Algorithms

Shortest Path Problems

• How can we find the shortest route between
two points on a map?
• Model the problem as a graph problem:
– Road map is a weighted graph:
vertices = cities
edges = road segments between cities
edge weights = road distances
– Goal: find a shortest path between two vertices
(cities)

University of Hyderabad
 Shortest Path Problems
• Input: t x
6
3 9
– Directed graph G = (V, E)
3
4
– Weight function w : E → R s 0
2 1
2 7
3
• Weight of path p = ⟨v0, v1, . . . , vk⟩5
k 5 11
6
w( p )  w( vi  1 , vi ) y z
i 1

• Shortest-path weight from u to v:

p
δ(u, v) = min w(p) : u v if there exists a path from u
to v
∞ otherwise
University of Hyderabad
• Shortest path u to v is any path p such that w(p) = δ(u,
 Variants of Shortest Paths
• Single-source shortest path
– G = (V, E) ⇒ find a shortest path from a given source
vertex s to each vertex v ∈ V
• Single-destination shortest path
– Find a shortest path to a given destination vertex t from
each vertex v
– Reverse the direction of each edge ⇒ single-source
• Single-pair shortest path
– Find a shortest path from u to v for given vertices u and
v
– Still have to solve the single-source problem
• All-pairs shortest-paths
– Find a shortest path from u to v for every pair of vertices
u and v University of Hyderabad
 Negative-Weight Edges
• s → a: only one path What if we have negative-
𝛅(s, a) = w(s, a) = 3 weight edges?
a b
• s → b: only one path 3
-4
-1

𝛅(s, b) = w(s, a) + w(a, b) = -1

3 4
c 6 d g
5 8
s 0 5 11 -∞
-3
• s → c: infinitely many paths 2 3 7
-∞ -∞
⟨s, c⟩, ⟨s, c, d, c⟩, ⟨s, c, d, c, d, c⟩ e -6 f

cycle ⟨c, d, c⟩ has positive weight (6 - 3 = 3)

⟨s, c⟩ is shortest path with weight 𝛅(s, b) = w(s, c) = 5

University of Hyderabad
 Negative-Weight Edges
• s → e: infinitely many paths: a b
-4
⟨s, e⟩, ⟨s, e, f, e⟩, ⟨s, e, f, e, f, e⟩ 3 -1
3 4
c d g
– cycle ⟨e, f, e⟩ has negative 5
6
8
s 0 5 11 -∞
weight: -3
2 3 7
3 + (- 6) = -3 -∞ -∞
-6 f
– can find paths from s to e with e

arbitrarily large negative h i

2
weights
– 𝛅(s, e) = - ∞ ⇒ no shortest path
∞ ∞
h, i, j not
-8 3 reachable

• Similarly: 𝛅(s, f) = - ∞,
exists between s and e ∞ from s

𝛅(s, h) = 𝛅(s, i) = 𝛅(s, j) = ∞

𝛅(s, g) = - ∞
j

University of Hyderabad
 Negative-Weight Edges
a b
• Negative-weight edges may form -4

3 4
negative-weight cycles c 6 d g
5 8
s 0
-3
2 3 7
• If such cycles are reachable from

the source: 𝛅(s, v) is not properly

e -6 f

defined for any node v on the cycle

– Keep going around the cycle, and get

w(s, v) = - ∞ for all v on the cycle

University of Hyderabad
 Cycles
• Can shortest paths contain cycles?
• Negative-weight cycles No!
• Positive-weight cycles: No!
– By removing the cycle we can get a shorter path
• Zero-weight cycles
– No reason to use them
– Can remove them to obtain a path with same
weight
• We will assume that when we are finding
shortest paths, the paths will have no cycles
University of Hyderabad
 Shortest-Path Representation
For each vertex v ∈ V:
t x
• d[v] = δ(s, v): a shortest-path 3
6
9
estimate 3
4
2 1
– Initially, d[v]=∞ s 0
2 7
3
– Reduces as algorithms progress 5

• 𝛑[v] = predecessor of v on a
5 11
6
y z

shortest path from s

– If no predecessor, 𝛑[v] = NIL
– 𝛑 induces a tree—shortest-path
tree
• Shortest paths & shortest path
University of Hyderabad
 Initialization
Alg.: INITIALIZE-SINGLE-SOURCE(V, s)
1. for each v ∈ V
2. do d[v] ← ∞
3. 𝛑[v] ← NIL
4. d[s] ← 0

• All the shortest-paths algorithms start

with INITIALIZE-SINGLE-SOURCE

University of Hyderabad
 Relaxation
• Relaxing an edge (u, v) = testing whether
we can improve the shortest path to v
found so far by going through u
If d[v] > d[u] + w(u, v)

⇒ update d[v] and 𝛑[v]

we can improve the shortest path to v

s s
u v u v
5
2
9
2 After relaxation:
5 6
d[v] ≤ d[u] +
RELAX(u, v, w) RELAX(u, v,w(u,
w) v)

u v u v
2 2
5 7 5 6
University of Hyderabad
 RELAX(u, v, w)
1. if d[v] > d[u] + w(u, v)

𝛑[v] ← u
2. then d[v] ← d[u] + w(u, v)
3.

