Hindawi

Journal of Mathematics
Volume 2022, Article ID 7660174, 9 pages
https://doi.org/10.1155/2022/7660174

Research Article
An Integer Linear Programming Model for Partially Ordered Sets

Elsayed Badr ,1 I.M. Selim,2 Hoda Mostafa,3 and Hala Attiya4

1
Scientific Computing Department, Faculty of Computers and Artificial Intelligence, Benha University, Benha, Egypt
2
Computing Science Department, Faculty of Computers and Artificial Intelligence, University Sadat City, Sadat City, Egypt
3
Mathematics and Computer Science Department, Faculty of Science Menoufia University, Shibin Al Kawm, Egypt
4
Basic Science Department, Faculty of Technology and Education, Beni-Suef University, Beni Suef, Egypt

Correspondence should be addressed to Elsayed Badr; badrgraph@gmail.com

Received 26 April 2022; Revised 30 July 2022; Accepted 23 August 2022; Published 10 September 2022

Academic Editor: Mehar Ali Malik

Copyright © 2022 Elsayed Badr et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Linear programming is an important approach that is used to represent a large class of combinatorial optimization problems. The
simplex algorithm is one of the algorithms for solving linear programming problems with exponential time complexity. For-
tunately, this algorithm solves real-world problems with polynomial time complexity. In particular, a new Integer Linear
Programming model (ILPM) is proposed for partially ordered sets. Robert Dilworth’s Decomposition theorem is formulated by
ILPM and proves its correctness using the paradigm of strong duality in linear programming. Finally, ILPM is run on fifteen
benchmark partially ordered sets for finding their width. The computational experiments show the validity of the proposed model.

1. Introduction complete knowledge of the issue of optimization to be

represented by a linear program. The second of these tools is
Linear programming is a very rich problem-solving model the complete knowledge of the general form of the linear
which is used to represent a large class of optimization program (objective function–constraints–the right-hand
problems. Optimization problems are defined as problems side of the constraints (constants)–non-negative–etc). The
whose solutions represent maximum values (such as max- third of these tools is the practice of formulating a set of
imum profit) or minimum values (such as minimum cost). optimization problems into a linear program version. For
There are various types of optimization problems such as graph labeling problems, Badr et al. [3–5] introduced new
constraint optimization problems and unconstraint opti- integer linear programming models (ILPMs) to find the
mization problems. On the other hand, there are many types upper bounds of the radio number, a radio mean number,
according to the nature of the variables of the optimization and radio mean square number for a given graph, respec-
problem, such as linear programming, integer linear pro- tively. The proposed ILPMs outperformed other algorithms
gramming, and nonlinear programming. There are various according to the running time. For the metric dimension of a
algorithms to solve linear programming problems, the most given graph, Chartrand et al [6] formulated the optimization
famous of which are the simplex algorithm and the interior problem of determining the metric dimension of a given
point algorithm. The simplex algorithm is one of the ten best graph (G). For the traveling salesman problem, there are
mathematics algorithms of the twentieth century [1]. By a different mathematical models [7–9] for the traveling
common numerical example, Klee and Minty [2] proved that salesman problems.
the simplex algorithm is not a polynomial time algorithm. The problem of efficient tasks scheduling is ubiquitous in
Fortunately, this algorithm solves real-world problems of many areas of computer science and engineering. We are
polynomial time complexity. often given a collection of tasks to complete, in which some
Casting optimization problems into a linear program jobs can be done in parallel, and others must be done in a
version is an art. The tools of this art are represented in the predefined sequence. In general, we can formulate these
2 Journal of Mathematics

