Duality in nonlinear fractional programming problem using fuzzy programming and genetic algorithm

International Journal of Soft Computing, Mathematics and Control (IJSCMC), Vol. 4, No.1, February 2015
DOI : 10.14810/ijscmc.2015.4103 19
DUALITY IN NONLINEAR FRACTIONAL
PROGRAMMING PROBLEM USING FUZZY
PROGRAMMING AND GENETIC ALGORITHM
Ananya Chakraborty
Assistant Professor, Department of Mathematics, Vemana Institute of Technology,
Kormangala 3rd
Block, Bangalore-34. Karnataka, India
ABSTRACT
In this paper we have considered nonlinear fractional programming problem with multiple constraints. A
pair of primal and dual for a special type of nonlinear fractional programming has been considered under
fuzzy environment. Exponential membership function has been used to deal with the fuzziness. Duality
results have been developed for the special type of nonlinear programming using exponential membership
function. The method has been illustrated with numerical example. Genetic Algorithm as well as Fuzzy
programming approach has been used to solve the problem.
KEYWORDS
Fuzzy Mathematical Programming, Nonlinear Fractional Programming, Exponential membership function,
Decision Analysis, Genetic Algorithm.
1. INTRODUCTION
Several factors in the real world imply the increase in use of nonlinear programming models.
There are several classes of problems where nonlinear programming had had a great impact for
example oil and petrochemical industries, nonlinear network problems and economic planning
models. The area where nonlinear programming can be used in these several classes of problems
is given in detail in (Ladson et al., 1980). The concept of fuzzy programming in decision making
problem was first proposed by (Bellmann and Zadeh, 1970). Many authors have applied fuzzy
programming approach in different area of linear programming (Zimmermann, 1978, Stancu-
Minasian et al., 1978, Chakraborty and Gupta, 2002). In (Jimenez, 2005), author has considered a
non linear programming problem with fuzzy constraints and the solution has been obtained by
muti objective evolutionary Algorithms. An interactive cutting plane algorithm for fuzzy multi
objective nonlinear programming problems has been presented in (Kanaya, 2010). In (Jameel,
2012), author has obtained accurate results for solving non linear programming using fuzzy
environment by using properties of fuzzy set and fuzzy number with linear membership function.
Fractional programs arise in management decision making as well as outside of it. They also
occur sometimes indirectly in modeling where initially no ratio is involved. The efficiency of a
system is sometimes characterized by a ratio of technical and/or economical terms. Maximizing
system efficiency then leads to a fractional program. List of frequently occurring objectives are
maximization of productivity, maximization of return on investment, maximization of return/risk,

