Professional Documents
Culture Documents
Journal of Applied Mathematics-1
Journal of Applied Mathematics-1
Applied Mathematics
ISSN: 1110-757X (Print)
Volume 2015, Isue 1, 2015
This work is licensed under a Creative Commons Attribution License. Which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Journal of
Applied Mathematics
ISSN: 1110-757X (Print)
Volume 2015, Isue 1, 2015
This work is licensed under a Creative Commons Attribution License. Which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Editorial Board
Saeid Abbasbandy, Imam Khomeini International University, Iran
Mina B. Abd-El-Malek, American University in Cairo, Egypt
Mohamed A. Abdou, Alexandria University, Egypt
Subhas Abel, Gulbarga University, India
Janos Abonyi, University of Pannonia, Hungary
Sergei Alexandrov, Russian Academy of Sciences, Russia
M. Montaz Ali, University of the Witwatersrand, South Africa
Mohammad R, Aliha, Iran University of Science and Technology, Iran
Carlos J. S. Alves, Technical University of Lisbon, Portugal
Mohamad Alwash, West Los Angeles College, USA
Gholam R. Amin, Sultan Qaboos University, Oman
Igor Andrianov, RWTH Aachen University, Germany
Boris Andrievsky, Institute for Problems of Mechanical Engineering of the RAS, Russia
Whye-Teong Ang, Nanyang Technological University, Singapore
Abul-Fazal M. Arif, King Fahd University of Petroleum & Minerals, Saudi Arabia
Sabri Arik, Isik University, Turkey
Ali R. Ashrafi, Kashan University, Iran
Allaberen Ashyralyev, Fatih University, Turkey
Francesco Aymerich, University of Cagliari, Italy
Jalel Azaiez, University of Calgary, Canada
Seungik Baek, Michigan State University, USA
Hamid Bahai, Brunel University, United Kingdom
Olivier Bahn, HEC Montral, Canada
Antonio Bandera, Universidad de Malaga, Spain
Jean-Pierre Barbot, ECS ENSEA, France
Mostafa Barigou, University of Birmingham, United Kingdom
Roberto Barrio, Universidad de Zaragoza, Spain
Henri Baudrand, Laboratoire LAPLACE-GRE, France
Alfredo Bellen, University of Trieste, Italy
Vasile Berinde, Universitatea de Nord din Baia Mare, Romania
Jafar Biazar, University of Guilan, Iran
Anjan Biswas, King Abdulaziz University, Saudi Arabia
Abdellah Bnouhachem, Universit Ibn Zohr, Morocco
Gabriele Bonanno, University of Messina, Italy
Walter Briec, University of Perpignan, France
James Robert Buchanan, Millersville University, USA
Humberto Bustince, Public University of Navarra, Spain
Xiao Chuan Cai, University of Colorado, USA
Piermarco Cannarsa, University of Rome Tor Vergata, Italy
Jinde Cao, Southeast University, China
Yijia Cao, Hunan University, China
Zhenfu Cao, Shanghai Jiao Tong University, China
Bruno Carpentieri, University of Groningen, The Netherlands
Ana Carpio, Universidad Complutense de Madrid, Spain
Alexandre Carvalho, University of So Paulo, Brazil
Song Cen, Tsinghua University, China
Wei-Der Chang, Shu-Te University, Taiwan
Tai-Ping Chang, National Kaohsiung First University of Science and Technology, Taiwan
Shih-sen Chang, Yunnan University of Finance And Economics, China
Shuenn-Yih Chang, National Taipei University of Technology, Taiwan
Kripasindhu Chaudhuri, Jadavpur University, India
Yuming Chen, Wilfrid Laurier University, Canada
Jianbing Chen, Tongji University, China
Rushan Chen, Nanjing University of Science and Technology, China
Xinkai Chen, Shibaura Institute of Technology, Japan
Editorial Workflow
The following is the editorial workflow that every manuscript submitted to the journal undergoes
during the course of the peer-review process.
The entire editorial workflow is performed using the online Manuscript Tracking System. Once a
manuscript is submitted for publication, the manuscript is checked by the journals editorial office
to ensure that it is suitable to go through the normal peer review process. Once this is done, the
manuscript is sent to an appropriate Editor based on the subject of the manuscript and the
availability of the Editors. All manuscripts shall be handled by an Editor who is not from the same
institute as any of the manuscripts authors, is not from the same country as any of the manuscripts
authors, and has not co-authored any papers with any of the manuscripts authors in the past three
years.
