Home > Articles > Programming > Algorithms

This chapter is from the book

Our shortest-paths algorithms are based on a simple operation known as relaxation. We start a shortest-paths algorithm knowing only the network's edges and weights. As we proceed, we gather information about the shortest paths that connect various pairs of vertices. Our algorithms all update this information incrementally, making new inferences about shortest paths from the knowledge gained so far. At each step, we test whether we can find a path that is shorter than some known path. The term "relaxation" is commonly used to describe this step, which relaxes constraints on the shortest path. We can think of a rubber band stretched tight on a path connecting two vertices: A successful relaxation operation allows us to relax the tension on the rubber band along a shorter path.

Our algorithms are based on applying repeatedly one of two types of relaxation operations:

  • Edge relaxation: Test whether traveling along a given edge gives a new shortest path to its destination vertex.

  • Path relaxation: Test whether traveling through a given vertex gives a new shortest path connecting two other given vertices.

Edge relaxation is a special case of path relaxation; we consider the operations separately, however, because we use them separately (the former in single-source algorithms; the latter in all-pairs algorithms). In both cases, the prime requirement that we impose on the data structures that we use to represent the current state of our knowledge about a network's shortest paths is that we can update them easily to reflect changes implied by a relaxation operation.

First, we consider edge relaxation, which is illustrated in Figure 21.5. All the single-source shortest-paths algorithms that we consider are based on this step: Does a given edge lead us to consider a shorter path to its destination from the source?

21fig05.gifFigure 21.5. Edge relaxation

The data structures that we need to support this operation are straightforward. First, we have the basic requirement that we need to compute the shortest-paths lengths from the source to each of the other vertices. Our convention will be to store in a vertex-indexed array wt the lengths of the shortest known paths from the source to each vertex. Second, to record the paths themselves as we move from vertex to vertex, our convention will be the same as the one that we used for other graph-search algorithms that we examined in Chapters 18 through 20: We use a vertex-indexed array spt to record the last edge on a shortest path from the source to the indexed vertex. These edges constitute a tree.

With these data structures, implementing edge relaxation is a straightforward task. In our single-source shortest-paths code, we use the following code to relax along an edge e from v to w:

if (wt[w] > wt[v] + e.wt())
  { wt[w] = wt[v] + e.wt(); spt[w] = e; }

This code fragment is both simple and descriptive; we include it in this form in our implementations, rather than defining relaxation as a higher-level abstract operation.

Definition 21.2 Given a network and a designated vertex s, a shortest-paths tree (SPT) for s is a subnetwork containing s and all the vertices reachable from s that forms a directed tree rooted at s such that every tree path is a shortest path in the network.

There may be multiple paths of the same length connecting a given pair of nodes, so SPTs are not necessarily unique. In general, as illustrated in Figure 21.2, if we take shortest paths from a vertex s to every vertex reachable from s in a network and from the subnetwork induced by the edges in the paths, we may get a DAG. Different shortest paths connecting pairs of nodes may each appear as a subpath in some longer path containing both nodes. Because of such effects, we generally are content to compute any SPT for a given digraph and start vertex.

Our algorithms generally initialize the entries in the wt array with a sentinel value. That value needs to be sufficiently small that the addition in the relaxation test does not cause overflow and sufficiently large that no simple path has a larger weight. For example, if edge weights are between 0 and 1, we can use the value V. Note that we have to take extra care to check our assumptions when using sentinels in networks that could have negative weights. For example, if both vertices have the sentinel value, the relaxation code just given takes no action if e.wt is nonnegative (which is probably what we intend in most implementations), but it will change wt[w] and spt[w] if the weight is negative.

Our code always uses the destination vertex as the index to save the SPT edges (spt[w].w() == w). For economy and consistency with Chapters 17 through 19, we use the notation st[w] to refer to the vertex spt[w].v() (in the text and particularly in the figures) to emphasize that the spt array is actually a parent-link representation of the shortest-paths tree, as illustrated in Figure 21.6. We can compute the shortest path from s to t by traveling up the tree from t to s; when we do so, we are traversing edges in the direction opposite from their direction in the network and are visiting the vertices on the path in reverse order (t, st[t], st[st[t]], and so forth).

21fig06.gifFigure 21.6. Shortest-paths trees

One way to get the edges on the path in order from source to sink from an SPT is to use a stack. For example, the following code prints a path from the source to a given vertex w:

EdgeStack P = new EdgeStack(G.V()); Edge e = st[v];
while (e != null) { P.push(e); e = st[e.()]); }
if (P.empty()) Out.print(P.top().v());
while (!P.empty())
  { Out.print("-" + P.top().w()); P.pop(); }

In a class implementation, we could use code similar to this to provide clients with a array that contains the edges of the path.