International Journal of Soft Computing, Mathematics and Control (IJSCMC), Vol.4, No. 1, February 2015
20
minimization of cost/time, maximization of output/input, Non-Economic Applications. There are
a number of management science problems which indirectly give rise to a fractional program; see
e.g. (Schaible, 1981, 1983, 1995). A new method has been given by (Borza et al. 2012) for
solving the linear fractional programming problems with interval coefficients in the objective
function.
An algorithm has been developed by (Saad et al., 2011) for multi objective integer nonlinear
fractional programming problem under fuzziness. In order to defuzzify the problem, he has
developed the concept of α-level set of the fuzzy number is given and for obtaining an efficient
solution to the problem (FMOINLFP), a linearization technique is presented to develop the
solution algorithm. In the paper (Biswas and Bose, 2012) author presented a fuzzy programming
procedure to solve nonlinear fractional programming problems in which the parameters involved
in objective function are considered as fuzzy numbers.
The duality theory for nonlinear multiobjective optimization problems in the field of the
optimization theory has intensively developed during the last decades. In (Rodder and
Zimmermann, 1980), a generalization of maxmin and minmax problems in a fuzzy environment
is presented and thereby a pair of fuzzy dual linear programming problems is constructed. An
economic interpretation of this duality in terms of market and industry is also discussed in that
paper. In (Bector and Chandra, 2002), a pair of linear programming primal-dual problem is
introduced under fuzzy environment and appropriate results were proved to establish the duality
relationship between them. In (Liu, 1995) a constructive approach has been proposed to duality
for fuzzy multiple criteria and multiple constraint level linear programming problems. (Biswas
and Bose, 2012) gives a parametric approach for the duality in fuzzy multi criteria and multi
constraint level linear programming problem. In (Gupta and Mehlawat , 2009), a study of a pair
of fuzzy primal–dual linear programming problems has been presented and calculated duality
results using an aspiration level approach using exponential membership function, while a
discussion of fuzzy primal dual linear programming problem with fuzzy coefficients has been
presented in (Bector and Chandra, 2002; Liu, 1995). In (Bector and Chandra, 2002), a pair of
linear programming primal-dual problem is introduced under fuzzy environment and appropriate
results were proved to establish the duality relationship between them. Also in (Chakraborty et al.
2014), author has presented a pair of linear primal – dual programming using linear and
exponential membership function using fuzzy programming approach and genetic algorithm
approach.
In this paper we have extended the work done by (Gupta and Mehlawat , 2009) in which he has
solved duality in convex fractional programming using linear membership function. We have
taken the same kind of problem and proved the duality theorems considering exponential
membership function for fuzzified objective function and constraints and then solved using
LINGO and Genetic Algorithm also and the results have been analysed.
The paper has been organized as follows: In section 2, the nonlinear fractional programming
problem and its dual has been defined. The fuzzified and the crisp formulation have been defined
for the primal and dual. In section 3, the necessary duality results have been developed. The
numerical example defined in (Gupta and Mehlawat , 2009) has been illustrated in section 4.
Finally, analysis of the solution and concluding remarks has been presented in section 5.
2. NON LINEAR FRACTIONAL PROGRAMMING PROBLEM
The nonlinear fractional programming problem (Gupta and Mehlawat , 2009) can be defined as:

21
2
( )
Max ( )
t
t
c x
f x
d x
=
Subject to Ax b≤
0x ≥
(1)
The dual form of the above problem can be written as
Min ( , ) t
g u v b u=
Subject to 2
2t
A u dv cv+ ≥
, 0u v ≥
(2)
Where A is m x n matrix, x, c, b are column vectors with n components and b is a column vector
of m components.
2.1 Fuzzified Non Linear Fractional Programming Problem
Fuzzified form of nonlinear fractional programming problem can be defined as
Find
n
x R∈
Subject to
2 ~
0
( )
( )
t
t
c x
f x Z
d x
= ≥
~
Ax b≤
0x ≥ (3)
Where Z0 is the aspiration level of the objective function of the primal problem. “
~
≤” and “
~
≥”
denotes the flexibility of the objective function and the multi constraints.
Fuzzified form of the dual of the problem can be defined as
Find
m
u R∈
Subject to
~
0g(u,v) t
u b w= ≤
~
2
2t
A u dv cv+ ≥
, 0u v ≥ (4)
Where w0 is the aspiration level of the objective function of the dual problem.
2.2 Crisp Formulations using Exponential Membership Function
The advantage of nonlinear membership function over linear membership function is already
described in (Gupta and Mehlawat , 2009; Chakraborty et al. 2014). Linear membership function
is most commonly used in fuzzy linear programming problem because it is simple and it is
defined by fixing two points. Also, linear membership function has been used in a vast range of
nonlinear problems (Bector and Chandra, 2002; Gupta and Mehlawat, 2009; Chakraborty et al.
2014). But in many practical situations linear membership function is not a suitable

22
representation. Furthermore, if the membership function is interpreted as the fuzzy utility of the
decision maker, used for describing levels of indifference, preference towards uncertainty, then a
nonlinear membership function provides a better representation than a linear membership
function. Moreover, unlike linear membership function, for nonlinear membership functions the
marginal rate of increase (or decrease) of membership values as a function of model parameters is
not constant- a technique that reflects reality better than the linear case (Gupta and Mehlawat ,
2009).
Let us consider the following form of exponential membership function (Gupta and Mehlawat ,
2009; Chakraborty et al. 2014) for the fuzzy nonlinear fractional programming problem
respectively.
( )
( )
2
0
2
0
( )
/
2
0 0
c
1
c
( ) z
1-
0
t
t
t
t
c x
z p
d x
t
t
x
if z
d x
xe e
x if p z
e d x
i
α
α
α
µ
   
