# rand() for floats

• 11-24-2003
frenchfry164
rand() for floats
How do I do rand() for floats? Obviously this won't work:
Code:

`float a = (float)rand() % 5.2;`
So how do I do it?
• 11-24-2003
JaWiB
Code:

`float R = (float)((rand()%100)/10)`
• 11-24-2003
frenchfry164
That kept giving me whole numbers for some reason. I don't get it, ten doesn't evenly divide into all numbers.
• 11-24-2003
JaWiB
D'oh! Should've been:
Code:

`float R =  float(rand()%1000) / 100 ;`

Or actually probably should've used a C++ cast...but anyways...

Edit: I should mention that someone else probably knows a better way, but at least this works :D
• 11-24-2003
axon
Only way I can think of is to seed rand to a float. Try this:
Code:

```        srand(.5);         float R = (float)((rand()%100)/.55);```
this seemed to work for me. If you only divide by a tenth, it gives an int...play around with it.
• 11-24-2003
frenchfry164
thx dude!

*gives a straight huggle

*wait a minute, I'm a guy......saying huggle? Man, I got issues.

edit: Okay! I got this worked out! You don't even need to seed with a float, just seed it with time as normal. Here is my function:
Code:

```inline float FloatRand(float MaxVal) {     return float(rand()%100)/((100/MaxVal)+0.1); }```
If you're wondering what the "+0.1" is for at the end, after some testing, I found this makes it more random. Before it would just randomize to 2 decimal places, this lets itself max out the decimal places.
• 11-24-2003
XSquared
Code:

```float FloatRand( float MaxVal ) {         return ( (float)rand( ) / (float)RAND_MAX ) * MaxVal; }```
• 11-25-2003
Cat
Quote:

Originally posted by XSquared
Code:

```float FloatRand( float MaxVal ) {         return ( (float)rand( ) / (float)RAND_MAX ) * MaxVal; }```

This is indeed the best way.

Do note that this can SOMETIMES return MaxVal; so the range of numbers is [0,MaxVal] -- it includes BOTH endpoints. If this is not what you want, you need to check for that condition.

I mention this because often you want a number, for example, in the range [0,1) -- and a value of exactly 1 might break everything, once every RAND_MAX times you call the function. Very difficult to debug. In this case, you could use something like:

Code:

```float r; while ( (r = FloatRand(1)) == 1); // now r is in the range [0,1).```
• 11-25-2003
XSquared
Code:

```float FloatRand( float MaxVal ) {         return ( (float)rand( ) / ( (float)RAND_MAX + 1.0f ) ) * MaxVal; }```