problems by constructing a poset of objects (representing Definition 3. A partially-ordered set, or poset, is a set of
tasks) in which some elements are “comparable” (jobs that elements P, along with a binary relation ≺, having the
must be done in predefined sequence), and others are property that ≺ is transitive and antisymmetric for all ele-
“incomparable” (jobs that can be done in parallel). If our ments in P.
goal is to devise an efficient task scheduling algorithm, we Note that, two elements ai and aj can also be incom-
might be interested in finding the largest subset of tasks that parable, that is, when we have both ai ⊀ aj and aj ⊀ ai . In this
can be completed in parallel (the maximum antichain case, tasks ai and aj may be completed in parallel.
problem). Or we may be interested in the related problem of We also define maximal and minimal elements in a
distributing all the tasks between the fewest number of poset:
subgroups (e.g., for parallel processing) such that all jobs in a An element ai is maximal in a subset S ⊆ P if it is not
subgroup must be completed in a predefined sequence (the smaller than, any other element in S, i.e. ai ⊀ aj for all aj ∈∈S .
minimum chain covering problem). Similarly, an element ai is minimal S if it is not greater than
For more details about the history of the decomposition any other element i.e. aj ⊀ ai for all ai ∈∈S.
of the partially ordered sets posets, the reader can refer to
[10–16]. On the other hand, for more details about how to Definition 4. A subset of elements � 􏼈a1 , . . . , ak 􏼉 ⊆ P forms
formulate a problem as a mathematical model the reader is an antichain if and only if ai ⊀ aj for all i≠ j. That is all
referred to [17–19]. Finally, for more details about the well elements in A are pairwise incomparable.
know optimal PMU placement the reader is referred to
[20–22]. Definition 5. A subset of elements C � 􏼈a1 , . . . , ak 􏼉⊆ Ƥ
In this paper, a new integer linear programming model forms a chain if and only if they satisfy for all
(ILPM) is proposed for posets. Robert Dilworth’s Decom- a1 ≺a2 ≺ . . . . . . ≺ak for some ordering of elements in C. That
position theorem is formulated by ILPM and proves its is all elements in C are pairwise comparable.
correctness using the paradigm of strong duality in linear For example, in a task scheduling problem, if a subset of
programming. Finally, ILPM is run on fifteen benchmark jobs forms an antichain. Then, all these jobs may be com-
partially ordered sets for finding their width. The compu- pleted in parallel. If a subset of jobs forms a chain, then these
tational experiments show the validity of the proposed jobs must be completed in the sequence determined by
model. a1 ≺a2 ≺ . . . . . . ≺ak . Furthermore, it is useful to describe a
The remaining of this paper is organized as follows: basic covering of a poset P by chains to be a collection of chains
definitions and notations are introduced in Section 2. In 􏼈C1 , . . . . . . , Ck 􏼉 such that every element of Ƥ is contained in
Section 3, Dilworth theorem and its mathematical formu- the union of these chains, that is, ⋃ Ci � P. This is related to
lation are proposed. In Section 4, the experimental results distributing all tasks on parallel processors 􏼈C1 , C2 . . . . . .􏼉 so
analysis of the proposed ILPM on fifteen benchmark posets that jobs assigned to any processor Ci are to be completed in
are provided. In Section 5, conclusions are made. a predefined sequence.

2. Preliminaries 3. Dilworth’s Theorem

Before presenting Dilworth’s Theorem, we need to have a With these definitions in hand, we can now formulate
notion of comparability between elements in a set. To Dilworth’s theorem [10] on the size of maximum antichains
formalize this notion, we define some properties of binary and minimum chain coverings in a poset.
relations to represent a partially ordered set (poset). We then
illustrate how subsets of elements in a poset can have chain Theorem 1 (Dilworth’s Theorem). İe size of the largest
or antichain properties. antichain in any poset P is equal to the size of the minimum
coverage of P by chains.
We can construct a proof of Dilworth’s theorem using
2.1. Binary Relations. Elements of a set can have relational the pigeonhole principle and Duality in linear programming.
properties such as transitivity or antisymmetry. If we let:
A � max {|A|:A is antichain of Ƥ} (size of largest anti-
chain in P), and B � min k: ∪ 􏼈C1 , . . . . . . , Ck 􏼉 � P (mini-
Definition 1. A binary relation,≺, on a set P is transitive if mum number of chains needed to cover P), then one way to
and only if (a ≺ b and b ≺ c) implies a ≺ c for all a, b, c ϵ P prove Dilworth’s theorem is to show A ≤ B and B ≤ A.
To prove A ≤ B, we use the pigeonhole principle.
Definition 2. A binary relation ≺ , on a set P is anti-sym-
metric if and only if (a ≠ b and a ≺ b) implies b ≺ a for all a, b Lemma 1. The size of any chain in a poset P is equal to at
ϵ P. most the minimum number of chains needed to cover P, i.e.
We use binary relations with these specific properties to A ≤ B.
represent a partial order on a set of elements, in which some
pairs of elements are comparable by ai ≺aj In context of a Proof. As above-mentioned, let B be the minimum number
task scheduling problem, this means that job.ai must be of chains C1 , . . . . . . , CB needed to cover P. Assume, for the
completed before. sake of contradiction, that an antichain A � a1 , a2 , . . . an in
Journal of Mathematics 3