− − −   
−   
−
≥
−
= − < <
( )
2
0
ct
t
x
f z p
d x
 
 
 
 
 
 
 
 
 
 ≤ −
 
  
{( )/ }
1
( )
1-
0
j j j j j
j
j j
b A x q
j j j j j
j j j
if A x b
e e
x if b A x b q
e
if A x b q
α α
α
µ
− − − −
−
≤

−
= < < +

 ≥ +








j=1, 2,…….m
(5)
Where α and jα are user defined parameters which determine the shape of the membership
function. p and qj (j = 1, 2, …..m) are subjectively chosen constant of admissible violations such
that p is associated with nonlinear fractional objective function and 'jq s are associated with m
linear constraints of the problem.
Therefore using exponential membership function the crisp formulation of the problem can be
given by
Max ߣ
Subject to
( )2
0 /
1-
t
t
c x
z p
d x
e e
e
α
α
α
λ
    − − − 
   −  
−
−
≤
{( )/ }
1-
j j j j j
j
b A x q
e e
e
α α
α
λ
− − − −
−
−
≤ j=1, 2,…m
ߣ ≤ 1, ߣ ≥ 0, x ≥ 0. (6)

23
or Max ߣ
Subject to
( )
2
0log{ (1 ) } t
t
c x
p e e z
d x
α α
λ α− −
 
− + ≤ − 
 
 
log{ (1 ) } ( )j j
j j j jq e e b A x
α α
λ α
− −
− + ≤ − , j =1, 2,…m
ߣ ≤ 1, ߣ ≥ 0, x ≥ 0. (7)
The exponential membership function for the dual problem objective function and nonlinear
constraints are given by
0
0
{( ( , ))/ }
0 0
0
1 if ( , )
( ) if ( , )<
1-
0 if ( , )
w g u v r
g u v w
e e
w w g u v w r
e
g u v w r
β β
β
µ
− − − −
−
 ≤

−
= < +
≥ +


 

 
 
 
2
t 2
{( 2 )/ }
t 2
t
1 if A 2
( ) if 2 <A 2
1-
0 if A
t
j j j j j
j
j j
A u d v c v s
j j j j j
u d v c v
e e
w c v s u d v c v
e
u
β β
β
µ
− + − − −
−
+ ≥
−
= − + <
2
2j j jd v c v s
 
 
  
 
 
 + ≤ −
   j = 1, 2,…,n
(8)
Where β and 'j sβ are user defined parameters which determine the shape of the membership
function. r and sj (j = 1, 2, ……., n) are subjectively chosen constant of admissible violations
such that r is associated with objective function and 'js s are associated with n linear
constraints of the dual problem.
Min (-η)
Subject to
{ }0
( ( , ))/
1-
w g u v r
e e
e
β β
β
η
− − − −
−
−
≤
2
{( 2 )/ }
1-
t
j j j j j
j
A u d v c v s
e e
e
β β
β
η
− + − − −
−
−
≤ j=1, 2,…n
η ≤ 1, η ≥ 0, u,v ≥ 0. (9)
or Min (-η)
Subject to ( )0
log{ (1 ) } ( , )r e e w g u vβ β
η β− −
− + ≤ −