• All the single-source shortest-paths

algorithms
– start by calling INIT-SINGLE-SOURCE
– then relax edges
• The algorithms differ in the order and
how many times they relax each edge

University of Hyderabad
 Bellman-Ford Algorithm
• Single-source shortest paths problem
– Computes d[v] and 𝛑[v] for all v ∈ V
• Allows negative edge weights and cycles
• Returns:
– TRUE if no negative-weight cycles are
reachable from the source s
– FALSE otherwise ⇒ no solution exists
• Idea:
– Traverse all the edges |V| – 1 times, every time
performing a relaxation step of each edge

University of Hyderabad
 BELLMAN-FORD(V, E, w, s)
t x
1. INITIALIZE-SINGLE-SOURCE(V, 5
∞ ∞
s) 6 -2
-3
8
2. for i ← 1 to |V| - 1 s 0
-4
7

3. do for each edge (u, v) ∈ E 7 2

∞ ∞
4. do RELAX(u, v, w) y
9
z
5. for each edge (u, v) ∈ E t 5 x
6
∞ ∞
6. do if d[v] > d[u] + w(u, v) 6 -2
-3
8
7. then return FALSE s 0 -4
7

8. return TRUE 7 2
∞
7 ∞
9
y z

University of Hyderabad
Example
(t, x), (t, y), (t, z), (x, t), (y, x), (y, z), (z, x), (z, s), (s, t), (s, y)

t 5 x t x
Pass 1 5
6 ∞ Pass 2
6 4
∞
11
6 -2 6 -2
-3 -3
8 7 8 7
s 0 s 0
-4 -4
7 2 7 2
7 ∞ 7 2
∞
9 9
y z y z
Pass 3 t 5 x Pass 4 t 5 x
2
6 4
∞
11 2
6 4
∞
11
6 -2 6 -2
-3 -3
8 7 8 7
s 0 s 0
-4 -4
7 2 7 2
7 2
∞ 7 ∞2
-2
9 9
y z y z

Figure 1
Example 1

University of Hyderabad
Example 1 Sol

University of Hyderabad
 Detecting Negative Cycles
• for each edge (u, v) ∈ E s b
• do if d[v] > d[u] + w(u, v) 0
2
∞
• then return FALSE -8 3
• return TRUE ∞
c

s b s b Look at edge (s, b):

2 2
0
-3 ∞
2 -6
-3 -1
2
d[b] = -1
-8 3 -8 3
∞
5 5
2 d[s] + w(s, b) = -4
c c
⇒ d[b] > d[s] + w(s,
b)
University of Hyderabad
 Shortest Path Properties
• Triangle inequality s
u v
For all (u, v) ∈ E, we have: 5
2
7

δ(s, v) ≤ δ(s, u) + w(u, v)

s
u v
2
5 6

• If u is on the shortest path to v we have the

equality sign

University of Hyderabad
 Shortest Path Properties
• Upper-bound property
We always have d[v] ≥ δ(s, v) for all v.
Once d[v] = δ(s, v), it never changes.
– The estimate never goes up – relaxation only
lowers the estimate
v 5 x v 5 x
6 4
∞
11 2
6 4
∞
11
6 -2 Relax (x, v) 6 -2
-3 -3
8 7 8 7
s 0 s 0
-4 -4
7 2 7 2
7 2
∞ 7 2
∞
9 9
y z y z
University of Hyderabad
 Shortest Path Properties
• No-path property
If there is no path from s to v then d[v] = ∞
always.
– δ(s, h) = ∞ and d[h] ≥ δ(s, h) ⇒ d[h] = ∞
a b
-4 h i
3 -1 2
3 4 ∞ ∞
c 6 d g h, i, j not
5 8
s 0 5 11 -∞ -8 3 reachable
-3 ∞
2
y from s
3 7

