CSE 332 Data Abstractions:

Graphs and Graph Traversals

Kate Deibel
Summer 2012

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 1

 Some course stuff and a humorous story

GRADES, MIDTERMS, AND IT

SEEMED LIKE A GOOD IDEA

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 2

The Midterm
It was too long—I admit that

If it helps, this was the first exam I have

ever written

Even still, I apologize

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 3

You All Did Great
I am more than pleased with your
performances on the midterm

The points you missed were clearly due to

time constraints and stresses

You showed me you know the material

Good job!

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 4

How Grades are Calculated
Many (if not most) CSE major courses use
curving to determine final grades
 Homework and exam grades are used as
indicators and are adjusted as necessary
 Example:
A student who does excellent on
homework and projects (and goes
beyond) will get a grade bumped up
even if his/her exam scores are poorer

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 5

My Experiences as a Teacher
Timed exams are problematic
 Some of the best students I have known did
not do great on exams

The more examples of student work that one

sees, the more learning becomes evident
 Even partial effort/incomplete work tells a lot
 Unfortunately, this means losing points

The above leads to missing points

 All students (even myself back in the day) care
about points

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 6

My Repeated Mistake
As a teacher, I should talk more about how
points get transformed into a final grade

I learned this lesson my first year as a TA…

… and indirectly caused the undergraduate

CSE servers to crash

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 7

It Seemed Like a Good Idea at the Time
At the annual CS education conference
(SIGCSE), there is a special panel about
teaching mistakes and learning from them

I contributed a story at SIGCSE 2009:

http://faculty.washington.edu/deibel/
presentations/sigcse09-good-idea/deibel-
seemed-good-idea-sigcse-2009.ppt

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 8

My Promises
I know you will miss points

If you do the work in the class and put in

the effort, you will earn more than a
passing grade

As long as you show evidence of learning,

you will earn a good grade regardless

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 9

What This Means For You
Keep up the good work

Do not obsess over points

The final will be less intense

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 10

 That was fun but you are here for learning…

BACK TO CSE 332 AND

GRAPH [THEORY]

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 11

Where We Are
We have learned about the essential ADTs
and data structures:
 Regular and Circular Arrays (dynamic sizing)
 Linked Lists
 Stacks, Queues, Priority Queues
 Heaps
 Unbalanced and Balanced Search Trees

We have also learned important algorithms

 Tree traversals
 Floyd's Method
 Sorting algorithms

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 12

Where We Are Going
Less generalized data structures and ADTs

More on algorithms and related problems

that require constructing data structures
to make the solutions efficient

Topics will include:

 Graphs
 Parallelism

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 13

Graphs
A graph is a formalism for representing
relationships among items
Han Luke
 Very general definition
 Very general concept
Leia
A graph is a pair: G = (V, E) V = {Han,Leia,Luke}
 A set of vertices, also known E = {(Luke,Leia),
as nodes: V = {v1,v2,…,vn} (Han,Leia),
(Leia,Han)}
 A set of edges E = {e1,e2,…,em}
 Each edge ei is a pair of vertices (vj,vk)
 An edge "connects" the vertices

Graphs can be directed or undirected

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 14
A Graph ADT?
We can think of graphs as an ADT
 Operations would inlude isEdge(vj,vk)
 But it is unclear what the "standard
operations" would be for such an ADT

Instead we tend to develop algorithms over

graphs and then use data structures that are
efficient for those algorithms

Many important problems can be solved by:

1. Formulating them in terms of graphs
2. Applying a standard graph algorithm

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 15

Some Graphs
For each example, what are the vertices and
what are the edges?
 Web pages with links
 Facebook friends
 "Input data" for the Kevin Bacon game
 Methods in a program that call each other
 Road maps
 Airline routes
 Family trees
 Course pre-requisites

Core algorithms that work across such domains

is why we are CSE
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 16
Scratching the Surface
Graphs are a powerful representation and
have been studied deeply

Graph theory is a major branch of research

in combinatorics and discrete mathematics

Every branch of computer science involves

graph theory to some extent

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 17

 To make formulating graphs easy and standard,
we have a lot of standard terminology for graphs

GRAPH TERMINOLOGY

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 18

Undirected Graphs
In undirected graphs, edges have no specific
direction D
 Edges are always "two-way" A
C

Thus, (u, v) ∊ E implies (v, u) ∊ E. B

 Only one of these edges needs to be in the set
 The other is implicit, so normalize how you
check for it

Degree of a vertex: number of edges

containing that vertex
 Put another way: the number of adjacent vertices

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 19

Directed Graphs
In directed graphs (or digraphs), edges have direction
D
D
A A
C or C
B B

2 edges here

Thus, (u, v) ∊ E does not imply (v, u) ∊ E.