If the Editor finds that the manuscript may not be of sufficient quality to go through the normal peer
review process, or that the subject of the manuscript may not be appropriate for the journals scope,
the manuscript shall be rejected with no further processing.
If the Editor finds that the submitted manuscript is of sufficient quality and falls within the scope of
the journal, they should assign the manuscript to a number of external reviewers, provided that no
conflict of interests exists between these reviewers and the manuscripts authors. The reviewers will
then submit their reports on the manuscripts along with their recommendation of one of the
following actions to the Editor:
Publish Unaltered
Consider after Minor Changes
Consider after Major Changes
Reject: Manuscript is flawed or not sufficiently novel
When all reviewers have submitted their reports, the Editor can make one of the following editorial
recommendations:
Publish Unaltered
Consider after Minor Changes
Consider after Major Changes
Reject
If the Editor recommends Publish Unaltered, the manuscript will undergo a final check by the
journals editorial office in order to ensure that the manuscript and its review process adhere to the
journals guidelines and policies. Once this is done, the authors will be notified of the manuscripts
acceptance, and the manuscript will appear in the Articles in Press section of the journals website.
If the Editor recommends Consider after Minor Changes, the authors are notified to prepare and
submit a final copy of their manuscript with the required minor changes suggested by the reviewers.
The Editor reviews the revised manuscript after the minor changes have been made by the authors.
Once the Editor is satisfied with the final manuscript, the manuscript can be accepted.
If the Editor recommends Consider after Major Changes, the recommendation is communicated
to the authors. The authors are expected to revise their manuscripts in accordance with the changes
recommended by the reviewers and to submit their revised manuscript in a timely manner. Once the
revised manuscript is submitted, the Editor can then make an editorial recommendation which can
be Publish Unaltered, Consider after Minor Changes, or Reject.
If the Editor recommends rejecting the manuscript, the rejection is immediate. Also, if the majority
of the reviewers recommend rejecting the manuscript, the rejection is immediate.
The editorial workflow gives the Editors the authority to reject any manuscript because of
inappropriateness of its subject, lack of quality, or incorrectness of its results. The Editor cannot
assign himself/herself as an external reviewer of the manuscript. This is to ensure a high-quality,
fair, and unbiased peer-review process of every manuscript submitted to the journal, since any
manuscript must be recommended by one or more (usually two or more) external reviewers along
with the Editor in charge of the manuscript in order for it to be accepted for publication in the
journal.
The name of the Editor recommending the manuscript for publication is published with the
manuscript to indicate and acknowledge their invaluable contribution to the peer-review process
and the indispensability of their contributions to the running of the journals.
The peer-review process is single blinded; that is, the reviewers know who the authors of the
manuscript are, but the authors do not have access to the information of who the peer reviewers are.
Every journal published by Hindawi has an acknowledgment page for the researchers who have
performed the peer-review process for one or more of the journal manuscripts in the past year.
Without the significant contributions made by these researchers, the publication of the journal
would not be possible.
Journal of
Applied Mathematics
ISSN: 1110-757X (Print)
Volume 2015, Isue 1, 2015
Table of Contents
Palmprint Authentication System Based on Local and Global Feature
Fusion Using DOST
N. B. Mahesh Kumar and K. Premalatha
01-11
12-20
21-30
31-44
45-51
52-61
62-70
71-80
81-89
90-98
Research Article
Palmprint Authentication System Based on Local and Global
Feature Fusion Using DOST
N. B. Mahesh Kumar and K. Premalatha
Bannari Amman Institute of Technology, Sathyamangalam, Erode, Tamilnadu 638 401, India
Correspondence should be addressed to N. B. Mahesh Kumar; mknbmaheshkumar@gmail.com
Received 23 September 2014; Revised 16 December 2014; Accepted 16 December 2014; Published 31 December 2014
Academic Editor: Qiankun Song
Copyright 2014 N. B. M. Kumar and K. Premalatha. Tis is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.
Palmprint is the region between wrist and fngers. In this paper, a palmprint personal identifcation system is proposed based on
the local and global information fusion. Te local and global information is critical for the image observation based on the results of
the relationship between physical stimuli and perceptions. Te local features of the enhanced palmprint are extracted using discrete
orthonormal Stockwell transform. Te global feature is obtained by reducing the scale of discrete orthonormal Stockwell transform
to infnity. Te local and global matching distances of the two palmprint images are fused to get the fnal matching distance of the
proposed scheme. Te results show that the fusion of local and global features outperforms the existing works on the available three
datasets.
