Random Numbers

[spudtheimpaler]

Hey, A uni project I'm on basically requires us to design a mars rover simulation. All command line. I'm on this function that actually creates the map, and am trying to set up whether or not the rover can move over each square on the grid. I've got a rand() in there with 1's or 0's, IE yes or no, but it's giving me a fairly even playing field, i was wondering if there was any way in biasing the results so, say, it was twice as likely to be a 1 than a 0.? Cheers fo any help (or any ideas of any kind )

code's below so far, though its pretty standard

Cheers

Mitch

P.s. - there are to be a number of objects on the land, be they rocks, pools, aliens, etc, thats what the ObjId is

P.p.s, i have thought about for example having the rand make four variables, and having 0, 1 and 2 equal passable land and 3 equalling non-passable land, was just looking to a better, more c-based way.

void make_map(int mapsizex, int mapsizey)
{

	location **map;

	int row, col;

	//create the map, dynamically allocating memory for each location

	//allocate memory for rows
    
	map=(location**)(malloc((mapsizex)*sizeof(location*)));
    
	//allocate memory for each row of the matrix
    
	for(row=0;row<(mapsizex);row++)
		map[row]=(location*)(malloc((mapsizey)*sizeof(location)));


	//Matrix of Structures complete

	  srand(time(NULL));
	  

	  for(row=0; row<mapsizex; row++)
		  for(col=0; col<mapsizey; col++)
			  map[row][col].ObjId=rand()%18;


	  for(row=0; row<mapsizex; row++)
		  for(col=0; col<mapsizey; col++)
			  printf("%d ", map[row][col].ObjId);

      printf("\n\n\n");
		  
	  for(row=0; row<mapsizex; row++)
		  for(col=0; col<mapsizey; col++)
			  map[row][col].Movable=rand()%2;

 	  for(row=0; row<mapsizex; row++)
		  for(col=0; col<mapsizey; col++)
			  printf("%d ", map[row][col].Movable);

}

[Thantos]

Ok say you want to bias 1 so its twice as likely as 0 to appear.

Instead of ranging your randoms from 0 to 1, change it to 0 to 2. If its >= 1 return 1 if not return 0.

Also FYI:

In Numerical Recipes in C: The Art of Scientific Computing
(William H. Press, Brian P. Flannery, Saul A. Teukolsky,
William T. Vetterling; New York: Cambridge University
Press, 1992 (2nd ed., p. 277)), the following comments are
made:
"If you want to generate a random integer between 1
and 10, you should always do it by using high-order
bits, as in

j=1+(int) (10.0*rand()/(RAND_MAX+1.0));

and never by anything resembling

j=1+(rand() % 10);

(which uses lower-order bits)."

From the man page for rand()

Edit: Forgot to mention to do what I mentioned you'll need to add a function that gets a random number.

[Dave_Sinkula]

Originally posted by spudtheimpaler
I've got a rand() in there with 1's or 0's, IE yes or no, but it's giving me a fairly even playing field, i was wondering if there was any way in biasing the results so, say, it was twice as likely to be a 1 than a 0.? Cheers fo any help (or any ideas of any kind )

Here is a quick hack of an idea.

#include <stdio.h>
#include <stdlib.h>
#include <time.h>

int foo(void) /* even odds */
{
   return rand() > RAND_MAX / 2;
}

int bar(void) /* twice as likely to be 1 as 0 */
{
   return rand() > RAND_MAX / 3;
}

void test(int (*function)(void), int cycles)
{
   int i, result[2] = {0,0};
   double x,y;
   for ( i = 0; i < cycles; ++i )
   {
      ++result[function()];
   }
   x  = result[0] * 100.0 / cycles;
   y  = result[1] * 100.0 / cycles;
   printf("0: %d (%.2f%%), 1: %d (%.2f%%)\n", result[0], x, result[1], y);
}

int main(void)
{
   srand(time(0));
   test(foo, 10000);
   test(bar, 10000);
   return 0;
}

/* my output
0: 4979 (49.79%), 1: 5021 (50.21%)
0: 3302 (33.02%), 1: 6698 (66.98%)
*/

[edit]A different spin on the above.

#include <stdio.h>
#include <stdlib.h>
#include <time.h>

int likelihood(int difficulty)
{
   return rand() < RAND_MAX / (difficulty + 1);
}

void test(int (*function)(int), int n)
{
   int i, result[2] = {0,0}, cycles = 1000000;
   printf("likelihood = 1:%d |", n);
   for ( i = 0; i < cycles; ++i )
   {
      ++result[function(n)];
   }
   for ( i = 0; i < sizeof result / sizeof *result; ++i )
   {
      double percent = result[i] * 100.0 / cycles;
      printf(" %d = %6d (%5.2f%%)%c", i, result[i], percent, ",\n"[i]);
      fflush(stdout);
   }
}

int main(void)
{
   int i;
   srand(time(0));
   for ( i = 1; i < 10; ++i )
   {
      test(likelihood, i);
   }
   return 0;
}

/* my output
likelihood = 1:1 | 0 = 499869 (49.99%), 1 = 500131 (50.01%)
likelihood = 1:2 | 0 = 666451 (66.65%), 1 = 333549 (33.35%)
likelihood = 1:3 | 0 = 749825 (74.98%), 1 = 250175 (25.02%)
likelihood = 1:4 | 0 = 800192 (80.02%), 1 = 199808 (19.98%)
likelihood = 1:5 | 0 = 833687 (83.37%), 1 = 166313 (16.63%)
likelihood = 1:6 | 0 = 856950 (85.69%), 1 = 143050 (14.30%)
likelihood = 1:7 | 0 = 875569 (87.56%), 1 = 124431 (12.44%)
likelihood = 1:8 | 0 = 888736 (88.87%), 1 = 111264 (11.13%)
likelihood = 1:9 | 0 = 899771 (89.98%), 1 = 100229 (10.02%)
*/

[/edit]

[WaltP]

Originally posted by Thantos
Also FYI:

In Numerical Recipes in C: ... the following comments are made:
"If you want to generate a random integer between 1 and 10, you should always do it by using high-order bits, as in
j=1+(int) (10.0*rand()/(RAND_MAX+1.0));

and never by anything resembling

j=1+(rand() % 10);

(which uses lower-order bits)."

From the man page for rand()

What I dislike about this type of suggestion is the requirement of blind acceptance. Why this is better should be part of the explanation. If someone asks you why you did it that way, and unacceptable reason to me is "the book said so."

[Thantos]

The reason why is in the quote, using rand() % 10 only uses the low order bits while the other method uses the high order bits. If someone wants it explained further they can do more research on it.

I posted it as mearly informational.

[spudtheimpaler]

Thank you all

Cheers Thantos, Thats prob what i'll end up doing (i think i may have mentioned that method in my pps though).

As for all the high end low end stuff, I've barely been programming 6 months, and its a uni course IE a couple of hours a week. Its interesting seeing that sort of info and I'm glad i know it though i couldn't possibly include it. My code has to be transportable and be able to be combined in another persons project. I think i'd blow them away

Cheers for your help guys

Mitch