Help with rand() definition [modified]

[Grimes]

Hello everybody I am looking for the rand() function definition, because I want to use it to create a similar function for a super secret project of mine :suss:. I have found a website on the Internet that claims the rand() function definition is something like:

unsigned long int next = 1;
int rand()
{
next = next * 1103515245 + 12345;
return (unsigned int)(next/65536) % 32768;
}

however if I use this function it does not give me the same value as the rand() function.

cout << (rand() %50) << "\n";
cout << (rand() %50) << "\n";

gives the first random value of 41 and second value of 17. (the srand() function was not used so these are the same every time I run the program) and when I use the definition I got from the website

long unsigned int Getal = 1;
Getal = (Getal*1103515245 + 12345);
cout << (Getal/65536) % 32768 % 50 << "\n";
Getal = (Getal*1103515245 + 12345);
cout << (Getal/65536) % 32768 % 50 << "\n";

the first output is 38 and the second 8. Maybe there is something wrong with my implementation, or can someone maybe help or tell me what the correct definition is. I am using Visual C++ 6 and use the rand() function found in the stdlib.h. Is the data types I used correct? Where can I find the definition in the stdlib.h? I tried looking there but there's only a line that says "_CRTIMP int __cdecl rand(void);" ?!? Anyway I would appreciate help thanks grimes

KOM UIT DAAAAA!!!

modified on Friday, June 12, 2009 1:04 PM

[Luc Pattyn]

Hi, I could come up with an infinite number of pseudo-random-number-generators; they all are periodic, i.e. after N calls they repeat the sequence; you would hope N to be large. The simple ones have a single state variable (next in your snippet) that is operated upon (e.g. with multiply and add; or with shift and exor) and result in a period of at most 2^B for a state variable holding B bits. I can't tell you which one is used in the standard libraries, maybe because that is super secret too. I suggest you Google rand.c to get a small collection. :)

Luc Pattyn [Forum Guidelines] [My Articles]

---

DISCLAIMER: this message may have been modified by others; it may no longer reflect what I intended, and may contain bad advice; use at your own risk and with extreme care.

---

[Lost User]

Grimes wrote:

I am looking for the rand() function definition

You failed to mention what compiler you are using. Your in luck, I have several installed. Visual Studio 6: C:\Program Files (x86)\Microsoft Visual Studio\VC98\CRT\SRC\RAND.C Visual Studio 2005: C:\Program Files (x86)\Microsoft Visual Studio 8\VC\crt\src\rand.c Visual Studio 2008: C:\Program Files (x86)\Microsoft Visual Studio 9.0\VC\crt\src\rand.c You could also take a look at the rand implementation in the GNU LibC library[^]. The rand functions are located in the \glibc-XX\stdlib folder. Best Wishes, -David Delaune

[Grimes]

The reason I said it is super secret is so that when the CIA/FBI/NSA kick down your door to arrest you as one of my accomplices, you can have deniability. I thought that there was a simple random function that most programmers, that are too lazy to make their own, use. Perhaps you know of a method to predict the next number of a randomizer function. This is why I wanted to know the definition of the rand() function so I can write a program to predict the next 'random' number in a series. but thanks anyway... by the way, after this conversation, your phone might be tapped...

KOM UIT DAAAAA!!!

[Grimes]

OK thank you. I have Visual Studio 6, and I found the elusive 'rand.c' file I will take a look at it, thanks for the help!

KOM UIT DAAAAA!!!

[Lost User]

If you use % (as in rand() % 50) on a random number, the distribution of its values will change - by a lot. I know that's not what you asked, so this isn't an answer, and for many uses of rand it doesn't even matter.. but it's something to keep in mind. Oh and % (1< can be replaced by `` & ((1< which is a lot faster - you probably knew that (everyone does) but I saw a `% 32768` ``

[Grimes]

I know that there will be many values of next for every possible random number. I calculated that for the random generator I have, that there will be almost 1311 values of 'next' for every random number. I got this by: (32768/50)*(65536/32768)=65536/50 = 1311. I was hoping to predict the next random number in a series by using these 1311 values to see which of those values would give the next random number. I order to do this I need the definition of the random function used and I also need x amount of previous random numbers in order to predict the next. It seems from everybodie's replies that there are many different random number generators, I was hoping that there was a common function used by most programmers. but thanks anyway...

KOM UIT DAAAAA!!!

[T2102]

Are you using this for encryption? If not, you might leverage mersenne twister's random //generates a random number on [0,0xffffffff]-interval The MT has a period of at least 2^19937 - 1 back a few years ago and is probably larger now. Mersenne Twister is basically for Monte-Carlo simulations - it is not cryptographically secure "as is". http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html[^] Are simply trying to set the seed on the rand() function or learn how to crack linear congruential generators?

modified on Saturday, June 13, 2009 5:03 AM

[Grimes]

This helps a LOT! I was looking for a number generator that can be so 'random' as this mersenne twister. I am looking at methods for cracking random number generators, do you know of any good source of literature on this subject? I have googled it, but don't know which info is good, and which is junk. I am also wondering how hard is it to crack a cryptographically secure number generator? Is it impossible or is it so difficult that it is impractical to even try. I mean, will you need thousands/millions or billions of iteration values before you can start to predict the next iteration? thanks

KOM UIT DAAAAA!!!

[T2102]

DES is not that secure for example. Web browser using 256 bit DES could be cracked in several hours using a cluster. When I was in college ten years ago, I read a book by Douglas R. Stinson; the latest version is http://www.amazon.com/Cryptography-Practice-Discrete-Mathematics-Applications/dp/1584885084/ref=sr_1_1?ie=UTF8&s=books&qid=1244886755&sr=8-1[^] You also will want to read about why Mersenne Twister is not secure; I am sure it's discussed on that project webpage or another site. The math is mainly Modern Algebra like Chinese Remainder Theorem and basic statistics.

modified on Saturday, June 13, 2009 6:11 AM