selforganizology, 2016, vol. 3, iss. 1
TRANSCRIPT
Selforganizology
Vol. 3, No. 1, 1 March 2016
International Academy of Ecology and Environmental Sciences
Selforganizology ISSN 2410-0080
Volume 3, Number 1, 1 March 2016 Editor-in-Chief WenJun Zhang Sun Yat-sen University, China International Academy of Ecology and Environmental Sciences, Hong Kong E-mail: [email protected], [email protected] Editorial Board Awf Al-Kassir (University of Extremadura, Spain) Ramiz M. Aliguliyev (National Academy of Sciences, Azerbaijan) Andre Bianconi (Sao Paulo State University (Unesp), Brazil) Chris Cannings (University of Sheffield, UK) Manuel De la Sen (University of Basque Country, Spain) Alessandro Ferrarini (University of Parma, Italy) Zeljko Kanovic (University of Novi Sad, Serbia) Xin Li (Northwest A&F University, China) Abdul Qadeer Khan (University of Azad Jammu & Kashmir, Pakistan) Zdzislaw Kowalczuk (Gdansk University of Technology, Poland) Muhammad Naeem (Abbottabad University of Science and Technology, Pakistan) Lev V. Nedorezov (University of Nova Gorica, Slovenia) Quanke Pan (Northeastern University, China) Sergio Valero Verdu (Universidad Miguel Hernandez de Elche, Spain) Hafiz Abdul Wahab (Hazara University, Pakistan) ZhiGuo Zhang (Sun Yat-sen University, China) Editorial Office: [email protected]
Publisher: International Academy of Ecology and Environmental Sciences
Address: Unit 3, 6/F., Kam Hon Industrial Building, 8 Wang Kwun Road, Kowloon Bay, Hong Kong
Tel: 00852-2138 6086 Fax: 00852-3069 1955 Website: http://www.iaees.org/ E-mail: [email protected]
Selforganizology, 2016, 3(1): 1-9
IAEES www.iaees.org
Article
A random network based, node attraction facilitated network
evolution method
WenJun Zhang
School of Life Sciences, Sun Yat-sen University, Guangzhou 510275, China; International Academy of Ecology and Environmental Sciences, Hong Kong E-mail: [email protected], [email protected]
Received 6 August 2015; Accepted 28 September 2015; Published online 1 March 2016
Abstract
In present study, I present a method of network evolution that based on random network, and facilitated by
node attraction. In this method, I assume that the initial network is a random network, or a given initial
network. When a node is ready to connect, it tends to link to the node already owning the most connections,
which coincides with the general rule (Barabasi and Albert, 1999) of node connecting. In addition, a node may
randomly disconnect a connection i.e., the addition of connections in the network is accompanied by the
pruning of some connections. The dynamics of network evolution is determined of the attraction factor of
nodes, the probability of node connection, the probability of node disconnection, and the expected initial
connectance. The attraction factor of nodes, the probability of node connection, and the probability of node
disconnection are time and node varying. Various dynamics can be achieved by adjusting these parameters.
Effects of simplified parameters on network evolution are analyzed. The changes of attraction factor can
reflect various effects of the node degree on connection mechanism. Even the changes of only will generate
various networks from the random to the complex. Therefore, the present algorithm can be treated as a general
model for network evolution. Modeling results show that to generate a power-law type of network, the
likelihood of a node attracting connections is dependent upon the power function of the node’s degree with a
higher-order power. Matlab codes for simplified version of the method are provided. Keywords network evolution; node attraction; connection probability; disconnection.
1 Introduction
In 1998, Watts and Strogatz presented a method for generating random graphs. Thereafter, Barabasi and Albert
(1999) proposed a general and known mechanism for network evolution. The algorithm developed by Cancho
and Sole (2001) can generate a variety of complex networks with diverse degree distributions. Zhang (2012a,
2012b, 2015, 2016) proposed a series of methods for network generation and evolution. In present study, I will
propose method of network evolution that based on random network, and facilitated by node attraction. Deep
analyses will be implemented to understand the properties of the method.
Selforganizology ISSN 24100080 URL: http://www.iaees.org/publications/journals/selforganizology/onlineversion.asp RSS: http://www.iaees.org/publications/journals/selforganizology/rss.xml Email: [email protected] EditorinChief: WenJun Zhang Publisher: International Academy of Ecology and Environmental Sciences
Selforganizology, 2016, 3(1): 1-9
IAEES www.iaees.org
2 Algorithm
I assume that the initial network is a random network (this is popular in nature. For example, the distribution
of particles with different sizes with which the water vapor attaches to generate droplets; the random
distribution of matter/nucleus of crystallization for the generation of stars/crystals, etc.), or a given initial
network. When a node is ready to connect, it tends to link to the node already owning the most connections
(the largest node), which coincides with the general rule of node connecting (Barabasi and Albert, 1999). In
addition, a node may randomly disconnect a connection, i.e., the addition of connections in the network is
accompanied by the pruning of some connections.
Assume there are totally v nodes in the network. Expected initial connectance (connectance=practical
connections/potential maximum connections) is c (initial condition) if the initial network is a random network
generated by the algorithm, expected final connectance is ce (termination condition), the attraction factor of
nodes is (t,a) (driving variable; (t,a)>0, where a is the node degree), the probability of node connection is
p(t,a) (driving variable), the probability of node disconnection is q(t,a,b) (driving variable, where b is the
connected node’s degree), maximum number of iterations is iter (termination condition), and the confidence
degree for detecting the statistic significance of network type is (auxiliary constant). (t,a), p(t,a), and
q(t,a,b) are time and node (in particular node degree) varying. The procedures are as follows
(1) Generate the initial network. In the situation of random initial network, assume the adjacency matrix
of the random network is d=(dij), i, j=1,2,…,v, where dij=dji, dii=0, and if dij=1 or dji=1, there is a connection
between nodes i and j. For each pair of i, j (i=1,2,…,v-1; j>i), generate a random value r, if r<c, dij=1 and
dji=1. Otherwise, the initial network is a given network.
(2) Let t=1. Calculate the degree of node, ai(t), i=1,2,…,v. The cumulative attraction strength of node 1 to
node i is
v
j
atj
i
j
atji
jj tatatp1
),(
1
),( )(/)()(
(3) Generate/disconnect connections. For the node i, i=1,2,…,v, generate a random value s, if s<p(t,ai),
the node i is ready to connect to one of the remaining nodes. Let p0(t)=0. For the node j, j=1, 2,…,v, ji,
generate a random value w. dij=1 and dji=1, if pi-1(t)w<pi(t). In the practical uses, the interval [pi-1(t), pi(t))
represents the mass or volume of the particle i, the gravity of the celestial body i, the personality charm of the
person i, the academic impact of the scientist i, etc.
For the node i, i=1,2,…,v, generate a random value g; if g<q(t,ai,b), one of the connections of the node i,
e.g., dij, is randomly disconnected, and let dij=0 and dji=0.
By doing so, a network at time t is generated. Various indices and methods, e.g., coefficient of variation
(CV), aggregation index (AI), and entropy (Zhang and Zhan, 2011; Zhang, 2012a) can be used to detect the
types and properties of the network.
(4) Calculate the connectance C of the network. Let t=t+1 and return (2), if C is greater than the expected
final connectance ce; otherwise, the algorithm terminates, if C is less than ce, or the maximum iterations iter
are achieved.
For convenience and simplicity, assume (t,a)=, p(t,a)=p, q(t,a,b)=q, i.e., the attraction factor of nodes,
the probability of node connection, and the probability of node disconnection are constants for any degree of
nodes at any time. Thus we obtain a simplified version of the algorithm.
The following are Matlab codes for simplified version of the algorithm (netEvolution.m)
2
Selforganizology, 2016, 3(1): 1-9
IAEES www.iaees.org
%Reference: Zhang WJ. 2016. A random network based, node attraction facilitated network evolution method.
Selforganizology, 3(1): 1-9
v=input('Total number of nodes in the network= ');
choice=input('Input the type (1: a random network generated by the algorithm; 2: a given network) of generating initial network:
');
if (choice==1) ci=input('Expected initial connectance (=practical connections/potential maximum connections; e.g., 0.05, etc)= ');
end
if (choice==2) adjstr=input('Input the file name of adjacency matrix of the given initial network (e.g., raw.txt, raw.xls, etc.
Adjacency matrix is d=(dij)v*v, where v is the number of nodes in the network. dij=1, if vi and vj are adjacent, and dij=0, if vi
and vj are not adjacent; i, j=1,2,…, v: ','s'); end
ce=input('Expected final connectance (=practical connections/potential maximum connections; e.g., 0.1, 0.15, etc)= ');
lamda=input('Attraction factor of nodes (lamda; e.g., 2, 4, etc. lamda>0)= ');
p=input('Probability of node connection (e.g., 0.1, 0.2)= ');
q=input('Probability of node disconnection (e.g., 0, 0.01)= ');
alpha=input('Confidential degree for detecting network type (e.g., 0.05, 0.01)= ');
iter=input('Permitted maximum iterations (e.g., 5000)= ');
adj=zeros(v);
degr=zeros(1,v);
prop=zeros(1,v);
z=zeros(v);
if (choice==1)
for i=1:v-1
for j=i+1:v
if (rand()<ci) adj(i,j)=1; adj(j,i)=1; end
end; end; end
if (choice==2) adj=load(adjstr); end;
degr=sum(adj);
fprintf('Initial adjacency matrix\n')
disp([adj])
fprintf('Initial degree distribution\n')
disp([degr])
t=1;
while (v>0)
propdegr=degr.^lamda;
prop(1)=propdegr(1)/sum(propdegr);
for i=2:v;
prop(i)=prop(i-1)+propdegr(i)/sum(propdegr);
end
node=zeros(1,v);
nu=0;
for i=1:v
if (rand()<p) nu=nu+1; node(nu)=i; end
end
for k=1:nu;
for i=1:v
3
Selforganizology, 2016, 3(1): 1-9
IAEES www.iaees.org
if (node(k)~=i) continue; end
lab=0;
ran=rand();
for j=1:v
if (j==i) continue; end
if (j==1) st=0; end
if (j>=2) st=prop(j-1); end
if ((ran>=st) & (ran<prop(j))) lab=1; adj(i,j)=1; adj(j,i)=1; break; end
end
if (lab==1) break; end
end; end
nodes=zeros(1,v);
for i=1:v
if (rand()<q) nodes(i)=1; end
end
for i=1:v-1
if (nodes(i)~=1) break; end
nuu=0;
for j=i+1:v
if (adj(i,j)==1) nuu=nuu+1; z(i,j)=nuu; end
end
np=round(rand()*nuu+0.5);
for j=i+1:v
if (z(i,j)==np) adj(i,j)=0; adj(j,i)=0; end
end; end
fprintf(['\n\nTime ' num2str(t)])
fprintf(['\n\nAdjacency matrix\n'])
disp([adj])
degr=sum(adj);
fprintf('\nDegree distribution\n')
disp([degr])
cnow=(sum(degr)/2)/((v^2-v)/2);
fprintf(['\nConnectance=' num2str(cnow) '\n'])
meann=mean(degr);
varr=(std(degr))^2;
fprintf(['\nEntropy=' num2str(varr-meann) '\n'])
num=0;
cv=varr/meann;
fprintf(['\nCoefficient of variation (CV)=' num2str(cv) '. '])
x2=cv*(v-1);
sig=chi2cdf(x2,v-1);
if (sig<=alpha) fprintf('The network is a random network according to CV.\n'); end
if ((sig>alpha) & (cv>1)) fprintf('The network is a complex network according to CV.\n'); num=num+1; end;
summ=sum(degr);
summa=sum(degr.*(degr-1));
4
IAEES
h=v*summ
fprintf(['\n
x2=h*(sum
sig=chi2cd
if (sig<=al
if ((sig>alp
if (num>=
if (cnow>=
if (t>=iter)
t=t+1;
end
3 Result
In presen
3.1 Dyna
Running
propertie
=0.05.
network
asymptot
the three
3.2 Para
3.2.1 Eff
Fixing v=
algorithm
for =0.0
ma/(summ*(sum
nAggregation in
mm-1)+v-summ
df(x2,v-1);
lpha) fprintf('Th
pha) & (h>1)) f
=2) fprintf('\nTh
=ce) break; end
) break; end;
s and Analys
nt study, for th
amics of netw
the simplifie
es of the netw
In this case,
increases wi
te. However,
indices incre
Fig. 1
ametrical ana
fects of the at
=30, c=0.05,
m, analyze th
05, 1, 2, 3, an
mm-1));
ndex (AI)=' num
m;
he network is a
fprintf('The netw
he network is a
d
sis
he convenien
work evoluti
ed version of
work in the ev
the connecte
ith the time.
aggregation
ease most qui
Property dynam
alysis
ttraction facto
ce=0.2, p=0.2
he effects of t
nd 4, connec
Selforg
m2str(h) '. '])
random netwo
work is a comp
complex netwo
nce, the initial
ion
the network e
volution. The
d nodes incre
Coefficient
index increa
ickly until all
mics of network
or of nodes2, q=0, =0.0
the attraction
ctance of the
ganizology, 201
rk according to
plex network ac
ork according to
l network is a
evolution alg
parameters w
ease quickly
of variation
ases quickly t
l nodes are co
k evolution (v=
05, and runnin
n factor of no
network incr
16, 3(1): 1-9
o AI.\n'); end;
cording to AI.\n
o all indices.\n')
a random netw
gorithm, Fig.
were set as v=
until all node
(CV) increas
to a climax a
onnected.
