Received November 7, 2018, accepted December 31, 2018, date of publication January 14, 2019, date of current version

February 4, 2019.
Digital Object Identifier 10.1109/ACCESS.2019.2892013

Mandelbrot and Julia Sets via Jungck–CR

Iteration With s–Convexity
YOUNG CHEL KWUN1 , MUHAMMAD TANVEER2 , WAQAS NAZEER 3, KRZYSZTOF GDAWIEC 4,

AND SHIN MIN KANG5,6

1 Department of Mathematics, Dong-A University, Busan 49315, South Korea
2 Department of Mathematics and Statistics, The University of Lahore, Lahore 54000, Pakistan
3 Division of Science and Technology, University of Education, Lahore 54770, Pakistan
4 Institute of Computer Science, University of Silesia, 41-200 Sosnowiec, Poland
5 Department of Mathematics and Research Institute of Natural Science, Gyeongsang National University, Jinju 52828, South Korea
6 Center for General Education, China Medical University, Taichung 40402, Taiwan

Corresponding authors: Waqas Nazeer ([email protected]) and Shin Min Kang ([email protected])
This work was supported by Dong-a University funds, Busan, South Korea.

ABSTRACT In today’s world, fractals play an important role in many fields, e.g., image compression or
encryption, biology, physics, and so on. One of the earliest studied fractal types was the Mandelbrot and
Julia sets. These fractals have been generalized in many different ways. One of such generalizations is the
use of various iteration processes from the fixed point theory. In this paper, we study the use of Jungck-CR
iteration process, extended further by the use of s-convex combination. The Jungck-CR iteration process
with s-convexity is an implicit three-step feedback iteration process. We prove new escape criteria for the
generation of Mandelbrot and Julia sets through the proposed iteration process. Moreover, we present some
graphical examples obtained by the use of escape time algorithm and the derived criteria.

INDEX TERMS Julia set, Jungck-CR iteration, Mandelbrot set, s-convex combination.

I. INTRODUCTION In 2004, Rani and Kumar [1], [2] introduced superior

In the 1970’s Benoit Mandelbrot introduced to the world new Julia and Mandelbrot sets using Mann iteration scheme.
field of mathematics. He named this field fractal geometry Chauhan et al. [3] introduced the relative superior Julia sets
(fractus -- from Latin divided, fractional). Benoit Mandel- using Ishikawa iteration scheme. Also, relative superior Julia
bro extended the work of Gaston Julia and introduced the sets, Mandelbrot sets and tricorn, multicorns by using the
Mandelbrot set; a set of all connected Julia sets. Mandel- S-iteration scheme were presented in [4] and [5]. Recently,
brot expanded the ideas of G. Julia, studied the Mandelbrot Ashish et al. [6] introduced Julia and Mandelbrot sets using
set by using the complex function z2 + c with using z as the Noor iteration scheme, which is a three-step itera-
a complex function and c as a complex parameter. Fractal tive procedure. The junction of a s-convex combination [7]
geometry breaks the way we see everything. It provides a and various iteration schemes was studied in many papers.
new idea of modelling natural objects, such as clouds, plants, Mishra et al. [8], [9] developed fixed point results in rel-
landscapes, galaxies. One of the fractal types studied by ative superior Julia sets, tricorn and multicorns by using
Mandelbrot were complex fractals, i.e, fractals defined in the Ishikawa iteration with s-convexity. Kang et al. [10]
the complex plane. Mandelbrot and Julia sets are examples introduced new fixed point results for fractal generation
of those fractals. The fractal structure of Mandelbrot and using the implicit Jungck-Noor orbit with s-convexity,
Julia sets have been demonstrated for quadratic, cubic and whereas Nazeer et al. [11] used the Jungck-Mann and
higher degree polynomials, by using Picard orbit which is an Jungck-Ishikawa iterations with s-convexity. The use of
application of one-step feedback process. Since then many Noor iteration and s-convexity was shown in [12], whereas
different generalizations of those sets were proposed. One Gdawiec and Shahid [13] presented complex fractals gener-
of the generalizations is the use of results from fixed point ated by the S-iteration with s-convexity.
theory, namely the use of various iteration processes instead In this paper we study the use of Jungck-CR iteration
of the Picard one that is used in the generation of Mandelbrot with s-convexity in the generation of Mandelbrot and Julia
and Julia sets. sets. We prove an escape criterion for the function of the

This work is licensed under a Creative Commons Attribution 3.0 License. For more information, see http://creativecommons.org/licenses/by/3.0/
VOLUME 7, 2019 12167
 Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

form zn − az + c. Moreover, we present some graphical a very important role in the generation of Mandelbrot and
examples of Mandelbrot and Julia sets via the Jungck-CR Julia sets.
iteration with s-convexity. The escape time algorithms for the generation of Man-
The rest of the paper is organized as follows. In Sec. II, delbrot and Julia sets are presented in Algorithm 1 and 2,
we briefly introduce notions used in the paper. Next, respectively.
in Sec. III, we extend the Jungck-CR iteration using the
s-convex combination and we prove the escape criteria for the Algorithm 1 Mandelbrot Set Generation
functions of the form Qc (z) = zn −az+c using the Jungck-CR
Input: Qc : C → C – polynomial function, A ⊂ C –
iteration with s-convexity. In Sec. IV we present some graph-
area, K – maximum number of iterations,
ical examples of Mandelbrot and Julia sets obtained with the
colourmap[0..C − 1] – colourmap with C
derived criteria. Finally, in Sec. V, we give some concluding
colours.
remarks.
Output: Mandelbrot set for the area A.
II. PRELIMINARIES 1 for c ∈ A do
Definition 1 (Julia Set [14]): Let f : C → C be a polyno- 2 R = calculate the escape threshold
mial of degree ≥ 2. Let Ff be the set of points in C whose 3 n=0
orbits do not converge to the point at infinity, i.e., 4 z0 = critical point of Qc
5 while n ≤ K do
Ff = {z ∈ C : {f n (z)}∞
 