24
2
log{ (1 ) } ( 2 )j j t
j j j j js e e A u d v c v
β β
η β
− −
− + ≤ + − , j=1, 2,…n
η ≤ 1, η ≥ 0, u,v ≥ 0.
(10)
3. THEOREMS ON DUALITY
We shall prove some duality theorems. In the theorems, Q= (q1, q2, ……,qm), S= (s1, s2, ……., sn )
are column vectors
Theorem 3.1 (Modified fuzzy weak duality theorem): Let (x, ߣ) be the feasible solution of crisp
primal problem (7) and (u, v, η ) be feasible solution of crisp dual problem (10). Then
2~
1 1
log{ (1 ) } log{ (1 ) } ( )j ji i tm n
t t t
t
i ji j
e e e e c x
u Q S x u b
d x
β βα α
λ η
α β
− −− −
= =
− + − +
+ ≤ −∑ ∑
Proof: Since (x, ߣ) be the feasible solution of crisp primal problem (7) and (w, η ) be feasible
solution of crisp dual problem (10), then
log{ (1 ) } ( )i i
i i i iq e e b A xα α
λ α− −
− + ≤ − , i=1, 2,…m
2
log{ (1 ) } ( 2 )j j t
β β
η β
− −
− + ≤ + − , j=1, 2,…n
Or
1
log{ (1 ) }i im
i i
e e
Q b Ax
α α
λ
α
− −
=
− +
≤ −∑ (11)
2
1
log{ (1 ) }
2
j jn
t
j j
e e
S A u dv cv
β β
η
β
− −
=
− +
≤ + −∑ (12)
Multiplying the equation (11) by transpose of u and taking transpose of equation (12) and
multiplying equation by x, we have
1
log{ (1 ) }i im
t t t
i i
e e
u Q u b u Ax
α α
λ
α
− −
=
− +
≤ −∑
2
1
log{ (1 ) }
2
j jn
t t t t t t
j j
e e
S x u Ax v d x v c x
β β
η
β
− −
=
− +
≤ + −∑
Adding the above two inequalities, we have
2
1 1
log{ (1 ) } log{ (1 ) }
2
j ji im n
t t t t t t t
i ji j
e e e e
u Q S x u b v d x v c x
β βα α
λ η
α β
− −− −
= =
− + − +
+ ≤ + −∑ ∑
Or 1
2
1
2
2
2
1 1
log{ (1 ) } log{ (1 ) } ( ) ( )
( )
( )
j ji i t tm n
t t t t t
tt
i ji j
e e e e c x c x
u Q S x u b d x v
d xd x
β βα α
λ η
α β
− −− −
= =
 − + − +
+ ≤ + − − 
 
∑ ∑

25
1
2
1
2
2
2
1 1
log{ (1 ) } log{ (1 ) } ( ) ( )
( )
( )
j ji i t tm n
t t t t
t t
i ji j
e e e e c x c x
u Q S x u b d x v
d x d x
β βα α
λ η
α β
− −− −
= =
  − + − +
+ ≤ − + −   
   
∑ ∑
2~
1 1
log{ (1 ) } log{ (1 ) } ( )j ji i tm n
t t t
t
i ji j
e e e e c x
u Q S x u b
d x
β βα α
λ η
α β
− −− −
= =
 − + − +
+ ≤ − 
 
∑ ∑
Remark: When ߣ=1, η =1 in the above inequality we get the objective function of primal
problem is imprecisely less than the objective function of the dual
2~ ( )t
t
t
c x
u b
d x
 
≤ 
 
which can be
defined as the standard fuzzified weak duality theorem in a multi objective linear programming
problem.
Theorem 3.2: Let (
^ ^
,x λ ) be feasible solution of crisp primal problem (7) and (
^ ^
,w η ) be feasible
solution of crisp dual problem (10) such that
(i)
^ ^ ^
2^ ^^
^
1 1
log{ (1 ) } log{ (1 ) } ( )j ji i tm n
T T t
ti ji j
e e e e c x
w Q S x u b
d x
β βα α
λ η
α β
− −− −
= =
− + − +
+ = −∑ ∑
(ii)
^ ^ ^
2 ^
0 0^
log{ (1 ) } log{ (1 ) } ( )
( )
t
t
t
e e e e c x
p r u b w z
d x
α α β β
λ η
α β
− − − −  
− + − +  + = − + −
 
 
(iii) The aspiration levels z0 and w0 satisfy 0 0( ) 0z w− ≤
Then (
^ ^
,x λ ) is optimal solution of primal problem and (
^ ^
,w η ) be optimal solution of dual
problem.
Proof: Let (x, ߣ) be feasible solution of crisp primal problem (7) and (w, η ) be feasible solution
of crisp dual problem (10). Then from Theorem 3 we have,
( )
2
~
1 1
log{ (1 ) } log{ (1 ) }
0
j ji i
tm n
t t t
t
i ji j
c xe e e e
u Q S x u b
d x
β βα α
λ η
α β
− −− −
= =
 