=30, c=0.05, ce=
ng the simpli
odeson net
reases more q
n'); num=num+
); end;
work generat
1 illustrates a
=30, c=0.05,
es are connec
ses quickly to
nd drops dow
=0.2, =3, p=0.2
ified version o
twork evoluti
quickly if i
w
+1; end;
ted by the alg
a case of the d
ce=0.2, =3, p
cted. Connect
o an approxi
wn thereafter
2, q=0, =0.05)
of the networ
ion. Fig. 2 in
s smaller. Co
www.iaees.org
gorithm.
dynamics of
p=0.2, q=0,
tance of the
imate upper
. Generally,
).
rk evolution
ndicates that
oefficient of
5
IAEES
variation
Table
achieved
the coeff
the final
3.2.2 Eff
Fixing v=
evolution
greater.
3.2.3 Eff
Fixing v=
network
more qui
greater.
3.2.4 Eff
Fixing v=
evolution
quickly i
F
n and aggrega
e 1 indicates
d). It can be fo
ficient of var
network incr
fects of the pr
=30, c=0.05,
n. Fig. 3 indi
fects of the pr
=30, c=0.05,
evolution. Fi
ickly if q is
fects of expec
=30, ce=0.2,
n. Fig. 5 indi
if c is greater.
Fig. 2 Effects o
ation index in
the results o
found that req
iation, aggreg
reases with robability of n
ce=0.2, =3,
icates that for
robability of n
p=0.2, ce=0
ig. 4 indicate
smaller. Coe
cted initial co
=3, p=0.2,
icates that for
. Coefficient
of the attraction
Selforg
creases more
of properties
quired iteratio
gation index,
node connect
q=0, analyze
r p=0.01, 0.0
node disconn
.2, =3, anal
es that for q=
efficient of va
nnectance c
q=0, analyze
r c=0.01, 0.0
of variation a
factor of nodes
ganizology, 201
e quickly if of network in
ons increase w
, and entropy
tion p
e the effects o
05, 0.1, and 0
nection q
lyze the effec
=0, 0.01, 0.02
ariation and
e the effects
03, 0.05, and
and aggregati
s
16, 3(1): 1-9
is greater.
n the last iter
with . With
y increase wit
of the probabi
0.2, the three
cts of the pro
2, and 0.03, c
aggregation
of the expec
0.08, connec
ion index incr
, fixing v=30
ration (expec
the same exp
th , which m
ility of node c
indices incre
obability of n
connectance o
index increas
cted initial co
ctance of the
reases more q
0, c=0.05, ce=0
w
cted final con
pected final c
means the co
connection pease more qu
node disconn
of the networ
ses more qui
onnectance cnetwork incr
quickly if c is
0.2, p=0.2, and q
www.iaees.org
nnectance is
onnectance,
mplexity of
on network
ickly if p is
nection qon
rk increases
ickly if q is
on network
reases more
s smaller.
q=0.
6
IAEES
Tab
t
Connectance
Entropy
Coefficient of
Variation (CV)
Network type (C
Aggregation Ind
Network type (A
Network type (O
judge)
Degree distribut
Fig. 3
ble 1 The chang
CV) Rand
dex (AI)
AI) Rand
Overall Rand
tion
5
2
4
8
7
4
4
5
7
9
3 Effects of the
ges of network p
0.01
11
0.20
-2.14
0.64
dom network C
0.94
dom network C
dom network C
7 5
10 5
7 4
5 6
8 2
6 6
8 7
6 8
7 7
3 6
7
8
3
6
4
1
5
6
8
2
probability of n
Selforg
properties for th
0.5
13
0.21
0.27
1.05
Complex network
1.01
Complex network
Complex network
7 8 5
8 5 11
3 5 5
6 5 8
4 8 6
1 6 11
5 2 4
6 8 6
8 4 9
2 6 10
node connection
ganizology, 201
he last iteration
1
16
0.20
6.92
2.19
k Complex ne
1.19
k Complex ne
k Complex ne
10 8
1 11
2 5
9 10
1 0
3 10
9 2
8 1
7 4
4 4
n pp
16, 3(1): 1-9
over , fixing
1
etwork Compl
etwork Compl
etwork Compl
9
10
3
2
9
10
4
2
7
9
4
4
5
6 1
1
6
11
8
11
23
, fixin
v=30, c=0.05, c
2
15
0.20
12.82
3.16
lex network C
1.35
lex network C
lex network C
8 1
2 10
2 6
10 6
5 8
3 2
3 4
7 7
4 6
3 2
4
8
4
4
10
3
3
4
5
22
ng v=30, c=0.05
w
ce=0.2, p=0.2, q
3
93
0.20
27.98
5.77
omplex network
1.79
omplex network
omplex network
2 3
10 3
6 5
29 2
0 2 4
4 6
5 2
2 4
6 4
2 3 7
5, ce=0.2, =3,
www.iaees.org
q=0, =0.05.
4
323
0.20
44.50
8.67
Complex netw
2.29
Complex netw
Complex netw
6 3
3 5
2 4
2 2
4 3
3 5
3 20
2 4
29 3
6 4
and q=0.
work
work
work
3
3
3
7
3
2
4
29
3
4
7
IAEES
Fig. 4 E
Fig
Effects of the pr
g. 5 Effects of e
robability of no
expected initial
Selforg
de disconnectio
connectance c
ganizology, 201
on qq
(c=0.01,0.03,0.
16, 3(1): 1-9
0.03), fixin
.05,0.08), fixing
ng v=30, c=0.0
g v=30, ce=0.2,
w
5, ce=0.2, p=0.2
=3, p=0.2, an
www.iaees.org
2, and =3.
nd q=0,
8
Selforganizology, 2016, 3(1): 1-9
IAEES www.iaees.org
3.3 Model’s universality
As mentioned above, the changes of attraction factor can reflect various effects of the node degree on
connection mechanism (Table 1, Fig. 2). The larger will lead to generate complex networks, e.g., exponential
law, power law networks, etc. 0 means a trend to generate the random network. Even the changes of only
will generate various networks from the random to the complex (Table 1). Therefore, the present algorithm can
be treated as a general model for network evolution.
Modeling results (Table 1) show that to generate power-law distributed node degrees (i.e., to generate a
power-law type of network), the likelihood of a node attracting connections is dependent upon the power
function of the node’s degree with a higher-order power.
4 Discussion
In present method, the dynamics of network evolution is determined of (t,a), p(t,a), q(t,a,b), and c. I
simplify all parameters as constants. However, more complex mechanism for network evolution can be
achieved by setting reasonable forms of (t,a), p(t,a), q(t,a,b), depending on the networks being studied. In
present study, I use a small range of parametrical values for parametrical analysis. More properties may be
found by broadening the range of parametrical values. In present algorithm, the addition of connections
coincides with the general rule of node connecting (Barabasi and Albert, 1999). However, the mechanism for
pruning of connections is still unknown (unknown q(t,a,b)), thus in the simplified version of the algorithm,
q=0 is a better choice.
Acknowledgment
I am thankful to the support of High-Quality Textbook Network Biology Project for Engineering of Teaching
Quality and Teaching Reform of Undergraduate Universities of Guangdong Province (2015.6-2018.6), from
Department of Education of Guangdong Province, Discovery and Crucial Node Analysis of Important
Biological and Social Networks (2015.6-2020.6), from Yangling Institute of Modern Agricultural
Standardization, and Project on Undergraduate Teaching Reform (2015.7-2017.7), from Sun Yat-sen
University, China.
References
Barabasi AL, Albert R. 1999. Emergence of scaling in random networks. Science, 286(5439): 509
Cancho RF, Sole RV. 2001. Optimization in Complex Networks. Santafe Institute, USA
Watts D, Strogatz S. 1998. Collective dynamics of ‘small world’ networks. Nature, 393:440-442
Zhang WJ, Zhan CY. 2011. An algorithm for calculation of degree distribution and detection of network type:
with application in food webs. Network Biology, 1(3-4):159-170
Zhang WJ. 2012a. Computational Ecology: Graphs, Networks and Agent-based Modeling. World Scientific,
Singapore
Zhang WJ. 2012b. Modeling community succession and assembly: A novel method for network evolution.
Network Biology, 2(2): 69-78
Zhang WJ, 2015. A generalized network evolution model and self-organization theory on community
assembly. Selforganizology, 2(3): 55-64
Zhang WJ. 2016. Selforganizology: The Science of Self-Organization. World Scientific, Singapore
9
Selforganizology, 2016, 3(1): 10-15
IAEES www.iaees.org
Article
A review on research of the predator-prey interactions
H. A. Wahab1, Hazrat Ali1, Muhammad Naeem2, Sarfraz Ahmad3, Saira Bhatti4, Muhammad Shahzad1,
Khalid Usman1 1Department of Mathematics, Hazara University, Manshera, Pakistan 2Department of IT, Abbottabad University of Science and Technology, Havelian, Abbottabad, Pakistan 3Department of Mathematics, Abbottabad University of Science and Technology, Havelian, Abbottabad, Pakistan 4Department of Mathematics, COMSATS Institute of Information Technology, Abbottabad, Pakistan
E-mail: [email protected],[email protected]
Received 13 July 2015; Accepted 20 August 2015; Published online 1 March 2016
Abstract
In this review article, we aim to study predator-prey interactions and their historical development from time to
time. A general overview of the predator-prey model, its evolution, and the physical meanings of this model
has been discussed here. A brief discussion on the classification of predators and process of predation it its
effect on the environment is presented in this review article. One important characteristic of the predation
process is that predators co-operate with each other against preys which becomes more visible when the prey
are strong enough to be over-powered easily.
Keywords predator-prey interactions; classification of predators; predators’ co-operation; research
developments.
1 Introduction to Predator Prey Interactions
The predator-prey models are the basis which helps to see the life and ecosystems in which the predator and
prey have certain role. This model explains how the predators interact with its prey. It explains the sustenance,
evaluation and alternative dispersion of some species in the case of failure to complete in the life in which the
stronger has advantage of dominate role. The predator-prey model is like the survival of the fittest-theory
(Zhang and Liu, 2015). The fittest are the stronger species targeting the weaker species and win life for
themselves and this evaluation of life for one species results in the numerical and sometime general extinction
of other weaker species. The weaker species remain in constraint struggle to achieve their security in the
diaspora where general fear of life remains ever present. The weaker species which become prey adopt many
measures to trick the predator to avoid being hunted.
Selforganizology ISSN 24100080 URL: http://www.iaees.org/publications/journals/selforganizology/onlineversion.asp RSS: http://www.iaees.org/publications/journals/ selforganizology /rss.xml Email: [email protected] EditorinChief: WenJun Zhang Publisher: International Academy of Ecology and Environmental Sciences
Selforganizology, 2016, 3(1): 10-15
IAEES www.iaees.org
Briefly predator prey interaction is like a win-loss interaction (Zhang, 2012), it is the victory for one and
loss for the other; and life for one and elimination of the other.
2 A General Predator-Prey Model
Let us consider two kinds of populations, that have sizes at a reference time t and shown by ( ),X t and ( ),Y t
respectively. Where X and Y might show population number or the size of the respective population. Any
change in the size of population with reference to time or based is described by the time derivatives;
dtdXX /
and dtdYY /
respectively. The general model of interacting populations is usually put in
terms of independent differential equations presented by
),,( yxXgX
(1)
).,( yxYhY
(2)
For instance, the time ' 't does not seem clearly in the functions of ( , ),Xg x y and ( , )Yh x y and the
functions of g and h show the respective per capita growth rate of the two species. One assumption that
comes to us about this is as follows:
( , )dg x y
dy
and
( , )dh x y
dx
The general name for this model is the predator-prey model of Kolmogorov (Freedman, 1980).
3 The Historical Model of Lotke-Volterra
This model was presented by Vito-Volterra and Alfred J. Lotka. Vito, who studied the population of different
species of fish in the Adriatic sea (Voltera, 1926). He saw increasing number of predators and decreasing
number of prey in the Adriatic fauna. Thus, he observed on the three fishing markets. He gave his assessment
that reducing number of fish (prey) was due to the hostile interaction between predators and prey. Alfred J.
