# Difference between revisions of "Shortest path"

Line 2: | Line 2: | ||

Shortest paths in undirected graphs can be computed by replacing each undirected edge with two arcs of the same weight, one going in each direction, to obtain a directed graph. | Shortest paths in undirected graphs can be computed by replacing each undirected edge with two arcs of the same weight, one going in each direction, to obtain a directed graph. | ||

+ | |||

+ | '''Theorem''': In a graph with no cycles of negative weight, the shortest path is no shorter than the shortest simple path. (On the other hand, in a graph with a negative-weight cycle, lengths of paths may be unbounded below.) | ||

+ | |||

+ | '''Proof''': We show that any path from <math>u</math> to <math>v</math> can be transformed into a simple path from <math>u</math> to <math>v</math> which is at least as short. Let the path be denoted <math>[u = s_0, s_1, ..., s_j = v]</math>. We proceed by induction on the number of pairs <math>(i,j)</math> (with <math>i \neq j</math>) such that <math>s_i = s_j</math>, which is countable. When there are zero such pairs, the path is already simple, and nothing needs to be done. Otherwise, we transform the path into another path with fewer such pairs but equal or lesser weight by removing the vertices <math>s_i, s_{i+1}, ..., s_{j-1}</math> from the path. The weight of the path therefore decreases by an amount equal to the weight of the cycle <math>s_i, s_{i+1}, ..., s_j = s_i</math>, which is nonnegative. <math>_\blacksquare</math> | ||

+ | |||

+ | '''Corollary''': Assuming our graphs have no cycles of negative weight, the restriction that the shortest paths be simple is immaterial. Therefore, we will assume in the foregoing discussion that our graphs have no cycles of negative weight, for the problem of finding shortest ''simple'' paths in graphs containing negative-weight cycles is NP-complete. | ||

+ | |||

+ | '''Corollary''': In a finite graph, a shortest path always exists. (To prove this we simply use the fact that the graph has a finite number of simple paths, and only simple paths need be considered. So one of them must be the shortest.) | ||

==Variations== | ==Variations== | ||

Line 10: | Line 18: | ||

==Approach== | ==Approach== | ||

− | All the shortest paths algorithms discussed in this article have the same basic approach. At their core, they compute not the shortest paths themselves, but the distances. Using information computed in order to compute the distances, one can easily then reconstruct the paths themselves. They begin with the knowledge that the distance from any vertex to itself is zero, and they overestimate all other distances they need. (By this it is meant that they find a number <math>d_{i,j}</math> for each pair <math>(i,j)</math> under consideration such that the distance from <math>i</math> to <math>j</math> is less than or equal to <math>d_{i,j}</math>.) At some point, all overestimates will be refined, perhaps gradually, perhaps at once, so that once the algorithm has terminated, they are exactly the correct distances. | + | All the shortest paths algorithms discussed in this article have the same basic approach. At their core, they compute not the shortest paths themselves, but the distances. Using information computed in order to compute the distances, one can easily then reconstruct the paths themselves. They begin with the knowledge that the distance from any vertex to itself is zero, and they overestimate all other distances they need. (By this it is meant that they find a number <math>d_{i,j}</math> for each pair <math>(i,j)</math> under consideration such that the distance from <math>i</math> to <math>j</math> is less than or equal to <math>d_{i,j}</math>.) If <math>(i,j) \in E</math>, then the initial overestimate for the distance from <math>i</math> to <math>j</math> is the weight of the edge <math>(i,j)</math>; otherwise it is infinite. At some point, all overestimates will be refined, perhaps gradually, perhaps at once, so that once the algorithm has terminated, they are exactly the correct distances. |

===Relaxation=== | ===Relaxation=== | ||

There are theoretically many ways to refine overestimates but a specific way, known as '''relaxation''', is used in all the algorithms discussed in this article. Relaxation can take place when three conditions are met: | There are theoretically many ways to refine overestimates but a specific way, known as '''relaxation''', is used in all the algorithms discussed in this article. Relaxation can take place when three conditions are met: | ||

# The currently best overestimate for the distance from some vertex <math>i</math> to some vertex <math>k</math> is <math>d_1</math>; | # The currently best overestimate for the distance from some vertex <math>i</math> to some vertex <math>k</math> is <math>d_1</math>; | ||

− | # The currently best overestimate for the distance from <math>k</math> to some vertex <math>j</math> is <math>d_2</math> | + | # The currently best overestimate for the distance from <math>k</math> to some vertex <math>j</math> is <math>d_2</math>, |

# The currently best overestimate for the distance from <math>i</math> to <math>j</math> is greater than <math>d_1+d_2</math>. (This includes the case in which it is infinite.) | # The currently best overestimate for the distance from <math>i</math> to <math>j</math> is greater than <math>d_1+d_2</math>. (This includes the case in which it is infinite.) | ||

