Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics

DOI: 10.54254/2753-8818/25/20240979

Analysis of practical applications using the linear

programming model

Xinyan Zhang
The University of Manchester, Manchester, United Kingdom

[email protected]

Abstract. A linear programming model is a mathematical optimization technique used to find

the best possible outcome in a mathematical model with linear relationships under a set of
constraints. This technique has various applications in engineering, management, economic
analysis, and military operations. The linear programming model is the most popular technique
for determining the best course of action, and it is essential for decision-making and resource
allocation in modern society. This research paper will introduce the standard form of the linear
programming model and explore the differences and solutions between three linear programming
models: 0-1 integer programming, mixed integer programming, and multi-objective
programming. In the case of introducing the application and optimization effect of different
models in real life, list research and compare the different applications and research scenarios of
the three models to derive the strengths, weaknesses, and characteristics of the different models
to achieve the effect of providing more targeted models for different real-life cases.

Keywords: Linear programming, 0-1 integer programming, multi-objective programming,

mixed integer programming.

1. Introduction
Following Gomory's 1958 proposal of the cutting plane method, integer programming developed into a
separate field [1]. Many techniques have been created during the past 30 years to address various issues.
Integer programming can be categorized as 0-1 integer programming (0-1 IP). Integer programming
issues with only 0 or 1 as the only possible values are known as 0-1 integer programming problems. The
0-1 variable can be used to statistically explain the relationships between discrete variables that are
logical, sequential, and mutually exclusive. Any integer programming that uses bounded variables can
be processed as 0-1 programming. Numerous industries, including telecommunications, finance,
manufacturing, and logistics, can leverage 0-1 integer programming. Researchers have created hybrid
approaches integrating 0-1 integer programming with other optimization techniques, including
heuristics, meta-heuristics (genetic algorithms), relaxation, and recurrent methods to handle large-scale
complex 0-1 integer programming problems. Creating effective cutting planes, branch-and-cut
algorithms, parallel computing techniques, and investigation of applications in newly developing fields
like data science and machine learning are all part of current research on 0-1 integer programming.
Mixed integer programming (MIP) belongs to the same category of integer programming as 0-1
integer programming. In contrast to 0-1 integer programming, if some of the decision variables are not
discrete, the problem is called a mixed-integer programming problem [2]. Because it offers a flexible

© 2023 The Authors. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0
(https://creativecommons.org/licenses/by/4.0/).

246
 Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics
DOI: 10.54254/2753-8818/25/20240979

and effective method for resolving substantial, complicated issues and has been employed in various
fields, including finance, energy, telecommunications, and bioinformatics, mixed-integer programming
is frequently used in systems analysis and optimization. Includes the creation of sophisticated cutting
plane techniques, hybrid algorithms that combine Mixed-integer programming with other optimization
techniques, and efficient parallel computing strategies for handling large-scale Mixed-integer
programming issues.
Multi-objective programming (MOP) problems are mathematical planning issues considering
various objective functions [3]. In the middle of the 20th century, multi-objective programming began
to develop. Mathematical decision-making and optimization are the roots of multi-objective
programming. It entails resolving issues that require simultaneous optimization of several incompatible
goals. Genetic algorithms, simulated annealing, particle swarm optimization, and other techniques and
algorithms have all been created to handle multi-objective optimization issues. Numerous disciplines
use multi-objective programming, including engineering, economics, finance, environmental
management, and logistics. For instance, it can be used in engineering to develop products that consider
elements like cost, performance, and safety. Multi-objective programming is a crucial tool for supporting
decisions. It enables decision-makers to examine several solutions and decide with knowledge by their
preferences and priorities [4]. Recent developments in multi-objective programming include
incorporating machine learning methodologies, creating hybrid algorithms, and using multi-objective
programming to solve challenging real-world issues like resource management and healthcare
optimization.
This paper will start by introducing the background and leading solutions of the three linear
programming models and then analyze the background, purpose, and results of three classical linear
programming examples. A case study will be used to compare the three different linear programming
and research objectives. It will conclude with research adjustments and future perspectives.