Lotka and Volterra presented many mathematical formulae in order to explain the Vito’s observation in the
form of mathematics (Kingsland, 1985). Lotka in his book provided predator-prey systems which consisted of
a plane population and herbivorous animals which dependent on the plants for food (Lotka, 1958).
4 Lotka-Volterra Model
This model is also called the predator prey equations, which is the non-linear, pair of first order differential
equations, given as
XYaXaX 21
(3)
21 YbXYbY
(4)
where 1 2 1 2, , ,a a b b are parameters representing the interaction of the two species.
This model was the first ever mathematical model on the population which achieved a balanced in a
11
Selforganizology, 2016, 3(1): 10-15
IAEES www.iaees.org
population in a very cyclical way. This model shows the eco-logical systems in which the predator and prey
interact resulting in the competition between the two (Hoppensteadt, 1977). This model has following
important variables;
X : The prey-population
Y : The predators-population
1a : The natural growth rate of prey in the predation absence
2a : The predator’s effect on the prey-population
1b : The prey’s effect on the predator-population
2b : The natural predator’s death rate in the absence of prey to eat.
The following are the important Assumptions:
1) If a predator is absent, the population of the prey will increase at the rate proportional to the
current population.
Therefore, 1
dXa X
dt , when 0Y .
2) If the prey is absent, then there are no chances for predator to live.
Thus 2
dYb Y
dt , where 2 0b
when 0X .
3) The number of encounters between a predator and the prey is proportional to the product of
their populations. Every encounter will increase the population of the predator and reduce the
population of the prey.
Thus the growth rate of predator increases in the form of XY , if the growth rate of prey decreases in the form
of 2a XY , where 2a is positive constant, and 1 1 2, ,a b b all are positive constants. 1a
and 2b show the
growth rate of prey and the death rate of predator, respectively. 2a
and 1b measure the level of effect of
interaction between the prey and predator.
5 The Model Equations and Their Physical Meanings
This model presents some assumptions as to how predator and prey populations operate/interact, and evolve or
disappear.
1) The prey population has always more than required food.
2) Food supply for predator depends on how much size the prey population has.
3) The rate of change of population remains proportional to its size, in this process environment becomes
unhelpful for any species.
12
Selforganizology, 2016, 3(1): 10-15
IAEES www.iaees.org
4) The predators have un-limited appetite and this process is cyclical. The generation of both species continues
to interact this way (Cooke et al., 1981).
6 The Predator and Prey Interaction
The earth inhabits millions of such species that assume the role of predators. They kill other herbivores or even
the carnivore predators in order to achieve the survival for them. The predator prey interaction results in the
continuation of life for some and elimination for the others (Begon et al., 1996). Charles Darwin’s theory of
survival of the fittest appears to be very relevant in the life and elimination of different species. The predators
may have herbivores as their target, or may be, have even small predators as targets, thus giving us the
glimpses of an ever insecure world for smaller species. This piece of work will focus on following important
points:
7 Defining a Predator
A predator is one which kills and eats other living beings in order to live. It happens through the physical
outburst of predators that aggressively attack the prey and win life for themselves. In fact, the predators are
said to be killing which is an effort to guarantee life for them (Begon et al., 1996).
8 Classification
Predators have a variety of types, and some of them are;
i. Some predators may target large-size preys and may not directly eat them but they dismember or chew them
before eating them.
ii. Some predators may eat their prey directly as a whole not in parts. Some examples of this include dolphin
which swallows a fish or a snake that devours a frog.
iii. Some predator animals may swallow prey in a whole or in parts depending on the situation.
iv. Some predators kill some other animals or insects but may not eat them. This way, classification goes a
long way in which different predators may choose to target or kill the preys differently.
9 The Process of Predation
The process of killing preys may take great deal of caution and careful step-wise planning by predator in order
to ensure that prey does not escape. The predators may hunt preys actively in which preys may be chased. The
chasing of prey by predator is a race for life for both in which the prey fears life and predators does not want to
let the target escape in order to win security against hunger.
Some predators may sit inactive, waiting for the prey to come close to them. They may act in such a way
that preys are led to believe that the predator is uninterested in them. This caution plan by predator may be a
great success; however, some preys may hesitate to come near to the predator as they are more cautions and
have less trust in the apparently good intentions of predators. This hack of trust by prey may save its life and
may shatter the confidence of predator in its tricky plans forcing active chasing (Alcock, 1998).
10 Why Predation?
Why animals so for predation? It is a very important question. The answer to that is present in the heart of the
argument; is the survival of these animals. The carnivorous animals, faced by ever-increasing hunger, have
found a natural principle for their survival in the middle of rich wildlife that serves as the tasty food for them.
The other besides life is the increasing competition between different species. Some species, although with
13
Selforganizology, 2016, 3(1): 10-15
IAEES www.iaees.org
greatest influence in the wildlife, may confront a threat to them should their population shrink. To this effect,
they embark on ruthless killings to make sure that other species have got small number around them. This
small number of other species may be considered synonymous with security of these predator species.
11 What Happens When Predator Does Not Take Place
If the predation does not take place, there may be two serious consequences;
1) Absence of predation would mean massive increase in the population of these species.
2) The second important consequence is that if predation does not take place, the herbivores increase in
population and finally, it affects the grass and green plants which will have more pressure on it.
12 Different Predators in Co-operation
One important characteristic of the predation process is that predators co-operate with each other against preys.
The co-operation becomes more visible when the prey species are strong enough to be over-powered easily
(Lloyd, 1965).
Moreover, the predators may confront two problems. The first problem may about the large size of prey.
The large size may undermine predator’s ability to successfully target the prey (Molles, 2002). The second
issue facing predators is about the tough terrain. The tough terrain may be the movement of predators and
preys affect understanding of location may benefit it. The speed of predator may be affected. They may also
get injured while chasing the prey (Ripple and Robert, 2004).
13 Recent Developments
We have introduced the quasi chemical approach to represent the different types of mechanisms between the
predator and its prey by taking a case study of foxes and rabbits and determined the modeled equations for
these mechanisms including the mechanism of circulation, mechanisms of attraction and repulsion and
mechanism of sharing place (Shakil et al., 2014)
In our second serried of papers, we developed the modeled equations for different types of mechanism of
the predator-prey interactions with the help of a quasi chemical approach while taking a special study case of
foxes and rabbits, these mechanisms include autocatalysis mechanism, pair wise interactions and the
mechanism of their movements to some free places (Shakil et al., 2015). The chemical reactions representing
the interactions obey the mass action law. The territorial animal like fox is assigned a simple cell as its territory.
Under the proper relations between coefficients, this system may demonstrate globally stable dynamics.
14 Future Plan
We aim to build up a concise study of complex biological systems and to tackle some key research questions
concerning our study case by proposing some new techniques and algorithms that are inspired by those complex
biological systems. After the developments of the models with the help of our quasi-chemical approach, we aim
to check the stability of all the mechanisms discussed in our study case (Shakil et al., 2015). We will then find
the solution of the mechanism in relation to the existence and uniqueness of those solutions.
References
Alcock J. 1998. Animal Behavior: An Evolutionary Approach. Sunderland, Sinauer Associates, Mass, USA
Begon MC, Townsend, Harper J. 1996. Ecology: Individuals, populations and communities. Blackwell
Science. London, UK
14
Selforganizology, 2016, 3(1): 10-15
IAEES www.iaees.org
Cooke D, Hiorns RW. 1981. The Mathematical Theory of the Dynamics of Biological Populations II.
Academic Press Inc, USA
Freedman HI. 1980. Deterministic Mathematical Models in Population Ecology. Monographs and Textbooks
in Pure and Applied Mathematics Vol 57. Marcel Dekker Inc, New York, USA
Hoppensteadt FC. 1977. Mathematical Methods of Population Biology. Courant Institute of Mathematical
Sciences New York University, New York, USA
Kingsland SE. 1985. Modeling Nature, Science and Its Conceptual Foundations. University of Chicago Press,
Chicago, USA
Lotka J. 1958. Elements of Mathematical Biology (Formerly published under the title Elements of Physical
Biology), Dover Publications Inc. New York, USA
Lloyd JE. 1965. Aggressive mimicry in Photuris: Firefly Femmes Fatales. Science, 149(3684): 653-654
Molles Manuel C. Jr. 2002. Ecology: Concepts and Applications. The McGraw-Hill Companies, New York,
USA
Shakil M Wahab HA, Naeem M, Bhatti S. 2014. A quasi chemical approach for the modeling predator-prey
interactions. Network Biology, 4(3): 130-150
Shakil M, Wahab HA, Naeem M, Bhatti S, Shahzad M. 2015. Modeling of the predator-prey interactions.
Network Biology, 5(2): 71-81
Shakil M, Wahab HA, Naeem M, Bhatti S, Shahzad M. 2015. The predator-prey models for the mechanism
of autocatalysis, pair wise interactions and movements to free places. Network Biology, 5(4): 169-179
Voltera V. 1926. Variations and Fluctuations of the Number of the Individuals in Animal Species Living
Together. In: Animal Ecology. McGraw-Hill, New York, USA
Wahab HA, Shakil M, Khan T, Bhatti S, Naeem M. 2013. A comparative study of a system of Lotka-Voltera
type of PDEs through perturbation methods. Computational Ecology and Software. 3(4): 110-125
Zhang WJ. 2012. Computational Ecology: Graphs, Networks and Agent-based Modeling. World Scientific,
Singapore
Zhang WJ, Liu GH. 2015. Coevolution: A synergy in biology and ecology. Selforganizology, 2(2): 35-38
15
Selforgani
IAEES
Article
Validi
in the
Syed Uz1Departme2Departme
E-mail: ssu
Received 1
Abstract
As we kn
way to a
approach
design. In
we will u
last we w
same des
Keyword
1 Introd
Accordin
definition
reasonab
irregular
pattern th
to detect
next pha
refactorin
flaws are
2011; Ta
over all
that is mo
SelforgaISSN 241URL: httRSS: httpEmail: sEditorinPublishe
izology, 2016, 3
ty proce
e design
air Ahmad1,ent of Informati
ent of Informati
uab.kk@gmail.
19 November 20
t
now that in th
address flaw
h has been us
n this paper w
use model bas
will validate t
sign.
ds validation
duction
ng Jennifer C
n by (Jane H
ble representa
or no associ
hat may affec
and remove
ases of the sy
ng to refine
e very difficu
ahvildari and
it is good to
odel has been
anizology 100080 p://www.iaees.op://www.iaees.oselforganizologynChief: WenJuner: International
3(1): 16-24
ess for re
model
, Muhammadion Technology
ion Technology
.com, Naeem@
015; Accepted 2
he start level
through refa
sed to refacto
we are going
sed on coupli
the model, the
; flaw; coupli
Campbell win
Hillston Septe
ation of the a
iation with ea
ct the quality
the flaws tha
ystem. When
the design f
ult to handle
Kontogianni
pick out flaw
n disturbed or
org/publicationorg/[email protected] n Zhang l Academy of Eco
efactored
d Naeem2 y, Hazara Unive
y, Abbottabad U
@hu.edu.pk
22 December 2
l, it is better t
actoring. All
or the non di
to address th
ing defined m
e validation n
ing; bad fixes
nter 2007 va
ember 19, 20
actual system
ach other (Ob
of software i
at may exist in
n a design fla
from that flaw
them all and
is, 2003; Men
w and refacto
r not.
ns/journals/selfos/journals/selfo
ology and Enviro
d couplin
ersity, Mansehr
University of Sc
2015; Published
to pick out fl
of technique
spatchable fl
he missing po
model, second
needs because
s; refactoring
lidation mea
03) Validatio
m “A design f
bjecteeing, 2
is called a de
n the design o
aw is found,
w (Yaowarat
d harder to u
ns et al., 200
or it as possib
organizology/onorganizology/rss
onmental Scienc
ng based
ra, Pakistan
cience and Tech
d online 1 Marc
law and refac
es cover the
law based on
oint as for fou
d will refactor
e to ensure th
g; dispatch.
ns “Are we
on is the task
flaw is define
015; Seidewi
esign flaw (M
of a system as
normally re
ttanaprasert a
understand, m
04; Bacchelli,
ble, and for c
nlineversion.asps.xml
ces
d non-dis
hnology, Abbot
ch 2016
ctored it, as s
area of refa
n coupling an
unded in prev
red it, third w
hat the model
building the
k of demonstr
ed as the cla
itz, 2003). In
Mekruksavanic
s early as pos
esearchers app
and Muencha
maintain the s
, 2010; Saxen
checking vali
p
w
spatchab
ttabad, Pakistan
soon as possib
actoring or v
nd validate th
vious works. F
will check bad
l has changed
right system
rating that th
sses or objec
n other words
ch, 2011). Th
ssible to do no
ply some tec
aisri, 2013).
system (Mek
na and Kuma
idate that mo
www.iaees.org
ble flaws
n
ble. It is best
validation, no
hat refactored
For example,
d fixes and at
d design or in
m?” the other
he model is a
cts having an
s, any design
his is advised
ot disturb the
chniques like
Later on the
kruksavanich,
ar, 2012). So
del to ensure
s
t
o
d
,
t
n
r
a
n
n
d
e
e
e
,
o
e
Selforganizology, 2016, 3(1): 16-24
IAEES www.iaees.org
2 Back Ground
Jane Hillston (September 19, 2003) define validation as: “Validation is the task of demonstrating that the
model is a reasonable representation of the actual system”. Validation concerned with building the right model.
