Time Limit: 6000/3000 MS (Java/Others)    Memory Limit: 262144/262144 K (Java/Others)
Total Submission(s): 707    Accepted Submission(s): 245

By YJQ我们令dp[i][j]表示在前i个数中，选出若干个数使得它们的gcd为j的方案数，于是只需要枚举第i+1个数是否被选中来转移就可以了

令第i+1个数为v，当考虑dp[i][j]的时候，我们令$dp[i+1][j] += dp[i]j,dp[i+1][gcd(j,v)] += dp[i]j

复杂度O(N*MaxV) MaxV 为出现过的数的最大值

其实有O(MaxV *log(MaxV))的做法，我们考虑记f[i]表示从这些数中选择若干个数，使得他们的gcd是i的倍数的方案数。假如有K个数是i的倍数，则f[i]=2^K-1，再用g[i]表示从这些数中选择若干个数，使得他们的gcd是i的方案数，则g[i]=f[i] - g[j] (对于所有j是i的倍数)。

由调和级数可以得到 复杂度为O(MaxV *log(MaxV))；

Problem Description

CA is a fine comrade who loves the party and people; inevitably she loves GCD (greatest common divisor) too.
Now, there are N different numbers. Each time, CA will select several numbers (at least one), and find the GCD of these numbers. In order to have fun, CA will try every selection. After that, she wants to know the sum of all GCDs.
If and only if there is a number exists in a selection, but does not exist in another one, we think these two selections are different from each other.

Input

First line contains T denoting the number of testcases.
T testcases follow. Each testcase contains a integer in the first time, denoting N , the number of the numbers CA have. The second line is N numbers.
We guarantee that all numbers in the test are in the range [1,1000].
1≤T≤50

Output

T lines, each line prints the sum of GCDs mod 100000007 .

Sample Input

Sample Output

Source

BestCoder Round #78 (div.2)

Recommend

wange2014   |   We have carefully selected several similar problems for you:   5659  5658  5657  5654  5653

复杂度O(N*MaxV) 的做的法：

#include <iostream>
#include <cstdio>
#include <cstring>
#include <algorithm>
#define LL long long
using namespace std;
LL dp[1010][1010];//dp[i][j]表示在前i个数//中，选出若干个数使得它们的gcd为j的方案数
int A[1010];
int gcd[1010][1010];
const LL MOD=1e8+7;
int __gcd(int x,int y)
{return (x!=0)?__gcd(y%x,x):y;
}
int main()
{int n,t;cin>>t;for(int i=1;i<=1000;i++){for(int j=1;j<=1000;j++){gcd[i][j]=__gcd(i,j);}}while(t--){cin>>n;memset(dp,0,sizeof(dp));for(int i=0;i<n;i++){scanf("%d",&A[i]);}sort(A,A+n);//可以确保当前数为所遍历的最大值；for(int i=0;i<n;i++){for(int j=1;j<=A[i];j++){(dp[i+1][j]+=dp[i][j])%=MOD;(dp[i+1][gcd[j][A[i]]]+=dp[i][j])%=MOD;}dp[i+1][A[i]]++;}LL ans=0;for(int i=1;i<=1000;i++)(ans+=dp[n][i]*i)%=MOD;printf("%I64d\n",ans);}return 0;
}

O(MaxV *log(MaxV))的做法：

#include <cstdio>
#include <algorithm>
#include <cstring>
#include <iostream>
#include <cmath>
#include <queue>
#include <stack>
#define LL long long
using namespace std;
LL dp[1005];
LL cnt[1005];
const LL MOD=1e8+7;
LL power(LL x,LL n)
{LL ans=1;while(n){if(n&1){(ans*=x)%=MOD;}n>>=1;(x*=x)%=MOD;}return ans;
}
int main()
{int x;int t,n;int maxn;scanf("%d",&t);while(t--){memset(dp,0,sizeof(dp));memset(cnt,0,sizeof(cnt));maxn=0;scanf("%d",&n);for(int i=0;i<n;i++){scanf("%d",&x);maxn=max(x,maxn);cnt[x]++;}LL ans=0,c;for(int i=maxn;i>=1;i--){c=0;for(int j=i;j<=maxn;j+=i){c+=cnt[j];(dp[i]=dp[i]-dp[j]+MOD)%=MOD;}(dp[i]+=(power(2,c)-1)%MOD)%=MOD;(ans+=i*dp[i])%=MOD;}printf("%I64d\n",ans);}return 0;
}