𝛅(s, h) = 𝛅(s, i) = 𝛅(s, j) = ∞

-∞ -∞ j
e -6 f

University of Hyderabad
 Shortest Path Properties
• Convergence property
If s u → v is a shortest path, and if d[u] =
δ(s, u) at any time prior to relaxing edge (u,
v), then
d[v] = δ(s, v) at all times afterward
u v • If d[v] > δ(s, v) ⇒ after
2
5
6 8
∞
11 relaxation:
5
d[v] = d[u] + w(u, v)
s 0 d[v] = 5 + 2 = 7
4
• Otherwise, the value remains
4
7 unchanged, because it must
University have been the shortest path
of Hyderabad
 Shortest Path Properties
• Path relaxation property
Let p = ⟨v0, v1, . . . , vk⟩ be a shortest path
from s = v0 to vk. If we relax, in order, (v0,
v1), (v1, v2), . . . , (vk-1, vk), even
intermixed with other relaxations, then d[vk ]
= δ(s, vvk). 1 2 v2
d[v2] = δ(s, v2)
∞
6
5 7
∞
11 v4
5 14
∞
v3 3
d[v1] = δ(s, v1) 4
s 0 ∞
11 d[v4] = δ(s, v4)

d[v3] = δ(s, v3)

University of Hyderabad
 All-Pairs Shortest Paths
• Given: 2
3 4
– Directed graph G = (V, E)
8
1 3
– Weight function w : E → R 2
1
• Compute: -4 7 -5

5 4
6
– The shortest paths between all
pairs of vertices in a graph
– Representation of the result: an
n × n matrix of shortest-
path distances δ(u, v)

University of Hyderabad
 All-Pairs Shortest Paths
• Assume the graph (G) is given as 2
4
adjacency matrix of weights 3
8
– W = (wij), n x n matrix, |V| = n 1
2
3

– Vertices numbered 1 to n -4 7
1
-5

0 if i = j 5
6
4

wij = weight of (i, j) if i ≠ j , (i, j) ∈ E

∞ if i ≠ j , (i, j) ∉ E
• Output the result in an n x n matrix
D = (dij), where dij = δ(i, j)
• Solve the problem using dynamic
programming University of Hyderabad
 Optimal Substructure of a Shortest
Path
at most m edges
• All subpaths of a
∞
11
j
shortest path are i
k
shortest paths
at most m - 1 edges
• Let p: a shortest path
from vertex i to j that • If i ≠ j: p = ip’ k→j
– p’ has at most m-1
contains at most m
edges
edges – p’ is a shortest path
• If i = j δ(i, j) =δ(i, k) + wkj
– w(p) = 0 and p has no of Hyderabad
University
 Recursive Solution
• lij(m) = weight of shortest path i j that
at most m edges
contains at most m edges
∞
11
j
i
• m = 0: lij (0)
= 0 if i = j k

∞ if i ≠ j
– No edges allowed

• m = 1: lij(1) wij L(1) = W

=
– The path between i and j is restricted to 1 edge

University of Hyderabad
 Recursive Solution
• lij(m) = weight of shortest path i j that
at most m edges
contains at most m edges
∞
11
j
i
k

• m > 1: lij(m) min

= { l ij
(m-1) ,min {l (m-1) + w } }
ik kj
1≤k≤n
= min {lik(m-1) + wkj}
1≤k≤n

– Shortest path from i to j with at most m – 1 edges

– Shortest path from i to j containing at most m
edges, considering all possible predecessors (k)
University of Hyderabad
of j
 Computing the Shortest Paths
• m = 1: lij(1) = wij L(1) = W
– The path between i and j is restricted to 1 edge

• Given W = (wij), compute: L(1), L(2), …, L(n-1), where

L(m) = (lij(m))
• L(n-1) contains the actual shortest-path weights
Given L(m-1) and W ⇒ compute L(m)
– Extend the shortest paths computed so far by one more
edge
• If the graph has no negative cycles: all simple
shortest paths contain at most n - 1 edges
(n+1) = lij
(n-1)
δ(i, j) = l (n-1)
and l
ij
(n)
l ...
ij , ij
University of Hyderabad
 The Floyd-Warshall Algorithm
• Given: 2
4
– Directed, weighted graph G = (V, 3
8
E) 1 3
2
– Negative-weight edges may be -4 1
-5
7
present 5 4
6
– No negative-weight cycles could
be present in the graph
• Compute:
– The shortest paths between all
pairs of vertices in a graph

University of Hyderabad
 The Structure of a Shortest
Path
3
• Vertices in G are given by 6
0.5
1 2
V = {1, 2, …, n} 3 4
2 1
• Consider a path p = ⟨v1, v2, …, 2