It is utilized to determine that a model is an accurate representation of the real system. Validation is usually
achieved through the calibration of the model, an iterative process of comparing the model to actual system
behavior and using the discrepancies between the two, and the insights gained, to improve the model. This
process is repeated until model accuracy is judged to be acceptable. A model is usually developed to analyze a
particular problem and may therefore represent different parts of the system at different levels of abstraction.
As a result, the model may have different levels of validity for different parts of the system across the full
spectrum of system behaviour For most models there are three separate aspects which should be considered
during model validation which are Assumptions, input parameter values and distributions and Output values
and conclusions.
There are three validation models or strategies for validating data:
- Rejecting bad data: creating a set of undesirable data and rejecting them. This model is also known as
“blacklist” approach.
- Accepting only known good data: data constrained by Five Primary Security Input Validation Attributes
which are: type, length, character set, format, reasonableness. Data is rejected unless it matches for known
good data. This model is also known as “white list” approach.
- Sanitizing data: sanitizing a defined set of dangerous data so that it does not pose a threat to the software
(PedramHayati 2008). It is important to know that how we can apply validation process to our model. Before
validation we will use metric to find different class and object number in model. Detail of metric use and
types are under:
Definition1: MetricModel
Model metrics are for estimating the size or the amount of information contained in a model.
We can use metrics according to pointed situation. Each different object and class relation has different metric,
as given below.
Table 1 Software metrics for UML models (Abbreviation UML Metric).
CBC Coupling between classes DIT Depth of inheritance tree NACM Number of actors in a model NACU Number of actors associated with a use case NAGM Number of the aggregations in a model NASC Number of the associations linked to a class NASM Number of the associations in a model NATC1 Number of the attributes in a class - unweight NATC2 Number of the attributes in a class - weighted NCM Number of the classes in a model NDM Number of the directly dispatched messages of a message NDM* Number of the elements in the transitive closure of the directly dispatched messages of a message NIM Number of the inheritance relations in a model NMM Number of the messages in a model NMRC Number of messages received by the instantiated objects of a class NMSC Number of messages sent by the instantiated objects of a class NMU Number of messages associated with a use case NOM Number of the objects in a model NOPC1 Number of the operations in a class - unweight NOPC2 Number of the operations in a class - weighted
17
Selforganizology, 2016, 3(1): 16-24
IAEES www.iaees.org
NPM Number of the packages in a model NSCU Number of system classes associated with a use case NSUBC Number of the subclasses of a class NSUBC* Number of the elements in the transitive closure of the subclasses of a class NSUPC Number of the super classes of a class NSUPC* Number of the elements in the transitive closure of the super classes of a class NUM Number of the use cases in a model
3 Model Metrics
1. Number of the packages in a model (NPM): This metric counts the number of packages in a model. Package
is a way of managing closely related modeling elements together. Also by using packages, naming conflicts
can be avoided.
2. Number of the classes in a model (NCM): A class in a model is an instance of the meta class “class”. This
metric counts the number of classes in a model. This metric is comparable to the traditional LOC (lines of code)
or a more advances McCabe’s cyclomatic complexity (MVG) metric for estimating the size of a system [7].
Thus, in OOP this metric can be used to compare sizes of systems.
3. Number of actors in a model (NAM): According to the UML specification [10], an actor is a special class
whose stereotype is “Actor”. This metric computes the number of actors in a model.
4. Number of the use cases in a model (NUM): The rationale behind the inclusion of this metric is that a use
case represents a coherent unit of functionality provided by a system, a subsystem, or a class.
5. Number of the objects in a model (NOM): In a similar manner that a class is an instance of the metaclass
“Class”, an object is an instance of a class.
6. Number of the messages in a model (NMM): A message is an instance of the metaclass “Message”.
Messages are exchanged between objects manifesting various interactions.
7. Number of the associations in a model (NASM): An association is a connection, or a link, between classes.
This metric is useful for estimating the scale of relationships between classes.
8. Number of the aggregations in a model (NAGM): An aggregation is a special form of association that
specifies a whole-part relationship between the aggregate (whole) and a component part.
9. Number of the inheritance relations in a model (NIM): This metric counts the number of generalization
relationships between classes existing in a model.
4 Related Work
Moha (2007) provided a systematic method to specify design defects accurately. Their approach is based on
detection and correction algorithms by using refactoring semi-automatically. To apply and validate these
algorithms on open-source object-oriented programs was used to show that method allows the systematic
description, detection, and correction of design defects with a reasonable precision.
Mekruksavanich (2011) proposed a methodology for detection of design flaws. Symbolic logic
representation and analytical learning technique are used to diagnose design flaws in simple way and to
extrapolate patterned rules for complex flaws. The methodology is validated by detecting design flaws in an
open-source system.
Saxena and Kuma (2012) helped to find the flaw in the design model and to remove it as early as possible.
They used the flaw pattern for finding the flaw. When design flaw is detected based in the design pattern, the
process exits after dispatching that flaw, the proposed approach was composed of model representation of
design model and flaws detection using flaw patterns. The design models of UML class and sequence
18
Selforganizology, 2016, 3(1): 16-24
IAEES www.iaees.org
diagrams were used as an input. It would be transformed to the proposed representation model. In detecting
flaws, flaw patterns are used in checking against the representation model. This study covered flaw patterns for
detecting Large Class, Refused Bequest, and Middle Man.
In Mohamed et al. (2011), authors used the approach of automatic flaw detection in design model. To find
the number of flaw number of classes and detection. Which was based on model qualities metrics and design
flaws, author suggest a new demarche allowing the mechanized finding of model refactoring opportunities and
the assisted model restructuration. Which focused on class and sequence diagrams. That developed a software
call’s M-Refactor for those works.
According to Trifu et al. (2004) authors used the flaw detection and correction. The process as problem
detected, developers obtain a list of design flaws together with their location in the system. The necessary
transformations that removed them were left to their own judgment and experience. The mapping between
specific design flaw and code transformations is removed.
In Kessentini (2011) authors used an approach to detect the flaw in design and correct the flaw in the
source code. Their approach support automatic generation of rules to detect defects by the help of genetic
programming. Using a genetic algorithm, adjustment solutions are found by combining refactoring operations
in such a way to reduce the number of detected defects. The detection system is physically specified. Projected
corrections fix, in standard, more than 74% of detected defects.
Alikacem and Sahraoui (2010) provide support for source code analysis. They proposed a rule-based
approach that allowed the specification and detection of flaws. The approach provided a new language to
describe flaws as sets of rules. The latter are translated into Jess’s rule format, and given as input to Jess
inference engine. The current work is an extension of our source code analysis platform and PatOIS, a metric
description language. A main advantage of his approach was its extensibility since the tool is not limited to a
set of predefined flaws. Existing flaws could be modified to a specific context and new ones could be added.
Budi et al. (2011) provided a framework that automatically labels classes as Boundary, Control, or Entity,
and detects design flaws of the rules associated with each stereotype. Their evaluation with programs
developed by both novice and expert developers show that his technique is able to detect many design flaws
accurately.
The main theme of authors in the paper is to find flaw through metric base and convert it into code may in
Java or C++. They defined such detection strategies for capturing around ten important flaws of object-
oriented design found in the literature and validated the approach experimentally on multiple large-scale case-
studies (Marinescu, 2004).
Marinescu (2003) focused on flaw detection through metric base and converted into object-oriented
system.
This paper presented a metrics-based approach for detecting design problems, which describes two
concrete techniques for the detection of two well-known design flaws found in the literature. By an experiment
it was showed that the proposed technique found indeed real flaws in the system and it suggests that, based on
the same approach.
Moha et al. (2008) used an approach propose a novel approach for defect removal in object-oriented
programs that combines the efficiency of metrics with the theoretical strength of formal concept analysis
Algorithm. They suggested a novel approach for defect deduction in object-oriented programs that combines
the usefulness of metrics with the hypothetical power of formal concept analysis, and case study of an exact
fault.
Simon et al. (2006) have worked for finding bad smells. With four typical refactoring’s and present both a
tool supporting the identification and case studies of its application. They showed that special kind of metrics
19
Selforganizology, 2016, 3(1): 16-24
IAEES www.iaees.org
can support these skewed perceptions and thus can be used as effective and efficient way to get support for the
decision where to apply which refactoring. They demonstrate this loom for four typical refactoring’s and
present both a tool supporting the classification and case studies of its function.
Syed et al. (2015) worked on Refactoring of non-dispatchable flaws in the design model based on
coupling. In this work he cover all the aspects which was blank in the above all works. But one thing which is
validation still remained in his work.
The entire above techniques draw backs have been covered by Syed el al. but, one thing still remains that
is: When the process has been done of dispatch and non-dispatchable flaw, how we will find that our model
has been in required position. For this we are going to use an approach to validate our model after passing
through the process as under.
5 Our Approach
The previous work was divided into six steps, now we are going to fix the previous error in approach. For this
we will add another one step in the existing approach. So the new step will be validation of model.
Step 1: Domain analysis and metrics identification
Step 2: Modeling and meta modeling
Step 3: Flaw detection and flaw pattern
Step 4: Option if flaw >=1 or no flaw found
Step 5: Condition check for dispatch and non-dispatchable flaw
Step 6: validation check
The systematic view of our approach is
Step 1 (Domain Analysis): We will do the analysis of the domain area, in this step we associate manually
with each design defect to detect them and set refactoring by using metric based identification to find class and
their association through coupling.
Step 2 (Modeling): In modeling, the Meta model for software modeling is important, because it forms the
basis for the UML definition. The UML specification document is indeed a Meta model for UML. That is, it
includes a set of statements that must not be false for any valid UML model. In the metric forms that shows
different sort of associations.
Step 3: (Flaw Detection): In this step, we find the flaws by using flaw detection patterns. Basically, a pattern
is a format which will identify a flaw in the model. We use coupling for the detection of flaws. Coupling
measures the strength of all relationships between functional units.
Step 4: (Flaw does exist or not): Using the above formula, if flaw exists then to be refactored or dispatched, if
not then exit. When the flaw =0, it will exit else if flaw≥1then condition shall be checked. That either flaw is
dispatchable or non-dispatchable, if dispatchable go to dispatch-able module otherwise non-dispatchable and
go for refactoring to refactor module.
Step 5: (Flaw checking of dispatch-able or non-dispatch-able): Condition checking whether the flaw is
dispatch able or non-dispatch able. The dispatch able flaw goes to dispatch able flaw module and the non-
dispatch able flaw goes to the non-dispatch able flaw module. First condition is to check that if flaw is dispatch
able, the flaw goes to the dispatch able module and removed there. The second condition, If flaw is non-
dispatch able and been removed through refactoring a tag Ref; attached to the refactored flaw as a comment for
the detected flaw module to understand that this flaw has refactored and didn’t need to catch it again. Both
from dispatch and non-dispatch able flaw modules the model goes again for rechecking flaws to the detection
flaw module. The cycle continues until all flaws are dispatched or refactored.
20
IAEES
Step 6:
dispatcha
for this c
2, where
step have
associatio
The
6 Class D
The clas
Graphics
Primitive
coordina
This
they mus
design p
recursive
the follow
(validation
able flaws. B
checking we w
we made me
e similar asso
ons.
e whole system
Diagram Exa
s diagram giv
s are constitu
es have a m
ate system.
s model has s
st be treated
pattern addres
ely grouped i
wing steps, le
of model):
ut we don’t k
will validate
etrics. Simply
ociations so m
m over view
ample
ven in Fig. 1
uted of geom
matrix attribu
some design
differently,
sses this type
in a part-who
eading to the
Selforga
here the ma
know that the
our model. T
y we check a
model will go
is indicated i
1 is a simple
metric Primi
ute representi
flaws; for ins
thus making
e of problem
ole hierarchy.
diagram pres
anizology, 2016
ajor work tak
e refactored m
The last step o
all the associa
o to exit state
n Fig.1.
model of a g
tives and su
ing how they
stance, as Prim
g the code un
m, where a st
. We will the
sented in Fig.
