Difference Between Greedy Method and Dynamic Programming

Analysis of AlgorithmsAlgorithmsGreedy AlgorithmMisc Algorithms

In this post, we will understand the differences between the greedy algorithm and dynamic programming methods.

Greedy algorithm

It is an algorithmic paradigm that builds up on a solution in parts, step by step. The next step is chosen such that it gives the most obvious and immediate benefit.

  • Problems that involve choosing local optimal values will help in choosing the global optimal values/solution to the problem. Such ate the problems associated with greedy algorithm.
  • There is no surety that a greedy algorithm would lead to an optimal solution.
  • An optimal choice is made at every stage of the problem, i.e the local optimal solution.
  • It is efficient in terms of memory usage since there is no question of having to go back or change previous solutions/values.
  • In general, they are quick in comparison to dynamic programming techniques.
  • Example: Dijkstra's shortest path algorithm that takes O(ELogV + VLogV) time.
  • The solution in a greedy algorithm is computed in a forward method, never visiting the previous values/solutions or changing them.

Dynamic Programming

It is an optimization technique that helps store the result of sub-problems so that they don't need to be re-computed when need in the future. They can just be extracted from the pre-computed set. It reduces the time complexity from exponential to polynomial complexity. 

  • For example: A recursive solution can be turned into a dynamic programming problem by computing.
  • In this, the decision made at every step is done by considering the current problem in hand, and the solution to previously solved sum-problem. This will be used to calculate the optimal value/solution.
  • It is guaranteed that a dynamic programming problem's solution would be an optimal one.
  • Here, the optimal solution chosen is a globally optimal one. It uses certain formula which would have been used to store previously calculated state values.
  • The dynamic programming table is required for memorization. This increases the memory complexity.
  • It is comparatively slower.
  • Example: Bellman Ford algorithm that takes O(VE) time.
  • Dynamic programming determines the solution using a bottom up or top down approach, by developing from smaller problems that have optimal solutions.
Published on 02-Mar-2021 05:02:51