Ƥ has size n > B. Then, by the pigeonhole principle, at least Claim 1. For all chains Ck ⊆P, at most one of the weights
two distinct elements ai , aj must be members of the same associated with elements in Ck is equal to 1. All other weights
chain so that, by Definition 5, we have ai aj or aj ai . must be 0.
However, by Definition 4, this means ai and aj are not in the
same chain, a contradiction. Thus, no antichain in P can Proof of Claim 1. For the sake of contradiction, assume that
have size greater than B, whereby it follows the maximum more than one element in chain C has a positive weight. We
size antichain in P is also no larger B. □ would then have 􏽐i: ai εCxi > 1 since all xi are non-negative
integers, thereby violating constrains of the linear program.
It follows that xi ∈ {0, 1} for all i. Now consider each xi as
an indicator variable for whether element ai appears in an
3.1. Casting as a Linear Program. To prove B ≤ A A is more antichain Amax , i.e. let xi � 1 if ai εAmax and xi � 0
difficult. We can begin by trying to cast the objective of otherwise.
finding the maximum size antichain of a poset as a linear Since at most one indicator variable in any chain is
program. positive, the solution to (1) satisfies the requirement that no
two elements chosen for the maximum antichain come from
the same chain.
3.1.1. Primal Form. Assign non-negative weights
→ Furthermore, under the indicator variable interpreta-
x � (x1 , . . . , xn ) respectively to each element a1 , . . . . . . a1
tion, the objective of maximizing z � 􏽐 xi corresponds to
ϵP. Each chain Ck ⊂ Ρ will have an associated constrain
choosing the maximum number of elements to include in an
inequality over the elements ai ∈ Ck that comprise it.
antichain of P, so we have Maxz � |Amax |, as desired. This
Consider the following linear program in canonical form:
proves Proposition 1 that an integer solution to (1) gives the
n
size of the maximum antichain in P.
max z � 􏽘 xi , Now, we prove the analogous proposition for the dual
i�1
form. □
􏽘 xi ≤ 1 for all chains Ck ⊆ Ρ . (1)
s.t i: ai ∈Ck →
Proposition 2. If yj εZ for all weights in y then solving dual
x1 ≥ 0 for i � 1. . . . . . . n. (2) gives the size of the minimum covering of P by chains.

Proof. The reasoning is analogous to that of Proposition 1.