If we simply want to print or otherwise process the edges on the path, going all the way through the path in reverse order to get to the first edge in this way may be undesirable. One approach to get around this difficulty is to work with the reverse network, as illustrated in Figure 21.6. We use reverse order and edge relaxation in single-source problems because the SPT gives a compact representation of the shortest paths from the source to all the other vertices, in an array with just V entries.

Next, we consider path relaxation, which is the basis of some of our all-pairs algorithms: Does going through a given vertex lead us to a shorter path that connects two other given vertices? For example, suppose that, for three vertices s, x, and t, we wish to know whether it is better to go from s to x and then from x to t or to go from s to t without going through x. For straight-line connections in a Euclidean space, the triangle inequality tells us that the route through x cannot be shorter than the direct route from s to t, but for paths in a network, it could be (see Figure 21.7). To determine which, we need to know the lengths of paths from s to x, x to t, and of those from s to t (that do not include x). Then, we simply test whether or not the sum of the first two is less than the third; if it is, we update our records accordingly.

21fig07.gifFigure 21.7. Path relaxation

Path relaxation is appropriate for all-pairs solutions where we maintain the lengths of the shortest paths that we have encountered between all pairs of vertices. Specifically, in all-pairs shortest-paths code of this kind, we maintain an array of arrays d such that d[s][t] is the shortest-path length from s to t, and we also maintain an array of arrays p such that p[s][t] is the next vertex on a shortest path from s to t. We refer to the former as the distances matrix and the latter as the paths matrix. Figure 21.8 shows the two matrices for our example network. The distances matrix is a prime objective of the computation, and we use the paths matrix because it is clearly more compact than, but carries the same information as, the full list of paths that is illustrated in Figure 21.3.

21fig08.gifFigure 21.8. All shortest paths

In terms of these data structures, path relaxation amounts to the following code:

if (d[s][t] > d[s][x] + d[x][t])
  { d[s][t] = d[s][x] + d[x][t]; p[s][t] = p[s][x]; }

Like edge relaxation, this code reads as a restatement of the informal description that we have given, so we use it directly in our implementations. More formally, path relaxation reflects the following:

If a vertex x is on a shortest path from s to t, then that path consists of a shortest path from s to x followed by a shortest path from x to t.

Proof: By contradiction. We could use any shorter path from s to x or from x to t to build a shorter path from s to t. ▪

We encountered the path-relaxation operation when we discussed transitive-closure algorithms, in Section 19.3. If the edge and path weights are either 1 or infinite (that is, a path's weight is 1 only if all that path's edges have weight 1), then path relaxation is the operation that we used in Warshall's algorithm (if we have a path from s to x and a path from x to t, then we have a path from s to t). If we define a path's weight to be the number of edges on that path, then Warshall's algorithm generalizes to Floyd's algorithm for finding all shortest paths in unweighted digraphs; it further generalizes to apply to networks, as we see in Section 21.3.

From a mathematician's perspective, it is important to note that these algorithms all can be cast in a general algebraic setting that unifies and helps us to understand them. From a programmer's perspective, it is important to note that we can implement each of these algorithms using an abstract + operator (to compute path weights from edge weights) and an abstract < operator (to compute the minimum value in a set of path weights), both solely in the context of the relaxation operation (see Exercises 19.55 and 19.56).

Property 21.1 implies that a shortest path from s to t contains shortest paths from s to every other vertex along the path to t. Most shortest-paths algorithms also compute shortest paths from s to every vertex that is closer to s than to t (whether or not the vertex is on the path from s to t), although that is not a requirement (see Exercise 21.18). Solving the source–sink shortest-paths problem with such an algorithm when t is the vertex that is farthest from s is equivalent to solving the single-source shortest-paths problem for s. Conversely, we could use a solution to the single-source shortest-paths problem from s as a method for finding the vertex that is farthest from s.

The paths matrix that we use in our implementations for the all-pairs problem is also a representation of the shortest-paths trees for each of the vertices. We defined p[s][t] to be the vertex that follows s on a shortest path from s to t. It is thus the same as the vertex that precedes s on the shortest path from t to s in the reverse network. In other words, column t in the paths matrix of a network is a vertex-indexed array that represents the SPT for vertex t in its reverse. Conversely, we can build the paths matrix for a network by filling each column with the vertex-indexed array representation of the SPT for the appropriate vertex in the reverse. This correspondence is illustrated in Figure 21.9.