Relaxation refines the best overestimate for the distance from <math>i</math> to <math>j</math> by setting it to <math>d_1+d_2</math>, which is better than its current value. | Relaxation refines the best overestimate for the distance from <math>i</math> to <math>j</math> by setting it to <math>d_1+d_2</math>, which is better than its current value. | ||

− | + | <!-- | |

'''Theorem''': Relaxation will never strictly underestimate the distance from <math>i</math> to <math>j</math>. | '''Theorem''': Relaxation will never strictly underestimate the distance from <math>i</math> to <math>j</math>. | ||

− | '''Proof''': Concatenating the shortest paths from <math>i</math> to <math>k</math> and <math>k</math> to <math>j</math> gives a path from <math>i</math> to <math>j</math> with length less than or equal to <math>d_1+d_2</math>. Therefore <math>d_1+d_2</math> is an overestimate. | + | '''Proof''': Concatenating the shortest paths from <math>i</math> to <math>k</math> and <math>k</math> to <math>j</math> gives a path from <math>i</math> to <math>j</math> with length less than or equal to <math>d_1+d_2</math>. Therefore <math>d_1+d_2</math> is an overestimate. <math>_\blacksquare</math> |

+ | |||

+ | '''Theorem''': Relaxation is impossible when all distances under consideration are correctly known. | ||

+ | |||

+ | '''Proof''': Using the variables defined above, this implies that the distance from <math>i</math> to <math>j</math> is in fact greater than the sum of the distances from <math>i</math> to <math>k</math> and <math>k</math> to <math>j</math>, which is impossible, as we can merely concatenate shortest paths from <math>i</math> to <math>k</math> and <math>k</math> to <math>j</math> to obtain a path from <math>i</math> to <math>j</math> of <math>d_1+d_2</math>, which is less. <math>_\blacksquare</math> | ||

+ | |||

+ | '''Theorem''': In a graph with a finite number of vertices, only a finite number of successive relaxations can take place, assuming the initial conditions are as defined above. | ||

+ | |||

+ | '''Proof''': We show by induction on the number of relaxations so far performed that if the current best overestimate on the distance from <math>i \in V</math> to <math>j \in V</math> is finite and equals <math>d</math>, then we can exhibit a path of weight <math>d</math> from <math>i</math> to <math>j</math>. Clearly this is true initially, when no relaxations have been performed, because we used empty paths and paths consisting of only one edge to set the initial overestimates. The inductive step is also obvious, because we can concatenate the already-known paths of length <math>d_1</math> and <math>d_2</math> from <math>i</math> to <math>k</math> and <math>k</math> to <math>j</math>, respectively, to obtain a path of length <math>d_1+d_2</math> from <math>i</math> to <math>j</math>. | ||

− | + | After a relaxation takes place, the path we have from <math>i</math> to <math>j</math> contains at least as many edges as the paths we have from <math>i</math> to <math>k</math> and from <math>k</math> to <math>j</math>. Now, at most a finite number of relaxations | |

+ | --> | ||

− | ''' | + | '''Theorem''': When the distances from <math>u</math> to all other vertices are all overestimated, but no relaxations are possible, then those distances are all known correctly. Contrapositively, if at there exists <math>v \in V</math> such that the distance from <math>u</math> to <math>v</math> is not correctly known, then relaxation must be possible somewhere in the graph. |

− | ''' | + | '''Proof''': By induction on the number of edges in some shortest path from <math>u</math> to <math>v</math> (which we can take to be simple). It is vacuously true when this is zero or one, because all paths of length zero or one were accounted for in the initial overestimates. Assume the path has at least two edges, and denote by <math>s</math> the last vertex on the path before <math>v</math>. If the distance from <math>u</math> to <math>s</math> or from <math>s</math> to <math>v</math> is not known, then by the inductive hypothesis, we are done. Otherwise, notice that the path contains two subpaths, one from <math>u</math> to <math>s</math> and one from <math>s</math> to <math>v</math> (the latter is trivial as it consists of a single edge), and that each of these must itself be a shortest path, otherwise we could replace it with a shorter path to obtain a shorter path from <math>u</math> to <math>v</math>, a contradiction. Now, as the distances from <math>u</math> to <math>s</math> and <math>s</math> to <math>v</math> are correctly known, and the correct distance from <math>u</math> to <math>v</math> is exactly the sum of the distances from <math>u</math> to <math>s</math> and <math>s</math> to <math>v</math> (as these values equal the weights of the aforementioned subpaths), and our overestimate of the distance from <math>u</math> to <math>v</math> is incorrect, it must be strictly greater than the sum of the distances from <math>u</math> to <math>s</math> and <math>s</math> to <math>v</math>. Hence <math>(u,v)</math> can be relaxed. <math>_\blacksquare</math> |

==Single-source shortest paths== | ==Single-source shortest paths== |

## Revision as of 21:56, 15 November 2010

The **shortest paths** problem is one of the most fundamental problems in graph theory. Given a directed graph , possibly weighted, and a set of pairs of vertices , the problem is to compute, for each , a **simple** path in from to (a list of vertices such that for all , ) such that no other simple path in from to has a lower total weight.

