Graph Assignment

Goals
Learn and implement the adjacency matrix structure an
Campus Tour

Kruskal’s minimum spanning tree algorithm
 Understand and use the decorator pattern and various JDSL
classes and interfaces
Your task
 Implement the adjacency matrix structure for representing a
graph
 Implement Kruskal’s MST algorithm
Frontend
 Computation and visualization of an approximate traveling
salesperson tour

© 2004 Goodrich, Tamassia Campus Tour 1 © 2004 Goodrich, Tamassia Campus Tour 2

Adjacency Matrix Structure Kruskal’s Algorithm

Edge list structure a v b The vertices are Algorithm KruskalMSF(G)
Input weighted graph G
Augmented vertex u w partitioned into clouds Output labeling of the edges of a
objects  We start with one cloud minimum spanning forest of G
 Integer key (index) per vertex Q ← new heap-based priority queue
associated with  Clouds are merged during for all v ∈ [Link]() do
vertex the execution of the l ← makeSet(v) { elementary cloud }
2D-array adjacency 0 u 1 v 2 w algorithm [Link](l)
for all e ∈ [Link]() do
array Partition ADT: [Link]([Link](), e)
 Reference to edge while ¬[Link]()
0 1 2 makeSet(o): create set {o}
object for adjacent 
e ← [Link]()
vertices 0 ∅ ∅ and return a locator for [u,v] ← [Link]()
Null for non object o A ← find([Link]())

1 ∅ B ← find([Link]())
nonadjacent  find(l): return the set of
vertices a 2 ∅ ∅ b the object with locator l if A ≠ B
setMSFedge(e)
 union(A,B): merge sets A { merge clouds }
and B union(A, B)

© 2004 Goodrich, Tamassia Campus Tour 3 © 2004 Goodrich, Tamassia Campus Tour 4
 Example Example (contd.)
G G G G
8 8 B 8 B 8
B 4 B 4 4 4
E E 9 E 9 E
9 6 9 6 5 6 6
1 5 F 5 F 1 F 1 5 F
C 11 3 1 C 11 3 C 11 3
C 11 3
2 2 2 2
7 H 7 H 7 D H 7 H
D D D
A A A 10 A
10 10 10
four steps

G G G G
8 8 B 8 B 8
B 4 B 4 4 4
E E 9 E 9 E
9 6 9 6 5 6 5 6
1 5 F 5 F 1 F 1 F
1 C 11 3 C 11 3
C 11 3 C 11 3
2 2 2 2
7 H 7 H 7 D H 7 D H
D D A A
A 10 A 10 10
10
© 2004 Goodrich, Tamassia Campus Tour 5 © 2004 Goodrich, Tamassia Campus Tour 6

Partition Implementation Analysis of Kruskal’s Algorithm

Partition implementation Amortized analysis Graph operations
 A set is represented the  Consider a series of k Partiton  Methods vertices and edges are called once
sequence of its elements ADT operations that includes  Method endVertices is called m times
A position stores a reference n makeSet operations
Priority queue operations

back to the sequence itself (for
operation find) Each time we move an
We perform m insert operations and m removeMin operations


element into a new sequence,
The position of an element in

the sequence serves as locator the size of its set at least Partition operations
for the element in the set doubles  We perform n makeSet operations, 2m find operations and no
 In operation union, we move  An element is moved at most more than n − 1 union operations
the elements of the smaller log2 n times
sequence into to the larger
Label operations
Moving an element takes O(1)
We set vertex labels n times and get them 2m times

sequence time


Worst-case running times  The total time for the series Kruskal’s algorithm runs in time O((n + m) log n) time
 makeSet, find: O(1) of operations is O(k + n log n) provided the graph has no parallel edges and is
 union: O(min(nA, nB)) represented by the adjacency list structure
© 2004 Goodrich, Tamassia Campus Tour 7 © 2004 Goodrich, Tamassia Campus Tour 8
 Decorator Pattern Traveling Salesperson Problem
Labels are commonly used in The decorator pattern extends A tour of a graph is a spanning cycle
graph algorithms the methods of the Position (e.g., a cycle that goes through all
 Auxiliary data
ADT to support the handling the vertices) 7 D
of attributes (labels) A traveling salesperson tour of a B 4
 Output
 has(a): tests whether the weighted graph is a tour that is 2
Examples position has attribute a 2 5 F
simple (i.e., no repeated vertices or C
DFS: unexplored/visited get(a): returns the value of edges) and has has minimum weight 8
 
6 3
label for vertices and attribute a
No polynomial-time algorithms are E
unexplored/ forward/back set(a, x): sets to x the value of A 1

known for computing traveling
labels for edges attribute a
salesperson tours
Dijkstra and Prim-Jarnik:  destroy(a): removes attribute Example of traveling

a and its associated value (for The traveling salesperson problem
distance, locator, and (TSP) is a major open problem in salesperson tour
cleanup purposes)
parent labels for vertices computer science
Kruskal: locator label for
The decorator pattern can be (with weight 17)

implemented by storing a  Find a polynomial-time algorithm
vertices and MSF label for computing a traveling salesperson
edges
dictionary of (attribute, value) tour or prove that none exists
items at each position
© 2004 Goodrich, Tamassia Campus Tour 9 © 2004 Goodrich, Tamassia Campus Tour 10

TSP Approximation
We can approximate a TSP tour
with a tour of at most twice the
weight for the case of Euclidean
graphs
 Vertices are points in the plane
 Every pair of vertices is connected
by an edge
 The weight of an edge is the
length of the segment joining the
points
Approximation algorithm
 Compute a minimum spanning tree
 Form an Eulerian circuit around the
MST
 Transform the circuit into a tour

© 2004 Goodrich, Tamassia Campus Tour 11