− + − +  + − − ≤
 
 
∑ ∑
(13)
From condition (i) and Equation (13) we have
( )
2
log{ (1 ) } log{ (1 ) }
1 1
tc xm ni i j je e e et t tu Q S x u b
ti j d xi j
α α β βλ η
α β
 
 − − − −− + − +  ∑ + − −∑
 = =
 
 
2^
^ ^
^ ^~ ^
^
1 1
log{ (1 ) } log{ (1 ) }j ji i
t
m n
t t t
ti ji j
c x
e e e e
u Q S x u b
d x
β βα α
λ η
α β
− −− −
= =
  
  − + − +   ≤ + − −
 
 
 
∑ ∑

26
Thus,
^ ^ ^ ^
( , , , )x wλ η is optimal to the problem whose maximum objective value is 0.
Max
( )
2
log{ (1 ) } log{ (1 ) }
1 1
ti j d xi j
α α β βλ η
α β
 
 − − − −− + − +  ∑ + − −∑
 = =
 
 
Subject to
( )
2
0log{ (1 ) }
t
t
c x
p e e Z
d x
α α
λ α− −
 
 − + ≤ −
 
 
log{ (1 ) } ( )i i
i i i iq e e b A xα α
λ α− −
− + ≤ −
(i = 1, 2, ……, m)
0log{ (1 ) } ( )t
r e e w u bβ β
η β− −
− + ≤ −
2
log{ (1 ) } ( 2 )j j t
β β
η β
− −
− + ≤ + −
(j = 1, 2, …….., n )
ߣ ≤ 1, ߣ ≥ 0, x ≥ 0, 1, , 0, 0u vη η≤ ≥ ≥ .
Adding condition (i) and (ii), we have
( )
2
log{ (1 ) } log{ (1 ) }
1 1
ti j d xi j
α α β βλ η
α β
 
 − − − −− + − +  ∑ + − −∑
 = =
 
 
+
^ ^ ^
2 ^
0 0^
log{ (1 ) } log{ (1 ) } ( )
( ) 0
t
t
t
e e e e c x
p r u b w z
d x
α α β β
λ η
α β
− − − −  
− + − +  + − − − − =
 
 
Or
log{ (1 ) } log{ (1 ) }
1 1
m ni i j je e e et tu Q S x
i ji j
α α β βλ η
α β
− − − −− + − +
∑ + ∑
= =
+
^ ^
0 0
log{ (1 ) } log{ (1 ) }
( ) 0
e e e e
p r z w
α α β β
λ η
α β
− − − −
− + − +
+ + − =
Each term in the above sum is non positive as
^ ^
, 1.λ η ≤
Therefore,
^
^
1
log{ (1 ) }i im
t
i i
e e
u Q
α α
λ
α
− −
=
− +
∑ = 0
^
x t ^
1
log{ (1 ) }
0
j jn
t
j j
e e
S x
β β
η
β
− −
=
− +
=∑
^
log{ (1 ) }
0
e e
p
α α
λ
α
− −
− +
=

27
^
log{ (1 ) }
0
e e
r
β β
η
β
− −
− +
=
0 0z w− =0
Since
log{ (1 ) }
0
e e
p
α α
λ
α
− −
− +
≤ and
log{ (1 ) }
0,
e e
r
β β
η
β
− −
− +
≤ because
^ ^
, 1λ η ≤ , we get
^
log{ (1 ) } log{ (1 ) }e e e e
p p
α α α α
λ λ
α α
− − − −
− + − +
≤
^
log{ (1 ) } log{ (1 ) }e e e e
r r
β β β β
η η
β β
− − − −
− + − +
≤
or
^
λ λ≤ and
^
η η≤
Thus, (
^ ^
,x λ ) is optimal solution of (7) and (
^ ^
,w η ) be optimal solution of (10).
4. NUMERICAL EXAMPLE
Let us consider the same numerical example which is defined in [20]. The primal dual nonlinear
fractional programming problem can be defined as
Min
2
1 2
1 2
(2 )
( )
2
x x
f x
x x
+
=
+
Subject to 1 22 6x x+ ≥
1 23 8x x+ ≥
1 2, 0x x ≥
Max 1 2( , ) 6 8g u v u u= +
Subject to 2
1 22 4 0u u v v+ + − ≤
2
1 23 2 2 0u u v v+ + − ≤
1 2, , 0u u v ≥
The fuzzified nonlinear fractional programming problem can be written as:
Find n
x R∈
Min
Subject to
2 ~
1 2
1 2
(2 )
( ) 1
2
x x
f x
x x
+
= ≤
+
~
1 22 6x x+ ≥

