Neural Computing and Applications

https://doi.org/10.1007/s00521-018-3404-6 (0123456789().,-volV)(0123456789().,-volV)

S.I. : EMERGING INTELLIGENT ALGORITHMS FOR EDGE-OF-THINGS COMPUTING

A novel method for solving the fully neutrosophic linear programming

problems
Mohamed Abdel-Basset1 · M. Gunasekaran2 · Mai Mohamed1 · Florentin Smarandache3

Received: 9 February 2018 / Accepted: 24 February 2018

© The Natural Computing Applications Forum 2018

Abstract
The most widely used technique for solving and optimizing a real-life problem is linear programming (LP), due to its
simplicity and efficiency. However, in order to handle the impreciseness in the data, the neutrosophic set theory plays a
vital role which makes a simulation of the decision-making process of humans by considering all aspects of decision (i.e.,
agree, not sure and disagree). By keeping the advantages of it, in the present work, we have introduced the neutrosophic LP
models where their parameters are represented with a trapezoidal neutrosophic numbers and presented a technique for
solving them. The presented approach has been illustrated with some numerical examples and shows their superiority with
the state of the art by comparison. Finally, we conclude that proposed approach is simpler, efficient and capable of solving
the LP models as compared to other methods.

Keywords Trapezoidal neutrosophic number · Linear programming · Neutrosophic set · Ranking function

1 Introduction overwhelmingly deceitful and this affects optimal solution

of LP problems. Probability distributions failed to transact
One of the most extremely used OR methods in real-life with inaccurate and unclear information. Also fuzzy sets
problems according to empirical surveys is linear pro- were introduced by Zadeh [5] to handle vague and
gramming [1–4]. It is a mathematical programming which imprecise information. But also fuzzy set does not repre-
contains a linear objective function and a group of linear sent vague and imprecise information efficiently, because it
equalities and inequalities constraints. The petroleum considers only the truthiness function. After then, Ata-
manufacture was the first and most productive application nassove [6] introduced the concept of intuitionistic fuzzy
of linear programming. Well-defined data which contain a set to handle vague and imprecise information, by con-
greater cost of information are required for LP problems. sidering both the truth and falsity function. But also intu-
But in real-life problems, the precision of data is itionistic fuzzy set does not simulate human decision-
making process. Because the proper decision is funda-
mentally a problem of arranging and explicate facts the
& Mohamed Abdel-Basset
concept of neutrosophic set theory was presented by
[email protected]
Smarandache, to handle vague, imprecise and inconsistent
M. Gunasekaran information [7–10]. Neutrosophic set theory simulates
[email protected]
decision-making process of humans, by considering all
Mai Mohamed aspects of decision-making process. Neutrosophic set is a
[email protected]
popularization of fuzzy and intuitionistic fuzzy sets; each
Florentin Smarandache element of set had a truth, indeterminacy and falsity
[email protected]
membership function. So, neutrosophic set can assimilate
1
Department of Operations Research, Faculty of Computers inaccurate, vague and maladjusted information efficiently
and Informatics, Zagazig University, Sharqiyah, Egypt and effectively [11, 12]. We now can say that NLP problem
2
University of California, Davis, Davis, USA is a problem in which at least one coefficient is represented
3
Math and Science Department, University of New Mexico, by a neutrosophic number due to vague, inconsistent and
Gallup, NM 87301, USA uncertain information. The NLP problems are more useful