1. Introduction
Te unique physical or behavioural characteristics of human
beings are of broad interest in the biometrics based methods
[1]. In the modern e-world, biometrics has great potentials
because of their high accuracy and convenience to use.
Researchers and scientists have systematically investigated
the use of a number of biometric characteristics like fngerprint [25], face [6], iris [7, 8], palmprint [9, 10], hand
geometry [11], hand vein [12], fnger knuckle-print [13, 14],
voice [15], and ear [16] in the development of computing
techniques. Te use of applications like automatic personal
authentication systems in day-today life results in reliable and
efective security control. Recently, hand-based biometrics
has been attracting signifcant attention in the modern eworld.
Te region between wrist and fngers is referred to as
palmprint. It has features like texture, wrinkles, principle
lines, ridges, and minutiae points that can be used for its
representation. Te critical properties of biometric characteristics such as universality, individuality, stability, and
collectability are satisfed by palmprint compared to other
biometric traits. It has relatively stable and unique features. It
is nonintrusive and data collection is very easy. It needs fewer
Image acquisition
Preprocessing
Database
Enroll/
verify/
identify
Decision
Matching
regions. Te global features can be extracted more consistently from the poor quality of the partial palmprints. It is
too space and time consuming to directly store and compare
the features pixel by pixel. Global feature based matching
overlaps two given templates with diferent transformation
parameters and estimates the similarity score between the
corresponding cells. Compared to local features, the global
features have less distinctness and hence been ofen exploited
together with other features or in the preprocessing stage
of palmprint matching. Both the local and global features
are fairly independent and capture current information and
hence it is realistic to improve the discriminating ability of
matching by fusing features. Te local and global features are
combined in the matching stage with accessible feature level
fusing strategies. Te combined local and global features can
improve the performance of the palmprint systems on large
scale databases. How to select features is pivotal for the efect
of feature combination. Te combination of irrelative features
will improve the accuracy or efciency of the palmprint
biometric systems. Te suitable hierarchical approach can
be employed to reduce the additional time or memory cost
required while fusing the local and global features.
In this work, a local-global feature fusion technique for
palmprint biometric system is proposed. Te instantaneous
phase information is extracted as the local feature by DOST.
Te global feature can be obtained by aggregating the scale
of the DOST. Te DOST can be reduced to the Fourier
transform of the whole image if the scale of the DOST is
increased to infnity. Hence, the local features cannot be
obtained but the fnest resolution for the global frequency
analysis of the image can be obtained. Terefore, Fourier
transform coefcients are obviously taken as the global
features. Te arrangement between intraclass palmprint ROIs
can also be refned by the global Fourier features. Te two
matching distances are calculated by matching the local and
the global information separately. Ten the two matching
distances are merged by fusion rule to get the fnal matching
2. Literature Review
In [4], a new BioHashing technique is proposed to generate
a cancellable template in the context of unordered set of
noisy minutiae features and shows that the work is efcient in
realistic context. In [17], an intersession variability modelling
(ISV) and joint factor analysis (JFA) for face authentication
are proposed to achieve a least signifcant error rate reduction
for measuring accuracy. Te line features in palmprint are
extracted by using sobel and morphological methods [10].
In [18], a new chaf point generation technique for the fuzzy
vault in biocryptosystems is proposed and shows that the
proposed technique achieves more chaf points compared to
Khalil-Hanis algorithm. In [19], the principle lines along with
isolated points are used as features. Te ridges containing
creases of the palmprint are eliminated and it has been
proposed in [20]. Te features like end points of principle
lines have been used in [9]. Te end points in the palmprint
are found to be directional invariant. Te crease points that
are related to diseases of a person have been suggested in
the palmprint biometrics system. Te directional line energy
features are also used to identify the palmprint of a person.
Further, the palmprint biometric systems that are based
on structural features are not invariant to occlusion of the
palmprint image.
3
(2) Te 2D-ST of an array of size has a computational complexity of O (4 + 4 log ) and storage
requirements of O (4 ).
(3) To eliminate this problem of 2D-ST, we use DOST
which is also a multiresolution technique for extraction of features from the palmprint images and is
based on the S-transform.
