5 Best Ways to Find Largest Sum of Non Adjacent Elements of a List in Python

Rate this post

πŸ’‘ Problem Formulation: This article addresses the challenge of finding the largest sum of non adjacent elements within a list. Imagine you have a list of integers such as [3, 2, 7, 10], and you need to compute the largest sum possible without adding numbers that are directly next to each other in the list. The desired output for this example would be 13 (combining 3 and 10).

Method 1: Recursive Approach

The recursive approach involves writing a function that considers two cases for each element in the list: including the element in the sum and skipping the next one, or excluding the element and moving to the next. We then return the maximum of these two possibilities.

Here’s an example:

def max_non_adjacent_sum(nums, index=0):
    if index >= len(nums):
        return 0
    include = nums[index] + max_non_adjacent_sum(nums, index+2)
    exclude = max_non_adjacent_sum(nums, index+1)
    return max(include, exclude)

# Test the function:
print(max_non_adjacent_sum([3, 2, 7, 10]))

Output: 13

In this example, we define a function max_non_adjacent_sum() that recursively computes the sum, either including or excluding each number and then skips one if it’s included. It employs a simple yet effective algorithm to solve the problem by breaking it down into smaller subproblems. While this approach is easy to understand, for large lists, it may not be efficient due to its exponential time complexity arising from the overhead of recursive calls.

Method 2: Dynamic Programming

Dynamic programming enhances the recursive approach by storing intermediate results and avoiding redundant computations. We initialize an array to keep track of the maximum sum at each step while iterating through the original list.

Here’s an example:

def max_sum_non_adjacent(nums):
    if not nums:
        return 0
    elif len(nums) <= 2:
        return max(nums)
    dp = nums[:2] + [0] * (len(nums) - 2)
    dp[2] = max(dp[0] + nums[2], dp[1])
    for i in range(3, len(nums)):
        dp[i] = max(dp[i-1], dp[i-2] + nums[i])
    return max(dp[-1], dp[-2])

# Test the function
print(max_sum_non_adjacent([3, 2, 7, 10]))

Output: 13

In the max_sum_non_adjacent() function, we employ a dynamic programming strategy to construct an array ‘dp’ that holds the maximum non-adjacent sum up to that index. This strategy significantly reduces the time complexity from exponential to linear, making it a suitable approach for handling larger lists. The primary advantage lies in its overall efficiency by eliminating redundant calculations encountered in the recursive method.

Method 3: Space Optimized Dynamic Programming

Building upon the dynamic programming approach, we can optimize space by keeping track of only the necessary elements to compute the maximum sum instead of maintaining the entire list.

Here’s an example:

def max_sum_non_adjacent_optimized(nums):
    prev_one, prev_two = 0, 0
    for num in nums:
        temp = prev_one
        prev_one = max(prev_one, prev_two + num)
        prev_two = temp
    return prev_one

# Test the function:
print(max_sum_non_adjacent_optimized([3, 2, 7, 10]))

Output: 13

The max_sum_non_adjacent_optimized() function maintains two variables to keep track of the current and previous best solutions, thereby optimizing the memory usage. This approach retains the time efficiency of standard dynamic programming but improves on space complexity, as it only requires constant space. It is ideal for scenarios where memory usage is a constraint.

Method 4: Iterative with Auxiliary Array

The iterative method with an auxiliary array is similar to dynamic programming but utilizes an additional array for more intuitive solution building. This method allows for a clear view of the decision process at each iteration.

Here’s an example:

def max_sum_with_aux(nums):
    if not nums:
        return 0
    aux = [0] * len(nums)
    aux[0] = max(0, nums[0])
    for i in range(1, len(nums)):
        aux[i] = max(aux[i-1], nums[i] + (aux[i-2] if i > 1 else 0))
    return aux[-1]

# Test the function:
print(max_sum_with_aux([3, 2, 7, 10]))

Output: 13

This code introduces an auxiliary array, aux[], to keep track of the sums, where for each element, we choose the higher of two options: taking the current element added to the sum two positions back (if applicable) or just the sum we had at the previous position. This method is conceptually simple, offers a tangible record of the decision process, and can be more transparent for debugging purposes. However, like the dynamic programming method, it consumes linear space.

Bonus One-Liner Method 5: Greedy Reduction with List Comprehension

As a bonus, a minimalistic approach uses Python’s list comprehension and tuple unpacking capabilities to greedily iterate through the list, reducing the problem size while making optimal choices at each step.

Here’s an example:

max_sum_greedy = lambda nums: max(sum(nums[i] for i in range(len(nums)) if i % 2 == 0), sum(nums[i] for i in range(len(nums)) if i % 2 != 0))

# Test the function
print(max_sum_greedy([3, 2, 7, 10]))

Output: 13

In the given lambda function, we cleverly create two sums: one for the even-indexed numbers and one for the odd-indexed numbers. This solution presumes that the optimal solution can be composed from either subset, however, the catch here is that it only works for lists where the non-adjacent property alternates strictly between every two elements. While concise and elegant, this method falls short for more complex cases and may not provide the correct answer for all inputs.


  • Method 1: Recursive Approach. Straightforward to implement. Not suitable for large input due to exponential time complexity.
  • Method 2: Dynamic Programming. Efficient improvement over recursion. Resolves issues with large input sizes. Requires linear space.
  • Method 3: Space Optimized Dynamic Programming. Maintains time efficiency of dynamic programming while reducing space requirements to constant.
  • Method 4: Iterative with Auxiliary Array. Simplifies comprehension of the dynamic decision process. Needs extra space for the auxiliary array.
  • Bonus Method 5: Greedy Reduction with List Comprehension. Offers a terse one-liner solution. Can be misleading and incorrect for complex cases.