n=0 is bounded}. (1) 6 zn+1 = Qc (zn )
7 if |zn+1 | > R then
Ff is called as filled Julia set of the polynomial f . The
8 break
boundary points of Ff are called the points of Julia set of the
polynomial f or simply the Julia set. 9 n=n+1
Definition 2 (Mandelbrot Set [15]): The Mandelbrot set 10 i = b(C − 1) Kn c
M consists of all parameters c for which the filled Julia set 11 colour c with colourmap[i]
of Qc (z) = z2 + c is connected, i.e.,
M = {c ∈ C : FQc is connected}. (2)
The Mandelbrot set M for the quadratic function Qc (z) =
z2 + c can be equivalently defined in the following way [16]: Algorithm 2 Julia Set Generation
Input: Qc : C → C – polynomial function, c ∈ C –
M = {c ∈ C : {Qnc (0)} does not tend to ∞ as n → ∞}, (3)
parameter, A ⊂ C – area, K – maximum number
We choose the initial point 0, because 0 is the only critical of iterations, colourmap[0..C − 1] – colourmap
point of Qc , i.e., Q0c (0) = 0. with C colours.
Definition 3 (Multicorn): Let Ac (z) = zm + c, where Output: Julia set for the area A.
c ∈ C. The multicorn M∗ for Ac is defined as the collection 1 R = calculate the escape threshold
of all c ∈ C for which the orbit of 0 under the action of Ac is 2 for z0 ∈ A do
bounded, i.e., 3 n=0
∗
M = {c ∈ C : |Anc (0)| 6 → ∞ as n → ∞} (4) 4 while n ≤ K do
5 zn+1 = Qc (zn )
Multicorn for m = 2 is called the tricorn. 6 if |zn+1 | > R then
To generate visualizations of Mandelbrot and Julia sets we 7 break
can use many different algorithms [17], [18], e.g., distance 8 n=n+1
estimator, potential function, escape time etc. In this paper
we use the escape time algorithm. The algorithm is based on 9 i = b(C − 1) Kn c
the number of iterations necessary to determine whether the 10 colour z0 with colourmap[i]
orbit sequence tends to infinity or not. To determine whether
the orbit escapes or not we use the escape criterion. For
instance, for the classical Mandelbrot and Julia sets, i.e., the Definition 4 (Picard Iteration [15]): Let X be a nonempty
sets defined by Qc (z) = z2 + c, the escape criterion is the set and f : X → X . For any point x0 ∈ X , the Picard’s
following: if there exists k ≥ 0 such that iteration is defined in the following way

|Qkc (z)| > max{|c|, 2}, (5) xk+1 = f (xk ), (6)

then Qnc (z) → ∞ as n → ∞. where k = 0, 1, . . ..

We call the right side of (5) the escape threshold. This Definition 5 (Jungck iteration [19]): Let S, T : X → X
threshold can be different for different functions Qc and plays be the two maps such that S is injective. For any x0 ∈ X the

12168 VOLUME 7, 2019

Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

Jungck iteration is defined in the following way Let us start with the definition of the Jungck-CR iteration.
Definition 9 (Jungck-CR iteration [19]): Let S, T : C →
S(xk+1 ) = T (xk ), (7) C be mappings, where S is injective, and let x0 ∈ C be a
where k = 0, 1, . . .. starting point. The Jungck-CR iteration is defined as follows:
Definition 6 (Jungck-Mann Iteration With s-Convexity

S(xk+1 ) = (1 − α)S(yk ) + αT (yk )

[11]): Let S, T : C → C be the two complex maps such
S(yk ) = (1 − β)T (xk ) + βT (uk ), (11)
that T is a complex polynomial of degree greater than 1 and
S(uk ) = (1 − γ )S(xk ) + γ T (xk ),


S is injective. For any x0 ∈ C the Jungck-Mann iteration with
s-convexity is defined in the following way where α ∈ (0, 1], β, γ ∈ [0, 1] and k = 0, 1, 2, . . ..
S(xk+1 ) = (1 − α) S(xk ) + α T (xk ),
s s
(8) In each of the three steps of the Jungck-CR iteration we
use a convex combination of two elements. In the literature
where α, s ∈ (0, 1], k = 0, 1, 2, . . .. we can find some generalizations of the convex combination.
Let us notice that for α = 1 Jungck-Mann iteration with One of such generalizations is the s-convex combination.
s-convexity reduces to the Jungck iteration. Definition 10 (s-Convex Combination [7]): Let z1 , z2 ,
Definition 7 (Jungck-Ishikawa Iteration With s-Convexity . . . , zn ∈ C and s ∈ (0, 1]. The s-convex combination is
[11]): Let S, T : C → C be the two complex maps such that defined in the following way:
T is a complex polynomial of degree greater than 1 and S is
λs1 z1 + λs2 z2 + . . . + λsn zn , (12)
injective. For any x0 ∈ C the Jungck-Ishikawa iteration with
Pn
s-convexity is defined in the following way where λk ≥ 0 for k ∈ {1, 2, . . . , n} and k=1 λk = 1.
( Let us notice that the s-convex combination for s = 1
S(xk+1 ) = (1 − α)s S(xk ) + α s T (yk ),
(9) reduces to the standard convex combination. This type of
S(yk ) = (1 − β)s S(xk ) + β s T (xk ), combination was successfully used in the generation of Man-
where α, s ∈ (0, 1], β ∈ [0, 1] and k = 0, 1, 2, . . .. delbrot and Julia sets [10], [11]. Moreover, it was also used
Let us notice that Jungck-Ishikawa iteration with in the generation of other type of fractals generated in the
s-convexity reduces to the: complex plane, namely in the methods that use root finding
of complex polynomials [20].
Jungck-Mann
Now, we will replace the convex combination in the
• iteration with s-convexity for β = 0,
Jungck-CR iteration with the s-convex one.
• Jungck iteration for β = 0 and α = 1.
Definition 11 (Jungck-CR Iteration With s-Convexity): Let
Definition 8 (Jungck-Noor Iteration With s-Convexity S, T : C → C be mappings, where S is injective, and let
[10]): Let S, T : C → C be the two complex maps such x0 ∈ C be a starting point. The Jungck-CR iteration with
that T is a complex polynomial of degree greater than 1 and s-convexity is defined as follows:
S is injective. For any x0 ∈ C the Jungck-Noor iteration with 
S(xk+1 ) = (1 − α) S(yk ) + α T (yk ),
s s
s-convexity is defined in the following way 
 S(yk ) = (1 − β) T (xk ) + β T (uk ),
s s (13)
S(xk+1 ) = (1 − α) S(xk ) + α T (yk ),
s s
S(uk ) = (1 − γ ) S(xk ) + γ T (xk ),
 