Fig. 1 Our app
6, 3(1): 16-24
kes palace, a
model has bee
of validation t
ations and co
e else model
graphical hie
ub graphs; th
y are scaled
mitives have
nnecessarily
tructure is co
erefore introd
2.
proach.
as we refacto
en changed fr
take place wi
ompare last st
will be rearra
erarchy for a
hey have a
d, rotated or
no inheritanc
complex. Fo
omposed of b
duce this patt
or the dispat
rom original
ith the compa
tep with step
anged accord
vector graph
method to b
translated in
ce relation w
ortunately, th
basic objects
tern in the m
www.iaees.org
tch and non-
shape or not,
arison of step
2, if both of
ding to step 2
hics program.
be displayed.
n the global
with Graphics,
he Composite
s that can be
model through
g
-
,
p
f
2
.
.
l
,
e
e
h
21
IAEES
1. Renam
2. Addin
3. Makin
4. Mergin
5. Finally
We
Renamin
(i.e. it do
The
no effect
Cre
to multip
empty.
Mer
same obj
invocatio
Fina
not mode
Whi
have an
object di
ming the Gra
ng an abstract
ng the class Pr
ng the Group
y, we can mo
need to justif
ng of a model
oes not alread
e added abstra
t on the mode
ating a gener
ple inheritanc
rging two ass
jects), when
on through an
ally, moved m
eled).
ile most of th
impact on ot
agrams (Gers
aphic class to
superclass na
rimitive a sub
p-Group and G
ove relevant m
fy why the be
l element doe
dy exist in the
act superclass
el.
ralization betw
ce, for instan
sociations is
the methods
n association
methods or a
hese transform
ther views, th
sonSuny, 200
Selforga
Group;
amed Graphi
bclass of Gra
Group-Primit
methods and a
ehavior prese
es not change
e model).
s has no attrib
ween two cla
nce) is introd
only allowed
invoked thro
is always foll
attributes to th
mations - nam
he merging o
01).
F
Fig. 3 Re
anizology, 2016
c to Group.
aphic.
tive aggregati
attributes up t
ervation cond
anything to t
butes or meth
asses does not
duced; in our
d when these
ough these as
lowed by an i
he superclass
mely element
of two associ
Fig. 2 Initial cla
estructured clas
6, 3(1): 16-24
ions into Gro
to Graphic.
ition holds fo
the model beh
hods. It is a \e
t introduce ne
case, Primit
two associat
ssociations h
invocation th
s will simply
t renaming an
iations may r
ss diagram.
s diagram.
oup-Graphic.
or these mode
havior; provi
empty" mode
ew behavior,
tive had no s
tions are disjo
ave the same
hrough the oth
be inherited
nd the additio
require chang
el transformat
ded the new n
l element; its
provided no
superclass an
oint (they do
e signature, a
her.
afterwards (
on of a superc
ges on collab
www.iaees.org
tions:
name is legal
s addition has
conflict (due
nd Graphic is
o not own the
and when the
(overriding is
class - do not
borations and
g
l
s
e
s
e
e
s
t
d
22
Selforganizology, 2016, 3(1): 16-24
IAEES www.iaees.org
7 Conclusion
As for performing this sort of process we will be able to refactor all the dispatch able and non-dispatchable
flaws, we will also be able to re-structure the model if the model has been misplaced. The model simply makes
meta models using metrics and the store that metric code. After performing all the operations for validation the
model again comes to modeling state. Here the previous and new model metric compared and find the
differences between them. If the model metrics same validate and exit, or else it restructure the model and go
to exit state.
References
Ahmad SU, Naeem M. 2015. Refactoring of non-dispatchable flaws in the design model based on coupling.
Selforganizology, 2015, 2(3): 46-54
Alikacem H, Sahraoui HA. 2010. Rule-Based System for Flaw Specification and Detection in Object-Oriented
Programs. 13th TOOLS Workshop on Quantitative Approaches in Object-Oriented Software Engineering.
1-11
Bacchelli, Lanza M. 2010. On the Impact of Design Flaws on Software Defects.IEEE 10th International
Conference on Quality Software (QSIC), 23-31 Budi A, Lucia, Lo D, Jiang L, Wang S. 2011. Automated
Detection of Likely Design Flaws in Layered Architectures, Research Collection School of Information
Systems, 1-6
Fowler M, Beck K. 1999. Refactoring: Improving the Design of Existing Code. Addison-Wesley Professional,
USA
Kessentini M, Kessentini W, Erradi A. 2011. Example-based Design Defects Detection and Correction.IEEE
19th International Conference Program Comprehension (ICPC). 81-90
Marinescu R. 2001. Detecting Design Flaws via Metrics in Object-Oriented Systems. 39th International
Conference and Exhibition on Technology of Object-Oriented Languages and Systems (TOOLS), 173-182
Marinescu R. 2004. Detection Strategies: Metrics-Based Rules for Detecting Design Flaws. International
IEEE20th Conference on Software Maintenance. 350-359
Mekruksavanich S. 2011. Design flaws detection in object-oriented software with analytical learning method.
International Journal of e-Education, e-Business, e-Management and e-Learning, 1(3): 210-216
Mens T, Tourw T. 2004. A survey of software refactoring. IEEE Transactions on Software Engineering, 30(2):
126-139
Moha N. 2007. Detection and Correction of Design Defects in Object-Oriented Architectures. Companion to
the 22nd ACM Special Interest Group ACM SIGPLAN Conference, 949-950
Moha N, Rezgui J, Gueheneuc Y, Valtchev P, Boussaidi G. 2008. Using FCA to Suggest Refactorings to
Correct Design Defects. 4th International Conference on Concept Lattices and Their Applications. LNCS
4923, 269-275
Mohamed M, Romdhani M, Ghedira K. 2011. M-REFACTOR: A new approach and tool for model refactoring.
American Resources Policy Network (ARPN). Journal of Systems and Software, 1(4): 1-6
Objecteeing. 2015. http://support.objecteering.com
Saxena V, Kumar S. 2012. Impact of coupling and cohesion in object-oriented technology. Journal of Software
Engineering and Applications, 671-676
Seidewitz E. 2003. What Models Mean. IEEE Software. IEEE Computer Society Press Los Alamitos, CA,
USA, 20(5): 26-32
23
Selforganizology, 2016, 3(1): 16-24
IAEES www.iaees.org
Simon F, Steinbrückner F, Lewerentz C. 2001. Metrics Based Refactoring. IEEE 5th European Conference on
Software Maintenance and Reengineering. 30-38
Sunyé, Gerson, Pollet D, Traon Y, Jézéquel J. 2001. Refactoring UML Models. UML2001—TheUnified
Modeling Language. Modeling Languages, Concepts and Tools. 134-148, Springer, Berlin, Heidelberg,
Germany
Tahvildari L, Kontogiannis K. 2003. Metric-Based Approach to Enhance Design Quality ThroughMeta-Pattern
Transformations. Proceedings of 7th European Conference on Software Maintenance and Reengineering.
183-192
Trifu A, Seng O, Genssler T. 2004. Automating Design Flaw Correction in Object-Oriented Systems.
Proceedings of 8th European Conference on Software Maintenance and Reengineering (CSMR 2004).174-
183
Yaowarattanaprasert N, Muenchaisri P. 2013. Graphical pattern matching approach for detecting design flaw
in design model. International Journal of Advanced Research in Computer Science and Software, 2: 1-5
24
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
Article
Non-traditional approach to fitting of time series of larch bud moth
dynamics: Application of Moran – Ricker model with time lags
L.V. Nedorezov Center of Interdisciplinary Investigations of Environmental Problems, Russian Academy of Sciences, nab. Kutuzova 14, Saint-Petersburg, 191187, Russia E-mail: [email protected]
Received 28 October 2015; Accepted 5 December 2015; Published online 1 March 2016
Abstract
In current publication analyses of time series of larch bud moth (Zeiraphera diniana Gn.) dynamics are
considered. For fitting of time series Moran - Ricker model with time lags was used. Estimations of model
parameters were provided with non-traditional approach: for every considered case feasible sets of points in
space of model parameters were determined where selected statistical criterions give required results for
deviations between theoretical (model) and empirical datasets. In all considered situations obtained results
were compared with other results obtained with least squared method. It was obtained that Moran – Ricker
model without time lag and with time lag in one year is not suitable for fitting of time series. Best
correspondences on quantitative and qualitative levels between model trajectories and empirical dataset were
found for cases with time lag in 3 years.
Keywords larch bud moth dynamics; time series; fitting; Moran - Ricker model with time lag.
1 Introduction
For estimation of ecological model parameters various methods and approaches can be used – from expert
estimations up to Bayesian approach (Bard, 1974; Borovkov, 1984; Draper and Smith, 1981; Wood, 2001 a, b
and others). For one and the same model which is used for fitting of one and the same time series we can get
various estimations for model parameters: expert estimations, estimations obtained with least square method,
obtained with method of maximum likelihood etc. Naturally, in some situations question about better
estimations does not arise, and in some situations question about correspondence of model and time series
does not arise too.
Let us assume that parameters of following discrete-time model must be estimated:
Selforganizology ISSN 24100080 URL: http://www.iaees.org/publications/journals/selforganizology/onlineversion.asp RSS: http://www.iaees.org/publications/journals/selforganizology/rss.xml Email: [email protected] EditorinChief: WenJun Zhang Publisher: International Academy of Ecology and Environmental Sciences
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
),(1 kk xFx . (1)
In (1) kx is population size at moment k , ...2,1,0k ; is vector of model parameters; F is
non-negative function (for non-negative values kx and admissible values of parameters). Let us additionally
to assume that at initial time moment 0k population size is equal to 0x and it is unknown amount which
must be estimated too.
Let }{ *kx , Nk ,...,1,0 , be empirical time series of population size changing in time; 1N is
sample size. Using this sample }{ *kx we have to estimate model parameters and initial population size
0x .
Use of least square method (LSM) (Bard, 1974; Borovkov, 1984) is based on assumption that best
estimations of model parameters can be found with minimizing of sum of squared deviations between
theoretical (model) and empirical datasets. If time series is approximated by model trajectory (global fitting;
Wood, 2001 a, b) loss-function can be presented in the following form:
N
kkk xxxxQ
0
2*00 ),(),( . (2)
In (2) )},({ 0xxk is model (1) trajectory obtained for fixed values of and 0x . Let’s also assume
that for certain point ),( **0
** x there is a global minimum in (2):
N
kkk
xxxxxQ
0
2*0
,
**0
** ),(min),(0
r
. (3)
Following a traditional approach (Bard, 1974; Borovkov, 1984; Draper and Smith, 1981) after
determination of estimations ),( **0
** x analysis of set of deviations }{ ke between theoretical and
empirical datasets must be provided:
***0
** ),( kkk xxxe . (4)
Model (1) is recognized to be suitable for fitting of considering time series if following conditions are
truthful: deviations }{ ke are values of independent stochastic variables with Normal distribution and with
zero average. Following these assumptions Kolmogorov – Smirnov, Lilliefors, Shapiro – Wilk or other tests
are used for checking of Normality of deviations (Bolshev and Smirnov, 1983; Lilliefors, 1967; Shapiro et al.,
1968). For checking of independence of stochastic variables Durbin – Watson and/or Swed – Eisenhart tests
are used (Draper and Smith, 1981; Hollander and Wolfe, 1973; Likes and Laga, 1985).
If in the sequence of residuals (4) serial correlation is observed it gives a background for conclusion that
26
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
considering model isn’t suitable for fitting and needs in modification. It means also that some of important
factors or processes were not taken into account within the framework of model. Similar conclusion about
model and its applicability to fitting can be made in situation when hypothesis about Normality of deviations
must be rejected (for selected significance level). In other words, final conclusion about suitability of model
for approximation of considering time series is based on analysis of properties of unique point ),( **0
** x in
the space of model parameters.
In our opinion, this is one of basic problems of LSM: a’priori it is impossible to exclude from
consideration a situation when nearest to ),( **0
** x points have required properties. One more problem of
LSM is absence of a background for assumption about Normality of deviations. This is very strong condition
and it must be reduced to two other conditions – symmetry of distribution with respect to origin and monotonic
behaviour of branches of density function (Nedorezov, 2012, 2015a, b).
Next problem of LSM is following: there are no criterions for selection of loss-function. This function
may have form (2) (with or without system of weights), it can be a sum of absolute values of deviations etc.
This freedom in selection of loss-function is based on absence of real correlation with biological problem, and
it is based on desire for obtaining of one point only but not set of points. Note that even in a case when model
(1) is a law of population dynamics with natural values of parameters, there are no reasons for assumption that
it must give minimum of any abstract loss-function.
Before finding of any special point in a space of parameters basic criterions for sets of deviations between
theoretical and empirical datasets must be determined. On the next step it will allow determination of a feasible
set – this is a set in the space of model parameters where all statistical criterions demonstrate required results.
After that it is possible to use any loss-functions within the limits of feasible set. It is possible but not
obligatory step – structure of feasible set depends on significance levels, and changing of these levels it will be
possible to find points with extreme properties. These extreme points give best approximation for time series
from the standpoint of selected criterions (but not from standpoint of loss-function).
