Unit 7:

Network Flows
Network Flow Problems
Ford Fulkerson algorithm for the max flow problem

An Introduction to Network Flow Problems

Network flow problems are a class of optimization problems that deal with the
efficient allocation of resources in a network. These problems have applications
in various fields, including transportation, telecommunications, logistics, and
supply chain management.
Network flow problems involve determining how items or data are sent from
sources to destinations through the network by minimizing or maximizing some
objective function.
The main component of the network flow problem is:
1. Network Structure: A network is represented as a graph consisting of
nodes and directed edges. Nodes can represent sources, intermediate
points, or sinks (destinations), and edges represent possible routes or
channels through which flows can be transmitted. Each edge has a
capacitance that limits the amount of current it can carry.
2. Flow: A flow represents the set of objects, data, or resources that move
from one node to another along the edges of a network. The goal is to
determine the flow through each edge in a way that satisfies certain
constraints and goals.
3. Objective function: Network flow problems typically involve optimizing
an objective function. Depending on the specific problem, this could be
minimizing transport costs, maximizing the flow between specific source-
destination pairs, or sharing the most efficient way to find resources. The
most common types of power supply problems are:
1. Maximum Flow Problem: The objective of this problem is to
maximize the flow from the source node to the sink node of the
network respecting the capacity constraints of each edge. It can be
considered as the most efficient way to move goods or information
from source to destination.
 2. Minimal cut problem: The goal is to find the smallest set of edges
whose removal would cut the source from the sink side of the
network. It is closely related to the maximum flow problem and has
applications in network reliability and network design.
3. Minimum Cost Flow Problem: This problem extends the
maximum flow problem by adding the cost of sending a flow
through each edge. The goal is to minimize the total cost of sending
current from the source to the sink, taking into account capacity
constraints.
4. Bilateral matching problem: In this case, you have two different
sets of nodes, and the goal is to find the maximum number of
pairwise connections (edges) between these sets, subject to certain
constraints. This problem is often used in scenarios such as work
assignments or finding optimal tasks.
5. Multiple Commodity Flow Problem: This generalization of the
maximum flow problem considers multiple types of commodities
flowing through the network simultaneously, each with its own
source, sink, and flow requirements. Network flow problems are
important in operations research, computer science, and engineering
and can be solved by various algorithms such as the Ford-Fulkerson
method, the Edmonds-Karp algorithm, and linear programming
techniques. These problems have real-world applications in
transportation, communication networks, water distribution, and
many other areas where the efficient allocation of resources is
critical.