(4) DOST have less computational and storage complexity in comparison to the S-transform because it uses
an orthonormal set of basis functions, while retaining
all the advantageous properties of S-transform.
(5) 2D-DOST provides a spatial frequency representation
of an image, with computational and storage complexity as O (2 + 2 log ) and O (2 ), respectively.
1 2
2 , 2
( + V , V + V ) ,
=22 V=22
here V = 2 1+2
2 (
+
),
21 21
V = 2 1+2
(1)
2
(a)
(b)
Figure 2: (a) Sample input image. (b) ROI image for PolyU palmprint database.
(a)
(b)
Figure 3: (a) Sample input image. (b) ROI image for COEP palmprint database.
(a)
(b)
Figure 4: (a) Sample input image. (b) ROI image for IIT Delhi palmprint database.
(IF) and the middle fnger (MF) and that between the MF
and the ring fnger (RF). Among the three gaps, the gap
between MF and RF is the toughest point. In order to
eliminate the translation error, the gap between MF and RF
is used as the reference point. Te maximum square region
extracted from the palm area is selected as the ROI. Te
square region is horizontally centered on the axis running
through the gap between MF and RF. Due to the regular and
controlled uniform illumination conditions during image
capture, the acquired hand image and its background contrast
in colour. Te sample input image is shown in Figures 2(a),
3(a), and 4(a) for PolyU palmprint database, COEP palmprint database, and IIT Delhi palmprint database. Global
thresholding has been applied to extract the hand from the
background. Te image is processed with opening and closing
morphological operations to remove any isolated small blobs
or holes. Contour-tracing algorithm is applied to extract the
contour of the hand image from the palmprint. Te ROI
extraction technique of IIT Delhi palmprint database, PolyU
palmprint database, and COEP palmprint database are same
as in [9, 26]. Te corresponding Figures 2(b), 3(b), and 4(b)
are shown in this paper which gives the details of ROI image.
5. Local-Global Feature
Extraction and Matching
0.1
(1 , 2 ) = (1 , 2 ) 11 1 22 2 = (1 ,2 ) (1 ,2 ) ,
1 ,2
(2)
where 1 = 1 , . . . , 1 . 2 = 2 , . . . , 2 , 1 =
(2/1 ) . 2 = (2/2 ) and the operator 1 2 denotes
1
2
1 =
2 =
. (1 ,2 ) and (1 ,2 ) are amplitude com1
2
0.08
0.066
0.044
0.022
0
0.02
6000
8800
400
600
400
200
200
0
= (1 ,2 ) (1 ,2 ) (1 ,2 ) ,
(3)
(1 , 2 ) (1 , 2 )
(4)
= (1 ,2 ) .
1
(1 , 2 ) 1 1 1 2 2 2 ,
1 , 2
(5)
1
2
where 1 2 denotes =
.
=
1
1
2
2
Te most signifcant property of POC is compared with
the normal correlation and it gives accuracy in pattern matching. Te distinct sharp peak is obtained when POC function
is the same. Te peak in the POC drops considerably if the
two palm print images are not the same. Figures 5, 6, and 7
show the POC function of ROI for PolyU palmprint database,
COEP palmprint database, and IIT Delhi palmprint database.
Tus, the POC function shows much higher perception
ability than the normal correlation function. In order to
match the two palmprint images, the height of the peak will
be used as a good resemblance measure.
0.1
0.15
0.08
0.1
0.06
0.04
0.05
0.02
0
0
0.02
6000
0.05
300
800
8
400
400
200
300
200
100
200
0
4400
200
600
100
0 0
0.06
0.15
0.044
0.1
0.022
0.05
0
0
0.022
6000
0.05
300
800
8
400
600
200
0
4
400
200
400
200
300
200
100
100
0
0
0
1
(,V)2(/ 1 +V/ 2 ) ,
1 2 = V=
0
0.15
0.1
0.05
0
0.05
300
400
200
300
100
200
100
(6)
0.9
0.9
Genuine acceptance rate (%)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
105
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
104
103
102
False acceptance rate (%)
101
0
105
100
Figure 11: ROC curve using local features of proposed DOST for
PolyU palmprint database.
101
100
Figure 13: ROC curve using local features of proposed DOST for
COEP palmprint database.
