Click here to Skip to main content
13,045,379 members (116,977 online)
Click here to Skip to main content
Add your own
alternative version

Tagged as


40 bookmarked
Posted 20 Apr 2014

Held–Karp algorithm implementation in C#

, 2 Dec 2014
Rate this:
Please Sign up or sign in to vote.
Implementation of the dynamic programming solution for the Travelling salesman problem

Code Repository


In this article we jump straight to the implementation of the algorithm, I assume reader is familiar with the Travelling salesman problem and Dynamic Programming.


I couldn't find a solid implementation of the algorithm, So I decided to do it myself, hope it helps somebody ;)

How it works

The gist of dynamic programming is two parts:

1) A base case that we already know the answer of (stopping condition)
2) Decreasing the problem domain and calling our algorithm again. (Recursion)

put very simply, and without going into much mathematical lingo, the algorithm takes two inputs, the vertex we are starting from v<code>ertex, and a list of vertices to visit vertices.

No we are faced with two situations, either:

  1. vertices is empty, then yay! no work to do, return the distance between starting point and <span style="font-size: 14px;">v</span><span style="font-size: 14px; color: rgb(153, 0, 0); font-family: Consolas, 'Courier New', Courier, mono;">ertex</span><span style="color: rgb(17, 17, 17); font-family: 'Segoe UI', Arial, sans-serif;">.</span>
  2. vertices is not empty, hmph! lets decrease our problem space:

a) consider each vertex newV in vertices as a starting point.

b) since we are considering newV as a starting point, we have to adjust the list of vertices to visit, by removing newV from vertices. (why would I plan a route to a city I am already in) let it be newVertices.

c) calculate cost of visiting newV + cost of visiting rest of vertices starting from newV.

(which is basically calling the algorithm again but with newV and newVertices as input)

d) return minimum result from step c.

Step by step walkthrough

lets say we have a bi-directional graph with the following distance matrix:



the starting vertex is (1), we wish to visit vertices {2,3,4}
lets say our function is called g, then what we want is:
g(1,{2,3, 4})

These are broken down into getting minimum of:
c12 + g(2,{3, 4})
c13 + g(3,{2,4})
c14 + g(4,{2,3})}

c12 is cost of getting from (1) to (2), c13 is cost of getting from (1) to (3),etc...
which are (in order)

g(2, {3,4})= min {

c23 + g(3,{4}),

c24 + g(4,{3})

} = 25

g(3, {4}) = c34 + g(4,{}) = 12 + 8 = 20

g(4, {3}) = c43 + g(3,{}) = 9 + 6 =15

g(3, {2,4})= min {

c32 + g(2,{4}),

c34 + g(4,{2})

} = 25

g(2, {4}) = c24 + g(4, {}) = 10 + 8 = 18

g(4, {2}) = c42 + g(2,{}) = 8 + 5 = 13

g(4, {2,3})= min {

c42 + g(2,{3}),

c43 + g(3,{2})

} = 23

g(3, {2}) = c32 + g(2,{}) = 13 + 5 = 18

g(2, {3}) = c23 + g(3, {}) = 9 + 6 = 15

Substituting in our first three formulas gives :

10 + 25,

15 + 25,

20 + 23

}=min {35, 40, 43}=35
To get the actual tour, we create a Tree

private class Node
    public int Value { get; set; }
    public Node[] ChildNodes { get; set; }
    public bool Selected { get; set; }

The tree data structure will mimic the branching of our algorithm, marking each time the Node (Vertex) with least cost as selected.

After the algorithm terminates, we traverse the tree, checking only vertices that were marked as selected.


private double GetMinimumCostRoute(int startVertex, HashSet<int> set, Node root)
            if (!set.Any())
                //source node is assumed to be the first
                root.ChildNodes = new Node[1] { new Node { Value = _vertices.First(), Selected = true } };
                return _adjacencyMatrix[startVertex, 0];

            double totalCost = double.MaxValue;
            int i = 0;
            int selectedIdx = i;
            root.ChildNodes = new Node[set.Count()];

            foreach (var destination in set)
                root.ChildNodes[i] = new Node { Value = destination };

                double costOfVistingCurrentNode = _adjacencyMatrix[startVertex, destination];

                var newSet = new HashSet<int>(set);
                double costOfVisitingOtherNodes = GetMinimumCostRoute(destination, newSet, root.ChildNodes[i]);
                double currentCost = costOfVistingCurrentNode + costOfVisitingOtherNodes;

                if (totalCost > currentCost)
                    totalCost = currentCost;
                    selectedIdx = i;


            root.ChildNodes[selectedIdx].Selected = true;

            return totalCost;



Note that the running time is exponential, There are at most O(n*2n) subproblems, and each one takes linear time to solve. The total running time is therefore O(n2*2n).

As always I hope I delivered a clear explanation and an adequate implementation ;)
Do let me know if you have any feedback/questions.


This article, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)


About the Author

Omar Gameel Salem
Software Developer
Australia Australia
Enthusiastic programmer/researcher, passionate to learn new technologies, interested in problem solving, data structures, algorithms, AI, machine learning and nlp.

Amateur guitarist/ keyboardist, squash player.

You may also be interested in...

Comments and Discussions

QuestionAlternative Pin
Magerusan Grigore Cosmin4-Aug-16 5:05
memberMagerusan Grigore Cosmin4-Aug-16 5:05 
Questioncode project Pin
Halima Azza21-Dec-15 22:07
memberHalima Azza21-Dec-15 22:07 
QuestionWhere is memorization step? Pin
Member 1155863226-Mar-15 11:03
memberMember 1155863226-Mar-15 11:03 
GeneralGood Job Pin
Emre Ataseven18-May-14 8:13
professionalEmre Ataseven18-May-14 8:13 
GeneralThanks Pin
Victor Nascimento16-May-14 15:03
memberVictor Nascimento16-May-14 15:03 
GeneralMy vote of 5 Pin
Philip Liebscher29-Apr-14 8:36
memberPhilip Liebscher29-Apr-14 8:36 

General General    News News    Suggestion Suggestion    Question Question    Bug Bug    Answer Answer    Joke Joke    Praise Praise    Rant Rant    Admin Admin   

Use Ctrl+Left/Right to switch messages, Ctrl+Up/Down to switch threads, Ctrl+Shift+Left/Right to switch pages.

Permalink | Advertise | Privacy | Terms of Use | Mobile
Web02 | 2.8.170713.1 | Last Updated 2 Dec 2014
Article Copyright 2014 by Omar Gameel Salem
Everything else Copyright © CodeProject, 1999-2017
Layout: fixed | fluid