History
The history of web streaming problems is a rich and fascinating journey spanning
several decades, marked by significant contributions from fields as diverse as
operations research, computer science, and mathematics. Here is a brief overview
of the main milestones in the development of network flow problems:
o Early works (1950s): The history of network current problems can be
traced back to the middle of the 20th century. In the 1950s, researchers
such as George Dantzig and T.C. Koopmans began working on linear
programming and optimization problems, laying the foundation for the
development of network flow theory.
o Ford-Fulkerson Algorithm (1956): The key work of L.R. Ford Jr. and
D.R. In 1956, Fulkerson introduced the Ford-Fulkerson algorithm, a
pioneering method for solving the maximum flow problem. This algorithm
laid the foundation for subsequent research on network flows.
o Edmonds-Karp algorithm (1972): In 1972, Jack Edmonds and Richard
M. Karp improved the Ford-Fulkerson algorithm by introducing the
Edmonds-Karp algorithm. This algorithm provides a polynomial-time
solution to the maximum flow problem using width-unit search to find
increasing paths.
o Duality theory (1970s): Scholars including George Dantzig and R.L.
Graves developed duality theory for network flow problems. Duality
theory provides valuable insight into the relationship between original
(initial) and dual (connected) problems and has applications to more
efficiently solve network flow problems.
o Least Cost Flow and Linear Programming (1970s-1980s): Extending
network flow problems to include costs or edge weights led to the
development of the minimum cost flow problem. A linear programming
technique was used to solve these problems efficiently.
o Algorithmic Progress (1980-1990): During this period, researchers have
made significant progress in developing efficient algorithms for solving
various network flow problems, including minimum cost flow, multivariate
flow, and transportation problems. This led to the practical applicability of
network streaming methods in real-world scenarios.
o Combinatorial Optimization (2000s to present): Network flow
problems remain a core topic in the field of combinatorial optimization.
Researchers have developed new algorithms, approximation techniques,
and heuristics to solve more complex variants of network flow problems
considering constraints such as power, time, and reliability.
 o Applications and Real Impact (Continued): Network flow models and
algorithms have found wide applications in transportation, logistics,
telecommunications, supply chain management, and other fields. They
play a key role in optimizing resource allocation, minimizing costs, and
improving the efficiency of various systems. The history of network flow
problems reflects the interdisciplinary nature of optimization and algorithm
development. Over the years, these problems have evolved and their
solutions have become an integral part of solving complex real-world
challenges. Researchers and practitioners continue to explore new
dimensions and applications of network theory to ensure its relevance in
today's era of interconnected systems and data-driven decision-making.
Advantages of Network Flow Problems
Network flow problems are a class of optimization problems that involve the
efficient allocation of resources over a network or graph. These problems have
many practical applications and have several advantages:
1. Modeling real-world scenarios: Network flow problems can model a
variety of real-world scenarios, such as transportation and distribution
systems, communication networks, and supply chain management. This
versatility makes them valuable in solving practical problems.
2. Optimization: Network flow problems aim to optimize the flow of
resources such as goods, information, or services through the network. This
optimization can lead to cost savings, improved efficiency, and better
utilization of resources.
3. Simplicity: Many network flow problems are relatively easy to understand
and formulate mathematically. They often include simple concepts such as
sources, sinks, capacities, and flow rates, making them available for
modeling as well as solutions.
4. Efficiency: Algorithms for solving network flow problems are well-
developed and very efficient. They can handle large networks with
thousands or even millions of nodes and edges, making them suitable for
complex real-world systems.
5. Optimal Solutions: Network flow algorithms ensure that an optimal
solution is found if one exists. This is critical in scenarios where cost
minimization or efficiency maximization is of utmost importance.
6. Sensitivity analysis: Network flow problems allow for sensitivity
analysis, which means you can evaluate how changes in parameters such
 as capacity or cost affect the optimal solution. This understanding can help
decision-makers make informed choices.
7. Resource Allocation: Network power issues enable accurate allocation of
resources. For example, they can help determine the most efficient way to
allocate trucks for deliveries, data routing, or production schedules in a
factory.
8. Multi-valvar flow: Some network flow problems, such as the multi-valvar
flow problem, allow modeling the simultaneous flow of multiple types of
resources through a network. This skill is essential for solving complex,
relational problems.
9. Versatile Applications: Flow network problems have applications in a
variety of fields, including logistics, telecommunications, finance, and
manufacturing. They are used to optimize everything from traffic
management to network planning.
10.Advances in Graph Theory: Research and development in network flow
problems have contributed to the development of graph theory and
optimization algorithms that have applications beyond network flow
problems.
11.Heuristic approaches: In cases where finding an exact solution is
computationally difficult, heuristics can provide reasonably good
approximations to network flow problems, enabling practical applications.
12.Network Algorithms: Network flow algorithms can be adapted for
network or dynamic scenarios where network characteristics change over
time. This flexibility is important for real-time decision-making and