The constrains, in (2), 􏽐j: ai ∈Cj yj ≥ 1 for all ele-
3.1.2. Dual Form. In addition, we can convert (1) to its dual
mentsai ∈ P, impose the following requirement on variables
form, where variables represent chains on P, and each el-
yj if we assume they are all non-negative integers: □
ement ai ∈ Ρ has an associated constrain inequality over the
collection of chains that contain it (note the symmetry).
Claim 2. For all ai ∈ P, at least one of weights associated
Assign non-negative weights.
→ with the set of chains containing ai is positive.
y � (y1 , . . . , yk ), respectively, to each chains
C1 , . . . Ck ⊆ P. The dual of (1) (henceforth denoted (2)) is
Proof of Claim 2. It follows immediately from the con-
k strains. If, for some element ai , none of the weights asso-
Minimize w � 􏽘 yj , ciated with chains that contain ai are positive, then
j�1 􏽐j: ai ∈Cj yj � 0≯1, thereby violating the constrain of the
􏽘 yj ≥ 1 for all elements ai ∈ P (2) linear program.
Subject to : j: ai ∈Cj Furthermore, since the linear program seeks to minimize
􏽐 yj with all yj as integers, we must have that yj � 1
yj ≥ 0 for j � 1, . . . , k. whenever yj is chosen positive. Thus, with yj ε{0, 1} for all j,
We begin by showing that if the primal (1) and (2) have we can again consider each yj as an indicator variable for
integer solutions, then the objective z in (1) gives the size of whether chain Cj in a minimum covering of P.
the maximum antichain in P, and the objective w in (2) gives Note again that for each element ai ∈ P, at least one
the size of the minimum coverage of P by chains. indicator variable is associated with a chain containing ai
equal 1, so the solution to (2) satisfies the requirement that
→ all elements aεP appear in the minimum chain covering.
Proposition 1. If xi εZ for all weights in x , then solving Furthermore, under the indicator variable interpre-
primal (1) gives the maximum antichain in P. tation, the objective of minimizing w � 􏽐 yj corre-
sponds to choosing the minimum number of chains
Proof. In the case that all xi are non-negative integers, we to include in a covering of P, so we have Min w �
wish to show that the objective of (1), z � 􏽐ni�1 xi , is equal to min􏼈c: C1 , . . . , Cc covers P􏼉, as desired. This proves
|Amax |, where Amax is any maximum antichain in P. Proposition 2 that an integer solution to (2) gives the
Observe that constrains in (1), 􏽐i: ai ∈Ck xi ≤ 1 for all chains minimum covering of P by chains.
Ck ⊆ P impose the following requirement on the variable xi if Now, we need to show that both the primal and dual
we assume they are all non-negative integers: □ linear programs indeed have integer solutions. □
4 Journal of Mathematics

