perfect randomness?

**Cat** · 01-18-2004

Originally posted by Lurker
345
7
19
44
.5
56,023.3
3.3

That isn't truly random? I would love to see a pattern for that.

Here's an equation for those numbers:

-467.25x^6 + 7475.1x^5 - 44369x^4 + 121333x^3 - 150883x^2 + 66573x + 345

Just use x=0,1,2,3,4,5,6 and you get exactly those numbers out.

**Zach L.** · 01-18-2004

I program with random numbers pretty often... I haven't seen any problems with using rand() yet... as long as you seed it, it should be fine... unless there's some kind of portablitiy problem or something...

It is not portability problems (though, rand() isn't guaranteed to be defined the same for all implementations). The problem is that the distribution of the numbers and other related statistics are generally poor (again, it's not the same for every implementation). For scientific or security applications, rand() is generally a poor choice because you may need a certain distribution, or you may need more randomness. If I remember correctly, many rand() implementations have relatively high correlation in the low order bits (that is, those bits are "less random").

**tu_user** · 01-18-2004

what is eexactly the diff btw rand() and srand()?

**WaltP** · 01-18-2004

Originally posted by Lurker
345
7
19
44
.5
56,023.3
3.3

That isn't truly random? I would love to see a pattern for that.

But they weren't calcuated by the rand() function. That's what is being discussed here.

**tu_user** · 01-18-2004

what is seed? and diff btw rand() and srand()?

**major_small** · 01-18-2004

srand() seeds the random number generator... usually you see srand(time(0)); near the top of a program that uses rand(). rand() returns a 'random' integer, and srand() seeds it... I'm not too sure how rand() works... I've heard different stories about it, but I think it uses the seed in some kind of calculation to get to the random number or something... IDK... using time(0) uses the current time as the seed, so it will be more random...

**Zach L.** · 01-18-2004

rand() is generally a linear congruential generator. It works basically like this:

x[i + 1] = (a * x[i] + c) % m

Where a, c, and m are constants, and the x's are the "random" numbers.

**major_small** · 01-18-2004

but where are the 'random' numbers comming from in the first place... and where are the constants defined?

**tu_user** · 01-18-2004

Thanku. Its a very complicated topic.

**Zach L.** · 01-18-2004

The constants are implementation defined. The first x is the seed value, which is then replaced by the next x. I am doing a horrible job explaining this in English. Here is some example code (note, the constants are not chosen for any particular reason, so it will be a bad generator):

Code:

unsigned int seed;

void srand(unsigned int s)
{
  seed = s;
}

unsigned int rand()
{
  seed = (37 * seed + 8) % 247;
  return seed;
}

**tu_user** · 01-18-2004

Well I was trying to generate some random numbers here but it seems that the values generated by the following code are not within my defined limit?

Code:

#include<iostream>
#include<conio.h>

using std::cout;
using std::cin;
using std::endl;

void genNum(void);	// prototype

int main()
{
	genNum();	// function call
}
void genNum()	// definition
{
	int value1, value2, result, sum;	// declare alias
	
	cout<< value1 << "+" << value2 << "=\t";	// prompt the user for input
	cin>> result;	// read input
	value1=1+rand()%6;	// generate random integer 1-10
	value2=1+rand()%6;	// generate random integer 1-10
	sum=value1+value2;	// calculate sum

	if(sum==result)
	{
			cout<< "Congratulations! correct";
	}
	else
			cout<< "No. Please try again";
}

**CornedBee** · 01-19-2004

BTW, for some reason (I think backwards compatibility) the MS implementation of rand is something like

Code:

unsigned int rand()
{
  seed = (calculation);
  return seed & 0x8fff;
}

**Stoned_Coder** · 01-19-2004

Use the Boost library it has random number generators for all occaisions including MT my personal favourite for simulations.

**cyberCLoWn** · 01-19-2004

What's wrong with the following:

Code:

#include <cstdlib>
#include <iostream>
using namespace std;

int main()
{
   int x;
   
   // Gives the rand() function a seed based on the time
   srand( time( 0 ) );
   
   for( int i = 1; i <= 10; i++)
      cout << ( 1 + rand() % 100 ) << '\n'; // shifted and scaled integers produced by 1+rand()%100
                                            // number range from 1 - 100 
   
   cin.get();
   return 0;
}