Let (u, v)  E mean u → v
 Call u the source and v the destination
 In-Degree of a vertex: number of in-bound edges
(edges where the vertex is the destination)
 Out-Degree of a vertex: number of out-bound edges
(edges where the vertex is the source)

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 20

Self-Edges, Connectedness
A self-edge a.k.a. a loop edge is of the form (u, u)
 The use/algorithm usually dictates if a graph has:
 No self edges
 Some self edges
 All self edges

A node can have a(n) degree / in-degree / out-

degree of zero
A graph does not have to be connected
 Even if every node has non-zero degree
 More discussion of this to come

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 21

More Notation D

For a graph G = (V, E): A

C
 |V| is the number of vertices
 |E| is the number of edges B

 Minimum? V = {A, B, C, D}
 Maximum for undirected? E = {(C, B), (A, B),
 Maximum for directed? (B, A), (C, D)}

If (u, v) ∊ E , then v is a neighbor of u (i.e., v

is adjacent to u)
 Order matters for directed edges:
u is not adjacent to v unless (v, u)  E

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 22

More Notation D

For a graph G = (V, E): A

C
 |V| is the number of vertices
 |E| is the number of edges B

 Minimum? 0
 Maximum for undirected? |V||V+1|/2  O(|V|
2
)
 Maximum for directed? |V|2  O(|V|2)

If (u, v) ∊ E , then v is a neighbor of u (i.e., v

is adjacent to u)
 Order matters for directed edges:
u is not adjacent to v unless (v, u)  E
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 23
Examples Again
Which would use directed edges?
Which would have self-edges?
Which could have 0-degree nodes?
 Web pages with links
 Facebook friends
 "Input data" for the Kevin Bacon game
 Methods in a program that call each other
 Road maps
 Airline routes
 Family trees
 Course pre-requisites
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 24
Weighted Graphs
In a weighted graph, each edge has a weight or cost
 Typically numeric (ints, decimals, doubles, etc.)
 Orthogonal to whether graph is directed
 Some graphs allow negative weights; many do not
Clinton 20
Mukilteo

Kingston 30 Edmonds

Bainbridge 35
Seattle

60
Bremerton

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 25

Examples Again
What, if anything, might weights represent for
each of these?
Do negative weights make sense?

 Web pages with links

 Facebook friends
 "Input data" for the Kevin Bacon game
 Methods in a program that call each other
 Road maps
 Airline routes
 Family trees
 Course pre-requisites

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 26

 Paths and Cycles
We say "a path exists from v0 to vn" if there is a
list of vertices [v0, v1, …, vn] such that (vi,vi+1) ∊ E for
all 0  i<n.
A cycle is a path that begins and ends at the
same node (v0==vn) Chicago
Seattle

Salt Lake City

San Francisco
Dallas
Example path (that also happens to be a cycle):
[Seattle, Salt Lake City, Chicago, Dallas, San Francisco, Seattle]
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 27
 Path Length and Cost
Path length: Number of edges in a path
Path cost: Sum of the weights of each edge
Example where
P= [ Seattle, Salt Lake City, Chicago, Dallas,
San Francisco, Seattle]
3.5 Chicago
Seattle
2 2 length(P) = 5
cost(P) = 11.5
2 Salt Lake City
2.5
2.5 2.5 Length is sometimes
called "unweighted cost"
3
San Francisco Dallas
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 28
Simple Paths and Cycles
A simple path repeats no vertices (except the
first might be the last):
[Seattle, Salt Lake City, San Francisco, Dallas]
[Seattle, Salt Lake City, San Francisco, Dallas, Seattle]

A cycle is a path that ends where it begins:

[Seattle, Salt Lake City, Seattle, Dallas, Seattle]

A simple cycle is a cycle and a simple path:

[Seattle, Salt Lake City, San Francisco, Dallas, Seattle]

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 29

Paths and Cycles in Directed Graphs
Example: D

C
A

 Is there a path from A to D? No

 Does the graph contain any cycles? No

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 30

Undirected Graph Connectivity
An undirected graph is connected if for all
pairs of vertices u≠v, there exists a path
from u to v

Connected graph Disconnected graph

An undirected graph is complete,

or fully connected, if for all pairs
of vertices u≠v there exists an
edge from u to v

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 31

Directed Graph Connectivity
A directed graph is strongly connected
if there is a path from every vertex to
every other vertex

A directed graph is weakly connected

if there is a path from every vertex to
every other vertex ignoring direction
of edges

A direct graph is complete or fully

connected, if for all pairs of vertices
u≠v , there exists an edge from u to v

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 32

Examples Again
For undirected graphs: connected?
For directed graphs: strongly connected?
weakly connected?