Lemma 2. For any distribution of chains in a poset, (1) has the value of 0. It follows that optimal solutions to each LP
an optimal solution where all variables take on integer values. associated with subsets of P1 will also have integer weights.
Thus, the inductive hypothesis holds for k � 1.
Proof. Proof is by the Well-ordering Principle and con- Now, suppose that an optimal solution for (2) on poset P
tradiction of an optimal solution to (1) with the fewest has a variable assignment yi � a that is not an integer (note
number of noninteger variables (xi weights). Let N be the set that the optimal value for this LP must be an integer
of elements in P corresponding to these noninteger weights. Min w � k ∈ Z). We can remove from P all elements in the
Some of these elements will be maximal in N, and others will chain Ci associated with variable yi to obtain a new poset
be minimal (recall definitions in Section 2). We first need to P/C � P′ ⊂ P.
verify that. □ Let w′ be the objective function value of LP2′ for this a
new poset P′ . We observe that Min w′ is at most k − a
Claim 3. No elements in N can be both minimal and because all elements comprising the chain with weight
maximal. yi � a were removed so that w ≤ w − a � k − a. Since
′
P′ ⊂ P is a strict subset of P, k − a is an integer by the
Proof of Claim 3. Suppose, for the sake of contradiction, that inductive hypothesis, so Min w′ � k − 1. Furthermore, the
an element am εN is both minimal and maximal. Then, am is inductive hypothesis tells us that P has a covering by k − 1
′
chains where all weights assignments to the chains are
incomparable to all other elements in N (recall definitions in
Section 2). It follows that is associated weight xm is subject integers. It follows that we get a covering of P � P′ ⋃ Ci by
only to a single constrain that xm ≤ 1 in (1), whereby xm k chains if we simply assign weight of yi � 1 to Ci ,
would be assigned a value of 1 to maximize the objective. achieving the desired objective the desired value of
This contradicts the assumption that xm εN is a not an in- Min w � Min w′ + yi � (k + 1) + 1 � k. This completes the
teger, thus proving the claim. inductive argument and proof of Lemma 3.
Since no elements in N are both minimal and maximal, it Taking Lemma 2 and 3 in conjunction with Proposition
follows that each chain C ∈ N has exactly one maximal 1 and 2 gives the desired result that, for all partially-ordered
element, and a different minimal element. sets P:
Suppose we add a small value ∈ > 0 to each maximal Solving (1) gives the size of the largest antichain in P
element in N, and subtract e from each minimal in N. Since
Solving (2) gives the size of the smallest covering of P by
all elements in N are nonintegers we ensure that the resulting
chains
assignment of variables is feasible by choosing ∈ > 0 small
enough so that 0 ≤ xi ≤ 1 is still satisfied by all xi , as required Again, by applying the result of strong duality [19] in
by the constraints of (1). linear programming, we conclude that primal (1) and dual
If we keep repeating this procedure, then at least one (2) have the same optimal value. Therefore, (1) s solution for
variable corresponding to a maximal/minimal element in N the size of the largest antichain in a poset and (2) solution for
will reach a boundary value of 0 or 1. This contradicts the the size of the minimum chain covering of a poset are equal,
assumption that N has the fewest number of noninteger proving Theorem 1 (Dilworth’s Theorem). □
variables in a feasible solution. Thus, all variable assignments
in an optimal solution to (1) must be integers, so Lemma 2
holds for primal (1). Next, we need to show that dual (2) also 3.2. A Polynomial-Size Linear Program. In this section, we
has integer solutions. □ introduce Proposition 3 on which the proposed integer
linear programming model is based.
Lemma 3. For any distribution of chains in a poset, (2) has
an optimal solution where all variables take on integer values.
Proposition 3. An assignment of indicator values {0, 1} to
→
x � (x1 , . . . , xn ) produces an antichain if and only if xi +
Proof. First, we apply the result of strong duality in linear
xj 1 for all edges (ai , aj ) ∈ E(G).
programming (see [19] for details). Which says that the
primal and dual forms of a linear program have the same
optimal value. We prove using induction that if k ϵ Z is the Proof. First, we show that any antichain in P satisfies the
optimal value to LP2 on a poset P, then P has a covering by k constraints. An antichain cannot contain any two elements
→
chains where all variables y in (2) take on integer values. that are comparable (Definition 4). Note that, any two
Inductive hypothesis: for any poset P having an integer comparable elements in P are connected by an edge in G.
optimal (2) objective value of k, the optimal solutions to the Thus, we may choose at most one vertex from each edge
LPs associated with strict subsets of P are integers. (ai , aj ) ∈ E(G), meaning that at most one of xi , xj equals 1,
Base case. Consider a post P1 in which all elements form from which it follows that xi + xj ≤ 1 for all edges
a single, with optimal objective value Min w � k � 1. From (ai , aj ) ∈ E(G).
the constraints of (2), this optimal solution will have a single Conversely, we show that any assignment of indicator
variable y1 (corresponding to the chain containing all ele- variables satisfying the constraints produces a valid
ments in P1 ) take on the value of 1, and the other y’s take on antichain.
Journal of Mathematics 5

6 5 4

1 2 3
Figure 1: A poset with 6 elements and its width equal to three.

4 3
5
5 6
4

3 4 1 3

1 2 6 2
1 2
(P1; W (P1) = 2; S (P1) = 1) (P2; W (P2) = 2; S (P2) = 3) (P3; W (P3) = 2; S (P3) = 2)

6 5 4
6 5 4 5 4

1 2 3 1 2 3
1 2 3
(P4; W (P4) = 3; S (P4) = 3) (P5; W (P5) = 3; S (P5) = 3) (P6; W (P6) = 3; S (P6) = 3)

5 4 6 7 4 3
6 7
5
8 4

1 3
2 3
1 2 5
1 2

(P7; W(P7) = 3; S(P7) = 3) (P8; W (P8) = 3; S (P8) = 4) (P9; W(P9) = 3; S(P9) = 3)

8 7 6 5
8 7 6 5 7 6 5 4

1 2 3
1 2 3 4 1 2 3 4

(P10; W (P10) = 4;S (P10) = 4) (P11; W (P11) = 4; S (P11) = 4) (P12; W (P12) = 4; S (P12) =4)

7 7 5 4 7 6 5 4
6
6 5 4

1 2 3 1 2 3

1 2 3
(P13; W (P13) = 4; S (P13) = 4) (P14; W (P14) = 4; S (P14) = 4) (P14; W (P15) = 4; S (P15) = 4)

Figure 2: 15-Benchmark partially ordered sets with their width and jump number.
6 Journal of Mathematics