123
 Neural Computing and Applications

than crisp LP problems because decision maker in his/her solutions. The first introduction of fuzzy programming
formulation of the problem is not forced to make a delicate theory was suggested by Tanaka et al. [22]. The first for-
formulation. The use of NLP problems is recommended to mulation and solving of FLP problems are presented by
avert unrealistic modeling. In this research, it is the first Zimmerman [23]. Tanaka and Asai [24] suggested an
time to present LP problems in a neutrosophic environment approach for getting the fuzzy optimal solution of FLP
with trapezoidal neutrosophic numbers. Two ranking problems. Verdegay solved FLP problems by depending on
functions are introduced according to the problem type, for fuzzification principle of objective [25]. The fuzzified
converting NLP problem to crisp problem. The proposed version of mathematical problems was examined by Her-
model was applied to both maximization and minimization rera et al. [26]. An FLP problem with fuzzy values of
problems. objective function coefficients were proposed by Zhang
The remaining part of this research is marshaled as et al. [27]. They converted FLP problems into multi-ob-
follows: We survey the pertinent fuzzy and intuitionistic jective problems. Another model of FLP problems with
FLP problems literature review in Sect. 2. The important fuzzy values of objective function coefficients and con-
concepts of neutrosophic set arithmetic are presented in straints was introduced by Stanciulescu et al. [28]. An FLP
Sect. 3. The formularization of NLP models is presented in model with symmetric trapezoidal fuzzy numbers was
Sect. 4. The proposed method for solving NLP problems is presented by Ganesan and Veeramani [29]. They obtained
presented in Sect. 5. Numerical examples are disbanded the optimal solution of a problem without converting it to
with the suggested method, a comparison of results with the crisp form. A revised version of Ganesan and Veera-
different researchers is illustrated and the drawbacks of mani method was proposed by Ebrahimnejad [30]. A
existing methods are listed in Sect. 6. Finally, conclusions ranking function for arranging trapezoidal fuzzy numbers
and future trends are clarified in Sect. 7. of FLP problems was introduced by Mahdavi and Naasseri
[31]. The idealistic stipulation for FLP problems was
derived by Wu [32], by presenting the concept of a non-
2 Literature review dominated solution of multi-objective programming. By
utilizing a defuzzification function, Wu [33] converted the
Linear programming problems in the fuzzy environment problem into optimization problems. The full FLP prob-
have classified into two groups which are, symmetric and lems were introduced by Lotfi et al. [34]. Some researchers
non-symmetric problems according to Zimmermann [13]. have proposed a ranking function for converting FLP
Objectives and constraints weight are equally significant in problems into its tantamount crisp LP model and then
symmetric FLP problems, but non-symmetric problem solving it by standard methods. The primal simplex method
weights of objectives and constraints are not equal [14]. was extended by Maleki et al. [35], for solving FLP
Another classification of FLP problems was introduced by problems. Tavana and Ebrahimnejad introduced a new
Leung [15]: (1) problems with crisp values of objective and approach for solving FLP problems with symmetric
fuzzy values of constraints; (2) problems with crisp values trapezoidal fuzzy numbers [36]. The fully intuitionistic
of constraints and fuzzy values of objectives; (3) problems FLP problems introduced by Bharati and Singh [37]
with fuzzy objectives and fuzzy constraints; and finally (4) depend on sign distance between triangular intuitionistic
robust programming problems. Three types of fuzzy linear fuzzy numbers. A ranking function was used by Sidhu and
programming models were proposed by Luhandjula [16], Kumar [38] for solving intuitionistic FLP problems.
which are flexible, mathematical and fuzzy stochastic Nagoorgani and Ponnalagu [39] introduced an accuracy
programming models. Another six models of FLP prob- function to defuzzify triangular intuitionistic fuzzy number.
lems was introduced by Lnuiguchi et al. [17], which are as The previous researches motivated us to propose a study
follows: flexible, possibility programming, possibility LP for solving NLP problems. There does not exist any
by using fuzzy max, possibility linear programming with researches which solve neutrosophic linear programming
fuzzy preference relations, possibility linear programming problems with trapezoidal neutrosophic numbers [40–45].
with fuzzy objectives and robust programming. An FLP
problem with equality and inequality constraints are
introduced by Kumar et al. [18]. Various approaches for 3 Preliminaries
disbanding FLP with inequality constraints were proposed
by several authors [19–21], by firstly converting FLP A review of important concepts and definitions of neutro-
problems to its equivalent crisp model and then get the sophic set is presented in this section.
optimal fuzzy solution of the original case. A large number
Definition 1 [43] A single-valued neutrosophic set N
of authors have deliberated different properties of FLP
through X taking the form N={〈x, TN (x), IN ð xÞ, FN ð xÞ〉:
problems and suggested various models for finding

123
Neural Computing and Applications

 
x∈X}, where X be a universe of discourse, TN (x): X→[0, ða1 ; a2 ; a3 ; a4 Þ; aA~; hA~; bA~ and B~ ¼
1], IN ð xÞ: X→[0, 1] and FN ð xÞ: X→[0, 1] with 0≤TN (x)  
ðb1 ; b2 ; b3 ; b4 Þ; aB~; hB~; bB~ are as follows:
+IN ð xÞ+FN ð xÞ≤3 for all x∈X. TN (x), IN ð xÞ and FN ð xÞ  
A~ þ B~ ¼ ða1 þ b1 ; a2 þ b2 ; a3 þ b3 ; a4 þ b4 Þ; aA~ ^ aB~; hA~ _ hB~; bA~ _ bB~
represent truth membership, indeterminacy membership  
A~  B~ ¼ ða1  b4 ; a2  b3 ; a3  b2 ; a4  b1 Þ; aA~ ^ aB~; hA~ _ hB~; bA~ _ bB~
and falsity membership degrees of x to N. 

1 1 1 1  
A~1 ¼ ; ; ; ; aA~; hA~; bA~ ; where A~ 6¼ 0
Definition 2 [43] The trapezoidal neutrosophic number A~
a4 a3 a2 a1
( 
ðca1 ; ca2 ; ca3 ; ca4 Þ; aA~; hA~; bA~ if ðc [ 0Þ
is a neutrosophic set in R with the following truth, inde- cA~ ¼  
ðca4 ; ca3 ; ca2 ; ca1 Þ; aA~; hA~; bA~ if ðc\0Þ
terminacy and falsity membership functions: 8 

> a1 a2 a3 a4
8
 >
>
> ; ; ; ; aA~ ^ aB~; hA~ _ hB~; bA~ _ bB~ if ða4 [ 0; b4 [ 0Þ
>
> x  a1 >
>
>
< 

b4 b3 b2 b1

> aA~
> ð a1  x  a2 Þ A~ a4 a3 a2 a1
< a2  a1 ¼
B~ >
; ; ; ; aA~ ^ aB~; hA~ _ hB~; bA~ _ bB~ if ða4 h0; b4 i0Þ
>
>
b4 b3 b2 b1
TA~ð xÞ ¼ aA~ ð a2  x  a3 Þ ; ð1Þ >
>
>


