Rand and Floor

[EpsilonOmega]

Hello,
I'm having some problem with the floor and rand functions.
What I want to do is create a random number 1-6 and make sure it's a hole number.
Right now I'm using:

d1 = rand()%6 +1;
d1 = floor(d1);

But it gives me a warning for changing 'int' from 'double'
Anyone know why and/or a better way?
If you could explain to me in dpeth that would be best.

EO

[helios]

floor() is a function designed to truncate (well, it's not quite truncation, but it's close enough) a floating point number d1 is apparently an int, so it makes no sense to call floor() on it. Plus, rand() will never return a fractional. Plus, % automatically truncates fractional operands.

[Hammurabi]

Yeah, you're misunderstanding both rand() and %.

rand() returns an integer from 0 to RAND_MAX (often a ridiculously low number indicating a small period and therefore a bad random function).

% returns the integer remainder after division.

So you definitely don't need floor here. Everything's an int already. But if you did need it, you would have to cast it's result back to an int:

d1 = (int) floor(d1);

[EpsilonOmega]

Ok, then. Also I found this code for a 'better' random number:

     int randomNumber(int hi)  //the correct random number generator for 0-hi
     {
          int value=((float)rand()/RAND_MAX)*(hi+1);
          if(value>hi)
          {
               value = 0;
          }
          return value;
     }

What does this do, how is it better and how do I use it?

[jsmith]

1) It generates random numbers, per the comment
2) The only way that it is better is that it generates a uniform distribution for all inputs hi, whereas
rand() % hi does not (for all inputs).
3) Call it; the comment says what it does.

[EpsilonOmega]

So I'd call it by:

randomNumber (6);

right?

[jloundy]

#include <ctime>

srand((unsigned)time(0))// sets the rand seed to the clock so its random every time
int rand_NUM = (rand()%6)+1; // creates random number 1 - 6 

[EpsilonOmega]

Wait; what does that do?

[jsmith]

It does what it says it does. The first line initializes the random number generator using (hopefully) a different starting seed. The second line generates a random number in the range [ 1, 6 ].

However there is a very slight bias in the numbers it generates due to the fact that MAX_INT is not evenly divisible by 6. Though for any application where rand() is sufficient the bias will not be a problem.

[EpsilonOmega]

So would I have to call line 3 every time I wanted to create a random number. Or just once?

Also could someone describe what everything does? I would prefer to use code that I know about, than to paste random snippets that might not work.

[jsmith]

Call srand() exactly once in your program.

srand() seeds the random number generator with the given value.

rand() implements an LCRNG (linear congruential random number generator). An LCRNG is one which has the formula:

RN(i) = ( c1 * RN(i-1) + c2 ) mod p

where c1 and c2 are carefully chosen constants (you have no control over them). p is 2^32 for 32-bit machines. RN(k) is the kth random number. RN(0) is the seed. srand() sets RN(0).

rand() then simply runs the above formula once, generating RN(i).

[EpsilonOmega]

Thanks, this helps a lot.