🚀 go-pugleaf

chrisw@leland.stanford.edu (Chris Wilkins) wrote:
>In article <4745p3$62h@umbc9.umbc.edu> schlein@umbc.edu (Jonas J. Schlein) writes:
>
>>>> I have 13 array locations. Let's call them: n[0], n[1], n[2],....,
>>>> n[12]. I want to RANDOMLY fill those locations with integers
>>>> from 0 to 12. 
>
>>> Why not do something like this:
>>> 
>>> for(i = 0; i < 13; i++)
>>>    n[i] = i;
>>> 
>>> for(i = 0; i < 13; i++) {
>>>    s = random_number_from_0_to_12();
>>>    swap(n, i, s);
>>> }
>
>>Actually if you go strictly by what my Algorithms Professor says is a
>>better way (he's a crypto man so I won't argue with him on that) for a
>>random distribution would be to replace your random_number_from_0_to_12()
>>with random_number_from_0_to_i(). Also you can then start the second
>>for loop at 1 to avoid the unnecessary initial sway of n[0] with itself.
>
>This sounds seriously wrong. I think what you meant is that you count DOWN 
>through the array swapping each element with one of the remaining elements or 
>itself (with equal probabilities). All this was covered in the shuffling 
>algorithm thread a couple of days ago.
>
>Chris
All too dangerous: just fill an extra array with N random numbers, and sort.
A bit decadent to sort on random, but very robust.
like :
*/
typedef struct iter_ {
	int num;
	int val;
	} ITER;
ITER array[N};
for(i=0; i <N; i++)	{
	array[i].num = i; /* or i+1 if you want ;-) */
	array[i].val = some_random();
	}
int iter_cmp(ITER *l, ITER *r)
{ if (l->val == r->val) return 0;
else return l->val - r->val; /* or the other way around */
}
qsort( array, N , sizeof array[0] , iter_cmp ); /* too lazy to cast ... */

for(i=0;i< N; i++)
	printf("%d : %d\n", i, array[i].num );

/*
Yours randomly,
Adriaan van kessel