adaptive systems.
Disadvantages of Network flow problems
1. Assumptions and Simplifications: Network flow models are often based
on simplifications and assumptions that may not fully capture the
complexity of real systems. For example, they may assume constant power,
deterministic flows, or linear relationships that may not hold in practice.
2. Fixed topology: Current network problems usually require a fixed
network topology. In reality, networks can evolve and change over time,
and dealing with dynamic changes can be difficult in traditional network
flow models.
3. Computational complexity: Although there are many efficient algorithms
for solving network flow problems, optimally solving certain variations or
large-scale cases can still be computationally challenging. This may
require the use of approximation algorithms or heuristics, which may lead
to suboptimal solutions.
4. Integer Limits: Some net current problems have integer limits, making
them more difficult to solve than linear or continuous problems. Integer
programming techniques can be used, but often add computational
complexity.
5. Data accuracy: The quality and accuracy of the data used in network flow
models can significantly affect the results. Inaccurate information, such as
incorrect capacity estimates or unreliable cost information, can lead to
suboptimal or unrealistic solutions.
6. Complexity of real-world scenarios: In some real-world scenarios, the
underlying issues may be more complex than traditional network flow
models can represent. For example, accounting for uncertainty, multiple
objectives or non-linear relationships can be difficult.
7. Limited flow problems: Network flow models are mainly designed to
solve resource allocation and flow optimization problems. They may not
be suitable for solving other types of optimization problems that have
different characteristics and constraints.
8. Sustainability: Optimal solutions derived from network flow models may
not always be sustainable under uncertainty or unexpected events. Real-
world systems often require solutions that can adapt to changing
conditions.
9. Human Factors: Network flow models may not take into account human
decision-making, behavioral aspects, or social aspects that may be crucial
in certain applications, such as traffic planning or disaster situations.
10.Implementation costs: Implementing solutions derived from network
flow models can sometimes be expensive or impractical due to factors such
as infrastructure requirements, technology limitations, or organizational
resistance to change.
11.Model maintenance: As real systems change; network flow models may
require constant updating and maintenance to remain relevant and accurate.
This ongoing effort can be resource intensive.
 12.Communication Overhead: In network flow models applied to
communication networks, there can be significant overhead in message
delivery and coordination, which can affect system performance.
Implementation of network flow problems
Network flow problems are mathematical optimization problems that involve the
transport or movement of goods, information, or resources through a network of
interconnected nodes and edges. These problems have many practical
applications in various fields. Here are some of the main application areas for
network flow problems:
1. Transport and logistics: Supply Chain Management: Network flow
problems are used to optimize the distribution of goods from suppliers
to consumers and to minimize transportation costs.
2. Vehicle Routing: These helps determine the most efficient routes for a
fleet of vehicles to deliver goods to multiple destinations. Airline
Planning: Optimizing flight schedules to minimize delays and improve
resource utilization.
3. Telecommunications and computer networks: Data Routing: In
computer networks, network routing algorithms help find the most
efficient routes for data packets to travel from source to destination.
4. Telecommunications network planning: Optimizing the layout of
network cables and equipment for efficient data transmission. Energy
and service costs: Electric Network Control: The control of electric
current through the electric network to ensure a stable and efficient
power supply.
5. Water Distribution: Optimize water flow through pipelines and tanks
to meet demand and minimize losses. Economy and finance: Portfolio
Optimization: Standard algorithms can be used to optimize
investment portfolios by allocating funds to different investments to
maximize returns.
6. Traffic Management: Traffic flow: managing traffic flows on the road
network to minimize congestion and travel times. Public transport:
optimizing bus and train timetables to improve public transport
services.
7. Management of Natural Resources: Forestry: Planning the harvest of
trees in a forest to maximize tree yield and preserve the ecosystem.
 Fishing: manage the allocation of fishing quotas to different areas or
vessels.
8. Health care: Hospital Scheduling: Assigning medical staff to different
shifts and departments in the hospital to effectively meet patient needs.
The blood supply chain: optimizing the distribution of blood products
to hospitals and clinics.
9. Sports Programming: Optimize the schedule of sports events,
tournaments, and leagues to minimize conflicts and maximize
viewership.
10.Environmental Management: Pollution Control: Identify the optimal
locations of pollution monitoring stations or the best ways to clean up
environmental pollution.
How Does the Implementation of Network Flow Problems Work?
Performing network flow troubleshooting usually involves the following steps.
1. Formulation of the problem, and layout of the problem: Clearly define the
problem you want to solve. Define the nature of the network, including
nodes and edges, and define goals and boundaries.
2. Network modeling: Create a mathematical model that represents the
structure of the network. This includes defining nodes (sources, sinks,
brokers) and edges (connections between nodes). To set parameters:
Specify values or parameters for network elements. This may include
capacities, costs, requirements, or weight associated with edges or nodes.
3. Selection of the objective function: Define the objective you want to
