse_thread/thread/dbb7f... > > I had asked what was the difference between what' the difference > > between random numbers and pseudo-random numbers. I'm not that sure > > I'm going to get a response since some of the posters are more > > concerned about using colorful metaphors to describe each other. > > Usually when the vegititation in comp.lang.c starts to act like a > > bunch of hyperactive 6th graders, they tend to ignore everyone else. > > Maybe I will get a response on comp.lang.c. > > Sorry, but perhaps you should try to understand a bit more what > different news groups are meant for before you call someone a > "hyperactive 6th grader" - they have different topics or we > wouldn't need them at all. And the difference between random > numbers and pseudo-random numbers isn't anything related in > any way to the topic of comp.lang.c, the C programming lan- > guage. On the other hand comp.programming has a much broader > topic, so it's on topic here, so here we go: > I still have some general misconceptions about how the newsgroups work. > Random numbers are numbers that are truely random - there's > no way you could ever predict the next one when you already > have gotten some from the generator and even if you got all > the specifications for the generator. Think e.g. of a gene- > rator that gets it's numbers from a source that's known to > be strictly random (throwing a coin, radioactive decay etc.). > > In contrast, pseudo-random numbers are generated via some > algorithm. One example of such a generator often used to > create pseudo-random numbers employs a rather simple for- > mula like > > x[i+1] = trunc(a * x[i] + b) > > where trunc() returns the digits after the decimal point. > Obviously, the numbers generated this way aren't really > random, and once you know the formula and 'a' and 'b' you > can easily predict with absolute certainty what x[i+1] will > be from x[i] (and if you only have a lot of consecutive > number you still might be able to work out what the algo- > rithm and its parameters are). > > Of course, you can use some more complicated formulas, > but that doesn't change anything about the basic non- > randomness of the numbers created (and, if I remember > the chapter of Knuth's "The Art of Computer Programming" > about pseudo-random generators correctly, trying to get > too clever can actually make things worse;-) > > Creating really random numbers is difficult since you need > some physical process that's truely random to produce them, > so you need some extra hardware (some CPUs nowadays come > with something like that, probably taking advantage of the > randomness of Schottky noise). Pseudo-random numbers, on > the other hand, are simple, cheap and fast to produce and > for many purposes they're completely sufficient and are > thus used in quite a number of applications. > > Regards, Jens Okay, that explanation makes sense. Thanks,
From: spinoza1111 on 14 Apr 2008 00:08 On Apr 14, 2:54 am, Chad <cdal...(a)gmail.com> wrote: > The question stems from the following url > > http://groups.google.com/group/comp.lang.c/browse_thread/thread/dbb7f... > > I had asked what was the difference between what' the difference > between random numbers and pseudo-random numbers. I'm not that sure > I'm going to get a response since some of the posters are more > concerned about using colorful metaphors to describe each other. > > Usually when the vegititation in comp.lang.c starts to act like a > bunch of hyperactive 6th graders, they tend to ignore everyone else. > Maybe I will get a response on comp.lang.c. A computer is going to execute a deterministic algorithm OR take the least signficant digits of a time field. Neither operation is random. You need something physical such as a cute girl who grabs the first ball to drop out of a pressurized air chamber. The popular technique of taking the least significant digits of a time field is problematic because it depends on the "randomness" of when the program was started. This will be usually nonrandom (many programs for example are started up during working hours and although this doesn't affect the least significant digits of the time in many cases, many manual startups may occur at integral times such that the least significant digits of the time can be calculated from the deterministic time the code takes). And if a program is started automatically at a specific time then the least significant digits are predictable. Cf Donald Knuth's discussion in Seminumerical Algorithms. Randomness is related to encoding in cryptography. The digits of pi are "random" until the "enemy", trying to predict your "random" next digit, figures out you are merely expanding pi. Modern encoding of credit card numbers uses a pseudo-random key calculation which is merely computationally hard but will stop being so if mathematicians figure out "Riemann", and when they do, all financial systems will collapse overnight (if this happens soon, it is sure to interact in synergy with the current credit collapse to make a real mess). The preceding paragraph is a layperson's view, so we'll let the other sixth-graders here chime in and tell you that I'm a booger and wrong too.