 s s
S(yk ) = (1 − β)s S(xk ) + β s T (uk ), (10)


S(uk ) = (1 − γ ) S(xk ) + γ T (xk ),
s s where α, s ∈ (0, 1], β, γ ∈ [0, 1] and k = 0, 1, 2, . . ..
Let us notice that the Jungck-CR iteration with
where α, s ∈ (0, 1], β, γ ∈ [0, 1] and k = 0, 1, 2, . . .. s-convexity does not reduce to any of the iterations: Picard,
Let us notice that Jungck-Noor iteration with s-convexity Jungck-Mann with s-convexity, Jungck-Ishikawa with s-
reduces to the: convexity, Jungck-Noor with s-convexity. Thus, using this
Jungck-Ishikawa iteration we will create completely new orbits and in con-
• iteration with s-convexity for γ = 0, sequence new fractal sets.
Jungck-Mann In Picard iteration we use only one mapping and in the
• iteration with s-convexity for γ = β = 0, Jungck-CR iteration with s-convexity we have two map-
• Jungck iteration for γ = β = 0 and α = 1. pings. Thus, if we want to replace Picard iteration with the
Jungck-CR iteration with s-convexity, then we need to handle
III. MAIN RESULT the case of different number of mappings in the iterations.
Let us notice that the Picard iteration is the iteration used in We handle this in a following way. Let Qc : C → C be a
the generation of complex fractals. In the literature we can polynomial function. We decompose Qc into two mappings S,
find results of replacing Picard iteration with other iterations, T in such a way that Qc = T −S and S is injective. In the case
e.g., with the ones presented in Sec. II [10], [11]. In this of multicorns we decompose Q∗c (z) = Qc (z) in a following
section we show how to extend the Jungck-CR iteration way: Q∗c = T − S, where T = Ac and S is injective.
using the s-convex iteration, and next the use of the extended Of course this type of decomposition restricts the choice
iteration in the generation of Mandelbrot and Julia sets. of the polynomial functions that can be used. Having the

VOLUME 7, 2019 12169

 Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

decomposition we also need to derive a new escape criterion Thus

for the mappings and (13). s2 βγ |z|
 
In the following subsections we prove escape criteria for |y0 | ≥ |z| −1 .
some classes of polynomials. 1 + |a|
In the third step of the iteration we have
A. ESCAPE CRITERION FOR THE QUADRATIC COMPLEX
|S(z1 )| = (1 − α)s S(y0 ) + α s T (y0 )
 
POLYNOMIAL  
Let Qc (z) = z2 − az + c, where a, c ∈ C. We decompose Qc ≥ (1 − sα)ay0 + sα(y20 + c)
 
in the following way: T (z) = z2 + c and S(z) = az.
Theorem 1: Assume that |z| ≥ |c| > 2(1+|a|) ≥ sα|y20 + c| − (1 − sα)|ay0 |.
sα , |z| ≥ |c| >
2(1+|a|) 2(1+|a|)
, |z| ≥ |c| > where α, β, γ , s ∈ (0, 1] and βγ |z|
 2 
sβ sγ Since |y0 | ≥ |z| s1+|a| − 1 , which implies |y20 | ≥
define {zk }k∈N as follows
 s2 βγ |z2 |. We get
S(zk+1 ) = (1 − α) S(yk ) + α T (yk )
 s s  
|az1 | ≥ s3 αβγ z2  − sα|c| − |a||z|
 
S(yk ) = (1 − β)s T (zk ) + β s T (uk ), (14)
S(uk ) = (1 − γ )s S(zk ) + γ s T (zk ) k = 0, 1, 2, . . . ,
  
≥ s3 αβγ z2  − |z| − |a||z|
  

where z0 = z. Then |zk | → ∞ as k → ∞.

 
= s3 αβγ z2  − (1 + |a|) |z|
 
Proof: Because T (z) = z2 + c, S(z) = az and z0 = z we
have = |z| (s3 αβγ |z| − (1 + |a|)).
|S(u0 )| = (1 − γ )s S(z) + γ s T (z)
 
Hence
 
= (1 − γ )s az + (1 − (1 − γ ))s (z2 + c) s3 αβγ |z|
   
|z1 | ≥ |z| −1 .
1 + |a|
Using binomials series up to linear terms of γ and 1 − γ ,
and condition s < 1 we get Since |z| > 2(1+|a|) 2(1+|a|)
|z| > 2(1+|a|)
sα , sβ , |z| > sγ , so
2(1+|a|) s αβγ |z|
3
|au0 | ≥ (1 − s(1 − γ ))|z2 + c| − (1 − sγ )|az| |z| > s3 αβγ and in consequence 1+|a| − 1 > 1. Therefore
3 αβγ |z|
≥ |(s − s(1 − γ ))(z2 + c) |−|(1 − sγ )az| . there exist λ > 0 such that s 1+|a| −1 > 1+λ. Consequently
|z1 | > (1+λ)|z|. In particular |z1 | > |z|. So we may apply the
Because |z| ≥ |c| , |a| ≥ 1 and sγ < 1 we obtain
  same argument repeatedly to find |zk | > (1 + λ)k |z|. Thus,
|au0 | ≥ sγ z2 | − sγ |c − (1 − sγ )|az| the orbit of z tends to infinity and this completes the proof. 
 