>
> >
:
a4 a3 a2 a1
; ; ; ; aA~ ^ aB~; hA~ _ hB~; bA~ _ bB~ if ða4 \0; b4 \0Þ
>
> Aa ~ ð a  x  a Þ b1 b2 b3 b4
: 2 3
8 
0 otherwise < ða1 b1 ; a2 b2 ; a3 b3 ; a4 b4 Þ; aA~ ^ aB~; hA~ _ hB~; bA~ _ bB~ if ða4 [ 0; b4 [ 0Þ
>
8   A~B~ ¼ ða b ; a b ; a b ; a b Þ; a ~ ^ aB~; hA~ _ hB~; bA~ _ bB~ if ða4 h0; b4 i0Þ
> 1 4 2 3 3 2 4 1 A 
>
> a2  x þ hA~ x  a01  0  :
ða4 b4 ; a3 b3 ; a2 b2 ; a1 b1 Þ; aA~ ^ aB~; hA~ _ hB~; bA~ _ bB~ if ða4 \0; b4 \0Þ
>
>   a1  x  a2
>
> a  a 0
>
< 2 1
h A~ Definition 4 A ranking function of neutrosophic numbers
IA~ð xÞ ¼  0
  ð a2  x  a3 Þ ð2Þ
>
> x  a þ h A~ 4 a  x   is a function Ɍ: N ðRÞ ! R, where N ðRÞ is a set of neu-
>
>
3
  a \x  a 0
>
> a04  a3
3 4 trosophic numbers defined on set of real numbers, which
>
:
1 otherwise convert each neutrosophic number into the real line.
8     
>
> a2  x þ bA~ x  a001  00  Let A~ ¼ ða1 ; a2 ; a3 ; a4 Þ; aA~; hA~; bA~ and B~ ¼
>
>   a1  x  a2  
>
> a  a 00 ðb1 ; b2 ; b3 ; b4 Þ; aB~; hB~; bB~ are two trapezoidal neutro-
>
< 2 1
bA~
    ð a2  x  a3 Þ sophic numbers, then
FA~ð xÞ ¼ 00 ð3Þ
>
> x  a þ b A~ 4 a  x    
>
>
3
  a \x  a 00
1. If Ɍ(A)[R ~ B~ then A~ [ B; ~
>
> a004  a3
3 4
 
>
: ~ ~ B;
1 otherwise 2. If Ɍ(A)\R B~ then A\ ~
 
~ ~
3. If Ɍ(A)=R B then A ¼ B: ~ ~
where aA~, hA~ and bA~ represent the maximum degree of
truthiness, minimum degree of indeterminacy, minimum
degree of falsity, respectively, aA~, hA~ and bA~ 2 ½0; 1:
Also a001  a1  a01  a2  a3  a04  a4  a004 . 4 Neutrosophic linear programming
The membership functions of trapezoidal neutrosophic problem (NLP)
number are presented in Fig. 1.
In this section, various types of NLP problems are
Definition 3 [43] The mathematical operations on two
presented.
trapezoidal neutrosophic numbers A~ ¼

Fig. 1 Truth membership, indeterminacy and falsity membership functions of trapezoidal neutrosophic number

123
 Neural Computing and Applications

The first type of NLP problem is the problem in which X

n

coefficients of objective function variables are represented Maximize/minimize Z~  c~j x~j

j¼1
by trapezoidal neutrosophic numbers, but all other
parameters are represented by real numbers. Subject to
ð7Þ
Xn Xn
aij x~j  ~ b~i ;
~ ; ;  i ¼ 1; 2; . . .; m;
Maximize/minimize Z~  c~j xj
j¼1
j¼1

Subject to j ¼ 1; 2; . . .; n; xj  0:
ð4Þ
Xn
Here, c~j ; x~j and b~i are trapezoidal neutrosophic numbers.
aij xj  ; ¼;  bi ; i ¼ 1; 2; . . .; m;
Here, x~j is defined as trapezoidal neutrosophic numbers, if
j¼1
authors want to obtain results in the form of neutrosophic
j ¼ 1; 2; . . .; n; xj  0:
numbers. But in reality, any manager or decision maker
In this type of problem, c~j is a trapezoidal neutrosophic want to obtain the crisp optimal solution of problem,
number. through considering vague, imprecise and inconsistent
The second type of NLP problem is the problem in information when defining the problem. So, if we obtain
which objective function variables and coefficients are the crisp value of decision variables, this problem can be
exemplified by real values but coefficients of constraints considered as another formulation of NLP (6).
variables and right-hand side are represented by trapezoidal
neutrosophic numbers.
X
n 5 Proposed NLP method
Maximize/minimize Z ¼ cj xj
j¼1 A new approach suggested to find the neutrosophic optimal
Subject to solution of NLP problems is introduced in this section.
Xn
Step 1 Let decision makers insert their NLP problem
~ b~i ;
~ ; ; 
a~ij xj  i ¼ 1; 2; . . .; m; j ¼ 1; 2; . . .; n; xj  0:
j¼1
with trapezoidal neutrosophic numbers. Because we always
want to maximize truth degree, minimize indeterminacy
ð5Þ
and falsity degree of information, and then inform decision
Here, both a~ij and b~i are trapezoidal neutrosophic numbers. makers to apply this concept when entering trapezoidal
The third type of NLP problem is the problem in which neutrosophic numbers of NLP model.
all parameters are represented by trapezoidal neutrosophic Step 2 Regarding to definition 4, we propose a ranking
numbers, except variables are exemplified only by real function for trapezoidal neutrosophic numbers.
values. Step 3 Let (~ a ¼ al ; am1 ; am2 ; au ; ; Ta~; Ia~; Fa~Þ be a trape-
Xn zoidal neutrosophic number, where al ; am1 ; am2 ; au ; are
Maximize / minimize Z~  c~j xj lower bound, first, second median value and upper bound
j¼1
for trapezoidal neutrosophic number, respectively. Also
Subject to Ta~; Ia~; Fa~ are the truth, indeterminacy and falsity degree of
ð6Þ
Xn trapezoidal number. If NLP problem is a maximization
a~ij xj  ~ b~i ;
~ ; ;  i ¼ 1; 2; . . .; m; problem, then:
j¼1
Ranking function for this trapezoidal neutrosophic
j ¼ 1; 2; . . .; n; xj  0: number is as follows:


al þau þ2ðam1 þam2 Þ
Here, c~j ; a~ij and b~i are trapezoidal neutrosophic numbers. Ɍða~Þ ¼ +confirmation degree.
2
The NLP problem may also be a problem with neutro-
sophic values for variables, coefficients in goal function Mathematically, this function can be written as follows:

l 
and right-hand side of constraints. a þ au þ 2ðam1 þ am2 Þ
Rða~Þ ¼ þ ðTa~  Ia~  Fa~Þ ð8Þ
2
If NLP problem is a minimization problem, then:
Ranking function for this trapezoidal neutrosophic
number is as follows:


al þau 3ðam1 þam2 Þ
Ɍða~Þ ¼ 2 +confirmation degree.
Mathematically, this function can be written as follows:

123
Neural Computing and Applications


al þ au  3ðam1 þ am2 Þ equivalent crisp number. Remember that confirmation
Rða~Þ ¼ þ ðTa~  Ia~  Fa~Þ: ð9Þ degree of each trapezoidal number is (1, 0, 0) according to
2
decision maker opinion as we illustrated previously at the
Step 4 According to the type of NLP problem, apply the beginning of example. Then, the crisp model of previous
suitable ranking function to convert each trapezoidal neu- problem will be as follows:
trosophic number to its equivalent crisp value. This lead to
convert NLP problem to its crisp model. Maximize Z ¼ 19x1 þ 20x2 þ 21x3
Step 5 Solve the crisp model using the standard method Subject to
and obtain the optimal solution of problem. 12x1 þ 13x2 þ 12x3  503;
ð11Þ
14x1 þ 13x3  487;
12x1 þ 15x2  491
6 Numerical examples
x1 ; x2 ; x3  0:
In this section, to prove the applicability and advantages of We can structure the standard form of previous problem
our proposed model of NLP problems, we solved the same (11) as follows:
problem which introduced by Ganesan and Veeramani [29]
and Ebrahimnejad and Tavana [36]. Maximize Z ¼ 19x1 þ 20x2 þ 21x3
The difference between fuzzy set and neutrosophic set is Subject to
that fuzzy set takes into consideration the truth degree only. 12x1 þ 13x2 þ 12x3 þ s4 ¼ 503;
But neutrosophic set takes into consideration the truth, ð12Þ
14x1 þ 13x3 þ s5 ¼ 487;
indeterminacy and falsity degree. The decision makers and
12x1 þ 15x2 þ s6 ¼ 491;
problem solver always seek to maximize the truth degree,
minimize indeterminacy and falsity degree. Then, in the x1 ; x2 ; x3 ; s4 ; s5 ; s6  0:
following example, we consider truth degree (T)=1, inde-
where s4 ; s5 ; s6 are slack variables.
terminacy (I) and falsity (F) degree=0, as follows ð1; 0; 0Þ The previous standard form can be solved by the sim-
for each trapezoidal neutrosophic number and this called plex approach. The initial tableau of simplex is presented
the confirmation degree of each trapezoidal neutrosophic in Table 1.
number. We should also note that, according to Ganesan, The coming variable in Table 2 is x3 and departing
Veeramani and Ebrahimnejad, Tavana each trapezoidal variable is s5 .
number is symmetric with the following form: The entering variable is x2 and leaving variable is s4 as
 
a~ ¼ al ; au ; a; a ; shown in Table 3.

where al ; au ; a; a represented the lower, upper bound and

first, second median value of trapezoidal number, respec-
tively. The median values of trapezoidal numbers accord-
ing to Ganesan, Veeramani and by Ebrahimnejad, Tavana
are with equal vales (α). Now let us apply our proposed Table 1 Initial simplex form
method on the same problem.
Basic variables x1 x2 x3 s4 s5 s6 RHS

6.1 Example 1 s4 12 13 12 1 0 0 503

s5 14 0 13 0 1 0 487
s6 12 15 0 0 0 1 491
Z 19 20 21 0 0 0 0
Maximize Z~  ð13; 15; 2; 2Þx1 þ ð12; 14; 3; 3Þx2 þ ð15; 17; 2; 2Þx3
Subject to
~ ð475; 505; 6; 6Þ;
12x1 þ 13x2 þ 12x3 
Table 2 First simplex form
~ ð460; 480; 8; 8Þ;
14x1 þ 13x3 
~ ð465; 495; 5; 5Þ; Basic variables x1 x2 x3 s4 s5 s6 RHS
12x1 þ 15x2 
x1 ; x2 ; x3  ~
~ 0: s4 − 12/13 13 0 1 − 12/13 0 695/13
ð10Þ x3 14/13 0 1 0 1/13 0 487/13
s6 12 15 0 0 0 1 491
Because this NLP problem is a maximization problem, then Z − 47/13 20 0 0 − 21/13 0 10,227/13
by using Eq. (8) each trapezoidal number will convert to its

123
 Neural Computing and Applications

Table 3 Optimal form

Basic variables x1 x2 x3 s4 s5 s6 RHS

x2 − 12/169 1 0 1/13 − 12/169 0 695/169

x3 14/13 0 1 0 1/13 0 487/13
s6 2208/169 0 0 − 15/13 180/169 1 429
Z − 371/169 0 0 − 20/13 − 33/169 0 869

