Skip to content

Latest commit

 

History

History

1712

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 

A split of an integer array is good if:

  • The array is split into three non-empty contiguous subarrays - named left, mid, right respectively from left to right.
  • The sum of the elements in left is less than or equal to the sum of the elements in mid, and the sum of the elements in mid is less than or equal to the sum of the elements in right.

Given nums, an array of non-negative integers, return the number of good ways to split nums. As the number may be too large, return it modulo 109 + 7.

 

Example 1:

Input: nums = [1,1,1]
Output: 1
Explanation: The only good way to split nums is [1] [1] [1].

Example 2:

Input: nums = [1,2,2,2,5,0]
Output: 3
Explanation: There are three good ways of splitting nums:
[1] [2] [2,2,5,0]
[1] [2,2] [2,5,0]
[1,2] [2,2] [5,0]

Example 3:

Input: nums = [3,2,1]
Output: 0
Explanation: There is no good way to split nums.

 

Constraints:

  • 3 <= nums.length <= 105
  • 0 <= nums[i] <= 104

Related Topics:
Binary Search

Solution 1. Binary Search

Turn array A into its prefix sum array.

Let i be the last index of the left part. So A[i] is the sum of the left part.

For each i, we use binary search to find the range of the last index of the mid part.

For mid >= left, we have left + mid >= 2 * left, so we need to find the first prefix sum that is >= 2 * left.

For mid <= right, we have mid <= total - left - mid, so mid <= (total - left) / 2, so we need to find the last prefix sum that is <= left + (total - left) / 2.

Note that the valid range of the last index of the mid part is [i + 1, N - 1), so we need to make sure our binary search is searching within this range.

// OJ: https://leetcode.com/problems/ways-to-split-array-into-three-subarrays/
// Author: github.com/lzl124631x
// Time: O(NlogN)
// Space: O(1) if we are allowed to change the input array; otherwise O(N)
class Solution {
public:
    int waysToSplit(vector<int>& A) {
        long N = A.size(), mod = 1e9 + 7, ans = 0;
        partial_sum(begin(A), end(A), begin(A));
        for (int i = 0; i < N - 2; ++i) {
            int j = lower_bound(begin(A) + i + 1, end(A) - 1, A[i] * 2) - begin(A);
            int k = upper_bound(begin(A) + j, end(A) - 1, A[i] + (A.back() - A[i]) / 2) - begin(A);
            ans = (ans + k - j) % mod;
        }
        return ans;
    }
};

Solution 2. Three Pointers

Turn array A into its prefix sum array.

Let i be the last index of the left part. So A[i] is the sum of the left part.

Given i, the last index of the mid part is a range. Let it be [j, k).

Key Point: When we increment i, j and k must increase monotonically.

To find j, we can increment j from i + 1 until left <= mid i.e. A[j] - A[i] >= A[i].

To find k, we can keep incrementing k from j while it's valid -- i.e. k <= N-2 and mid <= right (A[k]-A[i] <= A[N-1]-A[k])

For each i, we add k - j to the answer.

Note that the valid range of the last index of mid is [i + 1, N - 2].

// OJ: https://leetcode.com/problems/ways-to-split-array-into-three-subarrays/
// Author: github.com/lzl124631x
// Time: O(N)
// Space: O(1) if we allow changing the input array; otherwise O(N)
class Solution {
public:
    int waysToSplit(vector<int>& A) {
        partial_sum(begin(A), end(A), begin(A));
        long mod = 1e9+7, ans = 0, N = A.size(), i = 0, j = 0, k = 0;
        for (; i < N; ++i) {
            j = max(i + 1, j); // `j` is at least one greater than `i`.
            while (j <= N - 2 && A[j] - A[i] < A[i]) ++j; // find the smallest `j <= N - 2` that satisfies `mid >= left`
            if (j > N - 2) break; // No room for the right part. Break
            k = max(k, j); // `k` is at least the same as `j`
            while (k <= N - 2 && A[k] - A[i] <= A.back() - A[k]) ++k; // keep incrementing `k` while it's valid -- `k <= N - 2` and `mid <= right`
            ans = (ans + k - j) % mod;
        }
        return ans;
    }
};

Or, we can calculate the sums on the fly without using prefix sums. In this way, we get real O(1) space.

// OJ: https://leetcode.com/problems/ways-to-split-array-into-three-subarrays/
// Author: github.com/lzl124631x
// Time: O(N)
// Space: O(1)
class Solution {
public:
    int waysToSplit(vector<int>& A) {
        long N = A.size(), mod = 1e9 + 7, left = 0, total = accumulate(begin(A), end(A), 0L), mid1 = A[0], mid2 = A[0], ans = 0;
        for (int i = 0, j = 0, k = 0; i < N - 2; ++i) {
            left += A[i];
            mid1 -= A[i];
            mid2 -= A[i];
            while (j <= i) mid1 += A[++j];
            while (j <= N - 2 && mid1 < left) mid1 += A[++j];
            if (j > N - 2) break;
            while (k < j) mid2 += A[++k];
            while (k <= N - 2 && mid2 <= total - mid2 - left) mid2 += A[++k];
            ans = (ans + k - j) % mod;
        }
        return ans;
    }
};

Discuss

https://leetcode.com/problems/ways-to-split-array-into-three-subarrays/discuss/999795/C%2B%2B-3-Pointers-O(N)-Time-O(1)-Space