In current publication this alternative approach is under consideration. Moran – Ricker model with time
lags is applied for fitting of larch bud moth (LBM) time series (Zeiraphera diniana Gn.; Baltensweiler, 1964,
1978). For every considering particular case (which are determined by length of time lags) obtained results
compare with results obtained with LSM and biological imaginations about LBM dynamics.
In our previous publication (Nedorezov and Sadykova, 2015) we applied Moran – Ricker model with
time lags for fitting of larch bud moth time series. But not all basic dynamic regimes were presented in pointed
out publication. In current publication we analyze wider spectrum of dynamic regimes which may have
relation to larch bud moth dynamics.
2 Basic Requirements to Model and Set of Deviations
Mathematical model can be recognized as suitable for approximation of time series if following requirements
are truthful:
1. Deviations between theoretical and empirical datasets must have a symmetric distribution with respect to
origin. Branches of density function must be monotonic curves – it must increase in negative part of straight
line, and it must decrease in right part.
Let }{ ke be a set of positive deviations, and }{ ke is a set of negative deviations (4) with sign minus.
Distribution has symmetry with respect to origin if and only if samples }{ ke and }{ ke have one and the
27
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
same distribution function. It allows using criterions for homogeneity – Kolmogorov – Smirnov test, Lehmann
– Rosenblatt test, and Mann – Whitney test ((Bolshev and Smirov, 1983; Hollander and Wolfe, 1973; Likes
and Laga, 1985). For testing of property of monotonic behavior of branches of density function Spearman rank
correlation coefficient was used.
2. Hypotheses about existence of serial correlation in sequences of residuals must be rejected. For testing of
this property Swed – Eisenhart test (Draper and Smith, 1981) and test “jump up –jump down” (Likes and Laga,
1985) were used below.
3. Behavior of model trajectory must correspond to behavior of time series. If for every increasing of values in
time series model demonstrates decreasing of values there are no backgrounds for conclusion about suitability of model for fitting of considering sample. Thus, quota q of cases when time series and model trajectory
demonstrate different changes must be rather small. In other words, Null hypothesis :0H 5.0q with
alternative hypothesis :1H 5.0q must be rejected.
Pointed out set of statistical criterions was used as for determination of points of feasible sets as for
finding points with strongest properties. For example, if hypothesis about symmetry of distribution cannot be
rejected with 5% significance level it does not mean that Null hypothesis must be accepted (p-value can be
rather small and close to 0.05). But if this hypothesis cannot be rejected with 95% significance level it means
that we have to accept it.
Changing of values of significance levels allows finding small number of points in a space of model
parameters with extreme properties. These points can be used as estimations for parameters. This is the first
basic idea of considering approach. The second idea is following: assuming that every point of feasible set can
be used as estimation of model parameters stochastic point with uniform distribution within the limits of
feasible set will allow obtaining a distribution of possible dynamic regimes. This distribution can also be used
as background for conclusion about population dynamics.
3 Larch Bud Moth Population Dynamics
3.1 Used time series and models
Regular observations of the changing of larch bud moth (Zeiraphera diniana Gn.) population densities in time
in Swiss Alps (Upper Engadine valley) had been started in 1949 (Auer, 1977; Baltensweiler, 1964, 1978). In
current publication time series on larch bud moth dynamics were used which can be free downloaded in
Internet (NERC Centre for Population Biology, Imperial College (1999) The Global Population Dynamics
Database, N 1407). Unit of measurement is number of larva per kilogram of branches. As it was pointed out in
GPDD, data were collected on 1800 m above sea level that corresponds to optimal zone of species living
(Isaev et al., 1984, 2001). Sample contains 38 values (first point corresponds to 1949).
Below we’ll follow one of basic concepts about larch bud moth dynamics: periodicity of population
fluctuations can be explained by influence of time lag in a reaction of self-regulation mechanisms (Isaev et al.,
1984, 2001; Berryman and Stark, 1985; Berryman, 1981, 2002; Sadykova and Nedorezov, 2013; Nedorezov
and Utyupin, 2011). Moran – Ricker model is one of well-known and well-studied models, and it has very rich
set of dynamic regimes (Moran, 1950; Ricker, 1954):
m
jjkjkk xaAxx
01 exp , 3,2,1,0m . (5)
In (5) all parameters are non-negative, k 0, constaA k . Initial values of population sizes must be
28
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
positive, 00 kx , 3,2,1,0k . Note, that it was proved that model (5) with 1m can give good
approximation for some time series of larch bud moth dynamics (McCallum, 2000).
3.2 Moran – Ricker model with 0m
Within the framework of traditional approach following estimations were obtained (Nedorezov, Sadykova,
2015): 7.465435min Q , 44.530 x , 0.517A , 1168.00 a . minQ is minimum value of
functional form (2)-(3). Obtaining values of stochastic points with uniform distribution in
]1,0[]1000,0[]100,0[ we get a possibility to analyze structure of feasible set . In Fig. 1 there is a
projection of feasible set onto plane ),( 0aA (for all criterions one and the same 5% significance level was
used). As one can see in fig. 1 LSM-estimation is far from a domain of maximum of point’s concentration:
705.18 A , 05.00 a approximately. Moreover, LSM-estimation doesn’t belong to .
For LSM-estimations analysis of properties of deviations }{ ke (4) allowed obtaining following results:
hypothesis about Normality of deviations must be rejected with 0.1% significance level (Shapiro – Wilk test).
Probability that distribution of deviations is symmetry is very small: 002507.0p for Wald – Wolfowitz
test and 025.0p for Kolmogorov – Smirnov test. Thus, within the framework of traditional approach we
have to conclude that Moran – Ricker model (5) with 0m is not suitable for fitting of considering time
series.
Fig. 1 Projection of feasible set for Moran – Ricker model without time lag onto plane ),( 0aA . Crest corresponds to point of minimum of functional form (2)-(3).
29
IAEES
Rem
0a it is
to all sta
A and
when we
(absence
Poin
symmetr
significan
Hypothe
significan
value for
branches
Hyp
Eisenhar
29, and v
Fig. 2 Asy5-cycle.
Valu
model b
greater th
Subsets
populatio
mark. Marked
possible to p
atistical criter
0a , and all p
e have small n
of symmetry
nt .5000 x
ry cannot be
nce level (L
sis about eq
nce level (cri
r t-test for co
s of density fu
pothesis abou
rt test) and wi
value of test i
ymptotic stable
ues of autoco
elong to clo
han 0.000715
of this 5-cy
on density, an
d in Fig. 1 po
point out initi
rions describe
possible valu
number of ne
y, existence o
6028. , A
rejected wit
Lehmann – R
quivalence of
itical value fo
onsidering po
unction must
ut absence of
ith 20% signi
s equal to 25
e dynamic regi
orrelation fun
ose interval
5. Thus, asym
ycle 1b ,…,
nd y is a bi
Selforga
oints corresp
ial value 00x
ed above. If p
ues of initial p
egative or po
of serial corre
0773.29 ,
th 99.9996%
Rosenblatt te
f Spearman r
or t-test is equ
oint 0.63913)
be accepted.
f serial correla
ificance level
).
ime for Moran
nction calcul
0,48115.0[
mptotic stable
5b are pre
irth rate, ky
anizology, 2016
ond to situat
which gives
point doesn’t
population siz
ositive deviati
lation etc.).
02.00 a
significance
est), with 96
rank correlat
ual to 0.513 f
. Thus, hypo
ation cannot
l (test “jump u
– Ricker mod
lated for 200
]999827.0 .
e dynamic re
esented on p
kk xx /1 .
6, 3(1): 25-40
tions when fo
s a model traj
t mark it mea
ze one of crit
ions) or when
63 has follo
level (Kolm
6.8% signific
tion coefficie
for sample siz
thesis about
be rejected w
up –jump dow
del with m
00 values of
Every fifth
egime can be
plane (log10
or fixed valu
jectory and se
ans that for fi
erions cannot
n one of crite
owing prope
mogorov – Sm
cance level
ent to 0 mus
ze 38 and 0.1
symmetry an
with 25.4% si
wn”; critical
0 . Sets 1b ,…
f trajectory af
value of au
classified as
)log, 10 yx
w
es of parame
et of deviatio
fixed values o
t be applied (
erions give ne
erties. Hypo
mirnov test),
(Mann – W
st be rejected
1% significan
nd monotonic
ignificance le
levels are equ
…, 5b are elem
fter 6102
utocorrelation
s (close to) fu
in Fig. 2
www.iaees.org
eters A and
ons satisfying
of parameters
(for example,
egative result
thesis about
with 99.7%
Whitney test).
d with o.1%
nce level; real
c behavior of
evel (Swed –
ual to 21 and
ments of fuzzy
free steps of
n function is
uzzy 5-cycle.
where x is
d
g
s
,
t
t
%
.
%
l
f
–
d
y
f
s
.
s
30
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
Let be a stochastic variable with uniform distribution on feasible set which is equal to length of
(asymptotic) cycle which is realized for respective point of the space of model parameters. In such a case
probability }{ kP is equal to quota of respective dynamic regimes among all dynamic regimes which can
be observed for points from feasible set. Distribution of stochastic variable can be as element of
background of final conclusion about population dynamic regime. For considering situation it was obtained
that probability of event }1000{ is equal to 0.935 approximately. Some other probabilities have
following values: 0184.0}4{ P , 0102.0}5{ P , 015.0}6{ P , 0029.0}8{ P
(other probabilities are much smaller). Probability of event }100029{ is equal to zero.
It allows concluding that regimes with strongest properties, most probable regimes, and regimes with
LSM-estimations don’t correspond to biological imaginations about larch bud moth fluctuations (Isaev et al.,
1984, 2001; Auer, 1977; Baltensweiler, 1964, 1978).
3.3 Moran – Ricker model with 1m
Within the framework of traditional approach following results were obtained (Nedorezov and Sadykova,
2015): 4.154148min Q , 1400 10148.4 x , 1274.00
1 x , 3.8A , 30 10205.2 a ,
02214.01 a . For these parameters asymptotic stable regime is 18-cycle with rather close values of cyclic
coordinates (Fig. 3). It looks like double 9-cycle, and in this occasion it corresponds to biological imagination
about larch bud moth dynamics (Auer, 1977; Baltensweiler, 1964, 1978). But analysis of deviations shows that
hypothesis about Normality must be rejected with 1% significance level; serial correlation is also observed in
sequence of residuals (Swed – Eisenhart test).
Within the boundaries of set ]1,0[]1,0[]1000,0[]100,0[]100,0[ 50000 points of feasible
set were found (with the help of stochastic variable with uniform distribution). Probability to find a point
in belonging to is equal to 5103356.5 approximately. Final picture is close to presented in fig. 1
(it is obvious that set for Moran – Ricker model with 0m is a subset for new one). New domains of
point’s concentration on the plane ),( 0aA were not identified.
The following point 705.4400 x , 333.530
1 x , 01.719A , 0318.00 a , 0017.01 a has
best characteristics with respect to symmetry. Value of Kolmogorov – Smirnov test is equal to 0.4425
(hypothesis about symmetry cannot be rejected with 98.7411% significance level); value of Lehmann –
Rosenblatt test is equal to 0.03305 (93.315%); value of Mann – Whitney test is equal to 0.0908 (92.04%). It
means that Null hypothesis about symmetry of distribution of residuals must be accepted.
31
IAEES
Fig
Like
rejected
correlatio
–jump d
significan
Valu
steps of
increasin
Fig. 4
(log10 x
plane (l
. 3 Asymptotic
Fi
e in previous
with 0.1% si
on cannot be
down” is equ
nce level and
ues of autoco
f model belo
ng, fast decrea
set of poin
)log, 10 yx . S
log,log 1010 x
stable dynamic
ig. 4 Asymptoti
case hypothe
ignificance le
rejected with
ual to 21: hy
d must be reje
orrelation fun
ong to close
asing of popu
nts belongin
Similar behav
)0 y were ob
Selforga
c regime (18-cy
ic stable dynam
esis about eq
evel (t-test is
h 67.44% sig
ypothesis abo
ected with 20%
nction calcula
e interval [
ulation size (d
ng to asymp
vior of autoco
bserved in ca
anizology, 2016
ycle) for Moran
mic regime for M
quivalence of
s equal to 0.7
gnificance lev
out absence
% level.
ated for 20000
07.0,033.0
during one st
ptotic stable
orrelation fun
ases when pro
6, 3(1): 25-40
n – Ricker mode
Moran – Ricker
Spearman ra
74483>0.513
vel (Swed – E
of serial cor
00 values of m
]76 . Dynam
tep of model)
e dynamic r
nction and si
obabilities of
el with 1m
model with m
ank correlatio
3). Hypothesi
Eisenhart test
rrelation cann
model traject
mics has irre
) with further
regime is p
imilar structu
f rejection of h
w
1 for LSM-esti
1m .
n coefficient
is about abse
t); value of t
not be reject
tory and after
gular nature
r monotonic i
presented on
ure of set of p
hypotheses a
www.iaees.org
mations.
to 0 must be
ence of serial
est “jump up
ted with 5%
r 200000 free
e: monotonic
increasing. In
n the plane
points on the
about absence
e
l
p
%
e
c
n
e
e
e
32
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
of serial correlation were minimized.