  Corollary 1: Suppose that
= sγ z2 | − sγ |c − |az| + sγ |az|
 
2(1 + |a|) 2(1 + |a|) 2(1 + |a|)
  |c| > , |c| > and |c| > ,
≥ sγ z2 | − |z − |a||z| sα sβ sγ
 
(15)
= |z| (sγ |z| − (1 + |a|)).
Thus then the Jungck-CR iteration with s-convexity escapes to
infinity.
sγ |z|
|u0 | ≥ |z| ( − 1). In the proof of theorem we used the facts that |z| ≥ |c| >
2(1+|a|) 2(1+|a|)
1 + |a|
sα , |z| ≥ |c| > sβ and |z| ≥ |c| > 2(1+|a|)
sγ .
In the second step of the iteration we have Hence the following corollary is the refinement of the escape
criterion discussed in the theorem.
|S(y0 )| = (1 − β)s T (z) + β s T (u0 )
 
  Corollary 2 (Escape Criterion): Let α, β, γ , s ∈ (0, 1]
≥ (1 − sβ)(z2 + c) + (1 − s(1 − β))(u20 + c) and
 
 
  2(1 + |a|) 2(1 + |a|) 2(1 + |a|)
≥ (1 − sβ)(z2 + c) + (s − s(1 − β))(u20 + c)
  |z| > max |c|, , , , (16)
sα sβ sγ
≥ sβ|u20 | − sβ|c| − |c| + sβ|c| + (1 − sβ)|z2 |
then there exist λ > 0 such that |zk | > (1 + λ)k |z| and |zk | →
≥ sβ|u20 | − |z|(because |1 − sβ| ≥ 0, |z| ≥ |c|). ∞ as k → ∞.

2 sγ |z|
2 Corollary 3: Suppose that
Since |z| > 2(1+|a|)
sγ , which implies |z| 1+|a| − 1 >  
2 2(1 + |a|) 2(1 + |a|) 2(1 + |a|)
|zm | > max |c|, , ,

sγ |z|
|z|2 . Hence |u0 |2 > |z|2 1+|a| − 1 > |z|2 > sγ |z|2 , and (17)
sα sβ sγ
1 + |a| ≥ 1. We get
for some m ≥ 0. Then there exist λ > 0 such that |zm+k | >
ay0 ≥ s2 βγ |z2 | − (1 + |a|) |z| . (1 + λ)k |zm | and |zk | → ∞ as k → ∞.

12170 VOLUME 7, 2019

Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

B. ESCAPE CRITERION FOR THE CUBIC COMPLEX In the third step of the iteration we have
POLYNOMIAL
|S(z1 )| = (1 − α)s S(y0 ) + α s T (y0 )
 
Let Qc (z) = z3 − az + c, where a, c ∈ C. We decompose Qc  
in the following way: T (z) = z3 + c and S(z) = az. ≥ (1 − sα)ay0 + sα(y30 + c)
 
 1
2
Theorem 2: Assume that |z| ≥ |c| > 2(1+|a|) sα , |z| ≥ ≥ sα|y30 + c| − (1 − sα)|ay0 |.
 1  1
2 2
|c| > 2(1+|a|) , |z| ≥ |c| > 2(1+|a|) where α, β, γ , s ∈ βγ |z|2
 2 
sβ sγ Since |y0 | ≥ |z| s 1+|a| − 1 , which implies |y30 | ≥
(0, 1] and define {zk }k∈N as follows
 s2 βγ |z3 |. We get
S(zk+1 ) = (1 − α) S(yk ) + α T (yk )
 s s  
|az1 | ≥ s3 αβγ z3  − sα|c| − |a||z|
 
S(yk ) = (1 − β) T (zk ) + β T (uk ),
s s (18)
S(uk ) = (1 − γ ) S(zk ) + γ T (zk ) k = 0, 1, 2, . . . ,
 s s
 
≥ s3 αβγ z3  − |z| − |a||z|
  
where z0 = z. Then |zk | → ∞ as k → ∞.  
= s3 αβγ z3  − (1 + |a|) |z|
 
Proof: Because T (z) = z3 + c, S(z) = az and z0 = z we  
have = |z| (s3 αβγ z2  − (1 + |a|)).
 
|S(u0 )| = (1 − γ )s S(z) + γ s T (z)
 
  Hence
= (1 − γ )s az + (1 − (1 − γ ))s (z3 + c)
 
!
s3 αβγ z2 
 
Using binomials series up to linear terms of γ and 1 − γ , |z1 | ≥ |z| −1 .
1 + |a|
and condition s < 1 we get
1 1
|au0 | ≥ (1 − s(1 − γ ))|z3 + c| − (1 − sγ )|az|
 
2(1+|a|) 2 2(1+|a|) 2
Since |z| > sα , |z| > sβ , |z| >
≥ |(s − s(1 − γ ))(z3 + c) |−|(1 − sγ )az| . 1
and in consequence s αβγ
3 |z|2

2(1+|a|) 2
, so |z|2 > 2(1+|a|) 1+|a| −
Because |z| ≥ |c| , |a| ≥ 1 and sγ < 1 we obtain sγ s3 αβγ
1 > 1. Therefore there exist λ > 0 such that s αβγ |z| 3 2
1+|a| − 1 >
 
|au0 | ≥ sγ z3 | − sγ |c − (1 − sγ )|az|
 
  1 + λ. Consequently |z1 | > (1 + λ)|z|. In particular |z1 | >
= sγ z3 | − sγ |c − |az| + sγ |az|
  |z|. So we may apply the same argument repeatedly to find
  |zk | > (1 + λ)k |z|. Thus, the orbit of z tends to infinity and
≥ sγ z3 | − |z − |a||z| this completes the proof. 
 