0.9
0.9
Genuine acceptance rate (%)
103
102
False acceptance rate (%)
Proposed work
Ordinal code
Palm code
Proposed work
Ordinal code
Palm code
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.1
0
105
104
104
103
102
False acceptance rate (%)
101
100
Proposed work
Ordinal code
Palm code
0
105
104
103
102
False acceptance rate (%)
101
100
Proposed work
Ordinal code
Palm code
Figure 12: ROC curve using global features of proposed DOST for
PolyU palmprint database.
Figure 14: ROC curve using global features of proposed DOST for
COEP palmprint database.
0.9
0.9
Genuine acceptance rate (%)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.1
0
105
0.8
104
103
102
False acceptance rate (%)
101
0
105
100
Figure 15: ROC curve using local features of proposed DOST for
IIT Delhi palmprint database.
101
100
Figure 17: ROC curve using both local and global features of
proposed DOST for PolyU palmprint database.
0.9
0.9
Genuine acceptance rate (%)
103
102
False acceptance rate (%)
Proposed work
Ordinal code
Palm code
Proposed work
Ordinal code
Palm code
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
105
104
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
104
103
102
False acceptance rate (%)
101
100
Proposed work
Ordinal code
Palm code
Figure 16: ROC curve using global features of proposed DOST for
IIT Delhi palmprint database.
0
105
104
103
102
False acceptance rate (%)
101
100
Proposed work
Ordinal code
Palm code
Figure 18: ROC curve using both local and global features of
proposed DOST for COEP palmprint database.
9
Table 1: Performance comparison of diferent palmprint verifcation
schemes for both local and global combination.
1
Genuine acceptance rate (%)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
105
104
103
102
False acceptance rate (%)
101
100
Proposed work
Ordinal code
Palm code
Figure 19: ROC curve using both local and global features of
proposed DOST for IIT Delhi palmprint database.
CRR (%)
EER (%)
(a) Performance for PolyU database
Palm code [9]
99.91
0.5338
Ordinal code [21]
100.00
0.0709
Proposed
100.00
0.0035
(b) Performance for COEP database
Palm code [9]
99.61
3.6732
Ordinal code [21]
99.84
1.7544
Proposed
100.00
1.2605
(c) Performance for IIT Delhi database
Palm code [9]
99.68
5.2108
Ordinal code [21]
100.00
1.1762
Proposed
100.00
1.0217
DI
5.5807
6.6785
7.2541
3.0734
3.3966
4.0228
1.800
2.2174
3.0168
Table 2: Computation time for key processes both local and global
combination.
3.5
Proposed
3
Proposed
2.5
Proposed
Extraction (ms)
Matching (ms)
(a) Systems speed for PolyU database
54.56
1.5
(b) Systems speed for COEP database
85.41
3.07
(c) Systems speed for IIT Delhi database
83.34
3.15
Total (ms)
56.06
88.48
86.49
(%)
1.5
0.5
0
0.1
0.2
0.3
0.4
Matching distance
0.5
Imposter
Genuine
1
100,
2
(7)
where 1 denotes the number of correct recognition of palmprint images and 2 is the total number of palmprint images.
Performance comparison of diferent palmprint verifcation
schemes is shown in Table 1. Table 2 shows the computation
time for key processes.
ROC curves are used to analyse the performance of
the palmprint biometric system. ROC is a plot of genuine
acceptance rate (GAR) against false acceptance rate (FAR)
and shown in Figures 11, 12, 13, 14, 15, 16, 17, 18, and 19
for PolyU palmprint database, COEP palmprint database,
and IIT Delhi palmprint database. Te FAR and FRR of a
palmprint biometric system can be obtained by adjusting the
various thresholds. It is known that both FAR and FRR are
10
2
,
DI = 1
(8)
(12 + 22 ) /2
3.5
2.5
(%)
1.5
8. Conclusion
0.5
0
0.1
0.2
0.3
0.4
Matching distance
0.5
Imposter
Genuine
Conflict of Interests
Te authors declare that there is no confict of interests
regarding the publication of this paper.
2.5
2
(%)
Acknowledgments
Te authors would like to thank Hong Kong Polytechnic
University, IIT-Delhi, and College of Engineering-Pune for
providing the palmprint databases.
1.5
References
0.5
0
0.1
0.2
0.3
0.4
Matching distance
0.5
Imposter
Genuine
11
[25] S. B. Nikam, P. Goel, R. Tapadar, and S. Agarwal, Combining
gabor local texture pattern and wavelet global features for fngerprint matching, in Proceedings of the International Conference on Computational Intelligence and Multimedia Applications
(ICCIMA 07), pp. 409416, Washington, DC, USA, 2007.
