Given a binary array nums, return the maximum number of consecutive 1’s in the array if you can flip at most one 0.

Thinking Process

Given a binary array nums, return the maximum number of consecutive 1’s in the array if you can flip at most one 0.

  • Define state: what subproblem does dp[i] (or dp[i][j]) represent?
  • Recurrence: how does the answer build from smaller indices?
  • Base cases first; optimize space if only prior row/layer is needed.
Sliding window a b c d e window expand right, shrink left when invalid

Common Approaches

Typical techniques for this pattern:

Approach Time Space Notes
1D DP (this problem) O(n) O(n) or O(1) Linear recurrence
2D DP O(nm) O(nm) or O(n) Grid or two-sequence problems
State machine DP O(n) O(1) Buy/sell, hold/not-hold states
Memoization (top-down) Same as DP O(n) Recursive + cache

Examples

Example 1:

Input: nums = [1,0,1,1,0]
Output: 4
Explanation: 
- If we flip the first zero, nums becomes [1,1,1,1,0] and we have 4 consecutive ones.
- If we flip the second zero, nums becomes [1,0,1,1,1] and we have 3 consecutive ones.
The maximum number of consecutive ones is 4.

Example 2:

Input: nums = [1,0,1,1,0,1]
Output: 4
Explanation: 
Flipping the zero at index 4 gives us [1,0,1,1,1,1] with 4 consecutive ones.

Constraints

  • 1 <= nums.length <= 10^5
  • nums[i] is either 0 or 1.

Algorithm Breakdown

Key Insight: Two-State DP

The solution uses two states to track different scenarios:

State 0 (dp0): Maximum consecutive ones ending at current position without flipping any 0

  • If current is 1: Extend sequence → dp0++
  • If current is 0: Reset sequence → dp0 = 0

State 1 (dp1): Maximum consecutive ones ending at current position with at most one flip

  • If current is 1: Extend sequence → dp1++ (no flip needed or already flipped)
  • If current is 0: Flip this 0, use previous dp0 → dp1 = dp0 + 1

State Transition Logic

class Solution:
    def findMaxConsecutiveOnes(self, nums):
        n = len(nums)
        rtn = 0
        dp0 = 0
        dp1 = 0

        for i in range(n):
            if nums[i]:
                dp1 += 1
                dp0 += 1
            else:
                dp1 = dp0 + 1
                dp0 = 0

            rtn = max(rtn, max(dp0, dp1))

        return rtn

Why This Works

  1. dp0 tracks pure sequences: Only counts consecutive 1s without any flips
  2. dp1 tracks sequences with one flip: Can use one flip to extend sequence
  3. When we see a 0:
    • We can’t extend dp0 (would require flip)
    • We can extend dp1 by flipping this 0 and using the previous dp0 sequence
  4. Result: Maximum of both states gives us the answer

Complexity

Time Complexity: O(n)

  • Single pass: Iterate through array once
  • Each iteration: O(1) operations
  • Total: O(n) where n = array length

Space Complexity: O(1)

  • Variables: Only dp0, dp1, and rtn (constant space)
  • No extra arrays: Space-efficient solution
  • Total: O(1)

Key Points

  1. Two-State DP: Track sequences with and without flip
  2. State Transitions: Clear logic for extending or resetting sequences
  3. Space Efficient: O(1) space, only track current states
  4. Single Pass: O(n) time, process each element once
  5. Optimal: Finds maximum consecutive ones with at most one flip

Detailed Example Walkthrough

Example: nums = [1,0,1,1,0,1]

Position:  0  1  2  3  4  5
nums:     [1, 0, 1, 1, 0, 1]

i=0: nums[0]=1
  dp0 = 1, dp1 = 1
  rtn = 1

i=1: nums[1]=0
  dp1 = dp0 + 1 = 1 + 1 = 2  (flip 0 at index 1)
  dp0 = 0  (reset)
  rtn = max(1, 2) = 2

i=2: nums[2]=1
  dp1 = 2 + 1 = 3  (extend with flip)
  dp0 = 0 + 1 = 1  (new sequence)
  rtn = max(2, 3) = 3

i=3: nums[3]=1
  dp1 = 3 + 1 = 4  (extend with flip)
  dp0 = 1 + 1 = 2  (extend without flip)
  rtn = max(3, 4) = 4

i=4: nums[4]=0
  dp1 = dp0 + 1 = 2 + 1 = 3  (flip 0 at index 4, use dp0 sequence)
  dp0 = 0  (reset)
  rtn = max(4, 3) = 4

i=5: nums[5]=1
  dp1 = 3 + 1 = 4  (extend with flip)
  dp0 = 0 + 1 = 1  (new sequence)
  rtn = max(4, 4) = 4

Result: 4

Visual Explanation:

nums:     [1, 0, 1, 1, 0, 1]
           ↑  ↑  ↑  ↑  ↑  ↑
           |  |  |  |  |  |
           |  |  |  |  |  └─ dp0=1, dp1=4
           |  |  |  |  └──── dp0=0, dp1=3 (flip at index 4)
           |  |  |  └─────── dp0=2, dp1=4
           |  |  └───────── dp0=1, dp1=3
           |  └──────────── dp0=0, dp1=2 (flip at index 1)
           └─────────────── dp0=1, dp1=1

Best sequence: [1,0,1,1] with flip at index 1 → [1,1,1,1] = 4 consecutive ones

Edge Cases

  1. All ones: [1,1,1,1] → Return length (4)
  2. All zeros: [0,0,0] → Return 1 (flip one zero)
  3. Single element: [1] → Return 1
  4. Single zero: [0] → Return 1 (flip it)
  5. Alternating: [1,0,1,0,1] → Return 3 (flip one zero)

Common Mistakes

  • Skipping edge cases (empty input, single element, boundaries).
  • Off-by-one errors in loops and index ranges.
  • Forgetting to handle the case when no valid answer exists.

Tags

Array, Dynamic Programming, Sliding Window, Medium

Key Takeaways

  • Define state: what subproblem does dp[i] (or dp[i][j]) represent?
  • Recurrence: how does the answer build from smaller indices?
  • Base cases first; optimize space if only prior row/layer is needed.

References

Template Reference