Shortest paths in undirected graphs can be computed by replacing each undirected edge with two arcs of the same weight, one going in each direction, to obtain a directed graph.

**Theorem**: In a graph with no cycles of negative weight, the shortest path is no shorter than the shortest simple path. (On the other hand, in a graph with a negative-weight cycle, lengths of paths may be unbounded below.)

**Proof**: We show that any path from to can be transformed into a simple path from to which is at least as short. Let the path be denoted . We proceed by induction on the number of pairs (with ) such that , which is countable. When there are zero such pairs, the path is already simple, and nothing needs to be done. Otherwise, we transform the path into another path with fewer such pairs but equal or lesser weight by removing the vertices from the path. The weight of the path therefore decreases by an amount equal to the weight of the cycle , which is nonnegative.

**Corollary**: Assuming our graphs have no cycles of negative weight, the restriction that the shortest paths be simple is immaterial. Therefore, we will assume in the foregoing discussion that our graphs have no cycles of negative weight, for the problem of finding shortest *simple* paths in graphs containing negative-weight cycles is NP-complete.

**Corollary**: In a finite graph, a shortest path always exists. (To prove this we simply use the fact that the graph has a finite number of simple paths, and only simple paths need be considered. So one of them must be the shortest.)

## Contents

## Variations

Three variations of the shortest path algorithm exist, and they are discussed in the following sections.

- In the
*single-pair shortest path*problem, there is only one pair in the problem set. In other words the shortest path is desired between a single pair of vertices. - In the
*single-source shortest paths*problem, the problem set is of the form . One vertex, , is designated the*source*, and we wish to find the shortest paths from the source to all other vertices. (To solve the analogous*single-destination shortest paths*problem, we merely reverse the directions of all edges, which reduces it to single-source.) - In the
*all-pairs shortest paths*problem, the problem set is ; that is, we wish to know the shortest paths from every vertex to every other vertex.

## Approach

All the shortest paths algorithms discussed in this article have the same basic approach. At their core, they compute not the shortest paths themselves, but the distances. Using information computed in order to compute the distances, one can easily then reconstruct the paths themselves. They begin with the knowledge that the distance from any vertex to itself is zero, and they overestimate all other distances they need. (By this it is meant that they find a number for each pair under consideration such that the distance from to is less than or equal to .) If , then the initial overestimate for the distance from to is the weight of the edge ; otherwise it is infinite. At some point, all overestimates will be refined, perhaps gradually, perhaps at once, so that once the algorithm has terminated, they are exactly the correct distances.

### Relaxation

There are theoretically many ways to refine overestimates but a specific way, known as **relaxation**, is used in all the algorithms discussed in this article. Relaxation can take place when three conditions are met:

- The currently best overestimate for the distance from some vertex to some vertex is ;
- The currently best overestimate for the distance from to some vertex is ,
- The currently best overestimate for the distance from to is greater than . (This includes the case in which it is infinite.)

Relaxation refines the best overestimate for the distance from to by setting it to , which is better than its current value.

**Theorem**: When the distances from to all other vertices are all overestimated, but no relaxations are possible, then those distances are all known correctly. Contrapositively, if at there exists such that the distance from to is not correctly known, then relaxation must be possible somewhere in the graph.

**Proof**: By induction on the number of edges in some shortest path from to (which we can take to be simple). It is vacuously true when this is zero or one, because all paths of length zero or one were accounted for in the initial overestimates. Assume the path has at least two edges, and denote by the last vertex on the path before . If the distance from to or from to is not known, then by the inductive hypothesis, we are done. Otherwise, notice that the path contains two subpaths, one from to and one from to (the latter is trivial as it consists of a single edge), and that each of these must itself be a shortest path, otherwise we could replace it with a shorter path to obtain a shorter path from to , a contradiction. Now, as the distances from to and to are correctly known, and the correct distance from to is exactly the sum of the distances from to and to (as these values equal the weights of the aforementioned subpaths), and our overestimate of the distance from to is incorrect, it must be strictly greater than the sum of the distances from to and to . Hence can be relaxed.