From: spinoza1111 on 14 Apr 2008 00:31
On Apr 14, 4:06 am, Chad <cdal...(a)gmail.com> wrote: > On Apr 13, 12:40 pm, j...(a)toerring.de (Jens Thoms Toerring) wrote: > > > > > > > Chad <cdal...(a)gmail.com> wrote: > > > The question stems from the following url > > >http://groups.google.com/group/comp.lang.c/browse_thread/thread/dbb7f.... > > > I had asked what was the difference between what' the difference > > > between random numbers and pseudo-random numbers. I'm not that sure > > > I'm going to get a response since some of the posters are more > > > concerned about using colorful metaphors to describe each other. > > > Usually when the vegititation in comp.lang.c starts to act like a > > > bunch of hyperactive 6th graders, they tend to ignore everyone else. > > > Maybe I will get a response on comp.lang.c. > > > Sorry, but perhaps you should try to understand a bit more what > > different news groups are meant for before you call someone a > > "hyperactive 6th grader" - they have different topics or we > > wouldn't need them at all. And the difference between random > > numbers and pseudo-random numbers isn't anything related in > > any way to the topic of comp.lang.c, the C programming lan- > > guage. On the other hand comp.programming has a much broader > > topic, so it's on topic here, so here we go: > > I still have some general misconceptions about how the newsgroups > work. Many have been dominated by computer thugs who can't publish their views anywhere else and use bullying to take over a newsgroup. They typically present themselves as a majority but when you count their actual supporters you realize that a very small and very meaningless integer count of supporters is calling itself a majority. For example, you cannot say anything critical about C here without the C kiddies jumping you. Nor can you question the wisdom of the C standard because a member of the team posts here and is oversensitive about the effort, which was a foolish effort (in my view) to make a silk purse out of a sow's ear by calling anything difficult undefined, and blaming the victim...actual programmers trying to use C. They typically "gang up" on posters they think are female, or African- American, as well as posters who can write and have done enough homework to make an original contribution. They inappropriately generalize from details, "errors" being their favorite theme. Outside of usenet, published computer authors, especially those who have a breezy style, are a favorite topic. The threats made by these types of people caused well-known Java author Kathy Sierra to cancel public appearances, and recently I succeeded in getting a neutral party at wikipedia to repair an article on well-known C/C++/C Sharp author Herb Schildt to remove links to turgid attacks on Herb's "errors"...pseudo-scientific attacks which clearly showed an interest, not in scientific or technical truth, but in destroying a man's reputation...behavior not characteristic at all of real professionals. I'm Edward Nilges, the author of "Build Your Own .Net Language and Compiler". Stand by for the explosion of infantile rage, because much of Fascism is simply the return to the nightmare of childhood. > > > > > > > Random numbers are numbers that are truely random - there's > > no way you could ever predict the next one when you already > > have gotten some from the generator and even if you got all > > the specifications for the generator. Think e.g. of a gene- > > rator that gets it's numbers from a source that's known to > > be strictly random (throwing a coin, radioactive decay etc.). > > > In contrast, pseudo-random numbers are generated via some > > algorithm. One example of such a generator often used to > > create pseudo-random numbers employs a rather simple for- > > mula like > > > x[i+1] = trunc(a * x[i] + b) > > > where trunc() returns the digits after the decimal point. > > Obviously, the numbers generated this way aren't really > > random, and once you know the formula and 'a' and 'b' you > > can easily predict with absolute certainty what x[i+1] will > > be from x[i] (and if you only have a lot of consecutive > > number you still might be able to work out what the algo- > > rithm and its parameters are). > > > Of course, you can use some more complicated formulas, > > but that doesn't change anything about the basic non- > > randomness of the numbers created (and, if I remember > > the chapter of Knuth's "The Art of Computer Programming" > > about pseudo-random generators correctly, trying to get > > too clever can actually make things worse;-) > > > Creating really random numbers is difficult since you need > > some physical process that's truely random to produce them, > > so you need some extra hardware (some CPUs nowadays come > > with something like that, probably taking advantage of the > > randomness of Schottky noise). Pseudo-random numbers, on > > the other hand, are simple, cheap and fast to produce and > > for many purposes they're completely sufficient and are > > thus used in quite a number of applications. > > > Regards, Jens > > Okay, that explanation makes sense. Thanks,- Hide quoted text - The discussion is in Seminumerical Algorithms. Knuth cheerfully admits to an error (showing an outdated manhood missing here) when in the early days he developed an elaborate "random number generator" that produced perfectly predictable patterns. The distance between primes will be "random" until "Riemann" is solved, I believe, but when you display it, it doesn't look random, which is very interesting. > > - Show quoted text -- Hide quoted text - > > - Show quoted text - |