  Corollary 4: Suppose that
= |z| (sγ z2  − (1 + |a|)).
 
 1  1
2(1 + |a|) 2 2(1 + |a|) 2
Thus |c| > , |c| > and
  sα sβ
sγ z2  1
|u0 | ≥ |z| ( − 1).

2(1 + |a|) 2
1 + |a| |c| > , (19)
sγ
In the second step of the iteration we have
then the Jungck-CR iteration with s-convexity escapes to
|S(y0 )| = (1 − β)s T (z) + β s T (u0 )
 
  infinity.
≥ (1 − sβ)(z3 + c) + (1 − s(1 − β))(u30 + c) In the proof of theorem we used the facts that |z| ≥ |c| >
 
 1  1
2(1+|a|) 2 2
, |z| ≥ |c| > 2(1+|a|) and |z| ≥ |c| >
 
≥ (1 − sβ)(z3 + c) + (s − s(1 − β))(u30 + c)
 
sα sβ
 1
2(1+|a|) 2
≥ sβ|u30 | − sβ|c| − |c| + sβ|c| + (1 − sβ)|z3 | sγ . Hence the following corollary is the refinement
≥ sβ|u30 | − |z|(because |1 − sβ| ≥ 0, |z| ≥ |c|). of the escape criterion discussed in the theorem.
1 3 Corollary 5 (Escape Criterion): Let α, β, γ , s ∈ (0, 1]
3 sγ |z|2
 
2
Since |z| > 2(1+|a|)
sγ , which implies |z| 1+|a| − 1 > and
 2
3 ( 1  1
|z|
|z|3 . Hence |u0 |3 > |z|3 sγ > |z|3 > sγ |z|3 , and

2(1 + |a|) 2 2(1 + |a|) 2
1+|a| − 1 |z| > max |c|, , ,
1 + |a| ≥ 1. We get sα sβ
1 )
ay0 ≥ s2 βγ |z3 | − (1 + |a|) |z| .

2(1 + |a|) 2
, (20)
Thus sγ
!
s2 βγ |z|2 then there exist λ > 0 such that |zk | > (1 + λ)k |z| and
|y0 | ≥ |z| −1 .
1 + |a| |zk | → ∞ as k → ∞.

VOLUME 7, 2019 12171

 Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

  1
Corollary 6: Suppose that Since |z| > 2(1+|a|) n−1
, which implies
sγ
n n
|z|n−1 |z|n−1
( 1  1  
|z|n sγ1+|a| > |z|n |n > |z|n sγ1+|a| >

2(1 + |a|) 2 2(1 + |a|) 2 −1 . Hence |u0 −1
|zm | > max |c|, , ,
sα sβ |z|n
> sγ |z|n , and 1 + |a| ≥ 1. We get
1 )
ay0 ≥ s2 βγ |zn | − (1 + |a|) |z| .

2(1 + |a|) 2
, (21)
sγ Thus
for some m ≥ 0. Then there exist λ > 0 such that
!
s2 βγ |z|n−1
|zm+k | > (1 + λ)k |zm | and |zk | → ∞ as k → ∞. |y0 | ≥ |z| −1 .
1 + |a|

C. ESCAPE CRITERION FOR HIGHER DEGREE COMPLEX In the third step of the iteration we have
POLYNOMIALS
|S(z1 )| = (1 − α)s S(y0 ) + α s T (y0 )
 
Let Qc (z) = zn − az + c, where a, c ∈ C. We decompose Qc
≥ (1 − sα)ay0 + sα(yn + c)
 
in the following way: T (z) = zn + c and S(z) = az. 0
  1 ≥ sα|yn0 + c| − (1 − sα)|ay0 |.
n−1
Theorem 3: Assume that |z| ≥ |c| > 2(1+|a|) sα , |z| ≥
Since |y0 | ≥ |z| s βγ
 2
|z|n−1

 1  1
− 1 , which implies |yn0 | ≥
 
n−1 n−1
|c| > 2(1+|a|)sβ , |z| ≥ |c| > 2(1+|a|)sγ where 1+|a|

α, β, γ , s ∈ (0, 1] and define {zk }k∈N as follows s2 βγ |zn |. We get

|az1 | ≥ s3 αβγ zn  − sα|c| − |a||z|
  
S(zk+1 ) = (1 − α) S(yk ) + α T (yk )
 s s
≥ s3 αβγ zn  − |z| − |a||z|
 
S(yk ) = (1 − β) T (zk ) + β T (uk ),
s s (22)
= s3 αβγ zn  − (1 + |a|) |z|
 
S(uk ) = (1 − γ ) S(zk ) + γ T (zk ) k = 0, 1, 2, . . . ,

 s s
 
= |z| (s3 αβγ zn−1  − (1 + |a|)).
 
where z0 = z. Then |zk | → ∞ as k → ∞.
Proof: Because T (z) = zn + c, S(z) = az and z0 = z we Hence
have !
s3 αβγ zn−1 
 
|S(u0 )| = (1 − γ )s S(z) + γ s T (z) |z1 | ≥ |z| −1 . (23)
 
1 + |a|
= (1 − γ )s az + (1 − (1 − γ ))s (zn + c)
 
  1   1
n−1 2(1+|a|) n−1
Using binomials series up to linear terms of γ and 1 − γ , Since |z| > 2(1+|a|)
sα , |z| > sβ , |z| >
and condition s < 1 we get   1
2(1+|a|) n−1
sγ , so |z|n−1 > 2(1+|a|)
s3 αβγ
and in consequence
|au0 | ≥ (1 − s(1 − γ ))|zn + c| − (1 − sγ )|az| s3 αβγ |z|n−1
− 1 > 1. Therefore there exist λ > 0 such that
1+|a|
≥ |(s − s(1 − γ ))(zn + c) |−|(1 − sγ )az| . s3 αβγ |z|n−1
1+|a| − 1 > 1 + λ. Consequently |z1 | > (1 + λ)|z|.
Because |z| ≥ |c| , |a| ≥ 1 and sγ < 1 we obtain In particular |z1 | > |z|. So we may apply the same argument
repeatedly to find |zk | > (1 + λ)k |z|. Thus, the orbit of z tends
|au0 | ≥ sγ zn | − sγ |c − (1 − sγ )|az|
 
to infinity and this completes the proof. 
= sγ zn | − sγ |c − |az| + sγ |az| Corollary 7: Suppose that
 
n 
≥ sγ z | − |z − |a||z|   1   1
2(1 + |a|) n−1 2(1 + |a|) n−1
  |c| > , |c| > and
= |z| (sγ zn−1  − (1 + |a|)). sα sβ
 