 Web pages with links

 Facebook friends
 "Input data" for the Kevin Bacon game
 Methods in a program that call each other
 Road maps
 Airline routes
 Family trees
 Course pre-requisites

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 33

Trees as Graphs
When talking about graphs, we
say a tree is a graph that is: D E
 undirected B
 acyclic
A
 connected
C
All trees are graphs, but NOT all
graphs are trees F

G H
How does this relate to the trees
we know and "love"?
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 34
Rooted Trees
We are more accustomed to rooted trees where:
 We identify a unique root
 We think of edges as directed: parent to
children
D E
Picking a root gives a unique B
rooted tree A
A
 The tree is simply drawn
B C
differently and with C
undirected edges D E F
F

G H G H

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 35

Rooted Trees
We are more accustomed to rooted trees where:
 We identify a unique root
 We think of edges as directed: parent to
children
D E
Picking a root gives a unique B
F
rooted tree
A G H C
 The tree is simply drawn
differently and with C A
undirected edges B
F

G H D E

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 36

Directed Acyclic Graphs (DAGs)
A DAG is a directed graph with no directed cycles
 Every rooted directed tree is a DAG
 But not every DAG is a rooted directed tree

 Every DAG is a directed graph

 But not every directed graph is a DAG

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 37

Examples Again
Which of our directed-graph examples do you
expect to be a DAG?

 Web pages with links

 Facebook friends
 "Input data" for the Kevin Bacon game
 Methods in a program that call each other
 Road maps
 Airline routes
 Family trees
 Course pre-requisites

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 38

Density / Sparsity
Recall:
In an undirected graph, 0≤|E|< |V|2
Recall:
In a directed graph, 0≤|E|≤|V|2

So for any graph, |E| is O(|V|2)

Another fact:
If an undirected graph is connected,
then |E| ≥ |V|-1 (pigeonhole principle)

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 39

Density / Sparsity
|E| is often much smaller than its maximum size

We do not always approximate as |E| as O(|V|2)

 This is a correct bound, but often not tight

If |E| is (|V|2) (the bound is tight), we say the

graph is dense
 More sloppily, dense means "lots of edges"

If |E| is O(|V|) we say the graph is sparse

 More sloppily, sparse means "most possible
edges missing"

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 40

 Insert humorous statement here

GRAPH DATA STRUCTURES

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 41

What’s the Data Structure?
Graphs are often useful for lots of data and questions
 Example: "What’s the lowest-cost path from x to y"

But we need a data structure that represents graphs

Which data structure is "best" can depend on:

 properties of the graph (e.g., dense versus sparse)
 the common queries about the graph ("is (u ,v) an
edge?" vs "what are the neighbors of node u?")

We will discuss two standard graph representations

 Adjacency Matrix and Adjacency List
 Different trade-offs, particularly time versus space

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 42

Adjacency Matrix
Assign each node a number from 0 to |V|-1
A |V| x |V| matrix of Booleans (or 0 vs. 1)
 Then M[u][v] == true means there is an
edge from u to v

D A B C D
A A F T F F
C
B T F F F
B
C F T F T
D F F F F

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 43

Adjacency Matrix Properties
Running time to:
 Get a vertex’s out-edges:
 Get a vertex’s in-edges:
 Decide if some edge exists:
 Insert an edge:
 Delete an edge: A B C D
A F T F F
Space requirements: B T F F F
C F T F T
D F F F F
Best for sparse or dense graphs?

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 44

Adjacency Matrix Properties
Running time to:
 Get a vertex’s out-edges: O(|V|)
 Get a vertex’s in-edges: O(|V|)
 Decide if some edge exists: O(1)
 Insert an edge: O(1)
A B C D
 Delete an edge: O(1)
A F T F F

Space requirements: B T F F F
O(|V|2) C F T F T
D F F F F
Best for sparse or dense graphs? dense

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 45

Adjacency Matrix Properties
How will the adjacency matrix vary for an
undirected graph?
 Will be symmetric about diagonal axis
 Matrix: Could we save space by using only
about half the array? A B C D
A F T F F
B T F F F
C F T F T
D F F T F

 But how would you "get all neighbors"?