28
~
1 23 8x x+ ≥
1 2, 0x x ≥ (14)
Max
~
1 2( , ) 6 8 1g u v u u= + ≥
Subject to
~
2
1 22 4 0u u v v+ + − ≤
~
2
1 23 2 2 0u u v v+ + − ≤
1 2, , 0u u v ≥ (15)
Let us take 1 2 1 22, 1, 1, 2, 2, 1p q q α α α= = = = = = for primal problem. Using exponential
membership function, the crisp model of (14) can be written as:
Min λ−
Subject to 2 2
1 2 1 2 1 24 4 ( 2 )log(0.865 0.1353)x x x x x x λ+ + ≥ + +
1 24 2 12 log(0.865 0.1353)x x λ+ − ≥ +
1 23 8 log(0.865 0.1353)x x λ+ − ≥ +
ߣ ≤ 1, ߣ ≥ 0
1 2, 0x x ≥ . (16)
Let us take 1 2 1 21, 1, 1, 2, 2, 1r s s β β β= = = = = = for dual problem (15). Using exponential
membership function, the crisp model can be written as:
Max η
Subject to 1 2log(0.865 0.1353) 12 16 2u uη + ≤ + −
2
1 2log(0.865 0.1353) 4 2 2 8u u v vη + ≤ − − − +
2
1 2log(0.865 0.1353) 3 2 2u u v vη + ≤ − − − +
0 1η≤ ≤ , 1 2, , 0u u v ≥ . (17)
5. SOLUTION AND ANALYSIS
Using LINGO software, the solution of the primal problem (16) is given by
Local optimal solution found.
Objective value: -1.000000
Infeasibilities: 0.000000
Extended solver steps: 5
Total solver iterations: 25
Model Class: NLP
Total variables: 3
Nonlinear variables: 3
Integer variables: 0

29
Total constraints: 5
Nonlinear constraints: 3
Total nonzeros: 11
Nonlinear nonzeros: 5
Variable Value
LAMBDA 1.000000
X1 1.234568
X2 0.1000000E+08
Minimum value of objective function is -1.
The solution of the dual problem (17) is given by
Local optimal solution found.
Objective value: 1.000000
Infeasibilities: 0.000000
Extended solver steps: 5
Total solver iterations: 37
Model Class: NLP
Total variables: 4
Nonlinear variables: 2
Integer variables: 0
Total constraints: 5
Nonlinear constraints: 3
Total nonzeros: 13
Nonlinear nonzeros: 5
Variable Value
ETA 1.000000
U1 0.1037300
U2 0.4665636E-01
V 0.1359680
Maximum value of the objective function is 1.
Genetic Algorithm gives global optimal solution. The NSGA used here is a real parameter GA
that works directly with the parameter values. NSGA (Nondominated Sorting Genetic Algorithm)
uses nitching, as well as nondominated sorting of the solutions in every generation to ensure that
the “good solutions get preference in selection for procreation”. The non-dominated sorting GA
uses a ranking selection method to emphasize good solutions and then a nitche building procedure
to maintain a stable sub population of good solutions. Since multi objective GAs can find
multiple pareto optimal solutions in one single run, the proposed technique is capable of finding
multiple solutions to the problems.
The parameters used to solve the problem (16) in Genetic Algorithm are as follows:
Number of objective functions : 1