2. Prerequisite knowledge
Develop a mathematical model for linear programming:
1. list the constraints and objective function
2. Draw the feasible domain represented by the constraints.
3. Find the optimal solution and the optimal value of the objective function in the feasible domain.

2.1. Standard linear programming and solution method

The standard form consists of three parts: 1. a function to be maximized. 2. problem constraints, 3. Non-
negative variables [5]. generally in the form of:
a11 x1 + a12 x2 ≤ b1
a21 x1 + a22 x2 ≤ b2 (1)
a31 x1 + a32 x2 ≤ b3
Other problems, such as minimization problems, problems with different forms of constraints, and
problems with negative variables, can be rewritten as standard types of their equivalent problems.
Therefore the standard form looks like [6]: Maximize 𝑐1 𝑥1 + 𝑐2 𝑥2 + ⋯ 𝑐𝑛 𝑥𝑛 , subject to
𝑎11 𝑥1 + 𝑎12 𝑥2 + ⋯ 𝑎1𝑛 𝑥𝑛 ≤ 𝑏1
𝑎21 𝑥1 + 𝑎22 𝑥2 + ⋯ 𝑎1𝑛 𝑥𝑛 ≤ 𝑏2
(2)
⋮
𝑎𝑚1 𝑥1 + 𝑎𝑚2 𝑥2 + ⋯ 𝑎𝑚𝑛 𝑥𝑛 ≤ 𝑏𝑚
𝑥1 , 𝑥2 , ⋯ , 𝑥𝑛 ≥ 0
It can also rewrite as matrix form:

247
 Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics
DOI: 10.54254/2753-8818/25/20240979

a11 a12 ⋯ a1n

a21 a22 ⋯ a2n
A=( ⋮ ⋱ ⋱ ⋮ ), (3)
am1 am2 ⋯ amn
b = (b1 , b2 , ⋯ , bn )T ,
c = (c1 , c2 , ⋯ , cn )T ,
x = (x1 , x2 , ⋯ , xn )T .
The simplex approach is typically used to solve problems using classic linear programming [7].
George Dantzig's Simplex Form Algorithm was first conceived as a method for gradually descending to
a vertex on a convex polyhedron [8]. As long as there is an optimal solution for this linear programming,
the optimal solution will be found after a finite number of steps of iteration. The basic idea of the simplex
method is to start from a vertex of the feasible set of linear programming and seek the next vertex in the
direction that makes the value of the objective function decrease. The three concerns that follow must
be resolved to use an iterative approach to discover the best linear programming solution: Three factors
can cause the value of the objective function to decrease more than others: 1. The optimal solution
discrimination criterion, which is the iteration termination criterion. 2. The base-switching operation,
which is the method of iterating from one base-feasible solution to another base-feasible solution, and
3. The selection of the base-entry columns, which is the selection of appropriate columns to carry out
the base-change operation, can result in a significant decrease in the value of the objective function [9].
The simplex method's general solution steps can be summed up as follows [10]:
(1) Find the initial basic feasible solution to the linear programming problem and express the system
of constraint equations as a canonical system of equations.
(2) The problem has no solution if the basic feasible solution is not possible, i.e., the constraints are
incompatible.
(3) If the basic feasible solution is present, start with it and use the optimality condition and feasibility
condition to introduce non-basic variables to replace a particular basic variable to find a different basic
feasible solution with a higher objective function value.
(4) Repeat steps 3 and 2 until the relevant test number satisfies the criterion for optimality (at this
point, the objective function value cannot be increased). This is the best way to solve the issue.
(5) The iteration is stopped if the objective function value of the issue is discovered to be unbounded
during the iteration procedure.
The amount of constraints has a major impact on how many iterations of the simplex approach are
needed to solve a linear programming problem. Today, the simplex method is used on computers to
handle issues involving general linear programming.