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 46
Adjacency Matrix Properties
How can we adapt the representation for
weighted graphs?
 Instead of Boolean, store a number in each cell
 Need some value to represent ‘not an edge’
 0, -1, or some other value based on how you
are using the graph
 Might need to be a separate field if no
restrictions on weights

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 47

Adjacency List
Assign each node a number from 0 to |V|-1
 An array of length |V| in which each
entry stores a list of all adjacent vertices
(e.g., linked list)

D
A B /
A
C
B A /
B
C D B /

D /

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 48

Adjacency List Properties
Running time to:
 Get a vertex’s out-edges: A B /

 Get a vertex’s in-edges: B A /

 Decide if some edge exists: C D B /

 Insert an edge: D /
 Delete an edge:

Space requirements:

Best for sparse or dense graphs?

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 49

Adjacency List Properties
Running time to:
 Get a vertex’s out-edges:
O(d) where d is out-degree of vertex
 Get a vertex’s in-edges:
O(|E|) (could keep a second adjacency list for this!)
 Decide if some edge exists:
O(d) where d is out-degree of source
 Insert an edge:
O(1) (unless you need to check if it’s already there)
 Delete an edge:
O(d) where d is out-degree of source

Space requirements: O(|V|+|E|)

Best for sparse or dense graphs? sparse

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 50

Undirected Graphs
Adjacency lists also work well for
undirected graphs with one caveat
 Put each edge in two lists to support
efficient "get all neighbors"

A B /
D
A B C / A /
C

B C D B /

D / C /

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 51

Which is better?
Graphs are often sparse
 Streets form grids
 Airlines rarely fly to all cities

Adjacency lists should generally be your

default choice
 Slower performance compensated by
greater space savings

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 52

 Might be easier to list what isn't a graph application…

APPLICATIONS OF
GRAPHS: TRAVERSALS

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 53

Application: Moving Around WA State

What’s the shortest way to get from

Seattle to Pullman?

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 54

Application: Moving Around WA State

What’s the fastest way to get from

Seattle to Pullman?

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 55

Application: Reliability of Communication

If Wenatchee’s phone exchange goes down,

can Seattle still talk to Pullman?

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 56

Application: Reliability of Communication

If Tacomas’s phone exchange goes down,

can Olympia still talk to Spokane?

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 57

Applications: Bus Routes Downtown

If we’re at 3rd and Pine, how can we get to

1st and University using Metro?
How about 4th and Seneca?
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 58
Graph Traversals
For an arbitrary graph and a starting node v,
find all nodes reachable from v (i.e., there
exists a path)
 Possibly "do something" for each node (print to
output, set some field, return from iterator, etc.)

Related Problems:
 Is an undirected graph connected?
 Is a digraph weakly/strongly connected?
 For strongly, need a cycle back to starting node

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 59

Graph Traversals
Basic Algorithm for Traversals:
 Select a starting node
 Make a set of nodes adjacent to current node
 Visit each node in the set but "mark" each
nodes after visiting them so you don't revisit
them (and eventually stop)
 Repeat above but skip "marked nodes"

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 60

In Rough Code Form
traverseGraph(Node start) {
Set pending = emptySet();
pending.add(start)
mark start as visited
while(pending is not empty) {
next = pending.remove()
for each node u adjacent to next
if(u is not marked) {
mark u
pending.add(u)
}
}
}
}

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 61

Running Time and Options
Assuming add and remove are O(1), entire
traversal is O(|E|) if using an adjacency list

The order we traverse depends entirely on how

add and remove work/are implemented
 DFS: a stack "depth-first graph search"
 BFS: a queue "breadth-first graph search"

DFS and BFS are "big ideas" in computer science

 Depth: recursively explore one part before going
back to the other parts not yet explored
 Breadth: Explore areas closer to start node first

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 62

Recursive DFS, Example with Tree
A tree is a graph and DFS and BFS are particularly
easy to "see" in one
A
DFS(Node start) {
mark and process start
B C for each node u adjacent to start
if u is not marked
D E F DFS(u)
}
G H
Order processed: A, B, D, E, C, F, G, H
 This is a "pre-order traversal" for trees
 The marking is unneeded here but because we
support arbitrary graphs, we need a means to
process each node exactly once

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 63

DFS with Stack, Example with Tree
DFS2(Node start) {
initialize stack s to hold start
A mark start as visited
while(s is not empty) {
B C next = s.pop() // and "process"
for each node u adjacent to next
D E F if(u is not marked)
mark u and push onto s
G H }
}

