给一个int类型数组,求出连续的子序列的最大和。
如:A[]={-1,1,2,-4,5,5,-2,5}, 最大子序列和为: 5+5+(-2)+5 = 13
方法一:暴力求解 ,时间复杂度O(n^3)
时间复杂度O(n^2)
public static int MaxSubsequenceSumOn2( int[] A) {
int len = A.length;
int thisSum = 0;
int maxSum = A[0];
for(int i = 0; i < len; i++){
thisSum = 0;
for(int j = i; j < len; j++){
thisSum += A[j];
if(thisSum > maxSum)
maxSum = thisSum;
}
}
return maxSum;
}
时间复杂度O(nlogn)
int Max3( int A, int B, int C )
{ /* 返回3个整数中的最大值 */
return A > B ? A > C ? A : C : B > C ? B : C;
}
int DivideAndConquer( int List[], int left, int right )
{ /* 分治法求List[left]到List[right]的最大子列和 */
int MaxLeftSum, MaxRightSum; /* 存放左右子问题的解 */
int MaxLeftBorderSum, MaxRightBorderSum; /*存放跨分界线的结果*/
int LeftBorderSum, RightBorderSum;
int center, i;
if( left == right ) { /* 递归的终止条件,子列只有1个数字 */
if( List[left] > 0 ) return List[left];
else return 0;
}
/* 下面是"分"的过程 */
center = ( left + right ) / 2; /* 找到中分点 */
/* 递归求得两边子列的最大和 */
MaxLeftSum = DivideAndConquer( List, left, center );
MaxRightSum = DivideAndConquer( List, center+1, right );
/* 下面求跨分界线的最大子列和 */
MaxLeftBorderSum = 0; LeftBorderSum = 0;
for( i=center; i>=left; i-- ) { /* 从中线向左扫描 */
LeftBorderSum += List[i];
if( LeftBorderSum > MaxLeftBorderSum )
MaxLeftBorderSum = LeftBorderSum;
} /* 左边扫描结束 */
MaxRightBorderSum = 0; RightBorderSum = 0;
for( i=center+1; i<=right; i++ ) { /* 从中线向右扫描 */
RightBorderSum += List[i];
if( RightBorderSum > MaxRightBorderSum )
MaxRightBorderSum = RightBorderSum;
} /* 右边扫描结束 */
/* 下面返回"治"的结果 */
return Max3( MaxLeftSum, MaxRightSum, MaxLeftBorderSum + MaxRightBorderSum );
}
int MaxSubseqSum3( int List[], int N )
{ /* 保持与前2种算法相同的函数接口 */
return DivideAndConquer( List, 0, N-1 );
}
时间复杂度O(n)
public static int MaxSubsequenceSum (int[] A) {
int len = A.length;
int thisSum = 0, maxSum = 0;
for(int i = 0; i < len; i++){
thisSum += A[i];
if(thisSum > maxSum)
maxSum = thisSum;
else if(thisSum < 0)
thisSum = 0;
}
return maxSum;
}
关键点是,连续和一旦小于0,清零之后,再往后加和 (因为要和最大嘛,前面的和小于0就没必要再加上前面子序列的和了)
测试用例:
static public void main(String[] args) {
int[] A = {-1,1,2,-4,5,5,-2,5};
// int[] A = {-1};
System.out.println(MaxSubsequenceSum(A)); //13
System.out.println(MaxSubsequenceSumOn2(A)); //13
}
方法三: 动态规划, 时间复杂度O(n)
用 dp[n
] 表示元素 n 作为末尾的连续序列的最大和,容易想到状态转移方程为dp[n] = max(dp[n-1] + num[n], num[n])
,从第 1 个元素开始,自顶向上求解:
c++:
int maxSubArray(vector<int>& nums) {
int* dp = new int[nums.size()]();
dp[0] = nums[0];
int result = dp[0];
for (int i = 1; i < nums.size(); i++) {
dp[i] = max(dp[i-1] + nums[i], nums[i]);
result = max(result, dp[i]);
}
return result;
}
这个问题当中,求解 dp[i]
只依赖 dp[i-1]
,因此可以使用变量来存储,简化代码:
int maxSubArray(int A[], int n) {
int result = INT_MIN;
int f = 0;
for (int i=0; i < n; i++) {
f = max(f + A[i], A[i]);
result = max(result, f);
}
return result;
}
参考: