exercise session 6 – shortest paths

Exercise Session 6 – Shortest PathsComputer Science II, D-ITET, ETH Zurich

Program Today

Feedback of last exercises

Shortest pathsDijkstraHeaps, DecreaseKey and Lazy DeletionRunning Time of the AlgorithmsDijkstra and Negative Edge Weights?

In-Class-Exercises

1

1. Feedback of last exercises

2

Depth-�rst-search and Breadth-�rst-search

A

B

C D

E

FG

H

Starting at ADFS: A,B,C,D,E, F,H,GBFS: A,B, F, C,H,D,G,E

There is no starting vertex where the DFS ordering equals the BFS ordering.

3

Depth-�rst-search and Breadth-�rst-search

A

B

C D

E

FG

H

Starting at ADFS: A,B,C,D,E, F,H,GBFS: A,B, F, C,H,D,G,EThere is no starting vertex where the DFS ordering equals the BFS ordering.

3

Depth-�rst-search and Breadth-�rst-search

Star: DFS ordering equals BFS ordering

A

B

C D

E

Starting at ADFS: A,B,C,D,EBFS: A,B,C,D,E

Starting at CDFS: C,A,B,D,EBFS: C,A,B,D,E

4

Depth-�rst-search and Breadth-�rst-search

Star: DFS ordering equals BFS ordering

A

B

C D

E

Starting at ADFS: A,B,C,D,EBFS: A,B,C,D,E

Starting at CDFS: C,A,B,D,EBFS: C,A,B,D,E

4

Topological Sorting

A B

C

D

E

Graph with cycles

Two minimal cycles sharing anedgeRemove edge =⇒ cycle-freeTopological Sorting by“removing” elements within-degree 0

5

Topological Sorting

A B

C

D

E

Graph with cyclesTwo minimal cycles sharing anedge

Remove edge =⇒ cycle-freeTopological Sorting by“removing” elements within-degree 0

5

Topological Sorting

A B

C

D

E

Graph with cyclesTwo minimal cycles sharing anedgeRemove edge =⇒ cycle-free

Topological Sorting by“removing” elements within-degree 0

5

Topological Sorting

A B

C

D

E

Graph with cyclesTwo minimal cycles sharing anedgeRemove edge =⇒ cycle-freeTopological Sorting by“removing” elements within-degree 0

5

Topological Sorting

A B

C

D

E

A Graph with cyclesTwo minimal cycles sharing anedgeRemove edge =⇒ cycle-freeTopological Sorting by“removing” elements within-degree 0

5

Topological Sorting

A B

C

D

E

A B Graph with cyclesTwo minimal cycles sharing anedgeRemove edge =⇒ cycle-freeTopological Sorting by“removing” elements within-degree 0

5

Topological Sorting

A B

C

D

E

A B

C

Graph with cyclesTwo minimal cycles sharing anedgeRemove edge =⇒ cycle-freeTopological Sorting by“removing” elements within-degree 0

5

Topological Sorting

A B

C

D

E

A B

C E

Graph with cyclesTwo minimal cycles sharing anedgeRemove edge =⇒ cycle-freeTopological Sorting by“removing” elements within-degree 0

5

2. Shortest paths

6

Shortest Paths

Given: G = (V,E, c), c : E → R, s, t ∈ V .

Path: s p t : 〈s = v0, v1, . . . , vk = t〉, (vi, vi+1) ∈ E (0 ≤ i < k)

Weight: c(p) := ∑k−1i=0 c((vi, vi+1)).

Weight of a shortest path from u to v:

δ(u, v) =

∞ no path from u to vmin{c(p) : u p

v} sonst

7

General Algorithm

1. Initialise ds and πs: ds[v] =∞, πs[v] = null for each v ∈ V2. Set ds[s]← 03. Choose an edge (u, v) ∈ E

Relaxiere (u, v):if ds[v] > d[u] + c(u, v) then

ds[v]← ds[u] + c(u, v)πs[v]← u

4. Repeat 3 until nothing can be relaxed any more.(until ds[v] ≤ ds[u] + c(u, v) ∀(u, v) ∈ E)

8

Algorithm Dijkstra(G, s)Input: Positively weighted Graph G = (V,E, c), starting point s ∈ V ,Output: Minimal weights d of the shortest paths and corresponding predecessor

node for each node.

foreach u ∈ V dods[u]←∞; πs[u]← ∅

ds[s]← 0; R← {s}while R 6= ∅ do

u← ExtractMin(R)foreach v ∈ N+(u) do

if ds[u] + c(u, v) < ds[v] thends[v]← ds[u] + c(u, v)πs[v]← uR← R ∪ {v}

9

Example

s

a

b

c

d

e

2

3

2

6

1

3

1

1

10

Example

s

a

b

c

d

e

2

3

2

6

1

3

1

1

0

∞

∞

∞

∞

∞s

M = {s}

R = {}

U = {a, b, c, d, e}

10

Example

s

a

b

c

d

e

2

3

2

6

1

3

1

1

0

∞

∞

∞

∞

∞ss

a

b

2

3

M = {s}

R = {a, b}

U = {c, d, e}

10

Example

s

a

b

c

d

e

2

3

2

6

1

3

1

1

0

∞

∞

∞

∞

∞ss

a

b

2

3

a c8

M = {s, a}

R = {b, c}

U = {d, e}

10

Example

s

a

b

c

d

e

2

3

2

6

1

3

1

1

0

∞

∞

∞

∞

∞ss

a

b

2

3

a c8

b d4

M = {s, a, b}

R = {c, d}

U = {e}

10

Example

s

a

b

c

d

e

2

3

2

6

1

3

1

1

0

∞

∞

∞

∞

∞ss

a

b

2

3

a c8

b d4

d

e5

7

M = {s, a, b, d}

R = {c, e}

U = {}

10

Example

s

a

b

c

d

e

2

3

2

6

1

3

1

1

0

∞

∞

∞

∞

∞ss

a

b

2

3

a c8

b d4

d

e5

7

e

6

M = {s, a, b, d, e}

R = {c}

U = {}

10

Example

s

a

b

c

d

e

2

3

2

6

1

3

1

1

0

∞

∞

∞

∞

∞ss

a

b

2

3

a c8

b d4

d

e5

7

e

6c

M = {s, a, b, d, e, c}

R = {}

U = {}

10

Implementation: Data Structure for R?

Relax for Dijkstra:if ds[u] + c(u, v) < ds[v] then

ds[v]← ds[u] + c(u, v)πs[v]← uif v 6∈ R then

Add(R, v) // Update of (v, d(v)) in the heap of Relse

DecreaseKey(R, v) // Update of a (v, d(v)) in the heap of R

11

DecreaseKey ?

Heap ( (a, 1), (b, 4), (c, 5), (d, 8) ) =(a,1)

(b,4)

(d,8)

(c,5)

after DecreaseKey(d, 3):

(a,1)

(d,3)

(b,4)

(c,5)

2 Probleme:Position of d unknown at �rst. Seach: Θ(n)Positions of the nodes can change during DecreaseKey

12

DecreaseKey ?

Heap ( (a, 1), (b, 4), (c, 5), (d, 8) ) =(a,1)

(b,4)

(d,8)

(c,5)

after DecreaseKey(d, 3):(a,1)

(d,3)

(b,4)

(c,5)

2 Probleme:

Position of d unknown at �rst. Seach: Θ(n)Positions of the nodes can change during DecreaseKey

12

DecreaseKey ?

Heap ( (a, 1), (b, 4), (c, 5), (d, 8) ) =(a,1)

(b,4)

(d,8)

(c,5)

after DecreaseKey(d, 3):(a,1)

(d,3)

(b,4)

(c,5)

2 Probleme:Position of d unknown at �rst. Seach: Θ(n)Positions of the nodes can change during DecreaseKey

12

Lazy Deletion !Heap ( (a, 1), (b, 4), (c, 5), (d, 8) ) =

(a,1)

(b,4)

(d,8)

(c,5)

Insert(d, 3):

(a,1)

(d,3)

(d,8) (b,4)

(c,5)

ExtractMin()→ (a, 1)(d,3)

(b,4)

(d,8)

(c,5)

ExtractMin()→ (d, 3)(b,4)

(d,8) (c,5)

Later ExtractMin()→ (d, 8) must be ignored

13

Lazy Deletion !Heap ( (a, 1), (b, 4), (c, 5), (d, 8) ) =

(a,1)

(b,4)

(d,8)

(c,5)

Insert(d, 3):(a,1)

(d,3)

(d,8) (b,4)

(c,5)

ExtractMin()

→ (a, 1)(d,3)

(b,4)

(d,8)

(c,5)

ExtractMin()→ (d, 3)(b,4)

(d,8) (c,5)

Later ExtractMin()→ (d, 8) must be ignored

13

Lazy Deletion !Heap ( (a, 1), (b, 4), (c, 5), (d, 8) ) =

(a,1)

(b,4)

(d,8)

(c,5)

Insert(d, 3):(a,1)

(d,3)

(d,8) (b,4)

(c,5)

ExtractMin()→ (a, 1)(d,3)

(b,4)

(d,8)

(c,5)

ExtractMin()

→ (d, 3)(b,4)

(d,8) (c,5)

Later ExtractMin()→ (d, 8) must be ignored

13

Lazy Deletion !Heap ( (a, 1), (b, 4), (c, 5), (d, 8) ) =

(a,1)

(b,4)

(d,8)

(c,5)

Insert(d, 3):(a,1)

(d,3)

(d,8) (b,4)

(c,5)

ExtractMin()→ (a, 1)(d,3)

(b,4)

(d,8)

(c,5)

ExtractMin()→ (d, 3)(b,4)

(d,8) (c,5)

Later ExtractMin()→ (d, 8) must be ignored13

Runtime Dijkstra

n := |V |, m := |E|n× ExtractMin: O(n log n)m× Insert or DecreaseKey: O(m log |V |)1× Init: O(n)Overal: O((n+m) log n). for connected graphs: O(m log n)

14

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS

O(m+ n) O(n2) O(n)DAG Top-Sort O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n)

O(n2) O(n)DAG Top-Sort O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2)

O(n)DAG Top-Sort O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2) O(n)DAG Top-Sort

O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2) O(n)DAG Top-Sort O(m+ n)

O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2) O(n)DAG Top-Sort O(m+ n) O(n2)

O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2) O(n)DAG Top-Sort O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra

O((m+ n) logn) O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2) O(n)DAG Top-Sort O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn)

O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2) O(n)DAG Top-Sort O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn)

O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2) O(n)DAG Top-Sort O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn) O(n logn)

general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

Quiz

n := |V |,m := |E|, c : E → R

problem method runtime dense sparsem ∈ O(n2) m ∈ O(n)

c ≡ 1 BFS O(m+ n) O(n2) O(n)DAG Top-Sort O(m+ n) O(n2) O(n)c ≥ 0 Dijkstra O((m+ n) logn) O(n2 logn) O(n logn)general Bellman-Ford1 O(m · n) O(n3) O(n2)

1will be covered later in class (dynamic programming)15

An Interesting Graph

s t32 16 8 4 2

5-5

10-10

19-19

36-36

69-69

Does Dijkstra work? (Try it out!)

16

Answer

Dijkstra (as we have presented it) works also for graphs with negative edgeweights, if no negative weight cycles are present. But Dijkstra may thenexhibit exponential running time!

17

Answer

Dijkstra (as we have presented it) works also for graphs with negative edgeweights, if no negative weight cycles are present. But Dijkstra may thenexhibit exponential running time!

17

Shortest Path in a Maze

18

A*-Algorithm(G, s, t, h)Input: Positively weighted Graph G = (V,E, c), starting point s ∈ V , end point

t ∈ V , estimate h(v) ≤ δ(v, t)Output: Existence and value of a shortest path from s to t

foreach u ∈ V do

d[u]←∞; f [u]←∞; π[u]← null

d[s]← 0; f [s]← h(s); R← {s}; M ← {}while R 6= ∅ do

u← ExtractMinf(R); M ←M ∪ {u}

if u = t then return successforeach v ∈ N+(u) with d[v] > d[u] + c(u, v) do

d[v]← d[u] + c(u, v); f [v]← d[v] + h(v); π[v]← uR← R ∪ {v}; M ←M − {v}

return failure19

Dijkstra and A*Edge data structure

Stores length and the destination nodeUsed for the edges in the graph, as well as for the back-edges of theshortest path

Graph data structureNodeP: Pointer to a nodestd::vector<Edge> get_adj(NodeP src): Returns a vector of Edgestarting from src

std::map<NodeP,Edge>

Maps a NodeP to an Edgem[u] Returns an edge

20

Dijkstra and A*Edge data structure

Stores length and the destination nodeUsed for the edges in the graph, as well as for the back-edges of theshortest path

Graph data structureNodeP: Pointer to a nodestd::vector<Edge> get_adj(NodeP src): Returns a vector of Edgestarting from src

std::map<NodeP,Edge>

Maps a NodeP to an Edgem[u] Returns an edge

20

Dijkstra and A*Edge data structure

Stores length and the destination nodeUsed for the edges in the graph, as well as for the back-edges of theshortest path

Graph data structureNodeP: Pointer to a nodestd::vector<Edge> get_adj(NodeP src): Returns a vector of Edgestarting from src

std::map<NodeP,Edge>

Maps a NodeP to an Edgem[u] Returns an edge

20

Dijkstra and A*

Node StructStores x and y coordinates

Manhattan Distance:d = |∆x|+ |∆y|

std::pair

Access with p.first und p.secondUsed to store nodes with their distances from s in the heap→ LazyDeletion

21

Dijkstra and A*

Node StructStores x and y coordinates

Manhattan Distance:d = |∆x|+ |∆y|

std::pair

Access with p.first und p.secondUsed to store nodes with their distances from s in the heap→ LazyDeletion

21

Dijkstra and A*

Node StructStores x and y coordinates

Manhattan Distance:d = |∆x|+ |∆y|

std::pair

Access with p.first und p.secondUsed to store nodes with their distances from s in the heap→ LazyDeletion

21

3. In-Class-Exercises

22

In-Class Programming Exercise

Implement lazy deletion in a heap.−→ CodeExpert

23

In-Class-Exercises (theory): Route planning

Exercise: You are givena directed, unweighted Graph G = (V,E),represented by an adjacency list,and a designated node t ∈ V (e.g., an emergency exit).

Design an algorithm,which computes for each node u ∈ V an outgoing edge in direction ofa shortest path to t.and has a running time of O(|V |+ |E|).

24

In-Class-Exercises: Route planning

Solution:1. Make a copy of the graph with edges having reverse direction:GT = (V,ET ), where ET = {(v, u) | (u, v) ∈ E}.Running time: O(|V |+ |E|).

2. Start a breadth-�rst search of GT , starting from t,and store all edges of the BFS-Tree.Running time: O(|V |+ |ET |) = O(|V |+ |E|).

3. Assign the stored edges (in reverse direction)to the discovered nodes. Running time: O(|V |).

25