Order processed: A, C, F, H, G, B, E, D
 A different order but still a perfectly fine
traversal of the graph

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 64

BFS with Queue, Example with Tree
BFS(Node start) {
initialize queue q to hold start
A mark start as visited
while(q is not empty) {
B C next = q.dequeue() // and "process"
for each node u adjacent to next
D E F if(u is not marked)
mark u and enqueue onto q
G H }
}

Order processed: A, B, C, D, E, F, G, H
 A "level-order" traversal

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 65

DFS/BFS Comparison
BFS always finds the shortest path (or
"optimal solution") from the starting node
to a target node
 Storage for BFS can be extremely large
 A k-nary tree of height h could result in a
queue size of kh

DFS can use less space in finding a path

 If longest path in the graph is p and highest
out-degree is d then DFS stack never has more
than d⋅p elements

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 66

Implications
For large graphs, DFS is hugely more
memory efficient, if we can limit the
maximum path length to some fixed d.

If we knew the distance from the start to the

goal in advance, we could simply not add any
children to stack after level d

But what if we don’t know d in advance?

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 67

Iterative Deepening (IDFS)
Algorithms
 Try DFS up to recursion of K levels deep.
 If fails, increment K and start the entire
search over

Performance:
 Like BFS, IDFS finds shortest paths
 Like DFS, IDFS uses less space
 Some work is repeated but minor
compared to space savings

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 68

Saving the Path
Our graph traversals can answer the standard
reachability question:
"Is there a path from node x to node y?"

But what if we want to actually output the path?

Easy:
 Store the previous node along the path:
When processing u causes us to add v to the
search, set v.path field to be u)
 When you reach the goal, follow path fields back to
where you started (and then reverse the answer)
 What's an easy way to do the reversal? A Stack!!

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 69

Example using BFS
What is a path from Seattle to Austin?
 Remember marked nodes are not re-enqueued
 Note shortest paths may not be unique
0 1 Chicago
Seattle

1 Salt Lake City

Austin

1
3
San Francisco
2
Dallas

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 70

Topological Sort
Problem: Given a DAG G=(V, E), output all the
vertices in order such that if no vertex appears
before any other vertex that has an edge to it
Example input: CSE 331 CSE 440

CSE 332
…
CSE 142 CSE 143 CSE 311

CSE 312
MATH CSE 341
126

CSE 351 CSE 333

Example output: CSE 352

 142, 126, 143, 311, 331, 332, 312, 341, 351,

333, 440, 352 Disclaimer: Do not use for official advising purposes!
(Implies that CSE 332 is a pre-req for CSE 312 – not true)

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 71

Questions and Comments
Terminology:
A DAG represents a partial order and a topological
sort produces a total order that is consistent with it

Why do we perform topological sorts only on DAGs?

 Because a cycle means there is no correct answer

Is there always a unique answer?

 No, there can be one or more answers depending
on the provided graph

What DAGs have exactly 1 answer?

 Lists

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 72

Uses Topological Sort
Figuring out how to finish your degree

Computing the order in which to

recalculate cells in a spreadsheet

Determining the order to compile files with

dependencies

In general, use a dependency graph to

find an allowed order of execution

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 73

Topological Sort: First Approach
1. Label each vertex with its in-degree
 Think "write in a field in the vertex"
 You could also do this with a data structure on
the side

2. While there are vertices not yet outputted:

a) Choose a vertex v labeled with in-degree of 0
b) Output v and "remove it" from the graph
c) For each vertex u adjacent to v, decrement in-
degree of u
- (i.e., u such that (v,u) is in E)

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 74

 Example
Output:

CSE 331 CSE 440

CSE 332
…
CSE 142 CSE 143 CSE 311

CSE 312
MATH CSE 341
126

CSE 351 CSE 333

CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed?
In-deg:

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 75

 Example
Output:

CSE 331 CSE 440

CSE 332
…
CSE 142 CSE 143 CSE 311

CSE 312
MATH CSE 341
126

CSE 351 CSE 333

CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed?
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 76

 Example
Output:
126
CSE 331 CSE 440

CSE 332
…
CSE 142 CSE 143 CSE 311

CSE 312
MATH CSE 341
126

CSE 351 CSE 333

CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 77

 Example
Output:
126
CSE 331 CSE 440
142

CSE 332
…
CSE 142 CSE 143 CSE 311

CSE 312
MATH CSE 341
126

CSE 351 CSE 333

CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1
0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 78
 Example
Output:
126
CSE 331 CSE 440
142
143
CSE 332
…
CSE 142 CSE 143 CSE 311