21fig09.gifFigure 21.9. All shortest paths in a network

In summary, relaxation gives us the basic abstract operations that we need to build our shortest paths algorithms. The primary complication is the choice of whether to provide the first or final edge on the shortest path. For example, single-source algorithms are more naturally expressed by providing the final edge on the path so that we need only a single vertex-indexed array to reconstruct the path, since all paths lead back to the source. This choice does not present a fundamental difficulty because we can either use the reverse graph as warranted or provide methods that hide this difference from clients. For example, we could specify a method in the interface that returns the edges on the shortest path in an array (see Exercises 21.15 and 21.16).

Accordingly, for simplicity, all of our implementations in this chapter include a method dist that returns a shortest-path length and either a method path that returns the first edge on a shortest path or a method pathR that returns the final edge on a shortest path. For example, our single-source implementations that use edge relaxation typically implement these methods as follows:

Edge pathR(int w) { return spt[w]; }
double dist(int v) { return wt[v]; }

Similarly, our all-paths implementations that use path relaxation typically implement these methods as follows:

Edge path(int s, int t) { return p[s][t]; }
double dist(int s, int t) { return d[s][t]; }

In some situations, it might be worthwhile to build interfaces that standardize on one or the other or both of these options; we choose the most natural one for the algorithm at hand.


  • 21.12 Draw the SPT from 0 for the network defined in Exercise 21.1 and for its reverse. Give the parent-link representation of both trees.

  • 21.13 Consider the edges in the network defined in Exercise 21.1 to be undirected edges such that each edge corresponds to equal-weight edges in both directions in the network. Answer Exercise 21.12 for this corresponding network.

  • 21.14 Change the direction of edge 0-2 in Figure 21.2. Draw two different SPTs that are rooted at 2 for this modified network.

  • 21.15 Write a method that uses the pathR method from a single-source implementation to put references to the edges on the path from the source v to a given vertex w in a Java Vector.

  • 21.16 Write a method that uses the path method from an all-paths implementation to put references to the edges on the path from a given vertex v to another given vertex w in a Java Vector.

  • 21.17 Write a program that uses your method from Exercise 21.16 to print out all of the paths, in the style of Figure 21.3.

  • 21.18 Give an example that shows how we could know that a path from s to t is shortest without knowing the length of a shorter path from s to x for some x.

InformIT Promotional Mailings & Special Offers

I would like to receive exclusive offers and hear about products from InformIT and its family of brands. I can unsubscribe at any time.


Pearson Education, Inc., 221 River Street, Hoboken, New Jersey 07030, (Pearson) presents this site to provide information about products and services that can be purchased through this site.

This privacy notice provides an overview of our commitment to privacy and describes how we collect, protect, use and share personal information collected through this site. Please note that other Pearson websites and online products and services have their own separate privacy policies.

Collection and Use of Information

To conduct business and deliver products and services, Pearson collects and uses personal information in several ways in connection with this site, including:

Questions and Inquiries

For inquiries and questions, we collect the inquiry or question, together with name, contact details (email address, phone number and mailing address) and any other additional information voluntarily submitted to us through a Contact Us form or an email. We use this information to address the inquiry and respond to the question.

Online Store

For orders and purchases placed through our online store on this site, we collect order details, name, institution name and address (if applicable), email address, phone number, shipping and billing addresses, credit/debit card information, shipping options and any instructions. We use this information to complete transactions, fulfill orders, communicate with individuals placing orders or visiting the online store, and for related purposes.


Pearson may offer opportunities to provide feedback or participate in surveys, including surveys evaluating Pearson products, services or sites. Participation is voluntary. Pearson collects information requested in the survey questions and uses the information to evaluate, support, maintain and improve products, services or sites, develop new products and services, conduct educational research and for other purposes specified in the survey.

Contests and Drawings

Occasionally, we may sponsor a contest or drawing. Participation is optional. Pearson collects name, contact information and other information specified on the entry form for the contest or drawing to conduct the contest or drawing. Pearson may collect additional personal information from the winners of a contest or drawing in order to award the prize and for tax reporting purposes, as required by law.


If you have elected to receive email newsletters or promotional mailings and special offers but want to unsubscribe, simply email information@informit.com.

Service Announcements

On rare occasions it is necessary to send out a strictly service related announcement. For instance, if our service is temporarily suspended for maintenance we might send users an email. Generally, users may not opt-out of these communications, though they can deactivate their account information. However, these communications are not promotional in nature.

Customer Service

We communicate with users on a regular basis to provide requested services and in regard to issues relating to their account we reply via email or phone in accordance with the users' wishes when a user submits their information through our Contact Us form.

Other Collection and Use of Information

Application and System Logs

Pearson automatically collects log data to help ensure the delivery, availability and security of this site. Log data may include technical information about how a user or visitor connected to this site, such as browser type, type of computer/device, operating system, internet service provider and IP address. We use this information for support purposes and to monitor the health of the site, identify problems, improve service, detect unauthorized access and fraudulent activity, prevent and respond to security incidents and appropriately scale computing resources.