  1
2(1 + |a|) n−1
Thus |c| > , (24)
  sγ
sγ zn−1 
|u0 | ≥ |z| ( − 1). then the Jungck-CR iteration with s-convexity escapes to
1 + |a|
infinity.
In the second step of the iteration we have Corollary 8 (Escape Criterion): Let α, β, γ , s ∈ (0, 1]
and
|S(y0 )| = (1 − β)s T (z) + β s T (u0 )
 
(   1   1
≥ (1 − sβ)(zn + c) + (1 − s(1 − β))(un0 + c)
  2(1 + |a|) n−1 2(1 + |a|) n−1
|z| > max |c|, , ,
sα sβ
≥ (1 − sβ)(zn + c) + (s − s(1 − β))(un + c)
 
0
 1 )
≥ sβ|un0 | − sβ|c| − |c| + sβ|c| + (1 − sβ)|zn |

2(1 + |a|) n−1
, (25)
≥ sβ|un0 | − |z|(because |1 − sβ| ≥ 0, |z| ≥ |c|). sγ

12172 VOLUME 7, 2019

Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

then there exist λ > 0 such that |zk | > (1 + λ)k |z| and
|zk | → ∞ as k → ∞.
Corollary 9: Suppose that
(   1   1
2(1 + |a|) n−1 2(1 + |a|) n−1
|zm | > max |c|, , ,
sα sβ
  1 )
2(1 + |a|) n−1
, (26)
sγ
for some m ≥ 0. Then there exist λ > 0 such that |zm+k | >
(1 + λ)k |zm | and |zk | → ∞ as k → ∞.
Theorem 4: Suppose that {zk }k∈N be the sequence of FIGURE 1. Example of Julia set for a quadratic function generated using
Jungck-CR iteration with s-convexity and |zk | → ∞ as k → Jungck-CR iteration with s-convexity.
  1   1
n−1 2(1+|a|) n−1
∞, then |z| ≥ |c| > 2(1+|a|)sα , |z| ≥ |c| > sβ ,
  1
n−1
|z| ≥ |c| > 2(1+|a|)
sγ where α, β, γ , s ∈ (0, 1].
Proof: Since {zk }k∈N be the sequence of Jungck-CR
iteration with s-convexity and |zk | → ∞ as k → ∞, then
it must be hold
|zk | ≥ (1 + λ)k |z|.
For k = 1, we have
|z1 | ≥ (1 + λ)|z|. (27)
But, since {zk }k∈N be the sequence of Jungck-CR iteration
with s-convexity, then from (23) FIGURE 2. Example of Julia set for a cubic function generated using
!
s3 αβγ zn−1  Jungck-CR iteration with s-convexity.
 
|z1 | ≥ |z| −1 . (28)
1 + |a|
Comparing (27) and (28), we have
s3 αβγ zn−1 
 
−1 = 1+λ
1 + |a|
s3 αβγ zn−1 
 
− 1 > 1,
1 + |a|
because λ > 0. This yields
  1
2(1 + |a|) n−1
|z| > .
s3 αβγ
  1
2(1+|a|) n−1
Consequently, we have |z| > sα , |z| > FIGURE 3. Example of Julia set for a quartic function generated using
Jungck-CR iteration with s-convexity.
  1   1
2(1+|a|) n−1 n−1
sβ and |z| > 2(1+|a|) sγ where n ≥ 2 and
α, β, γ , s ∈ (0, 1] and for complex fractal generation |z|
must be greater and equal to |c|, because for any given point IV. GRAPHICAL EXAMPLES
|z| < |c|, we have to compute the Jungck-CR orbit with s- In this section we present some graphical examples of Man-
convexity of z. If for some k, |zk | lies outside the circle of delbrot and Julia sets obtained with the Jungck-CR iteration
  1   1 with s-convexity. To generate the images we used the escape
n−1 2(1+|a|) n−1
radius max |c|, 2(1+|a|)sα , sβ ,
time algorithm with the escape criteria derived in Sec. III. The
  1  The algorithms were implemented in Mathematica.
2(1+|a|) n−1
sγ , we guarantee that the proposed orbit escapes. In the first example we present some Julia sets generated
Hence, z is not in the Julia sets and also, is not in Mandelbrot by using the Jungck-CR iteration with s-convexity. In all
sets. On the other hand, if |zk | never exceeds this bound (i.e. the examples the same maximum number of iterations was
|z| ≥ |c| holds), then by definition of the Julia sets, Mandel- preformed, and it was equal to 50. The resulting images are
brot sets and Multicorns, |zk | lies in Julia sets, Mandelbrot presented in Figs. 1–3 and the parameters used to generate
sets and Multicorns. This completes the proof.  them were the following:

VOLUME 7, 2019 12173

 Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

FIGURE 4. Example of Mandelbrot set for a quadratic function generated FIGURE 7. Example of multicorn for a quadratic function generated using
using Jungck-CR iteration with s-convexity. Jungck-CR iteration with s-convexity.

FIGURE 8. Example of multicorn for a cubic function generated using

FIGURE 5. Example of Mandelbrot set for a cubic function generated
Jungck-CR iteration with s-convexity.
using Jungck-CR iteration with s-convexity.