Table 1: A comparison between the standard width and directed acyclic graph (DAG).
Proposed model MATLAB
Poset n Standard w(P)
w(P) CPU time
P1 4 2 2 0.073318
P2 6 2 2 0.089859
P3 6 2 2 0.101679
P4 6 3 3 0.102334
P5 6 3 3 0.104611
P6 5 3 3 0.108036
P7 7 3 3 0.123854
P8 8 3 3 0.146553
P9 5 3 3 0.166726
P10 8 4 4 0.207229
P11 8 4 4 0.328825
P12 7 4 4 0.471679
P13 7 4 4 0.47268
P14 7 4 4 0.546337
P15 7 4 4 0.649021

If xi + xj ≤ 1 for all edges (ai , aj ) ∈ E(G), then most one 0 0 0 0 1 1

indicator variable xi , xj is positive because xi ∈ {0, 1}. ⎡⎢⎢⎢ ⎤⎥⎥
⎢⎢⎢ 0 0 0 1 0 1 ⎥⎥⎥⎥⎥
Hence, at most one vertex (element in P) on each edge is ⎢⎢⎢⎢ ⎥⎥⎥
chosen to be in the antichain. Since all pairwise comparable ⎢⎢⎢ 0
⎢ 0 0 1 1 0 ⎥⎥⎥⎥⎥
elements in P are connected by edges in our construction of A � ⎢⎢⎢⎢⎢ ⎥⎥⎥. (4)
⎢⎢⎢ 0
⎢⎢⎢ 0 0 0 0 0 ⎥⎥⎥⎥⎥
E(G), it follows that no two elements chosen in the antichain ⎢⎢⎢ ⎥⎥⎥
will be comparable, satisfying Definition 4. This proves ⎢⎢⎣ 0 0 0 0 0 0 ⎥⎥⎥⎥⎦
Proposition 3.6. □ 0 0 0 0 0 0
The m × n matrix A∗ ∈ Rm×n is generated from matrix A:
1 0 0 0 1 0
⎡⎢⎢⎢ ⎤⎥
3.2.1. Formulation of the Problem as a Mathematical Model. ⎢⎢⎢ 1 0 0 0 0 1 ⎥⎥⎥⎥⎥
For any poset P with element set 􏼈a1 , . . . , an 􏼉, we can ⎢⎢⎢ ⎥⎥⎥
⎢⎢⎢ ⎥⎥⎥
construct an n × n comparable incidence matrix A, having ⎢
⎢ 0 1 0 1 0 0 ⎥⎥⎥
A∗ � ⎢⎢⎢⎢⎢ ⎥⎥⎥. (5)
entry A(i, j) � 1 if elements ai , aj ∈ Pare comparable and ⎢⎢⎢⎢ 0 1 0 0 1 0 ⎥⎥⎥⎥⎥
A(i, j) � 0 otherwise. Now, associated with each element ⎢⎢⎢ ⎥
⎢⎢⎢ 0 0 1 1 0 0 ⎥⎥⎥⎥⎥
ai ∈ P an indicator variable xi ∈ {0, 1} that corresponds to ⎢⎣ ⎥⎦
using or not using that element in an antichain of P. 0 0 1 0 1 0
Now, we can formulate the problem of the decomposed
poset P as follows: Each row in the matrix A∗ ∈ Rm×n represents an in-
n equality based on Proposition 3.7.
max z � 􏽘 xi , Finally, the complete integer linear programming model
i�1 (3) is as the following:
∗ 6
Subject to: A x ≤ b : x1 , x2 , . . . , xn ∈ {0, 1}, max z � 􏽘 xi ,
i�1
where A∗ ∈ Rm×n is the m × n matrix which is generated
from the matrix A (based on Proposition 3) such that m is x1 0 0 0 x5 0≤1
the number of ones in the matrix A, x ∈ Rn and b ∈ Rm (all x1 0 0 0 0 x6 ≤ 1
elements in vector b equal to 1).
0 x2 0 x4 0 0≤1 (6)
Subject to: 0 x2 0 0 x5 0≤1
Example 1. We show how to formulate the problem of
decomposed partially ordered set (as shown in Figure 1) as 0 0 x3 x4 0 0≤1
an integer linear programming model. 0 0 x3 0 0 0 ≤ 1
The n × n comparable incidence matrix A that is cor-
responding to the post P. x1 , x2 , x3 , x4 , x5 , x6 ∈ {0, 1}.
Journal of Mathematics 7

4.5

3.5

2.5

1.5

0.5

0
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15

Standard W (P)
Upper Bound W (P)
Figure 3: Shows that the relation between posets and standard width, (DAG).