CSE 312
MATH CSE 341
126

CSE 351 CSE 333

CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 0 0 0
0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 79
 Example
Output:
126
CSE 331 CSE 440
142
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311

CSE 312
MATH CSE 341
126

CSE 351 CSE 333

CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 0 0
0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 80
 Example
Output:
126
CSE 331 CSE 440
142
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311
331
CSE 312
MATH CSE 341
126

CSE 351 CSE 333

CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 0 0
0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 81
 Example
Output:
126
CSE 331 CSE 440
142
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311
331
CSE 312 332
MATH CSE 341
126

CSE 351 CSE 333

CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 1 0 0 0
0 0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 82
 Example
Output:
126
CSE 331 CSE 440
142
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311
331
CSE 312 332
MATH CSE 341
126 312
CSE 351 CSE 333
CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 1 0 0 0
0 0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 83
 Example
Output:
126
CSE 331 CSE 440
142
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311
331
CSE 312 332
MATH CSE 341
126 312
341
CSE 351 CSE 333
CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 1 0 0 0
0 0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 84
 Example
Output:
126
CSE 331 CSE 440
142
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311
331
CSE 312 332
MATH CSE 341
126 312
341
CSE 351 CSE 333
351
CSE 352

Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x x x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 1 0 0 0 0
0 0 0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 85
 Example
Output:
126
CSE 331 CSE 440
142
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311
331
CSE 312 332
MATH CSE 341
126 312
341
CSE 351 CSE 333
351
CSE 352
333
Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x x x x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 1 0 0 0 0
0 0 0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 86
 Example
Output:
126 352
CSE 331 CSE 440
142
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311
331
CSE 312 332
MATH CSE 341
126 312
341
CSE 351 CSE 333
351
CSE 352
333
Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x x x x x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 1 0 0 0 0
0 0 0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 87
 Example
Output:
126 352
CSE 331 CSE 440
142 440
143
CSE 332
… 311
CSE 142 CSE 143 CSE 311
331
CSE 312 332
MATH CSE 341
126 312
341
CSE 351 CSE 333
351
CSE 352
333
Node: 126 142 143 311 312 331 332 333 341 351 352 440
Removed? x x x x x x x x x x x x
In-deg: 0 0 2 1 2 1 1 2 1 1 1 1
1 0 1 0 0 1 0 0 0 0
0 0 0
July 23, 2012 CSE 332 Data Abstractions, Summer 2012 88
Running Time?
labelEachVertexWithItsInDegree();
for(i=0; i < numVertices; i++) {
v = findNewVertexOfDegreeZero();
put v next in output
for each w adjacent to v
w.indegree--;
}

What is the worst-case running time?

 Initialization O(|V| + |E|) (assuming adjacency list)
 Sum of all find-new-vertex O(|V|2) (because each O(|V|))
 Sum of all decrements O(|E|) (assuming adjacency list)
 So total is O(|V|2 + |E|) – not good for a sparse graph!

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 89

Doing Better
Avoid searching for a zero-degree node every time!
 Keep the “pending” zero-degree nodes in a list, stack, queue,
bag, or something that gives O(1) add/remove
 Order we process them affects the output but not
correctness or efficiency

Using a queue:
 Label each vertex with its in-degree,
 Enqueue all 0-degree nodes
 While queue is not empty
 v = dequeue()
 Output v and remove it from the graph
 For each vertex u adjacent to v, decrement the in-degree
of u and if new degree is 0, enqueue it

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 90

Running Time?
labelAllWithIndegreesAndEnqueueZeros();

for(i=0; i < numVertices; i++) {

v = dequeue();
put v next in output
for each w adjacent to v {
w.indegree--;
if(w.indegree==0)
enqueue(w);
}
}
 Initialization: O(|V| + |E|) (assuming adjacency list)
 Sum of all enqueues and dequeues: O(|V|)
 Sum of all decrements: O(|E|) (assuming adjacency list)
 So total is O(|E| + |V|) – much better for sparse graph!

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 91

More Graph Algorithms
Finding a shortest path is one thing
 What happens when we consider
weighted edges (as in distances)?

Next time we will discuss shortest path

algorithms and the contributions of a
curmudgeonly computer scientist

July 23, 2012 CSE 332 Data Abstractions, Summer 2012 92