Re: Parallel program to calculate PI

Richard Heathfield wrote:
> Spiros Bousbouras said:
> 
>> On 13 May, 17:00, Richard Heathfield <r...@[EMAIL PROTECTED]
> wrote:
>>> Spiros Bousbouras said:
>>>
>>> <snip>
>>>
>>>> And of course you must know that your random numbers
>>>> will be uniformly distributed within the square.
>>> That's easy. Use the following random point generator:
>>>
>>> #include <assert.h>
>>>
>>> void rndpt(unsigned long int *x,
>>>            unsigned long int *y,
>>>            unsigned long int max) /* max = side of square */
>>> {
>>>   static unsigned long int n = 0;
>>>   assert(x != NULL && y != NULL);
>>>   *x = n % max;
>>>   *y = n++ / max;
>>>   n %= max;
>>>   return;
>>>
>>> }
>>>
>>> If you call this i * max times, where max is a constant and i is an
>>> integer, the distribution of the random points will be uniform.
>> Since *y will always get the value 0 I don't think
>> so.
> 
> Since it won't, I do.

     Try running it for a while and making a histogram of
the y outputs.

     (The bug is in the line preceding the return.)

-- 
Eric.Sosman@[EMAIL PROTECTED]