CPU Time
0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15

CPU Time
Figure 4: Shows that the relation between posets and CPU Time.

We can use an available software package such as (MAT- (n), and various heights h(P)) Figure 2 contains fifteen
LAB, Python, and LINGO) for solving the above mathematical benchmark posets with their width and jump number.
model which returns the solution is equal to three. The used environment is a 64-bit Windows 8.1 Op-
erating System with Core (TM) i5 CPU M 460 @2.53 GHz,
4. Computational Experiments 4.00 GB of memory, and MATLAB v7.01 Software
Package.
In this section, we evaluate the proposed ILPM by fifteen From Table 1 and Figure 3, it is obvious that the results of
benchmark posets. These posets are characterized by some the proposed mathematical model (ILPM) coincide with the
characteristics such as (various widths w(P), various sizes standard results for all fifteen benchmark posets.
8 Journal of Mathematics

It is also evident from Table 1 and Figure 4 that the Theodore S. Motzkin), pp. 159–175, Academic Press, Cam-
execution time of ILPM increases as the width of the poset bridge, MA, USA, 1972.
increases. That is, the running time of ILPM is directly [3] E. Badr, S. Almotairi, A. Eirokh, A. Abdel-Hay, and
proportional to the width of the post. Fortunately, the be- B. Almutairi, “An integer linear programming model for
havior of the simplex algorithm used to solve ILPM is a solving radio mean labeling problem,” IEEE Access, vol. 8,
pp. 162343–162349, 2020.
polynomial time to solve real-world applications, regardless
[4] E. M. Badr and M. I. Moussa, “An upper bound of radio
of being exponential time as proven in [2]. For this reason, k-coloring problem and its integer linear programming
we have done our best to create a poset in the form of the model,” Wireless Networks, vol. 26, no. 7, pp. 4955–4964,
examples provided by Klee and Minty in 1972 in which they 2020.
proved that the simplex algorithm is exponential time, but [5] E. Badr, S. Nada, M. M. Ali Al-Shamiri, A. Abdel-Hay, and
we have failed to do. A. ELrokh, “A novel mathematical model for radio mean
Because the examples presented by Klee and Minty square labeling problem,” Journal of Mathematics, vol. 2022,
depend on (right-hand side constants, the coefficients of the Article ID 3303433, 9 pages, 2022.
constraint matrix, and the coefficients of the objective [6] G. Chartrand, L. Eroh, M. A. Johnson, and O. R. Oellermann,
function), rarely, do these problems exist as real-world “Resolvability in graphs and the metric dimension of a graph,”
problems. We conclude from this that the presented model Discrete Applied Mathematics, vol. 105, pp. 99–113, 2000.
will be implemented in a polynomial time. It is well-known [7] Ö. Salehi, A. Glos, and J. A. Miszczak, “Unconstrained binary
models of the travelling salesman problem variants for
that the complexity of the simplex algorithm is O(n2) for
quantum optimization,” Quantum Information Processing,
real-world applications in the average case where n is the vol. 21, no. 2, p. 67, 2022.
number of elements of a poset. [8] G. Gutin and A. Punnen, The Traveling Salesman Problem and
its Variations. Combinatorial Optimization, Kluwer Academic
5. Conclusion Press, Dordrecht, Netherlands, 2002.
[9] C. Papalitsas, T. Andronikos, K. Giannakis,
A new integer linear programming model (ILPM) has been G. Theocharopoulou, and S. Fanarioti, “A QUBO model for
proposed for posets. Robert Dilworth’s Decomposition the traveling salesman problem with time windows,” Algo-
theorem is formulated by ILPM and proved its correctness rithms, vol. 12, no. 11, p. 224, 2019.
using the paradigm of strong duality in linear programming. [10] R. P. A. Dilworth, “A decomposition theorem for partially