FIGURE 9. Example of multicorn for a quartic function generated using

FIGURE 6. Example of Mandelbrot set for a quartic function generated Jungck-CR iteration with s-convexity.
using Jungck-CR iteration with s-convexity.

are presented in Figs. 4–6 and the parameters used to generate

• Fig. 1: Qc (z) = z2 + 2 z + c, c = 0.95 + 0.5i, A = them were the following:
[−3.5, 3.5]2 , α = 0.5, β = 0.7, γ = 0.8, s = 0.6, • Fig. 4: Qc (z) = z2 + 2 z + c, A = [−35, 12] × [−12, 12],
• Fig. 2: Qc (z) = z3 + 32 z + c, c = −2.295i, A = α = 0.9, β = 0.9, γ = 0.9, s = 0.8,
[−3.2, 3.2] × [−3.4, 3.4], α = 0.8, β = 0.7, γ = 0.6, 3
• Fig. 5: Qc (z) = z3 + 2 z + c, A = [−4, 4] × [−8, 8],
s = 0.7, α = 0.8, β = 0.7, γ = 0.6, s = 0.7,
• Fig. 3: Qc (z) = z4 + iz + c, c = −0.03 + 0.81i, A = • Fig. 6: Qc (z) = z4 + iz + c, A = [−4, 4]2 , α = 0.9,
[−2.7, 2.7]2 , α = 0.9, β = 0.1, γ = 0.8, s = 0.9. β = 0.1, γ = 0.8, s = 0.9.
In the second example we present some Mandelbrot sets In the last example examples of multicorns generated via
generated by using the Jungck-CR iteration with s-convexity. the Jungck-CR iteration with s-convexity are presented. Sim-
In all the examples the same maximum number of iterations ilar to the example with the Mandelbrot sets the maximum
was preformed, and it was equal to 100. The resulting images number of iteration is the same for all examples, and it is equal