Analysis of points of feasible set shows that cycles of various lengths can correspond to considering time
series: 1-cycle (regime of asymptotic stabilization or rather small fluctuations near stable level with variance
which is less than 10-40), 2,…, 7, 11, 15, 16, 19, 27, 30, 56. 8-cycles and 9-cycles were not found. Cycles with
lengths 16 and 27 don’t look like double or triple 8-cycle or 9-cycle: in these cases dynamic regimes with very
long phases of population increasing were observed.
For considering case 0164.0}1{ P , 0131.0}2{ P . Other probabilities are much less
these values. Biggest probability is equal to 0.9546 for event }1000{ . Thus, in both considered cases
( 0m and 1m ) dynamic regimes don’t correspond to existing imaginations about population dynamics.
3.4 Moran – Ricker model with 2m
Let’s consider situation when time lag is equal to 2. Minimum of squared deviations (3) 1.92099min Q
was observed for the following values of model parameters: 1500 10638.5 x , 160
1 10653.2 x ,
07208.002 x , 076.20A , 3
0 10155.8 a , 31 106994.1 a , 02097.02 a . Analysis of
deviations showed that for Kolmogorov–Smirnov test p < 0.05; for Lilliefors test p < 0.01; for Shapiro – Wilk
test p < 10-5 (Bolshev and Smirnov, 1983; Shapiro et al., 1968; Lilliefors, 1967). Thus, model (5) with
LSM-estimations cannot be used for fitting of empirical time series.
Within the boundaries of set : 1.00 0 kx , 400 A , 03.00 ka , 2,1,0k , – 50000
points of feasible set were found. Probability to find in a point belonging to is equal to
000915.0 approximately.
Point 06564.000 x , 02832.00
1 x , 0211.002 x , 607.27A , 01207.00 a ,
000138.01 a , 01257.02 a of the feasible set has following properties: hypothesis about symmetry of
deviations cannot be rejected with 99.9991% significance level (Kolmogorov – Smirnov test), with 99.7%
(Lehmann – Rosenblatt test), and with 98.81% (Mann – Whitney test). Hypothesis about equivalence of
Spearman rank correlation coefficient to 0 must be rejected with 0.1% significance level (t-test is equal to
0.7947). Hypothesis about absence of serial correlation cannot be rejected with 10.37% significance level
(Swed – Eisenhart test); value of test “jump up – jump down” is equal to 22: hypothesis about absence of serial
correlation cannot be rejected with 20% significance level. This point with extreme properties doesn’t
correspond to biological imaginations about larch bud moth dynamics: asymptotic stable dynamic regime is
strong 7-cycle.
Another point with extreme properties is following: 0098.000 x , 0902.00
1 x , 0585.002 x ,
758.27A , 0174.00 a , 00655.01 a , 0181.02 a . Hypothesis about symmetry of deviations
cannot be rejected with 99.9996% significance level (Kolmogorov – Smirnov test), with 99.7% (Lehmann –
Rosenblatt test), and with 89.31% (Mann – Whitney test). Hypothesis about equivalence of Spearman rank
33
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
correlation coefficient to 0 must be rejected with 0.1% significance level (t-test is equal to 0.8947). Hypothesis
about absence of serial correlation cannot be rejected with 28.37% significance level (Swed – Eisenhart test);
value of test “jump up – jump down” is equal to 20: hypothesis about absence of serial correlation cannot be
rejected with 5% significance level.
This point with extreme properties doesn’t also correspond to biological imaginations about larch bud
moth dynamics: asymptotic stable dynamic regime is non-periodic. Values of autocorrelation function belong
to close interval ]4088.0,2284.0[ . Asymptotic stable regime constructed for 20000 points (after 106 free
steps of model) on the plane )log,(log 1010 yx is presented in Fig. 5a. Approximation of considering time
series by initial part of model trajectory is presented in Fig. 5b. Visual analysis of these pictures shows that we
have not a background for conclusion that in this case Moran – Ricker model gives good fitting. In particular,
number of extreme points of trajectory (Fig. 5b) is much bigger than number of extreme points of time series.
Asymptotic trajectory of model is rather difficult, and we cannot say it has relation to larch bud moth dynamics
or not.
Analysis of points of feasible set shows that cycles of various lengths belong to this set. Like in previous
case biggest probability is equal to 0.9 for event }1000{ . For 1-cycle probability is equal to 0.01634; for
9-cycle 01202.0}9{ P , and for 8-cycle 00892.0}8{ P . In a group of regimes which are close
to 9-cycle, it is possible to point out strong 9-cycles, double 9-cycles (cycles of the length 18), and fuzzy
9-cycles with close coordinates (every ninth step values of autocorrelation function is bigger than 0.999). It
allows concluding that within the boundaries of feasible set it is possible to find dynamic regimes which are
close to biological imaginations about larch bud moth dynamics. But quota of these regimes is rather small.
3.5 Moran – Ricker model with 3m
Let us consider the situation with 3m . Minimum 9.92018min Q of squared deviations (3) was
observed for the following values of model parameters: 1400 1097.4 x , 160
1 1061.1 x ,
1602 102.4 x , 4837.10
3 x , 8439.19A , 0081.00 a , 0018.01 a , 0186.02 a ,
0021.03 a . Analysis of deviations showed that for Kolmogorov–Smirnov test 05.0p ; for Lilliefors
test 01.0p ; for Shapiro– Wilk test 510p (Bolshev and Smirnov, 1983; Shapiro et al., 1968;
Lilliefors, 1967). Thus, Moran – Ricker model with LSM-estimations of model parameters cannot be applied
for fitting of considering time series. Asymptotic stable dynamic regime is not cycling with length of cycle in
1000 years or less.
34
IAEES
Fig. 5 Asyapproxima
Wit
0 ka
point bel
Poin
0.01 a
about sym
test), wit
equivalen
ymptotic stable ation of time ser
thin the bo
02.0 , k
longing to
nt 0.000 x
00541, 2 a
mmetry of d
th 97.432% (
nce of Spearm
dynamic regimries (solid line)
oundaries of
3,2,1,0 , –
is equal to
969 , 01 x
00052.0 ,
deviations can
(Lehmann – R
man rank cor
Selforga
me for Moran –) by initial part
f set :
50000 poin
002023.0
0914.0 , 02x
015.03 a
nnot be rejec
Rosenblatt te
rrelation coef
anizology, 2016
(a)
(b)
Ricker model of model trajec
00 0 kx
nts of feasible
approximate
0802.002 ,
584 of the
ted with 99.9
st), and with
fficient to 0 m
6, 3(1): 25-40
with 2mctory (broken lin
1.0 , 0k
e set we
ely.
798.103 x
feasible set
9979% signif
97.66% (Ma
must be reject
on the plane (ne) (b).
2,1, , 0 x
ere found. Pr
84 , 20A
has followin
ficance level
ann – Whitne
ted with 0.1%
w
log,(log10 x
203 x , 0
obability to f
5404.0 , 0a
ng properties
(Kolmogoro
ey test). Hypo
% significance
www.iaees.org
)g10 y (a) and
400 A ,
find in a
00788.0 ,
s: hypothesis
ov – Smirnov
othesis about
e level (t-test
d
,
a
,
s
v
t
t
35
IAEES
is equal t
level (Sw
serial cor
Asy
outbreak
rate decr
increases
of mode
Kolmogo
forest ins
28-years
Fig. 6 A(log10 x
Sim
to 0.9295). H
wed – Eisenha
rrelation cann
ymptotic dyna
k trajectories o
reases in dom
s in domain w
el trajectory
orov type; Isa
sects. Thus, w
cycle.
Asymptotic stab)log, 10 yx (a)
milar propertie
Hypothesis ab
art test); valu
not be rejecte
amic regime
of various len
main where bir
where it’s les
was theoreti
aev et al., 198
we have goo
ble dynamic r), and approxim
es (Fig. 7) ar
Selforga
out absence o
ue of test “jum
ed with 5% sig
with pointed
ngths. Moreo
rth rate is big
s than one (p
ically explain
84, 2001) and
od backgroun
regime (strongmation of time se
e observed fo
anizology, 2016
of serial corre
mp up – jump
gnificance lev
d out paramet
over, consider
gger than one
parts of trajec
ned in our p
d can be obse
nd for conclu
(a)
(b)
g 28-cycle) foreries (solid line
or fuzzy 16-c
6, 3(1): 25-40
elation canno
p down” is eq
vel and must
ters is strong
ring regime h
e (parts of traj
ctory 9-11, 18
publications
erved for vari
sion that larc
r Moran – Re) by initial part
cycle which i
ot be rejected
qual to 21: hy
be rejected w
28-cycle (Fig
has following
jectory 13-15
8-20, and 26-
(within the
ious trajector
ch bud moth
Ricker model wt of model traje
is realized for
w
with 31.63%
ypothesis abou
with 20% leve
g. 6a). It con
g important fe
5 and 22-24),
1). Such kind
framework
ries of outbre
dynamics co
with 3m ctory (broken li
r following e
www.iaees.org
% significance
ut absence of
el.
nsists of three
features: birth
and this rate
d of behavior
of model of
ak species of
orresponds to
on the planeine) (b).
extreme point
e
f
e
h
e
r
f
f
o
e
t
36
IAEES
0.000 x
0.01 a
rate decr
it’s less t
every 8th
Hyp
(Kolmog
test). Hy
significan
with 13.6
hypothes
Ana
previous
probabili
9{ P
other pro
In a grou
cycles of
other dyn
0231 , 01x
00375 , 2a
reases in dom
than one. Auth value of argu
pothesis abo
gorov – Smirn
ypothesis abou
nce level (t-t
68% significa
sis about abse
alysis of poin
cases bigge
ity is much
08148.0}9
obabilities ar
up of regimes
f same length
namic regime
0768.00
51048.4
main where bi
tocorrelation
ument.
ut symmetry
nov test), wi
ut equivalenc
test is equal t
ance level (S
ence of serial
nts of feasibl
st probability
less than th
8 , and it is
re following
s which are c
hs etc. It allow
es which are c
Selforga
, 0.002 x
5 , 0.03 a
irth rate is big
function is b
y of deviati
th 99.7% (Le
ce of Spearm
to 0.8982). H
Swed – Eisen
correlation c
le set sh
y is observed
he same one
bigger than
g: }8{ P
lose to 9-cyc
ws concludin
close to biolo
anizology, 2016
0315 , 03x
136 of the f
gger than one
bigger than 0
ions cannot
ehmann – Ro
man rank corre
Hypothesis abo
hart test); va
cannot be reje
hows that cyc
d for the eve
e of previous
all other pro
02248.0 ,
le, it is possi
ng that within
ogical imagin
(a
6, 3(1): 25-40
2691.0 ,
feasible set. L
e (Fig. 7a), an
.995 every 16
be rejected
osenblatt test)
elation coeffi
out absence o
alue of test “j
ected with 20
cles of variou
ent 100{
s cases. Big
obabilities for
, }10{ P
ble to point o
n the boundar
ations about l
a)
, 3.19A
Like for previ
nd this rate in
6th value of a
d with 99.99
), and with 8
icient to 0 mu
of serial corre
ump up – jum
% significanc
us lengths be
}00 and is e
g probability
r sub-events
0229.0} ,
out strong 8-
ries of feasibl
larch bud mo
w
64 , 0 a
ious dynamic
ncreases in d
argument, and
999% signif
89.48% (Man
ust be rejecte
elation canno
mp down” is
ce level.
elong to this
equal to 0.77
is observed
of 100{
}18{ P
, 9-, and 10-
le set it is po
oth dynamics.
www.iaees.org
00904.0 ,
c regime birth
omain where
d bigger 0.95
ficance level
nn – Whitney
ed with 0.1%
ot be rejected
s equal to 22:
s set. Like in
759. But this
d for 9-cycle
}00 . Several
02566.0 .
cycles, fuzzy
ssible to find
.