Finally, ILPM is run on fifteen benchmark partially ordered ordered sets,” Annals of Mathematics, vol. 51, no. 1,
sets for finding their width. The computational experiments pp. 161–166, 1950.
showed the validity of the proposed model. On the other [11] G. B. Dantzig and D. R. Fulkerson, “Minimizing the number
of tankers to meet a fixed schedule,” Naval Research Logistics
hand, CPU Time has been calculated for fifteen benchmark
Quarterly, vol. 1, no. 3, pp. 217–222, 1954.
partially ordered sets. From computational experiments, it [12] G. B. Dantzig and A. J. Hoffman, “Dilworth’s theorem on
was concluded that the CPU time increases as the width partially ordered sets,” in Linear Inequalities and Related
increases. In future work, the proposed ILPM will be Systems, pp. 207–214, Princeton University Press, Princeton,
evaluated by using special posets such as interval posets, NJ, USA, 1956.
lattice, crown posets, and k-tower poset. ILPM will be ap- [13] R. P. Dilworth, “Some combinatorial problems on partially
plied on the jump-critical ordered sets [23]. ordered sets Combinatorial Analysis,” in Book: The Dilworth
Theorems: Selected Papers of Robert P. Dil Worth, Springer
Data Availability Science + Business Media, Berlin, Germany, 1990.
[14] D. R. Fulkerson, “Note on Dilworth’s decomposition theorem
The data used to support the findings of this study are for partially ordered sets,” Proceedings of the American
available from the corresponding author upon request. Mathematical Society, vol. 7, no. 4, pp. 701-702, 1956.
[15] L. R. Ford and D. R. Fulkerson, Flows in Networks, Princeton
University Press, Princeton, NJ, USA, 1962.
Conflicts of Interest [16] P. C. Fishburn, Interval Graphs and Interval Orders, Wiley,
Hoboken, NY, USA, 1985.
The authors declare that they have no conflicts of interest. [17] E. M. Badr and S. Almotairi, “On a dual direct cosine simplex
type Algorithm and its computational behavior,” Mathe-
Acknowledgments matical Problems in Engineering, Article ID 7361092, 8 pages,
2020.
The authors are thankful to Runpeng Liu for his suggestions [18] E. M. Badr and H. elgendy, “A Hybrid water cycle—particle
that make this manuscript better. swarm optimization for solving the fuzzy underground water
confined steady flow,” Indonesian Journal of Electrical Engi-
References neering and Computer Science, vol. 19, no. 1, p. 492, 2020.
[19] M. S. Bazaraa, J. J. Jarvis, and H. D. Sherali, Linear Pro-
[1] A. Barry, “Cipra “the best of the 20th century: editors name top gramming and Network Flows, John Wiley, Hoboken, NY,
10 algorithms” from,” SIAM News, vol. 33, 2000. USA, 2010.
[2] V. Klee and G. J. Minty, “How good is the simplex algorithm?” [20] N. P. Theodorakatos, “Optimal phasor measurement unit
Inequalities III (Proceedings of the Third Symposium on In- placement for numerical observability using branch-and-
equalities Held at the University of California, Los Angeles, bound and a binary-coded genetic algorithm,” Electric Power
Calif., September 1–9, 1969, Dedicated to the Memory of Components and Systems, vol. 47, pp. 357–371, 2019.
Journal of Mathematics 9

[21] N. P. Theodorakatos, “Fault location observability using

phasor measurement units in a power network through de-
terministic and stochastic algorithms,” Electric Power Com-
ponents and Systems, vol. 47, no. 3, pp. 212–229, 2019.
[22] N. P. Theodorakatos, N. M. Manousakis, and G. N. Korres,
“Optimal placement of phasor measurement units with linear
and non-linear models,” Electric Power Components and
Systems, vol. 43, no. 4, pp. 357–373, 2015.
[23] E. Badr and M. Moussa, “On jump-critical ordered sets with
jump number four,” Journal of Advances in Applied &
Computational Mathematics, vol. 1, pp. 8–13, 2014.