2.2. 0-1 integer programming and solution method

In 0-1 integer programming, the only possible values for the decision variables are 0 and 1. When
making the choice, 1 equals "yes" and 0 equals "no." In most cases, the branch and bound method of
hidden enumeration is used to solve 0-1 integer programming.
The main concept behind the branch and bound method is to compute a target lower term (for
minimum value problems) or a target upper term (for maximum value problems) for the set of solutions
within each subset, which is referred to as delimitation. This process is repeated until the entire space of
feasible solutions has been divided into smaller and smaller subsets, called branches. Pruning is the
process of stopping further branching after each subset whose boundaries go beyond the given objective
value of the feasible set, allowing for the disregard of numerous subsets [11].
The branch-and-bound method's primary steps are as follows (using minimization as an example):
1. If minimization is the goal of the problem, set as the value of the best solution.
2. In accordance with the Branching rule, a node is chosen from the unfathomed nodes (local
solutions) and split into numerous new nodes in the node's subsequent stratum.
3. Determine each freshly branched-out node's Lower bound.

248
 Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics
DOI: 10.54254/2753-8818/25/20240979

4. Each node is subject to the fathom condition; if its lower bound value exceeds or is equal to the Z
value, the node is no longer taken into consideration. This node contains a feasible solution that has the
lowest lower bound value. If this condition is true, the feasible solution is compared to the Z value, and
if the former is smaller, the Z value is modified to reflect the feasible solution's value [11].
5. If there are any unfathomed nodes, move on to step 2. If there are no more unfathomed nodes, the
process ends and the best solution is found.
The branch and bound method has the advantage that finding the best solution can be done more
quickly and with fewer subproblems to check. However, a large number of sub-node boundaries and
accompanying consumption matrices must be stored. It uses up a lot of memory [11].

2.3. Mixed integer programming and solution method

Partial decision variables with integer values are referred to as "mixed integer programming" problems.
This type of problem can be solved using the branch and bound method and the simplex method
mentioned above, as well as the memory cut in the plane, cut method.
The primary procedures in memory cut are:
1. relaxation of the overall variables
2. Use the simplex approach to solve the linear programming problem resulting from the relaxation
of the integer programming problem.
3. Cut the solution space while ensuring that the feasible solution is not cut off and that the solution
from the preceding linear programming is not present in the solution space after cutting.
4. After cutting, keep using linear relaxation and keep going through the stages above until you reach
the definitive answer.
It is important to keep in mind that repeating the aforementioned processes will produce many
distinct ideal solutions. If the intersection of the viable half-spaces of the constraint space is the empty
set, then the problem lacks a workable solution [12].

2.4. Multi-objective programming and solution method

Multi-objective programming involves solving several objectives simultaneously. Since these objectives
mostly interact with each other, when one objective is optimized, the other ones to be optimized may
worsen. In contrast to the other planning mentioned above, multi-objective planning cannot find an
optimal solution; the multi-objective planning generation goal is to find the best state within a certain
range.
The weighting sum method, the priority method, and the ideal point method are the three basic
approaches to solving multi-objective planning problems.

2.4.1. The weighting sum method. Since there are different preferences for different objectives, it is
possible to add weights to each objective function in multi-objective planning, thus transforming
multi-objective planning into single-objective planning. This method is suitable for real-world
environments where expert scoring is available.This is expressed as [13]:
min Z = ∑ni=1 wi fi, (4)

lb ≤ x ≤ ub
Aeq ∗ x = beq
subject to: { (5)
Ax ≤ b
∑𝑛𝑖=1 𝑤𝑖 = 1

2.4.2. The priority method. The main idea of the priority method is to categorize the importance of the
objectives into different levels, first seek the optimal value of the high-priority objective function, then
seek the low-priority objective function while making sure that the high-priority objective receives at
least the best quality.
Take an example of double-goal programming:

249
 Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics
DOI: 10.54254/2753-8818/25/20240979

min Z1 = f1 (x)
min Z2 = f2 (x)
lb ≤ x ≤ ub
subject to {Aeq ∗ x = beq (6)
Ax ≤ b
Assuming that the priority of f1 (x) is higher than the priority of f2 (x), then first find the optimal
solution of f1 (x) the solution Z1 . Subsequently solve for the optimal value of f2 (x) while ensuring that
f1 (x) is optimal:
min Z2 = f2 (x)
lb ≤ x ≤ ub
Aeq ∗ x = beq
subject to { (7)
Ax ≤ b
f1 (x) = Z1
By analogy, all multi-objective planning models with more than 2 objectives can be solved according
to the above ideas.

