Operations Research

Unit-III (Part-1)

TechniquesNetworks
Linear Programming: Networks
 Network Analysis
 Network Terminology
 Network Problems
 Examples of Network Problems
 Minimum Spanning Tree
 Simple Greedy Procedure
 Kruskal’s Algorithm
 Directed and Undirected Arcs, Networks
 Shortest Route Problem
 Dijkstra's Algorithm
 Network Analysis
 Network Analysis.- OR technique(s) for the solution of problems based on
the visual and conceptual representation of the interrelationships between
the components of a system.
 A network (graph) consists of a set of points and a set of lines connecting
certain points. The points are called nodes (vertices). The lines are called
arcs (edges, branches).
 The arcs may have certain flow. If the flow can go in only one direction in
the arc, the arc is said to be directed arc, if not the arc is undirected (or
links).
• Examples of Network Applications

Nodes Arcs Flow

Intersections Roads Vehicles
Airports Air Routes Aircraft
Switching Points Wires, Channels Messages
Pumping Stations Pipes Fluid
Work Centers Material Handling Jobs
Systems
 Network Terminology
 A network that has only directed arcs is said to be a
directed network. If all the arcs are undirected, the
network is undirected.
 A path between two nodes is a sequence of arcs
connecting them. A directed path from node i to node j
allows flow from node i to j. An undirected path from i
to j allows flow in either direction.
 Network Terminology
 A path that begins and ends at the same node is called a
cycle. A cycle can be directed or undirected if the path
is directed or undirected.
 Two nodes are connected if the network contains at
least one undirected path between them
 A connected network is a network where every pair of
nodes is connected.
 A tree is a connected network with no cycles. A
spanning tree is a tree connecting all the nodes of a
network.
 Arc capacity is the maximum amount of flow that can be
carried on a directed arc.
 In a supply node the flow out exceeds the flow in. The
opposite occurs in a demand node. In a transshipment
node, flow in = flow out.
 Network Problems
 Some problems typically solved by using network
methods:
1. Minimum Spanning Tree
2. Shortest Route
3. Maximum Flow
4. Critical Path
 Examples of Network Problems
 Minimum Spanning Tree- The objective is to connect all of the
nodes of a network with links that minimize the total “cost”
(“cost” could be time, distance, etc.)
• Common Examples:
 Design of computer networks
 Design of a network of roads
 Shortest Route- The objective is to minimize the “cost” to go
from a Node A (the origin) to a Node B (the destination)
• Common Examples:
 Selection of highways to go from City A to City B such that the total
traveled distance is minimized.
 Deciding what branches of a pipeline to build to connect City A to
City B such that the total cost is minimized.
 Examples of Network Problems
 Maximum Flow- The objective is to maximize the flow that
can be sent from one node (the source) to another (the sink)
• Common Examples:
 Traffic management
 Logistics systems
 Critical Path- The objective is to minimize the time it takes to
complete the total project
• Common Example:
 Project Management (CPM & PERT Networks)
 Minimum Spanning Tree
 A minimum spanning tree is the set of the arcs of a
network that have the shortest total length while providing
a route (path) between each pair of nodes.
 More formally:
• Let G = (V,E) represent a network, where V is the node set
and E is the arc set and w(e) is the cost (or weight) of
constructing edge eE.
• Problem: find the set of edges that form a spanning tree T
for G where the total weight
w(T )   w(e)
eT
is as small as possible.
• For a spanning tree to be a minimum spanning tree, it has to
satisfy the path optimality conditions
 Applications?
 Constructing highways or railroads spanning
several cities
 Laying pipelines connecting offshore drilling sites,
refineries, and consumer markets
 Designing computer networks
 Others?
 Simple Greedy Procedure
 Procedure:
1. Select any node arbitrarily and then connect it to the
nearest distinct node.
2. Identify the unconnected node that is the closest to a
connected node, and then connect these two nodes (ties
may be broken arbitrarily). Repeat this until all nodes have
been connected.
 Example1
Select any node
7
6 Connect the closest node
3
8 4
7 Select the closest node to
any of the connected nodes
6 6 8
1 Select the closest node to
7 any of the connected nodes
8

2 7
2

V13, V37, V27, V67, V68,=1 w(T) = 7+6+2+6+4=25

 Example2
 Apply the greedy procedure to solve the minimum
spanning tree for the graph shown below:

2
3
9 1
8
8 6 8
5
2 7
 Kruskal’s Algorithm
(a more refined greedy algorithm)
 A more efficient algorithm (and also optimal) for
obtaining the minimum spanning tree
 Sort all the arcs in non-decreasing order of cost
 Define a set, LIST, that is the set of arcs chosen as
part of a minimum spanning tree
 Initially LIST is empty
 Examine the arcs in the sorted order one by one and
check whether adding the arc we are currently
examining to LIST creates a cycle with the arcs
already in LIST. If not, add arc to LIST, otherwise
discard it.
 Terminate procedure when the cardinality of LIST
equals the number of vertices less one. The result is
minimum spanning tree T*
 Example3
7 Order Arc
3 6
1 2-7
Iteration List
8 4
7 0  2 6-8

6 6 8 1 3 3-7
1 V27
7 2
V27 ,V68 4 6-7
8
3 V27 ,V68 ,V37 5 3-6
2 7
2 4
V27 ,V68 ,V37 ,V67 6 1-3
5
From the sorted list, select the least cost arc V27 ,V68 ,V37 ,V67 ,V13 7 7-8
6
From the sorted list select the least cost arc 8 1-6

Selecting this arc, will create a 9 2-3

cyclewe skip it

Since the number of arcs is n-1 (number of nodes –1), we stop

 Directed and Undirected Arcs
 Directed arcs- used to represent cases when flow is
allowed only one direction. Examples: traffic in one-way
streets, water flow in channels, etc. They are graphically
represented with arrows.