12174 VOLUME 7, 2019

Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

to 100. The generated images of multicorns are presented [7] M. Pinheiro, ‘‘S-convexity—Foundations for analysis,’’ Differ.
in Fig. 7–9 and the parameters used to generate them were Geometry–Dyn. Syst., vol. 10, pp. 257–262, 2008. [Online]. Available:
http://www.mathem.pub.ro/dgds/v10/D10.htm
the following: [8] M. Mishra, D. Ojha, and D. Sharma, ‘‘Fixed point results in tricorn &
• Fig. 7: Qc (z) = z2 + 2 z + c, A = [−35, 14] × [−9, 9], Multicorns of Ishikawa iteration and S-convexity,’’ Int. J. Adv. Eng. Sci.
Technol., vol. 2, no. 2, pp. 156–159, 2011.
α = 0.9, β = 0.9, γ = 0.9, s = 0.8, [9] M. Mishra, D. Ojha, and D. Sharma, ‘‘Some common fixed point results in
• Fig. 8: Qc (z) = z3 + 23 z + c, A = [−4, 4] × [−6, 6], relative superior Julia sets with Ishikawa iteration and S-convexity,’’ Int. J.
α = 0.8, β = 0.7, γ = 0.6, s = 0.7, Adv. Eng. Sci. Technol., vol. 2, no. 2, pp. 175–180, 2011.
[10] S. Kang, W. Nazeer, M. Tanveer, and A. Shahid, ‘‘New fixed point results
• Fig. 9: Qc (z) = z4 + iz + c, A = [−3.5, 3.5]2 , α = 0.9, for fractal generation in Jungck Noor orbit with S−convexity,’’ J. Function
β = 0.1, γ = 0.8, s = 0.9. Spaces, vol. 2015, Jul. 2015, Art. no. 963016.
[11] W. Nazeer, S. Kang, M. Tanveer, and A. Shahid, ‘‘Fixed point results in the
generation of Julia and Mandelbrot sets,’’ J. Inequalities Appl., vol. 2015,
V. CONCLUSIONS p. 298, Sep. 2015.
In this paper, we study the Jungck-CR iteration with [12] S. Cho, A. Shahid, W. Nazeer, and S. Kang, ‘‘Fixed point results for fractal
generation in Noor orbit and S-convexity,’’ SpringerPlus, vol. 5, p. 1843,
s-convexity in the generation of fractals (i.e. Julia sets, Oct. 2016.
Mandelbrot sets, tricorn and Multicorns) for non-linear [13] K. Gdawiec and A. Shahid, ‘‘Fixed point results for the complex fractal
dynamics. We proved escape criterion to generate Mandelbrot generation in the s-iteration orbit with S-convexity,’’ Open J. Math. Sci.,
vol. 2, no. 1, pp. 56–72, 2018.
sets, Julia sets, tricorn and multicorns using this type of [14] M. Barnsley, Fractals Everywhere, 2nd ed. Boston, MA, USA: Academic,
iteration for complex quadratic, cubic and nth degree complex 1993.
polynomials. We presented the use of iteration other than [15] R. Devaney, A First Course in Chaotic Dynamical Systems: Theory and
Experiment. New York, NY, USA: Addison-Wesley, 1992.
the Picard one in the generation of Mandelbrot and Julia [16] L. Xiangdong, Z. Zhiliang, W. Guangxing, and Z. Weiyong, ‘‘Composed
sets. Moreover, we presented some graphical examples which accelerated escape time algorithm to construct the general Mandelbrot set,’’
showed that with the use of the Jungck-CR iteration with Fractals, vol. 9, no. 2, pp. 149–153, 2001.
[17] V. Drakopoulos, ‘‘Comparing rendering methods for Julia sets,’’ J. WSCG,
s-convexity we are able to obtain many diverse shapes of vol. 10, nos. 1–3, pp. 155–161, 2002.
fractal sets. The colors of figures depend upon the number [18] V. Drakopoulos, N. Mimikou, and T. Theoharis, ‘‘An overview of parallel
of iterations and input parameters. The input parameters are visualisation methods for Mandelbrot and Julia sets,’’ Comput. Graph.,
vol. 27, no. 4, pp. 635–646, 2003.
different for each figure, so color difference appeared. [19] N. Hussain, V. Kumar, and M. Kutbi, ‘‘On rate of convergence of Jungck-
The Jungck-CR iteration with s-convexity does not reduce type iterative schemes,’’ Abstract Appl. Anal., vol. 2013, Apr. 2013,
to neither Picard, nor Jungck-Mann, nor Jungck-Ishikawa, Art. no. 132626.
[20] K. Gdawiec, ‘‘Fractal patterns from the dynamics of combined polynomial
nor Jungck-Noor, nor any other iteration studied in the lit- root finding methods,’’ Nonlinear Dyn., vol. 90, no. 4, pp. 2457–2479,
erature on the generation of complex fractals, so the results 2017.
of this paper open new class of complex fractals. Moreover, [21] Y. Sun, R. Xu, L. Chen, and X. Hu, ‘‘Image compression and encryption
scheme using fractal dictionary and Julia set,’’ IET Image Process., vol. 9,
the obtained complex fractals could further extend the capa- no. 3, pp. 173–183, 2015.
bilities of the algorithms that use Mandelbrot and Julia sets, [22] Y. Sun, L. Chen, R. Xu, and R. Kong, ‘‘An image encryption algorithm
e.g., they can expand the domain dictionary used in fractal utilizing Julia sets and Hilbert curves,’’ PLoS ONE, vol. 9, no. 1, p. e84655,
2014.
image compression [21] or broaden the space for the initial [23] K. Gdawiec and W. Kotarski, ‘‘Polynomiography for the polynomial infin-
keys used in image encryption [22]. ity norm via Kalantari’s formula and nonstandard iterations,’’ Appl. Math.
In our further work we will try to derive the escape criteria Comput., vol. 307, pp. 17–30, Aug. 2017.
in the Jungck-CR iteration with s-convexity for functions of
other classes than the polynomial one, e.g., trigonometric.
YOUNG CHEL KWUN received the Ph.D. degree
Moreover, in the fixed point literature we can find many in mathematics from Dong-A University, Busan,
different iteration methods that can be used in the study of South Korea, where he is currently a Professor.
Julia and Mandelbrot sets. A review of the explicit iterations He is also a Mathematician from South Korea.
and their dependencies can be found in the paper by Gdawiec He has published over 100 research articles in dif-
ferent international journals. His research interests
and Kotarski [23]. are nonlinear analysis, decision theory, and system
theory and control.
REFERENCES
[1] M. Rani and V. Kumar, ‘‘Superior Julia set,’’ J. Korea Soc. Math. Edu. Ser.
D, Res. Math. Edu., vol. 8, no. 4, pp. 261–277, 2004.
[2] M. Rani and V. Kumar, ‘‘Superior mandelbrot set,’’ J. Korea Soc. Math. MUHAMMAD TANVEER received the M.Sc.
Edu. Ser. D, Res. Math. Edu., vol. 8, no. 4, pp. 279–291, 2004. degree in mathematics from Government college
[3] Y. Chauhan, R. Rana, and A. Negi, ‘‘New Julia sets of Ishikawa iterates,’’ University, Faisalabad, Pakistan, in 2008, and the
Int. J. Comput. Appl., vol. 7, no. 13, pp. 34–42, 2010. M.Phil. degree in mathematics from Lahore Leads
[4] S. Kang, A. Rafiq, A. Latif, A. Shahid, and A. Faisal, ‘‘Fractals through University, Lahore, Pakistan, in 2014. He is cur-
modified iteration scheme,’’ Filomat, vol. 30, no. 11, pp. 3033–3046, 2016. rently pursuing the Ph.D. degree with the Uni-
[5] S. Kang, A. Rafiq, A. Latif, A. Shahid, and Y. Kwun, ‘‘Tricorns and mul- versity of Lahore, Lahore. He has published over
ticorns of S-iteration scheme,’’ J. Function Spaces, vol. 2015, Jan. 2015, 20 research articles in different international jour-
Art. no. 417167. nals. His main research interest includes fixed
[6] Ashish, M. Rani, and R. Chugn, ‘‘Julia sets and Mandelbrot sets in Noor point results in fractal generation via different
orbit,’’ Appl. Math. Comput., vol. 228, pp. 615–631, Jan. 2014. Jungck type iteration.

VOLUME 7, 2019 12175

 Y. C. Kwun et al.: Mandelbrot and Julia Sets via Jungck–CR Iteration With s–Convexity

WAQAS NAZEER received the Ph.D. degree SHIN MIN KANG received the Ph.D. degree
in mathematics from the Abdus Salam School in mathematics from Dong-A University, Busan,
of Mathematical Sciences, Government College South Korea. He is currently a Professor with
University, Lahore, Pakistan. He is currently an Gyeongsang National University, South Korea.
Assistant Professor with the University of Edu- He is also a Mathematician from South Korea.
cation, Lahore. He is also a Mathematician from He has published over 200 research articles in dif-
Pakistan. During his studies, he was funded by the ferent international journals. His research interests
Higher Education Commission of Pakistan. He has include fixed point theory, nonlinear analysis, and
published over 100 research articles in different variational inequality.
international journals. His research interests are
analysis and graph theory. He received the Outstanding Performance Award
for the Ph.D. degree.

KRZYSZTOF GDAWIEC received the M.Sc.

degree in mathematics, and the Ph.D. degree in
computer science from the University of Silesia,
Poland, in 2005 and 2010, respectively, and the
D.Sc. degree in computer science from the Warsaw
University of Technology, Poland, in 2018. He is
currently an Assistant Professor with the Insti-
tute of Computer Science, University of Silesia.
He is an author of several journal and conference
publications. His main research interests include
applications of fractal geometry, pattern recognition, and computer graphics.
He is a member of the Polish Mathematical Society and SIGGRAPH.

12176 VOLUME 7, 2019