30
Objective function # 1 : Minimize
CROSSOVER TYPE : Binary GA (Single-pt)
STRATEGY : 1 cross - site with swapping
Population size : 100
Total no. of generations : 100
Cross over probability : 0.9000
Mutation probability : 0.1000
String length : 60
Number of variables
Binary : 3
Integer : 0
Enumerated : 0
Continuous : 0
TOTAL : 3
Epsilon for closeness : 0
Sigma-share value : 0.2320
Sharing Strategy : sharing on Parameter Space
Lower and Upper bounds :
0.0000 <= ߣ <= 1.0000
0.0000 <= x1 <= 2.0000
0.0000 <= x2 <= 2.0000
Table 1 gives a set of solution of problem (16).
Table 1: Primal Solution
ߣߣߣߣ x1 x2
0.942 1.952 1.946
0.992 1.673 1.976
0.962 1.794 1.974
0.942 1.884 1.916
0.977 1.514 1.922
The parameters used to solve the problem (17) in Genetic Algorithm are as follows:
Number of objective functions : 1
Objective function # 1 : Maximize
CROSSOVER TYPE : Binary GA (Single-pt)
STRATEGY : 1 cross - site with swapping
Population size : 100
Total no. of generations : 100
Cross over probability : 0.9000
Mutation probability : 0.1000
String length : 40
Number of variables
Binary : 4

31
Integer : 0
Enumerated : 0
Continuous : 0
TOTAL : 4
Epsilon for closeness : 0
Sigma-share value : 0.2810
Sharing Strategy : sharing on Parameter Space
Lower and Upper bounds :
0.0000 <= u1 <= 2.0000
0.0000 <= u2 <= 2.0000
0.0000 <= v <= 2.0000
0.0000 <= η <= 1.0000
Using Genetic algorithm, a set of solutions of dual problem (17) is given in the form of
table in Table 2.
Table 2: Dual Solution
Η u1 u2 V
0.956 1.863 1.961 1.844
0.956 1.867 1.607 0.956
1.000 1.836 0.741 1.273
0.979 0.716 1.118 1.994
0.953 1.935 1.075 1.894
From the above table, we can clearly see that the solution obtained satisfies the fuzzy constraints
as well as the fuzzy objective in both primal and dual problem. Also, it has been observed that
difference between the tolerance limit and aspiration level is less. In other words, the decision
makers have been provided with enough flexibility to choose satisfying solutions that maximize
or minimize their utility functions. Genetic Algorithm gives multiple pareto optimal solutions in
one single run. Therefore, it provides set of solutions to the decision maker. The following graph
clearly shows comparison between the primal and dual objective function of different sets of
solutions.

32
6. CONCLUSIONS
In this paper, an approach has been presented for a specific kind of fuzzy nonlinear fractional
programming problem by constructing a pair of fuzzy non linear fractional primal and dual
problems. Crisp form of the above primal and dual problems has been obtained by using
exponential membership function. Duality results have been established to prove the duality
relationship between above primal and dual problem and are illustrated by an example. The
duality results fully satisfy the aspiration levels or the tolerance levels of the objective functions
and the system constraints made by the decision maker. The difference between the achieved
level and the allowable limit of the satisfying solutions of the decision maker is very less. The
numerical example has also been solved by Genetic Algorithm Approach. Genetic Algorithm
gives multiple pareto optimal solution. The decision maker can choose any optimal solution
according to the convenience. The results of the present paper encourage us to apply duality
results in variety number of fields of optimization problem.
REFERENCES
[1] Bellman, R. E., Zadeh, L. A., (1970) Decision Making in a Fuzzy Environment, Management
Science, 17, 141-164.
[2] Bector, C. R. and Chandra, S., (2002) On duality in linear programming under fuzzy environment,
Fuzzy Sets and Systems, 125, 317-325.
[3] Biswas, A. and Bose, K., (2012), Application of Fuzzy Programming Method for Solving Nonlinear
Fractional Programming Problems with Fuzzy Parameters, Mathematical Modelling and Scientific
Computation Communications in Computer and Information Science 283 104-113.
[4] Borza, M., Rambely, A. S. and Saraj, M., (2012) Solving linear fractional programming problems
with interval coefficients in the objective function. A New Approach, Applied Mathematical
Sciences, 6, 69, 3443 – 3452.
[5] Chakraborty, M., Gupta, S., (2002) Fuzzy Mathematical Programming for Multi-objective Linear
Fractional Programming Problem, Fuzzy Sets and Systems, 125, 335-342.
0.942
0.992
0.962
0.942
0.977
0.956 0.956
1
0.979
0.953
0.91
0.92
0.93
0.94
0.95
0.96
0.97
0.98
0.99
1
1.01
1 2 3 4 5
λ/µ
Solution Number
Figure 1: Primal and Dual objective function of different solution
using GA
Primal
Dual