6.2 Comparisons between our proposed model need to approximate all not symmetric trapezoidal
and other existing models numbers into the closest symmetric numbers. This
approximation will make obtained results which are
By comparing proposed model results with Ebrahimnejad not delicate.
and Tavana [36] results of the same problem, we noted 6. The big drawback of Ebrahimnejad and Tavana fuzzy
that: model is the taking of truthiness function only. But in
real life, the decision-making process takes the
1. Our proposed model results are better than Ebrahim-
following form “agree, not sure and disagree.” We
nejad and Tavana results. Let us look at the optimal
treated this drawback in our model by using neutro-
tableau of our proposed model as shown in Table 3, it
sophic. Since, beside the truth function, we take into
is obvious that the objective function value equal 869
account the indeterminacy and falsity function.
but in Ebrahimnejad and Tavana, the objective func-
tion equal 635 by knowing that, the problem is a Also by comparing our model with Ganesan and
maximization problem. To make this more obvious, let Veeramani at the same problem, we also noted that:
us introduce the optimal tableau of Ebrahimnejad and
1. Our model is more simple and efficient than Ganesan
Tavana model as presented in Table 4.
and Veeramani model.
2. Ebrahimnejad and Tavana proposed their model to
2. Since obtained results of Ebrahimnejad, Tavana and
solve only symmetric trapezoidal numbers. But our
Ganesan and Veeramani are equals then, our results are
model can solve symmetric and non-symmetric
also better than Ganesan and Veeramani model.
numbers.
3. Our model represents reality efficiently than Ganesan
3. When entering symmetric trapezoidal numbers of
and Veeramani model, because we consider all aspects
Ebrahimnejad and Tavana, it take the following form:
of decision-making process in our calculations (i.e., the
 
a~ ¼ al ; au ; a; a , and they did not utilize the value of a truthiness, indeterminacy and falsity degree).
in their calculations of ranking function for obtaining the 4. Ganesan and Veeramani model represented to solve
equivalent crisp value, so let us ask ourselves a question only the symmetric trapezoidal numbers. Our model
“what is the rule of a?”. But in our proposed model, we can solve both the symmetric and non-symmetric.
take all values into considerations. Our ranking function Also, by comparing our model with Kumar et al. [18] for
has not any missing values of trapezoidal numbers, and solving the same problem we founded that:
then it is very accurate and comprehensive.
1. In their model, they convert the FLP problem to its
4. As we know,al ; au ; a; a represented the lower, upper tantamount crisp model. But their model has more
bound, first and second median value of trapezoidal variables and constraints.
number, respectively. Because two values of a are 2. Their models increase the complexity of solving linear
equals, then the triangular numbers will be more programming problem by simplex algorithm.
logical than trapezoidal numbers. 3. Our model reduces complexity of problem, by reduc-
5. To solve a problem with not symmetric trapezoidal ing the number of constraints and variables.
numbers using Ebrahimnejad and Tavana method, we 4. Their model is a time-consuming and complex, but our
model is not.
Table 4 Ebrahimnejad and Tavana optimal tableau 5. Also our model represents reality efficiently and better
Basis x1 x2 x3 s4 s5 s6 RHS than their model.

x2 − 12/169 1 0 1/13 − 12/169 0 730/169 By solving the previous example according to Saati
x3 14/13 0 1 0 1/13 0 470/169 et al. [44] proposed method, then the model will be as
s6 1848/169 0 0 − 15/13 180/169 1 70,170/169 follows:
Z 42/13 0 0 1 52/169 0 634.6

123
Neural Computing and Applications

Maximize Z ¼ 13x1 þ 12x2 þ 15x3 Table 5 Initial simplex form

Subject to Basic variables x1 x2 s3 s4 s5 RHS
12xl1 þ 13xl2 þ 12xl3  475; s3 33 46 1 0 0 90,041
12xu1 þ 13xu2 þ 12xu3  505; s4 53 15 0 1 0 48,046

1 þ 13x2 þ 12x3  6;
12xm1 m1 m1 s5 44 34 0 0 1 56,031
Z 25 48 0 0 0 0
1 þ 13x2 þ 12x3  6;
12xm2 m2 m2

14xl1 þ 13xl3  460;

14xu1 þ 13xu3  480; Maximize Z ¼ 25x1 þ 48x2
Subject to
1 þ 13x3  8;
14xm1 m1

ð14; 15; 17; 18Þx1 þ ð25; 30; 34; 38Þx2

1 þ 13x3  8;
14xm2 m2
ð13Þ ~ ð44; 980; 45; 000; 45; 030; 45; 070Þ

12xl1 þ 15xl2  465;
12xu1 þ 15xu2  495; ð21; 24; 26; 33Þx1 þ ð4; 6; 8; 11Þx2 ð14Þ
~ ð23; 980; 24; 000; 24; 050; 24; 060Þ