 Undirected Arcs (link)- used to represent cases when flow

is allowed in both directions. Examples: traffic in two-way
streets, pipes allowing flow in both direction, etc. They are
graphically represented with a simple straight line.
 Directed and Undirected
Networks
 Directed Network
3
1 6
8

2 4
7

 Undirected Network
3
1 6

2 4 5
 Paths
Directed
Directed Path
Cycle
3
1 6

2
4
7

Undirected
Path

Undirected
Cycle
 Connected Network
and Spanning Tree
3
3 1 6
1 6

7
7

2 Another
2 4
4 5 Spanning
5
Tree
Spanning
Tree

Suppose: G=(V,E) is a connected graph with vertex set V and

edge set E, and T is a subset of E such that:
•Every vertex of G is connected by an edge in T
•The edges in T form a tree (i.e. there are no cycles)
then T is a spanning tree for G.
 Capacitated Network
8 3 4
1 6
7
2
4 3 8
2 2 3
3
Supply 2 2
(source) 4 3 7
Node
4 Demand
3
(sink)
Node
5

Transshipment
Node
 Networks and Graphs
 There is a close relationship between networks and
graphs
 Commonly, a network is defined as a graph whose arcs
have a certain flow. This is usually accompanied by
“parameters” which describe costs or bounds.
 Usually, when we talk about graphs (graph theory) we
are interested in the topological (spatial relationship)
properties related to the graphical representation of the
problem.
 We are more interested in the solution of engineering
problems involving some kind of flow in the arcs of the
graph (distance, cost, etc) and typically there will be
parameters associated with the graph. Thus, the term
“network” rather than “graph” is most appropriate.
However, we will rely on some results and concepts from
graph theory to frame the discussion of the problem at
hand.
 Shortest Route Problem
 Objective: To determine the shortest route (path) from the source node
to a destination node.

 Alternatively, we could define the shortest path problem as determining

how to send 1 unit of flow as cheaply as possible from the source node
to another node in an incapacitated network.

 Corresponding to each arc (i,j), there is a non-negative number dij

called the distance from i to j (dij =  if we cannot get from i to j
directly).

 Example: Find the shortest route between Nodes 1 and 8.

4
3 6

3 2
2 1
1 2 7
5 5 8
1
6
1 2 3 6

2 5 4 3 7
 Applications

 Want to send a truck from a DC to a customer.

 Want to send a message across a telecommunications
network.
 What does “shortest” mean in these cases?
 Dijkstra's Algorithm
 Dijkstra’s algorithm is an efficient method to find the
shortest path between two nodes.

 Some of the characteristics of the algorithm:

• Assigns a temporary (unsolved) or permanent (solved)
label to each node in the network.

• The temporary (unsolved) label is an upper bound on the

shortest distance from the source (origin) node to that
node.

• The shortest route from the source to a node is given by

the permanent label.
 Steps of the Shortest Path (Dikjstra’s) Algorithm
 Objective of the nth iteration: Find the nth nearest node to the
origin.
 Input for the nth iteration: n-1 nearest nodes to the origin (solved
at the previous iteration), including their shortest path and the
distance from the origin (call these solved nodes).
 Candidates for the nth nearest node: Each solved node that is
directly connected by a link to one or more unsolved nodes
provides one candidate - the unsolved node with the shortest
connecting link (ties provide additional candidates).
 Calculation of the nth nearest node: For each such solved node
and its candidate, add the distance between them and the
distance of the shortest path from the origin to the solved node.
The candidate with the smallest such total distance is the nth
nearest node (ties provide additional solved nodes), and its
shortest path is the one generating this distance.
 Assumptions of Dijsktra’s Algorithm

 The arc lengths are integers

 The network contains a directed path from node s (the
source) to every other node in the network
 The network does not contain a negative cycle (you
cannot get back to the origin)
 The network is directed
 Basic Idea of the Algorithm
 Maintains a directed out-tree T rooted at s that spans the
nodes with finite distance labels.
 Maintains this tree using predecessor indices and the
following invariant property:
• Every tree arc (i,j) satisfies the condition d(j)=d(I)+cij with
respect to the current distance labels
 The operation of selecting a minimum temporary
distance label is termed the node selection operation.
 The operation of checking whether the current labels for
nodes I and j satisfy d(j)>d(I)+cij and, if so, then setting
d(j)=d(I)+cij is the called the distance update operation
 Dijkstra’s Algorithm Notation
 Finds shortest path from the source node to all other
nodes in the network
 Notation
• d(i) is the distance label for node i
• S is the set of permanently labeled nodes
• S’ is the set of temporarily labeled nodes
• s is the source node
• cij is the cost (distance) on arc (i,j)
• T is the shortest path tree on G
• pred(j) = i if (i,j)T
• A(i) is the arc adjacency list of node i