2.4.3. The ideal point method. By solving n single objective problems: min Zi (x), Suppose that the
single optimal solution to each single-objective problem is Zi ,then 𝑍 ∗ = (𝑍1 , 𝑍2 , . . . , 𝑍𝑛 ) can be called
a single ideal point in the value domain, which corresponds to the single optimal solution of each
single-objective problem, and if the multi-objective problem can reach this point, then the optimal
solution can be obtained [14].
However, since it is difficult to reach the optimal solution among multi-objective optimization, we
can only find the closest Z to 𝑍 ∗ as an approximate solution within some range of feasible points. The
most direct way is to minimize the distance of each planning problem from its corresponding optimal
solution, the equation looks like:

min Z = √[f1 (x) − Z1 ]2 + [f2 (x) − Z2 ]2 + ⋯ [fn (x) − Zn ]2

lb ≤ x ≤ ub
subject to {Aeq ∗ x = beq (8)
Ax ≤ b

3. Application analysis

3.1. 0-1 integer programming application: a university timetabling problem

The university timetable problem is a typical instance of 0-1 integer programming being used to
establish a workable schedule for classes, professors, and students while satisfying a variety of
constraints and objectives [15].

3.1.1. Consideration of factors. 1. Course scheduling: the curriculum includes both core and elective
courses that need to be assigned to specific slots and rooms to ensure that courses with conflicting
schedules are not assigned to the same slot. 2. Professor Availability: Courses are scheduled according
to the professors' specialisations and preferences [15]. 3. Time Allocation: Minimise overlap between
courses and free time between each session [15]. 4. Classroom consistency: all class periods for a course
should be held in the same classroom. The Usage of 0-1 Integer Programming in the Problem of
Developing a University Timetable.
Decision Variables: 0-1 variables are used to indicate whether a particular class or course is assigned
to a particular time and room. If a class is assigned to a specific time slot and room, then the variable
has a value of 1; otherwise, it has a value of 0.

250
 Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics
DOI: 10.54254/2753-8818/25/20240979

Objective function: The goal is to minimise conflicts, room usage or teacher workload [15]. This can
be expressed as an objective function to be minimised or maximised.
Constraints: the constraints of course scheduling, classroom consistency, professor availability, and
time allocation must be met.

3.1.2. Result. Scheduling using 0-1 integer programming optimizes the use of resources, such as
classrooms, teachers, and time, while reducing expenses. Additionally, scheduling issues between
students and teachers are decreased while satisfaction among both groups is raised [16].

3.2. Mixed Integer Programming Application: Production Line Optimisation

Mixed integer programming is frequently used in operations and manufacturing management for
production line optimization. Its goal is to boost revenue while boosting productivity and cutting
expenses [16].

3.2.1. Consideration of factors. To maximize productivity, cost, and resource utilization in a production
line, a number of factors must be taken into account. These elements consist of: 1. Production planning:
To achieve a balanced production line rate and a balanced distribution of work, determine the sequence
and timing of production operations for each product and build a production plan appropriately. 2.
Resource allocation: Effectively distribute scarce resources, including as labor, materials, and machines,
across various production activities while taking into account issues like the space needed for equipment,
the simplicity of moving semi-finished goods, and the competing uses of tools. 3. Manage inventory
levels to strike a balance between inventory expenses and the risk of stock-outs and production lags.