1 þ 15x2  5;
12xm1 m1

ð17; 21; 22; 26Þx1 þ ð12; 14; 19; 22Þx2

1 þ 15x2  5;
12xm2 m2
~ ð27; 990; 28; 000; 28; 030; 28; 040Þ

xl1 þ xu1  0;
~
~ 0:
x~1 ; x~2 
xl2 þ xu2  0;
xl3 þ xu3  0; By using Eq. (8), each trapezoidal number will convert to
its equivalent crisp number. Then, the crisp model of pre-
1 þ x1  0;
xm1 m2
vious problem will be as follows:
2 þ x2  0;
xm1 m2
Maximize Z ¼ 25x1 þ 48x2
3 þ x3  0:
xm1 m2
Subject to
As an effect, the numbers of constraints and variables are 33x1 þ 64x2  90; 041;
increased, and this lead to increase complexity of problem, ð15Þ
53x1 þ 15x2  48; 046;
increase the space of recording binary bits and also
increase computational time when solving it by simplex 44x1 þ 34x2  56; 031;
method. If the numbers of constraints of the original x1 ; x2  0:
problem are increased, then the solution will become very
We can structure the standard form of previous problem
difficult to apply. But our proposed method solves the same
(15) as follows:
problem with less variables and constraints, and then, with
less complexity and also less computational time when Maximize Z ¼ 25x1 þ 48x2
solving by simplex method. Subject to
33x1 þ 64x2 þ s3 ¼ 90; 041
6.3 Example 2 ð16Þ
53x1 þ 15x2 þ s4 ¼ 48; 046
In this example, we solve a NLP problem with trapezoidal 44x1 þ 34x2 þ s5 ¼ 56; 031
neutrosophic numbers. The order of element for trapezoidal x1 ; x2 ; s3 ; s4 ; s5  0:
neutrosophic numbers is as follows: lower, first median
where s3 ; s4 ; s5 are slack variables.
value, second median value and finally the upper bound.
The previous standard form can be solved by the sim-
The decision makers’ confirmation degree about each value
plex approach. The initial tableau of simplex is presented
of trapezoidal neutrosophic number is (0.9, 0.1, 0.1). This
in Table 5. The entering variable in Table 6 is x2 and
example belongs to the second classification of NLP
leaving variable is s3 .
problems as listed in Sect. 4.
The coming variable is x1 and departing variable is s5 as
in Table 7.

123
 Neural Computing and Applications

Table 6 First simplex form Table 8 Optimal simplex form

Basic variables x1 x2 s3 s4 s5 RHS Basis x1 x2 s3 s4 RHS

x2 33/64 1 1/64 0 0 1406.89 s3 0 − 7/3 1 − 2/3 10

s4 2897/64 0 − 15/64 1 0 26,942.6 x1 1 4/3 0 − 1/3 2
s5 847/32 0 − 17/32 0 1 8196.72 Z 0 2 0 2 12
Z 0.25 0 − 0.75 0 0 67,530.8

Table 7 Optimal form Table 9 Departments

Basic variables x1 x2 s3 s4 s5 RHS Products Wiring Drilling Assembly Inspection Unit profit

x2 0 1 2/77 0 − 3/154 1247.21 P1 0.5 3 2 0.5 ~

9$
s4 0 0 571/847 1 − 1.71015 12,925 P2 1.5 1 4 1 f
12$
x1 1 0 − 17/847 0 32/847 23,845/77 P3 1.5 2 1 0.5 f
15$
Z 0 0 − 631/847 0 − 8/847 67,608.2
P4 1 3 2 0.5 f
11$

6.4 Example 3
x1 ¼ 1:429 and x2 ¼ 0:429. It is obvious that two approach
Let us introduce another type of problems in this example results are nearly equal, but our proposed method has
and making a comparison with other research at the same several advantages over their method:
example. 1. We obtain the results which also obtained by Saati
By solving the same problem which introduced by Saati et al. [44] but with easy and simple method.
et al. [44]: 2. Number of constraints in our model is the same of the
Minimize Z ¼ 6x1 þ 10x2 original model, but when Saati solved their model, the
number of variables and constraints is significantly
Subject to
increased. Since in Saati et al. [44] model, number of
~ ð5; 8; 3; 13Þ;
2x1 þ 5x2  ð17Þ constraints of the previous problem becomes 20
~ ð6; 0; 4; 16Þ;
3x1 þ 4x2  constraints when they trying to solve the previous
~
~ 0:
x1 ; x2 
problem.
3. Due to the big increase in number of variables and
Let confirmation degree is (1, 0, 0) according to our constraints of Saati model, the complexity of solving
assumptions and note that, here the order of trapezoidal the problem by simplex will increase and computa-
neutrosophic number is as follows: lower bound, first, tional time will increase sure.
second median value and finally the upper bound, respec- 4. Their proposed approach is difficult to apply in large
tively. Let us use Eq. (9) for transforming the previous scale of problems.
model to its crisp model as follows: 5. Also their approach does not represent vague, incon-
sistent information efficiently.
Minimize Z ¼ 6x1 þ 10x2
Subject to
2x1 þ 5x2   6; : ð18Þ
6.5 Case study
3x1 þ 4x2  6; A company for electronic industries manufactures four
x1 ; x2  0: technical products for aerospace companies that conclude
NASA contracts. The outputs must get through four parts
The previous problem can be solved by the simplex
before they are shipped. These departments are: Wiring,
approach. The optimal tableau of simplex method is pre-
Drilling, Assembly and finally Inspection. The required
sented in Table 8.
time for each unit manufactured and its profit is presented
From the previous table, the value of objective function
in Table 9. The minimum production quantity for fulfilling
=12, x1 ¼ 2 and x2 ¼ 0:
contracts monthly is presented in Table 10. The objective
When Saati et al. [35] solved the previous example, the
of company is to produce products in such quantities for
results are nearly equal with our result. Since the value of Z
maximizing the total profits.
according to their model is equal to 12.857, the value of

123
Neural Computing and Applications

Table 10 Time capacity and

Departments Capacity (in hours) Products Minimum production level
minimum production level
Wiring g
1500 P1 g
150
Drilling g
2350 P2 g
100
Assembly g
2600 P3 g
300
Inspection g
1200 P4 g
400