Web Analytics

Pearson may use third party web trend analytical services, including Google Analytics, to collect visitor information, such as IP addresses, browser types, referring pages, pages visited and time spent on a particular site. While these analytical services collect and report information on an anonymous basis, they may use cookies to gather web trend information. The information gathered may enable Pearson (but not the third party web trend services) to link information with application and system log data. Pearson uses this information for system administration and to identify problems, improve service, detect unauthorized access and fraudulent activity, prevent and respond to security incidents, appropriately scale computing resources and otherwise support and deliver this site and its services.

Cookies and Related Technologies

This site uses cookies and similar technologies to personalize content, measure traffic patterns, control security, track use and access of information on this site, and provide interest-based messages and advertising. Users can manage and block the use of cookies through their browser. Disabling or blocking certain cookies may limit the functionality of this site.

Do Not Track

This site currently does not respond to Do Not Track signals.


Pearson uses appropriate physical, administrative and technical security measures to protect personal information from unauthorized access, use and disclosure.


This site is not directed to children under the age of 13.


Pearson may send or direct marketing communications to users, provided that

  • Pearson will not use personal information collected or processed as a K-12 school service provider for the purpose of directed or targeted advertising.
  • Such marketing is consistent with applicable law and Pearson's legal obligations.
  • Pearson will not knowingly direct or send marketing communications to an individual who has expressed a preference not to receive marketing.
  • Where required by applicable law, express or implied consent to marketing exists and has not been withdrawn.

Pearson may provide personal information to a third party service provider on a restricted basis to provide marketing solely on behalf of Pearson or an affiliate or customer for whom Pearson is a service provider. Marketing preferences may be changed at any time.

Correcting/Updating Personal Information

If a user's personally identifiable information changes (such as your postal address or email address), we provide a way to correct or update that user's personal data provided to us. This can be done on the Account page. If a user no longer desires our service and desires to delete his or her account, please contact us at customer-service@informit.com and we will process the deletion of a user's account.


Users can always make an informed choice as to whether they should proceed with certain services offered by InformIT. If you choose to remove yourself from our mailing list(s) simply visit the following page and uncheck any communication you no longer want to receive: www.informit.com/u.aspx.

Sale of Personal Information

Pearson does not rent or sell personal information in exchange for any payment of money.

While Pearson does not sell personal information, as defined in Nevada law, Nevada residents may email a request for no sale of their personal information to NevadaDesignatedRequest@pearson.com.

Supplemental Privacy Statement for California Residents

California residents should read our Supplemental privacy statement for California residents in conjunction with this Privacy Notice. The Supplemental privacy statement for California residents explains Pearson's commitment to comply with California law and applies to personal information of California residents collected in connection with this site and the Services.

Sharing and Disclosure

Pearson may disclose personal information, as follows:

  • As required by law.
  • With the consent of the individual (or their parent, if the individual is a minor)
  • In response to a subpoena, court order or legal process, to the extent permitted or required by law
  • To protect the security and safety of individuals, data, assets and systems, consistent with applicable law
  • In connection the sale, joint venture or other transfer of some or all of its company or assets, subject to the provisions of this Privacy Notice
  • To investigate or address actual or suspected fraud or other illegal activities
  • To exercise its legal rights, including enforcement of the Terms of Use for this site or another contract
  • To affiliated Pearson companies and other companies and organizations who perform work for Pearson and are obligated to protect the privacy of personal information consistent with this Privacy Notice
  • To a school, organization, company or government agency, where Pearson collects or processes the personal information in a school setting or on behalf of such organization, company or government agency.


This web site contains links to other sites. Please be aware that we are not responsible for the privacy practices of such other sites. We encourage our users to be aware when they leave our site and to read the privacy statements of each and every web site that collects Personal Information. This privacy statement applies solely to information collected by this web site.

Requests and Contact

Please contact us about this Privacy Notice or if you have any requests or questions relating to the privacy of your personal information.

Changes to this Privacy Notice

We may revise this Privacy Notice through an updated posting. We will identify the effective date of the revision in the posting. Often, updates are made to provide greater clarity or to comply with changes in regulatory requirements. If the updates involve material changes to the collection, protection, use or disclosure of Personal Information, Pearson will provide notice of the change through a conspicuous notice on this site or other appropriate way. Continued use of the site after the effective date of a posted revision evidences acceptance. Please contact us if you have questions or concerns about the Privacy Notice or any objection to any revisions.

Last Update: November 17, 2020