33
[6] Chakraborty, A., Tiwari, S. P., Chattopadhyay, A. and Chatterjee, K. (2014), Duality in Fuzzy Multi
objective linear programming with multi constraint, International Journal of Mathematics in
Operations Research, Vol. 6, No. 3, pp. 297-315.
[7] Gupta, P. and Mehlawat, M. K., (2009) Duality for a convex fractional programming under fuzzy
environment, International Journal of Optimization: Theory Methods and Applications, 1, 3, 291-301.
[8] Gupta, P. and Mehlawat, M. K., (2009), Bector- Chandra type duality in fuzzy linear programming
with exponential membership functions, Fuzzy Sets and Systems, 160, 22, 3290-3308.
[9] Jameel, A. F., Sadhegi, A., (2012), Solving Nonlinear Programming Problem in Fuzzy Environment,
Int. J. Contemp. Math. Sciences 7, 4, 159 – 170.
[10] Jimenez, F., Sanchez, G., Cadenas, J. M., Gomez-Skarmeta, A. F., Verdegay, J. L., (2005),
Computational Intelligence, Theory and Applications, Advances in Soft Computing, 33, 713-722.
[11] Kanaya, Z. A. (2010) An Interactive Method for Fuzzy Multiobjective Nonlinear Programming
Problems, JKAU: Sci. 22 1 103-112; DOI: 10.4197 / Sci. 22-1.8.
[12] Lasdon, L. S, Waren, A. D., (1980), Feature Article: Survey of Nonlinear Programming Applications,
Operations Research 28(5): 1029-1073.
[13] Liu, Y. J., Shi, Y. and Liu, Y. H. (1995) Duality of fuzzy MC2 linear programming: a constructive
approach, J. Math. Anal. Appl., 194, 389-413.
[14] Rodder, W. and Zimmermann, H. J. (1980), Duality in fuzzy linear programming. In: (A. V. Fiacco,
K. O. Kortane K Eds.). External methods and System Analysis. Berlin, New York, 415-429.
[15] Saad, O. M., M. S. Biltagy and T. B. Farag, An algorithm for Multi objective Integer nonlinear
fractional programming problem under fuzziness, Gen. Math. Notes, 2 1 (2011) 1-17.
[16] Schaible, S., (1981) Fractional programming: applications and algorithms, European Journal of
Operational Research, 7, 2, 111-120.
[17] Schaible, S., (1995), Fractional programming, in R. Horst and P.M. Pardalos (eds.), Handbook of
Global Optimization, Kluwer Academic Publishers, Dordrecht-Boston-London, 495-608.
[18] Schaible, S. and Ibaraki, T., (1983), Fractional programming, European Journal of Operational
Research 12, 3, 325-338.
[19] Stancu-Minasian, I. M., Pop, B., (1978) On a Fuzzy Set Approach to Solving Multiple Objective
Linear Fractional Programming Problem, Fuzzy Sets and Systems, 134, 397-405.
[20] Zimmermann, H. J., (1978) Fuzzy programming and linear programming with several objective
functions, Fuzzy Sets and Systems, 1, 45-55.
Authors
Ananya Chakraborty is Assistant Professor in the Department of Mathematics of Vemana
Institute of Technology, Bangalore, India. She has done her Ph.D from Indian School of
Mines, Dhanbad, Jharkhand, India. She is having almost 10 years of research experience. She
has published many papers in national/ international journals. She has presented many papers
in different conferences. Her current research interest includes Fuzzy programming, Operations Research.
She is also a member of International association of computer science and information technology

Duality in nonlinear fractional programming problem using fuzzy programming and genetic algorithm

Recommended

More Related Content

What's hot (16)

Similar to Duality in nonlinear fractional programming problem using fuzzy programming and genetic algorithm (20)

Recently uploaded (20)

Duality in nonlinear fractional programming problem using fuzzy programming and genetic algorithm