5
vl ⟩
– An intermediate vertex of p is

any vertex in the set {v2, v3, …, vl-

1 }

– E.g.: p = ⟨1, 2, 4, 5⟩: {2, 4}

University of Hyderabad
 The Structure of a Shortest
Path
• For any pair of vertices i, j ∈ V, consider all
paths from i to j whose intermediate
vertices are all drawn from a subset {1, 2,
…, k}
– Find p, a minimum-weight
p1 path from these
pu
paths i j

pt

No vertex on these paths has index > k

University of Hyderabad
 Example
dij(k) = the weight of a shortest path from
vertex i to vertex j with all intermediary
vertices drawn from {1, 2, …, k}
• d13(0) = 6

= 6
3
• d13 (1) 6
0.5
1 2
5 4
• d13 (2)
= 3
2 1
5
• d13 (3)
=
4.5
• d13(4) =
University of Hyderabad
 A Recursive Solution
• k is not an intermediate vertex of path p k
– Shortest path from i to j with intermediate i j

vertices from {1, 2, …, k} is a shortest path

from i to j with intermediate vertices from {1,
2, …, k - 1}

• k is an intermediate vertex of path p p1 ∞

k p2
j
– p1 is a shortest path from i to k i

– p2 is a shortest path from k to j

– p1 and p2 are shortest paths from i to k and k
to j respectively, with vertices from {1, 2, …,
k - 1} University of Hyderabad
 A Recursive Solution
dij(k) = the weight of a shortest path
from vertex i to vertex j with all
intermediary vertices drawn from {1,
2, …, k}
• k=0
• dij(k) = wij

University of Hyderabad
 A Recursive Solution
dij(k) = the weight of a shortest path
from vertex i to vertex j with all
intermediary vertices drawn from {1,
2, …, k}
• k≥1
• Case 1: k is not an
intermediate vertex of path k

p dij(k-1) i j

• dij(k) =

University of Hyderabad
 A Recursive Solution
dij(k) = the weight of a shortest path
from vertex i to vertex j with all
intermediary vertices drawn from {1,
2, …, k}
• k≥1
• Case 2: k is an
∞
k
intermediate vertex of path
i
j

p dik(k-1) + dkj(k-1)
• dij(k) =

University of Hyderabad
 Computing the Shortest Path
Weights
• dij(k) = wij if k = 0
min {dij(k-1) , dik(k-1) + dkj(k-1) } if
k≥1
• The final solution: D(n) = (dij(n)):
dij = 𝛅(i, j) ∀ i, j ∈ V
(n) j j

+ (k, j)

i i
(i, k)
D(k-1) D(k)

Compute D(0), D(1), D(2),…, D(n)

University of Hyderabad
 FLOYD-WARSHALL(W)
1. n ← rows[W]
2. D(0) ← W
3. for k ← 1 to n
4. do for i ← 1 to n
5. do for j ← 1 to n
6. do dij(k) ← min (dij(k-1), dik(k-1) +
dkj(k-1))
7. return D(n)

• Running time: Θ(n3)

University of Hyderabad
 Example dij(k) = min {dij(k-1) , dik(k-1) + dkj(k-1) }
D(0) = W D(1)
2 1 2 3 4 5 1 2 3 4 5
3 4 1 0 3 8 ∞ -4 1 0 3 8 ∞ -4
8
1 3 2 ∞ 0 ∞ 1 7 2 ∞ 0 ∞ 1 7
2

-4
1
-5
3 ∞ 4 0 ∞ ∞ 3 ∞ 4 0 ∞ ∞
7
5 4 4 2 ∞ -5 0 ∞ 4 2 5 -5 0 -2
6
D(2) 5 ∞ ∞ ∞ 6 0 5 ∞ ∞ ∞ 6 0
1 2 3 4 5 D(3) D(4)
1 0 3 8 4 -4 0 3 8 4 -4 0 3 -1 4 -4
2 ∞ 0 ∞ 1 7 ∞ 0 ∞ 1 7 3 0 -4 1 -1
3 ∞ 4 0 5 11 ∞ 4 0 5 11 7 4 0 5 3
4 2 5 -5 0 -2 2 -1 -5 0 -2 2 -1 -5 0 -2
5 ∞ ∞ ∞ 6 0 ∞ ∞ ∞ 6 0 8 5 1 6 0
CS 477/677 - Lecture 24
University of Hyderabad
University of Hyderabad
Example 1

University of Hyderabad
Solution

University of Hyderabad