嗯，这算法应该对的，就是慢了点。

这个Judge现在有问题，全部的CODE都判成编译错，我觉得这个代码应该没问题

#include <stdio.h>

int main()
{
    int k,n;
    while(scanf(\"%d %d\",&k,&n)!=EOF){
        int res=0,t=1;
        while(n){
            if(n%2){
                res+=t;
            }
            t*=k;
            n/=2;
        }
        printf(\"%d\n\",res);
    }    
}    

/*---------------------------------------------------------------------------
File name: NKU1021-[NKPC2]?S???.c
Author: HJin (email: fish_sea_bird [at] yahoo [dot] com )
Created on: 8/12/2007 21:29:53
Environment: Windows XP Professional SP2 English +
Visual Studio 2005 v8.0.50727.762

Modification history:
===========================================================================

I think I discussed this problem with Leeco before. Bottom of line is:

write out the binary representation of n.

Following is my code submitted to Nankai online judge.
*/

long long b[] = {1, 3, 9, 27, 81, 243, 729, 2187, 6561, 19683, 59049, 177147, 531441, 1594323, 4782969, 14348907, 43046721, 129140163, 387420489, 1162261467, 3486784401, 10460353203, 31381059609, 94143178827, 282429536481, 847288609443, 2541865828329, 7625597484987, 22876792454961, 68630377364883, 205891132094649, 617673396283947};

int main()
{
int k;
unsigned n;

while( scanf("%u", &n) != -1 )
{
--n;
k=0;

while(n)
{
if(n&1)
printf("%lld\n", b[k]);

++k;
n >>= 1;
}
}

return 0;
}

这个算法应该没有错啊