3.2.2. The usage of mixed integer programming in the problem of Production Line Optimization.
Decision variables: Define decision variables, some of which may be 0 or 1, to indicate choices such as
whether to produce a particular product in a particular time period or whether to allocate resources to a
particular task. Continuous variables may indicate the number of products to be produced as well as the
sequence of production processes.
Objective function: Develop an objective function for minimization or maximization. This function
can be a measure of production cost, production profit, production time, etc.
Constraints: These include consideration of factors as well as capacity constraints, resource
availability, demand requirements, and maintenance schedules. These constraints ensure that the
production plan meets the actual constraints and requirements.

3.2.3. Result. By optimizing the production line, a mixed integer program can effectively improve the
efficiency of the production line, the balance rate of the production line, and the production line process,
thereby increasing the production of products. At the same time, it can improve the profit and increase
the order quantity to increase the competitiveness in the market.

3.3. Multi-objective programming application: water resource allocation

In order to establish thorough evaluation statistics of water resource allocation options and to choose the
most reasonable water resource allocation option, multi-objective programming is a significant and
useful application approach in the field of water resource allocation [17].

3.3.1. Consideration of factors. 1. Water resource development: maximizing environmental

sustainability through the protection of water habitats, particularly riverine wetlands, and seasonal
influences on water resources [17]. 2. water demand: satisfy a range of demands, including
environmental, industrial, agricultural, and domestic ones. Measures to lessen the effects of seasonal
flooding and droughts on infrastructure, agriculture, and communities [17]. 3. Existing projects: Projects
relating to water supply, water conservation, wastewater reuse, and the possibility of water supply in
reservoirs and plants.

251
 Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics
DOI: 10.54254/2753-8818/25/20240979

3.3.2. The usage of multi-objective programming in the problem of water resource allocation. Decision
variables: water resources' added value, global water supply, per-person water supply, water supply
assurance, water shortage, and ecological satisfaction [17].
Objective function: Economic benefits and total investment; water supply; water Supply guarantee
and scarcity; environmental impact of public water.
Constraints: Including consideration of factors and local development and living standards.
Target weight: Six indicators are given the target weights: value added of water resources, total water
supply, water supply per person, water supply that is assured, water shortage, and ecological satisfaction
[17].

3.3.3. Result. Improved decision-making that takes into account many goals at once improves
sustainability, boosts resistance to droughts and floods, and reduces costs by better allocating ecological
and environmental needs.

3.4. Comparative analysis

This section compares and contrasts 0-1 IP, MIP, and MOP in terms of application objects, research
scenarios, method characteristics, benefits, and drawbacks.
Table 1. Comparison of 0-1 integer programming, mixed integer programming, objective multi-
objective programming.
0-1 integer multi-objective
mixed integer programming
programming programming
Tackling optimizing
Used for portfolio issues with many
Application decisions, and the Some variables must be 0 or 1; competing objectives
object decision variable is 0 others can take continuous values. requires several factors
or 1. to be considered at
once.
-Scheduling design:
-Environmental
selecting or
planning: balancing
excluding courses or
-Production Planning: Optimise objectives such as
schedules for a
production planning through pollution reduction
specific time.
decisions such as machine setup and minimize cost in
-Backpack problem:
(discrete) and productivity development plans.
selecting or rejecting
Research continuum (continuous). -Portfolio
items to maximize a
scenario -Supply Chain Optimisation: optimization:
particular value or to
Balancing inventory levels, labor, maximizing profit and
meet a weight limit.
and machine allocation(discrete) minimizing risk.
-Item selection:
with transport costs, productivity, -Product design:
deciding which
and profitability(continuous). Balancing cost, profit,
items to take on or
and performance in
exclude to optimize
engineering design.
a set of criteria.
Branch and bound Branch and bound, Linear Weighted sum method,
Methods method, cutting programming relaxation, cutting idea point method,
plane algorithms. plane algorithms. priority method.
The constraints and Various problem types
objectives are Provides flexibility and breadth in can be handled and
Advantages incorporated into a real-world problems with precise help decision makers
unified framework, solutions. understand the
and exact optimal inherent trade-offs to