The confirmation degree of previous information Maximize Z ¼ 27x1 þ 34x2 þ 43x3 þ 31x4
according to decision makers’ opinions is (0.9, 0.1, 0.1). Subject to
Let number of units of p1 produced=x1 ,
0:5x1 þ 1:5x2 þ 1:5x3 þ x4  4251;
Let number of units of p2 produced=x2 ,
Let number of units of p3 produced=x3 , 3x1 þ x2 þ 2x3 þ 3x4  6901;
Let number of units of p4 produced=x4 . 2x1 þ 4x2 þ x3 þ 2x4  7501;
The formulation of previous problem is as follows: 0:5x1 þ x2 þ 0:5x3 þ 0:5x4  3451; ð20Þ
Maximize Z~  9x 12x2 þ f
~ 1þf 15x3 þ f
11x4 x1  426;
Subject to x2  276;
g
0:5x1 þ 1:5x2 þ 1:5x3 þ x4  1500; x3  876;
g x4  1176:
3x1 þ x2 þ 2x3 þ 3x4  2350;
x1 ; x2 ; x3 ; x4  0:
g
2x1 þ 4x2 þ x3 þ 2x4  2600;
g By solving the previous model using simplex approach, the
0:5x1 þ x2 þ 0:5x3 þ 0:5x4  1200; ð19Þ
results are as follows:
g
x1  150;
x1 ¼ 426;
g
x2  100;
x2 ¼ 343;
g
x3  300; x3 ¼ 876;
g
x4  400: x4 ¼ 1176;
~
x1 ; x2 ; x3 ; x4  0: Z ¼ 97; 288:

Note that the values of each neutrosophic number repre-

sented by a trapezoidal neutrosophic number as follows:
7 Conclusions and research directions
9~ ¼ ð6; 8; 9; 12Þ; f
12 ¼ ð9; 10; 12; 14Þ;
f ¼ ð12; 13; 15; 17Þ; f
15 11 ¼ ð8; 9; 11; 13Þ; By applying the neutrosophic set concept to the linear
g ¼ ð120; 130; 150; 170Þ; programming problems, we treated imprecise, vague and
150
inconsistent information efficiently. We also have a better
g ¼ ð70; 80; 100; 120Þ;
100 representation of reality through considering all aspects of
g ¼ ð270; 280; 300; 320Þ; 400
300 g ¼ ð370; 380; 400; 420Þ; the decision-making process. We proposed two ranking
functions for converting trapezoidal neutrosophic numbers
g ¼ ð1200; 1300; 1500; 1700Þ;
1500 to its equivalent crisp values. The first ranking function is
g ¼ ð2200; 2250; 2350; 2400Þ
2350 for maximization problems and the second-ranking func-
g ¼ ð2200; 2400; 2600; 2800Þ; tion is for minimization problems. After using the suit-
2600
able ranking function and transforming the problem to its
g ¼ ð1000; 1100; 1200; 1300Þ:
1200 equivalent crisp model, then we solve the problem using
the standard methods. By comparing our proposed model
By using Eq. (8), the previous problem transform to the with other existing fuzzy models, we concluded that our
following crisp model as follows: proposed model is simpler, efficient and achieve better
results than other researchers. It is also revealed that pro-
posed method is equivalently applied for solving with the
symmetric and non-symmetric trapezoidal numbers.