optimize, such as cost minimization, flow maximization, or fulfillment.
Convert this target into a mathematical function.
4. Format restrictions: Define the constraints that must be met. Common
constraints include edge capacity constraints, flow maintenance at nodes,
and demand fulfillment.
5. Choice of algorithm: Choose the appropriate network flow algorithm or
method based on the characteristics of the problem, the size of the network,
and the specific objective. Common algorithms include the Ford-Fulkerson
algorithm, the Edmonds-Karp algorithm, the minimum cost flow
algorithm, and others.
6. To solve the problem: Apply the selected algorithm to the formulated
problem to find the optimal solution or its approximation. The algorithm
 iteratively adjusts the flow along the edges to arrive at an optimal or
feasible solution.
7. Interpretation of results: Examine the results of the optimization process to
determine the flow values for each edge, the value of the objective function
(such as cost or flow), and other relevant information.
8. Applying the solution: If the network flow problem is part of a real
application, implement the solution in practice. For example, adjust
transport routes, production schedules, or network settings based on
optimized flow.
9. Management and maintenance: Constantly monitor the network and the
factors that can affect it. Make necessary adjustments to maintain flow
efficiency and effectiveness. It is important to note that the specific steps
and details may vary depending on the type of network flow problem and
the chosen algorithm. In addition, large or complex networks often use
software tools and optimization solutions to automate the troubleshooting
process.
Question: Minimize the cost of transporting goods from multiple suppliers
to multiple customers while meeting capacity constraints.
Steps:
1. Draw a network with nodes representing suppliers, customers, and the
transportation network connecting them.
2. Assign capabilities and costs to edges representing transport routes.
3. Define an objective function to minimize the total cost of
transportation.
4. Define limits, including capacity limits for provider nodes and demand
limits for customer nodes.
5. Select the algorithm as the minimum cost flow algorithm.
6. Solve the problem of finding the optimal flow of goods through the
network.
7. Implement the recommended transportation plan.
8. This is a simplified overview, but it illustrates the general process for
implementing network flow problems in real-world scenarios.
 Problems Arise in Network Flow
Network flow problems are a class of optimization problems that involve the
movement of goods, information, or resources through a network. These
problems can be complex and various challenges and problems can arise in
solving them. Here are some of the more common problems you may encounter
with web streaming issues:
1. Infeasibility: a network flow problem may be infeasible if there are
constraints that cannot be satisfied simultaneously. For example, if supply
exceeds demand or if capacity constraints are violated, this can make the
problem impossible.
2. Unbounded: In some cases, a network flow problem may not have a
bounded solution. This can happen if there are no capacity constraints or if
the problem is not properly formulated.
3. Cyclic presence: Network cycles can complicate the resolution process. In
some network flow problems, it is desirable to eliminate loops, while in
others, they must be carefully managed to avoid infinite loops in the
algorithms.
4. Negative arc costs: If the problem involves cost minimization (such as in
the cost minimization flow problem) and the arc costs are negative, this can
lead to problems such as finding optimal solutions with negative infinite
costs.
5. Integer flow: Integer flow problems, where the values of flow must be
integers, can be more difficult to solve. Linear programming techniques
may not be directly applicable and special integer programming algorithms
may be required.
6. Non-integer capacities: In some practical cases, network edge capacities
may not be integers. This can cause problems in modeling and solving the
problem.
7. Complex networks: Real-world networks can be very large and complex,
making finding solutions computationally expensive. This can lead to
scalability issues and algorithm inefficiency.
8. Multiple Goods Flow: When you are dealing with multiple types of goods
with different flow requirements, the problem becomes more complex, and
ensuring that all goods reach their destination can be difficult.
 9. Severity and Uncertainty: Real networks often face uncertainty, such as
changes in supply or demand. Incorporating resilience into network flow
models can be difficult.
10.Algorithmic challenges: Depending on the specific formulation of the
problem, finding optimal solutions can be computationally intensive.
Special algorithms such as the Ford-Fulkerson algorithm, the Edmonds-Karp
algorithm, or network simplex methods can be used, and their convergence or
termination can sometimes be problematic.
Ford-Fulkerson Algorithm
Initially, the flow of value is 0. Find some augmenting Path p and increase
flow f on each edge of p by residual Capacity cf (p). When no augmenting
path exists, flow f is a maximum flow.
Example:

Let us consider the first augmenting path as S→A→ D→T

Now we consider another augmenting path arbitrarily S→ C→ D→ T
Updated graph is:

Next augmented path selected is:

Selected backward edge with non-zero value of flow

So bottle neck capacity is 4
All nodes other than source and sink should have same inflow and outflow.
Updated graph is:

Next we choose the augmenting path again:

Updated graph is:
Next we choose the augmenting path as:
Now we can not suggest any other augmenting path. So the maximum flow
through this network is 19.
Now we calculate the complexity of this algorithm:

Total complexity of entire algorithm is:

Example: Each Directed Edge is labeled with capacity. Use the Ford-Fulkerson
algorithm to find the maximum flow.

Solution: The left side of each part shows the residual network Gf with a shaded
augmenting path p,and the right side of each part shows the net flow f.