252
 Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics
DOI: 10.54254/2753-8818/25/20240979

Table 1. (continued).

make informed choices by

solutions can be
weighting different
found.
objectives.
The effectiveness of the
Assigning weights or
It is limited to binary formulation depends on the
priorities to objectives is
decisions which quality of the model
often subjective, and the
cannot accurately formulation. Poorly
choice of objective weights
represent all formulated models can lead
can significantly impact the
Disadvantages situations in the real to sub-optimal or infeasible
results. Significant
world. Problem solutions. Due to the
computational resources
solving can become complexity of the variables,
are required for large-scale
lengthy for large more sophisticated
or highly non-linear
problems. equipment may be required
problems.
for the calculations.

3.5. Discussion
This chapter uses classic examples of class scheduling, production line optimization, and water
balancing to analyze the application of different linear programming models in real-life cases. This
chapter briefly describes the applications, research scenarios, methodological features, advantages, and
disadvantages of the three different linear programming methods to provide more relevant and effective
modeling for different real-world applications. There are more than three types of linear programming
and many different application areas. However, one or more linear programming methods will be more
efficient and accurate than the others in arriving at the optimal solution in different real-world scenarios.

4. Conclusions
In many disciplines, including operations research, economics, engineering, and management, linear
programming is crucial for tackling various optimization issues. By selecting the suitable model for the
job, linear programming can be more effective. Similar to integer programming, more and more
optimization techniques will be used with linear programming in the future. Technical difficulties are
also included in the development of linear programming. More significant and complicated problems
can be addressed faster by using faster and more efficient linear programming solvers that are being
developed thanks to algorithms, hardware, and parallel computing improvements. In conclusion, linear
programming is still an essential and adaptable method for optimization in various fields. Its bright
prospects for the future are anticipated to fuel its ongoing significance and expansion as computer power,
integration with other optimization techniques, and applications in data-driven decision-making all
develop.

References
[1] Gomory R E 1958 Bulletin of the American Mathematical Society 64 275278
[2] Kantor I, Robineau J, Bütün H, Maréchal F 2020 Frontiers in Energy Research 8 49
[3] Alves M J, Clí maco J 2001 Encyclopedia of Optimization Springer Boston MA
[4] Karimi N, Feylizadeh M R, Govindan K, Bagherpour M 2022 Expert systems with applications
196 116663
[5] Matousek J and Gartner B 2007 Springer-Verlag, Berlin Heidelberg 1-6
[6] Omar Antolín C 2022 Standard Form for Linear Programs
[7] Guoqing H 2020 Optimization cbe cornell edu
[8] Dantzig G B 1987 Trchnical Report Sol 87-5
[9] Guogen X, Housui Z, Zhiyong H 2018 OptimizationBeijing University of Aeronautics and
Astronautics Press 73

253
 Proceedings of the 3rd International Conference on Computing Innovation and Applied Physics
DOI: 10.54254/2753-8818/25/20240979

[10] Fanghuai W, Ning X 2011Haitian Publishing House 130

[11] Subham D, Branch, Bound A 2020 Baeldung on Computer Science. 23 21
[12] István M 2023 Mixed Integer Linear Programming: Formal Definition and Solution Space.
[13] Xiujuan L, Zhongke S 2004 J. of Systems Engineering and Electronics 15 2 142-146
[14] Weixiong C, Yifei R, Jian L, Lixin Y 2022 J. of Beijing Jiaotong University 46 04 23-30 2022
[15] Bakir M, Akif, and Aksop C 2018Hacettepe Journal of Mathematics and Statistics 37 1 41-55
[16] Wenqi L, Wei L 2018 Fujian Computer 34 12 46-47
[17] Cuizhu Z, Zhitao L, Qiuping Y, Yin Z, Yaming H, Jianyu L 2022 Mechanical and Electrical
Engineering Technology 51 05 198-203

254