123
 Neural Computing and Applications

Limitation of proposed research More involvements 20. Hashemi SM, Modarres M, Nasrabadi E, Nasrabadi MM (2006)
from more companies will make our research better. Fully fuzzified linear programming, solution and duality. J Intell
Fuzzy Syst 17:253–261
21. Allahviranloo T, Lotfi FH, Kiasary MK, Kiani N, Alizadeh L
(2008) Solving fully fuzzy linear programming problem by the
ranking function. Appl Math Sci 2:19–32
Compliance with ethical standards 22. Tanaka H, Okuda T, Asai K (1973) Fuzzy mathematical pro-
gramming. Trans Soc Instrum Control Eng 9:607–613
Conflict of interest We declare that we do not have any commercial 23. Zimmermann H-J (1978) Fuzzy programming and linear pro-
or associative interest that represents a conflict of interest in con- gramming with several objective functions. Fuzzy Sets Syst
nection with the work submitted. 1:45–55
24. Tanaka H, Asai K (1984) Fuzzy solution in fuzzy linear pro-
gramming problems. IEEE Trans Syst Man Cybern 14(2):325–
328
References 25. Verdegay JL (1984) A dual approach to solve the fuzzy linear
programming problem. Fuzzy Sets Syst 14:131–141
1. Kivijärvi H, Korhonen P, Wallenius J (1986) Operations research 26. Herrera F, Kovacs M, Verdegay J (1993) Optimality for fuzzified
and its practice in Finland. Interfaces 16:53–59 mathematical programming problems: a parametric approach.
2. Lilien GL (1987) MS/OR: a mid-life crisis. Interfaces 17:35–38 Fuzzy Sets Syst 54:279–285
3. Tingley GA (1987) Can MS/OR sell itself well enough? Inter- 27. Zhang G, Wu Y-H, Remias M, Lu J (2003) Formulation of fuzzy
faces 17:41–52 linear programming problems as four-objective constrained
4. Selhausen HMZ (1989) Repositioning OR’s products in the optimization problems. Appl Math Comput 139:383–399
market. Interfaces 19:79–87 28. Stanciulescu CV, Fortemps P, Installé M, Wertz V (2003) Mul-
5. Zadeh LA (1965) Fuzzy sets. Inf Control 8:338–353 tiobjective fuzzy linear programming problems with fuzzy deci-
6. Atanassov KT (1986) Intuitionistic fuzzy sets. Fuzzy Sets Syst sion variables. Eur J Oper Res 149:654–675
20:87–96 29. Ganesan K, Veeramani P (2006) Fuzzy linear programs with
7. Broumi S, Bakali A, Talea M, Smarandache F, Vladareanu L trapezoidal fuzzy numbers. Ann Oper Res 143:305–315
(2016) Shortest path problem under triangular fuzzy neutrosophic 30. Ebrahimnejad A (2011) Some new results in linear programs with
information. In: 2016 10th international conference on software, trapezoidal fuzzy numbers: finite convergence of the Ganesan and
knowledge, information management and applications (SKIMA), Veeramani’s method and a fuzzy revised simplex method. Appl
pp 169–174 Math Model 35:4526–4540
8. Broumi S, Smarandache F, Talea M, Bakali A (2016) Single 31. Mahdavi-Amiri N, Nasseri S (2006) Duality in fuzzy number
valued neutrosophic graphs: degree, order and size. In: 2016 linear programming by use of a certain linear ranking function.
IEEE international conference on fuzzy systems (FUZZ-IEEE), Appl Math Comput 180:206–216
pp 2444–2451 32. Wu H-C (2008) Optimality conditions for linear programming
9. Broumi S, Bakali A, Talea M, Smarandache F (2016) Isolated problems with fuzzy coefficients. Comput Math Appl 55:2807–
single valued neutrosophic graphs. Neutrosophic Sets and Sys- 2822
tems 11:74–78. https://doi.org/10.5281/zenodo.571458 33. Wu H-C (2008) Using the technique of scalarization to solve the
10. Broumi S, Talea M, Smarandache F, Bakali A (2016) Decision- multiobjective programming problems with fuzzy coefficients.
making method based on the interval valued neutrosophic graph. Math Comput Model 48:232–248
In: Future technologies conference (FTC), pp 44–50 34. Lotfi FH, Allahviranloo T, Jondabeh MA, Alizadeh L (2009)
11. Deli I, Şubaş Y (2017) A ranking method of single valued neu- Solving a full fuzzy linear programming using lexicography
trosophic numbers and its applications to multi-attribute decision method and fuzzy approximate solution. Appl Math Model
making problems. Int J Mach Learn Cybern 8:1309–1322 33:3151–3156
12. Deli I, Şubaş Y (2017) Some weighted geometric operators with 35. Maleki HR, Tata M, Mashinchi M (2000) Linear programming
SVTrN-numbers and their application to multi-criteria decision with fuzzy variables. Fuzzy Sets Syst 109:21–33
making problems. J Intell Fuzzy Syst 32:291–301 36. Ebrahimnejad A, Tavana M (2014) A novel method for solving
13. Zimmermann H-J (2012) Fuzzy sets, decision making, and expert linear programming problems with symmetric trapezoidal fuzzy
systems, vol 10. Springer, New York numbers. Appl Math Model 38:4388–4395
14. Amid A, Ghodsypour S, O’Brien C (2006) Fuzzy multiobjective 37. Bharati S, Singh S (2015) A note on solving a fully intuitionistic
linear model for supplier selection in a supply chain. Int J Prod fuzzy linear programming problem based on sign distance. Int J
Econ 104:394–407 Comput Appl 119:30–35
15. Leung Y (2013) Spatial analysis and planning under imprecision. 38. Sidhu SK, Kumar A (2016) A note on “Solving intuitionistic
Elsevier, Amsterdam fuzzy linear programming problems by ranking function”. J Intell
16. Luhandjula M (1989) Fuzzy optimization: an appraisal. Fuzzy Fuzzy Syst 30:2787–2790
Sets Syst 30:257–282 39. Nagoorgani A, Ponnalagu K (2012) A new approach on solving
17. Inuiguchi M, Ichihashi H, Tanaka H (1990) Fuzzy programming: intuitionistic fuzzy linear programming problem. Appl Math Sci
a survey of recent developments. In: Stochastic versus fuzzy 6:3467–3474
approaches to multiobjective mathematical programming under 40. Lopez D, Gunasekaran M (2015) Assessment of vaccination
uncertainty. Springer, Dordrecht, pp 45–68 strategies using fuzzy multi-criteria decision making. In: Pro-
18. Kumar A, Kaur J, Singh P (2011) A new method for solving fully ceedings of the fifth international conference on fuzzy and neuro
fuzzy linear programming problems. Appl Math Model 35:817– computing (FANCCO-2015). Springer, Cham, pp 195–208
823 41. Varatharajan R, Manogaran G, Priyan MK, Balaş VE, Barna C
19. Buckley JJ, Feuring T (2000) Evolutionary algorithm solution to (2017) Visual analysis of geospatial habitat suitability model
fuzzy problems: fuzzy linear programming. Fuzzy Sets Syst based on inverse distance weighting with paired comparison
109:35–53

123
Neural Computing and Applications

analysis. Multimed Tools Appl 1–21. https://doi.org/10.1007/ 44. Saati S, Tavana M, Hatami-Marbini A, Hajiakhondi E (2015) A
s11042-017-4768-9 fuzzy linear programming model with fuzzy parameters and
42. Manogaran G, Lopez D (2017) Spatial cumulative sum algorithm decision variables. Int J Inf Decis Sci 7:312–333
with big data analytics for climate change detection. Comput 45. Abdel-Basset M, Mohamed M, Zhou Y, Hezam I (2017) Multi-
Electr Eng 59:1–15 criteria group decision making based on neutrosophic analytic
43. Mohamed M, Abdel-Basset M, Zaied AN, Smarandache F (2017) hierarchy process. J Intell Fuzzy Syst 33(6):4055–4066
Neutrosophic integer programming problem, Neutrosophic Sets
Syst 15:3–7. https://doi.org/10.5281/zenodo.570944

123