,
h
e
5
l
y
%
d
:
n
s
e
l
.
y
d
37
IAEES
Fig. 7 A(log10 x
Thu
regimes
3m , r
on quant
4 Conclu
One of
trajectori
with resp
conclusio
In c
with time
of model
space of
of statisti
checking
jump dow
statistica
For
of every
points w
series fro
distributi
hypothes
hypothes
Null hyp
hypothes
Ana
Asymptotic stab)log, 10 yx (a)
us, use of Mo
which are c
regimes with
itative level b
usion
very import
ies is determ
pect to origin
on that model
current public
e lags the fol
l parameters p
model param
ical criterions
g of symmetry
wn” for chec
al tests.
every consid
feasible set w
with extreme p
om the stand
ion cannot b
sis (because p
sis cannot be
pothesis cann
sis.
alysis of poin
ble dynamic r); (b) – the same
oran – Ricke
lose to consi
h same charac
but not on qu
ant stages in
mination of pr
n or if serial
l is not suitab
cation for fitt
lowing proce
points of feas
meters which
s includes Ko
y of distribut
cking of absen
dered case (fo
were found, a
properties is
d point of sel
be rejected w
p-value can b
rejected with
not be rejecte
nts with extre
Selforga
regime (fuzzy e regime but wi
er model with
idering time
cteristics wer
alitative leve
n analysis o
roperties of s
correlation is
ble for fitting
ting of larch
edure was use
sible sets wer
correspond to
olmogorov –
tion of deviat
nce/existence
or different v
and for every
based on fol
lected statisti
with 5% sign
be rather sma
h 20% signifi
ed with 99%
eme propertie
anizology, 2016
(b
16-cycle) forithout lines betw
h time lag in
series on qu
re not found (
l).
of correspond
set of deviati
s observed in
of considered
bud moth tim
ed. First of al
re determined
o deviations s
Smirnov, Ma
tions. It also i
e of serial cor
alues of time
y set points w
llowing idea:
ical criterion
nificance lev
all, for examp
ficance level.
significance
es allowed ob
6, 3(1): 25-40
b)
r Moran – Rween nearest po
n three years
uantitative an
(it is possible
dence betwe
ons. If distrib
n a sequence
d time series.
me series by
ll, with the he
d. Feasible se
satisfying to
ann – Whitne
includes Swe
rrelation in s
e lag in Mora
ith extreme p
: these points
ns. For exam
vel it doesn’t
ple, 0.0p
But stronges
level: in suc
btaining that
icker model woints of cycle.
, 3m , all
nd qualitative
e to point out
en empirical
bution of dev
of residuals
.
trajectories o
elp of Monte
et was define
certain statist
ey, and Lehm
ed – Eisenhar
equences of
an – Ricker m
properties we
s give best fi
mple, if hypot
t mean that
06 ). Stronger
st result corre
ch a situation
best results a
w
with 3m
lowed obtain
e levels. For
t regimes wh
l time series
viations is no
it gives a bac
of Moran – R
Carlo metho
ed as a set of
tical criterion
mann – Rosenb
rt test and tes
residuals, and
model) about
ere determined
itting of cons
thesis about
we have to
r result is fol
esponds to sit
n we have to
are observed
www.iaees.org
on the plane
ning dynamic
r cases when
hich are close
s and model
ot symmetric
ckground for
Ricker model
ods in a space
points in the
ns. Collection
blatt tests for
st “jump up –
d some other
50000 points
d. Finding of
sidering time
symmetry of
accept Null
llowing: Null
tuation when
o accept Null
for situation
e
c
n
e
l
c
r
l
e
e
n
r
–
r
s
f
e
f
l
l
n
l
n
38
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
when time lag is in three years. In this situation correspondence of theoretical and empirical datasets is
observed on quantitative (all statistical criterions are satisfied with extreme values of significance levels) and
qualitative (behavior of model trajectory corresponds to behavior of empirical time series) levels. Provided
analysis allows concluding that asymptotic stable dynamic regime of larch bud moth is strong 28-cycle (cycle
in 28 years with three maximum points) or fuzzy 16-cycle (cycle in 16 years).
References
Auer C. 1977. Dynamik von Larchenwicklerpopulationen langs des Alpenbogens. Mitteilungen der
Eidgenossischen Anstalt fiir forstliches Versuchswesen, 53: 71-105
Berryman AA. 1981. Population Systems: A General Introduction. Plenum Press, New York, USA
Berryman AA. 2002. Population regulation, emergent properties, and a requiem for density dependence. Oikos,
99(3): 600-606
Berryman AA, Stark RW. 1985. Assessing the risk of the forest insect outbreaks. Z. angew. Entomology, 99(2):
199-208
Baltensweiler W. 1964. Zeiraphera griceana Hubner (Lepidoptera, Tortricedae) in the European Alps. A
contribution to the problem of cycles. Canadian Entomologist, 96(5): 792-800
Baltensweiler W. 1978. Ursache oder Wirkung? Huhn oder Ei?. Mitteilungen der Schweizerischen
Entomologischen Geselschaft, 51: 261-267
Bard Y. 1974. Nonlinear Parameter Estimation. Academic Press Inc., New York, USA
Bolshev LN, Smirnov NV. 1983. Tables of Mathematical Statistics. Nauka, Moscow, USSR
Borovkov АА. 1984. Mathematical Statistics. Nauka, Moscow, USSR
Draper NR, Smith H. 1981. Applied Regression Analysis. Wiley and Sons Inc., New York, USA
Hollander MDA, Wolfe DA. 1973. Nonparametric Statistical Methods. John Wiley and Sons Inc., New York,
London, Sydney, Toronto
Isaev AS, Khlebopros RG, Nedorezov LV, Kondakov YuP, Kiselev VV. 1984. Forest Insect Population
Dynamics. Nauka, Novosibirsk, USSR
Isaev AS, Khlebopros RG, Nedorezov LV, Kondakov YuP, Kiselev VV, Soukhovolsky VG, 2001. Population
Dynamics of Forest Insects. Nauka, Moscow, Russia
Likes J, Laga J. 1985. Basic Tables of Mathematical Statistics. Finance and statistics, Moscow, USSR
Lilliefors HW. 1967. On the Kolmogorov–Smirnov test for normality with mean and variance unknown.
Journal of American Statistical Association, 64: 399-402
McCallum H. 2000. Population parameters estimation for ecological models. Blackwell Sciences Ltd.,
Brisbane, Australia
Moran PAP. 1950. Some remarks on animal population dynamics. Biometrics, 6(3): 250-258
Nedorezov LV. 2012. Chaos and Order in Population Dynamics: Modeling, Analysis, Forecast. LAP Lambert
Academic Publishing, Saarbrucken, Germany
Nedorezov LV. 2015a. Paramecia aurelia dynamics: non-traditional approach to estimation of model
parameters (on an example of Verhulst and Gompertz models). Ecological Modelling, 317: 1-5
Nedorezov LV. 2015b. Lotka-Vo0lterra Model of Competition Between Two Species and Gause Experiments:
Is There Any Correspondence? Biofizika, 60(5): 1039-1040
Nedorezov LV, Sadykova DL. 2015. Dynamics of larch bud moth populations: application of Moran - Ricker
models with time lags. Ecological Modelling, 297: 26-32
Nedorezov LV, Utyupin YuV. 2011. Continuous-Discrete Models of Population Dynamics: An Analytical
39
Selforganizology, 2016, 3(1): 25-40
IAEES www.iaees.org
Overview. State Public Scientific-Technical Library, Novosibirsk, Russia
Ricker WE, 1954. Stock and recruitment. Journal of Fishery Res. Board of Canada, 11(5): 559-623
Sadykova DL, Nedorezov LV. 2013. Larch bud moth dynamics: can we explain periodicity of population
fluctuations by the time lag dependence in birth rate? Population Dynamics: Analysis, Modelling,
Forecast, 2(4): 154-181
Shapiro SS, Wilk MB, Chen HJ. 1968. A comparative study of various tests of normality. Journal of the
American Statistical Association, 63: 1343-1372
Wood SN. 2001a. Minimizing model fitting objectives that contain spurious local minima by bootstrap
restarting. Biometrics, 57: 240-244
Wood SN. 2001b. Partially specified ecological models. Ecological Monographs, 71: 1-25
40
Selforganizology, 2016, 3(1): 41-42
IAEES www.iaees.org
Book Review
A review on the book, Selforganizology: The Science of Self-
Organization GuangHua Liu Guangdong AIB Polytech College, Guangzhou 510507, China
E-mail: [email protected]
Received 15 January 2016; Accepted 20 January 2016; Published online 1 March 2016
Abstract
The book, Selforganizology: The Science of Self-Organization, authored by WenJun Zhang and published by
World Scientific, was briefly reviewed in present report.
Keywords selforganizology; self-organization; book; review.
This invaluable book is the first of its kind on "selforganizology", the science of self-organization. It covers a
wide range of topics, such as the theory, principle and methodology of selforganizology, agent-based
modelling, intelligence basis, ant colony optimization, fish/particle swarm optimization, cellular automata,
spatial diffusion models, evolutionary algorithms, self-adaptation and control systems, self-organizing neural
networks, catastrophe theory and methods, and self-organization of biological communities, etc. Readers will
have an in-depth and comprehensive understanding of selforganizology, with detailed background information
provided for those who wish to delve deeper into the subject and explore research literature. This book is a
valuable reference for research scientists, university teachers, graduate students and high-level undergraduates
in the areas of computational science, artificial intelligence, applied mathematics, engineering science, social
science and life sciences.
Major contents of the book include
Organization and Organizational Theory
Selforganizology: The Science of Self-organization
Agent-based Modeling
Intelligence Principles
Catastrophe Theory and Methods
Self-adaptation and Control Systems
Cellular Automata and Spatial Diffusion Models
Artificial Neural Networks
Selforganizology URL: http://www.iaees.org/publications/journals/selforganizology/onlineversion.asp RSS: http://www.iaees.org/publications/journals/ selforganizology /rss.xml Email: [email protected] EditorinChief: WenJun Zhang Publisher: International Academy of Ecology and Environmental Sciences
Selforganizology, 2016, 3(1): 41-42
IAEES www.iaees.org
Ant Colony Optimization
Fish and Particle Swarm Optimization
Synergy, Coevolution, and Evolutionary Algorithms
Synergy: Correlation Analysis
Community Succession and Assembly
Mathematical Foundations
The readers of the book will be research scientists, university teachers, graduate students and high-level
undergraduates in the areas of computational science, artificial intelligence, applied mathematics, engineering
science, social science and life sciences.
Reference
Zhang WJ. 2016. Selforganizology: The Science of Self-Organization. World Scientific, Singapore
(http://www.worldscientific.com/worldscibooks/10.1142/9685)
42
Selforganizology
The Selforganizology (ISSN 2410-0080) is an open access (BOAI definition), peer/open reviewed online journal that considers scientific articles in all different areas of selforganizology. The goal of this journal is to keep a record of the state-of-the-art research and promote the research work in these fast moving areas. The topics to be covered by Selforganizology include, but are not limited to:
Innovative theories and methods on self-organization. Applications of evolution-, interaction-, behavior-, organization-, intelligence- and
feedback-based theories, such as coevolution theory, coextinction theory, community succession theory, correlation analysis, parrondo’s paradox, game theory, neural networks, artificial intelligence, behavioral theory, organization theory, and automation theory, in self-organization.
Simulation and modeling of self-organization systems. Algorithms of self-organization, including intelligence computation (swarm intelligence
algorithms, genetic algorithms, etc.), cellular automata, self-adaptation and automation, etc.
Various self-organization phenomena in nature.
We are particularly interested in short communications that clearly address a specific issue or
completely describe a new self-organization phenomenon.
Authors can submit their works to the email box of this journal, [email protected]. All
manuscripts submitted to this journal must be previously unpublished and may not be considered
for publication elsewhere at any time during review period of this journal. Authors are asked to
read and accept Author Guidelines and Publication Ethics & Malpractice Statement before
submitting manuscripts.
In addition to free submissions from authors around the world, special issues are also accepted.
The organizer of a special issue can collect submissions (yielded from a research project, a
research group, etc.) on a specific research topic, or submissions of a scientific conference for
publication of special issue.
Editorial Office: [email protected]
Publisher: International Academy of Ecology and Environmental Sciences
Address: Unit 3, 6/F., Kam Hon Industrial Building, 8 Wang Kwun Road, Kowloon Bay, Hong Kong
Tel: 00852-2138 6086 Fax: 00852-3069 1955
E-mail: [email protected]
Selforganizology ISSN 2410-0080
Volume 3, Number 1, 1 March 2016
Articles
A random network based, node attraction facilitated network
evolution method
WenJun Zhang 1-9
A review on research of the predator-prey interactions
H. A. Wahab, Hazrat Ali, Muhammad Naeem, et al. 10-15
Validity process for refactored coupling based non-dispatchable flaws
in the design model
Syed Uzair Ahmad, Muhammad Naeem 16-24 Non-traditional approach to fitting of time series of larch bud moth dynamics: Application of Moran – Ricker model with time lags
L.V. Nedorezov 25-40
Book Review A review on the book, Selforganizology: The Science of Self- Organization GuangHua Liu 41-42
IAEES http://www.iaees.org/