Basic Random Numbers
by bkenwright@xbdev.net
Have you ever wondered how srand() and rand() work in your stdlib.h C
library's? Well I'll show you how in just a sec, and you'll see just how
easy it is...and of course how insecure it is. But for most cases, rand()
is perfect....
rand() in the stdlib.h C library uses the "Generator to linear congruenze"
rule... and uses the formula:
where m would be 231-2 for a 32 bit machine, a and b
are constants.
| code:
code.cpp |
|
//------------------------------------------------------------//
// Pseudo-random number generation //
//------------------------------------------------------------//
static
unsigned long
int next = 1;
int
rand(void)
{
next =
next * 1103515245 + 12345;
return (unsigned
int)(next/65536) % 32768;
}
void
srand(unsigned int
seed)
{
next =
seed;
}
//------------------------------------------------------------//
// Program Entry Point //
//------------------------------------------------------------//
void
main()
{
printf("Random Numbers:\n");
srand(1);
for(int i=0;
i<10; i++)
{
int iRand = rand();
printf("%d \n", iRand);
}//
End for loop i
}//
End of main()
|
| Output: |
Random Numbers:
16838
5758
10113
17515
31051
5627
23010
7419
16212
4086 |
As you can see where generating a pseudo random number, which is
different each time in our loop. Its interesting to note, that if you run
the program a second time, you'll get an identical output as you'd
expect...which is why you would usually seed your random number generator with
some rand value to start of with, for example the time: "srand( (unsigned)time(
NULL ) );" which would use <time.h> from the standard C library's.
Now to proove that the C code I've shown above is just as good
as the one in the <stdlib.h> one...I'll repeat the experiment with the one
included in the standard C library's...and see what result I get:
| code:
code_stdlib.cpp |
|
#include
<stdio.h> // need this for printf(..)
#include
<stdlib.h> // for our srand() and rand()
//------------------------------------------------------------//
// Program Entry Point //
//------------------------------------------------------------//
void
main()
{
printf("Random Numbers using the <stdlib.h> C algorithm:\n");
srand(1);
for(int i=0;
i<10; i++)
{
int iRand = rand();
printf("%d \n", iRand);
}//
End for loop i
}//
End of main() |
| Output: |
Random Numbers using the <stdlib.h> C algorithm:
41
18467
6334
26500
19169
15724
11478
29358
26962
24464 |
Acckkk...its different than the value we have for our random
numbers...why?... well its simply a difference of the values that your compiler
has used... I'm using the Visual Studio compiler and, if I open up the rand.c
file deep..deep within the installation I find this:
|
int
__cdecl rand (void)
{
return(((holdrand = holdrand * 214013L +
2531011L) >> 16) & 0x7fff);
} |
See its not that much more different than ours.... but its named
its seed value different and used a different set of constant variables.
Now a few things to note about the improved Visual C
version..its using simple optimisation tricks...such as binary shift instead of
multiplication :) And if the modulous is a power of 2, then you can use a
AND operation etc...which I urge you to look into, as its a great tip to know.
A possible further optimisation would be maybe to make it inline...or use _fastcall
instead of __cdecl.
| 
