I was reading the tutorial located here:

http://faq.cprogramming.com/cgi-bin/...&id=1073086407

and there appear to be some errors. Here is the relevant part:

Removing the std:: specifiers, the random number code for the first method is:Quote:

This code shows two methods to use the high order bits of rand() by calculating the high value of the range with RAND_MAX and then dividing the call to rand() by the result. The point where 1 is added to RAND_MAX is important to note, this forces the range of random numbers to be 1 through 10 instead of 0 through 9.Code:`#include <iostream>`

#include <cstdlib>

#include <ctime>

int main()

{

std::srand ( (unsigned int)std::time ( 0 ) );

std::cout<< std::rand() / ( RAND_MAX / 10 + 1 ) <<std::endl;

std::cout<< (int)( (double)std::rand() / ( RAND_MAX + 1 ) * 10 ) <<std::endl;

return 0;

}

rand() / (RANDMAX / 10 + 1)

Let's suppose RANDMAX is 500. If rand() returns 1, then the result is:

1 / (500 / 10 + 1) = 1/51

which because of integer arithmetic produces 0. As a consequence, if rand() returns any number less than 51, the answer is 0, and it creates the mapping:

0 - 50 ---> 0

If rand() returns 51, then the result is 1. Due to integer arithmetic, the result will continue to be 1 until the numerator reaches 102, giving this mapping:

0 - 50 -----> 0

51 - 101 --> 1

If you continue figuring out the ranges, you will come up with this:

0 - 50 -----> 0

51 - 101 --> 1

102 - 152 -> 2

153 - 203 -> 3

204 - 254 -> 4

..

..

..

459 - 500 -> 9

First of all, you can see that the numbers produced are 0-9, not 1-10 as the tutorial seems to imply. In addition, all the ranges of numbers contain 51 numbers except the range 459 - 500, which contains only 42 numbers. That means the number 9 is less likely to be produced by the calculation, which means the calculation will not produce random numbers between 0 and 9.

The second method seems to suffer from similar problems. If RAND_MAX is equal to 500 and rand() returns 1, then the calculation produces:

1.0 / 501 * 10 = .02

and when you cast that to an int, you get 0.

The ranges are a little better, but the first range has 51 numbers, while the other ranges have 50 numbers. That makes it more likely the number 0 will be produced over any other number.

0 - 50 -----> 0

51 - 100 --> 1

101 - 150 -> 2

151 - 200 -> 3

201 - 250 -> 4

..

..

..

451 - 500 -> 9

Also, when I test out the code, I get the same number everytime, which must mean the seed is the same everytime I run the code. Why doesn't that work?