[26] G. S. Badrinath and P. Gupta, Stockwell transform based palmprint recognition, Applied Sof Computing, vol. 11, no. 7, pp.
42674281, 2011.
[27] PolyU Palmprint Database, Te Hong Kong Polytechnic
University, July 2010, http://www4.comp.polyu.edu.hk/biometrics.
[28] COEP Palm Print Database (College of Engineering Pune),
http://www.coep.org.in/.
[29] Palmprint Image Database version 1, IIT, Delhi, India, 2007,
http://www4.comp.polyu.edu.hk/csajaykr/IITD/Database
Palm.htm.
Research Article
A Comparison of Evolutionary Computation Techniques
for IIR Model Identification
Erik Cuevas,1,2 Jorge Glvez,1 Salvador Hinojosa,1 Omar Avalos,1
Daniel Zaldvar,1,2 and Marco Prez-Cisneros3
1
Departamento de Electronica, Universidad de Guadalajara, CUCEI, Avenida Revolucion 1500, 44430 Guadalajara, JAL, Mexico
Centro Universitario Azteca, Unidad de Investigacion, Avenida Juarez 340, 44280 Guadalajara, JAL, Mexico
3
CUTONALA, Avenida Nuevo Periferico 555, Ejido San Jose Tateposco, 48525 Tonala, JAL, Mexico
2
1. Introduction
System identifcation is a complex optimization problem
which has recently attracted the attention in the feld of
science and engineering. System identifcation is important
in the disciplines of control systems [1], communication [2],
signal processing [3], and image processing [4].
In a system identifcation confguration, an optimization
algorithm attempts to iteratively determine the adaptive
model parameters to get an optimal model for an unknown
plant by minimizing some error function between the output
of the candidate model and the output of the plant. Te optimal model or solution is attained when such error function
is efectively reduced. Te adequacy of the estimated model
depends on the adaptive model structure, the optimization
algorithm, and also the characteristic and quality of the inputoutput data [5].
13
error surface for the IIR model is mostly multimodal with
respect to the flter coefcients. Tis may result in leading
traditional gradient-descent approaches into local optima [9].
Te difculties associated with the use of gradient based
optimization methods for solving several engineering problems have contributed to the development of alternative
solutions. Evolutionary computation techniques (ECT) such
as the particle swarm optimization (PSO) [10], artifcial bee
colony (ABC) [11], electromagnetism-like method (EM) [12],
cuckoo search (CS) [13], and fower pollination algorithm
(FPA) [14] have received much attention regarding their
potential as global optimization methods in real-world applications. Inspired by the evolution process and survival of
the fttest in the biological world, ECT are search methods
that are diferent from traditional optimization methods.
Tey are based on a collective learning process within a
population of candidate solutions. Te population in ECT is
usually arbitrarily initialized, and each iteration (also called
a generation) evolves towards better and better solution
regions by means of randomized processes where several
operators are applied to each candidate solution. ECT have
been applied to many engineering optimization problems and
have proven to be efective for solving some specifc problems,
including multimodal optimization, dynamic optimization,
noisy optimization, and multiobjective optimization [1517].
Hence, they are becoming increasingly popular tools to solve
various hard optimization problems.
As an alternative to gradient based techniques, the
problem of IIR modelling has also been handled through
evolutionary computation techniques. In general, they have
been demonstrated to yield better results than those based on
gradient algorithms with respect to accuracy and robustness
[9]. Such approaches have produced several robust IIR identifcation systems by using diferent evolutionary computation
techniques such as PSO [18], ABC [19], EM [20], and CS [21],
whose results have been individually reported.
ECT are ofen designed to meet the requirements of
particular problems because no single optimization algorithm can solve all problems competitively [22]. Terefore,
when new alternative algorithms are proposed, their relative
efciency must be appropriately evaluated. Many eforts [23
25] have also been devoted to comparing ECT to each other.
Typically, such comparisons have been based on synthetic
numerical benchmark problems with most studies verifying
if one algorithm outperforms others over a given set of
benchmarks functions overlooking any statistical test. However, few comparative studies of various ECT considering the
application context are available in the literature. Terefore, it
is very important to discuss and compare the performance of
ECT methods from an application point of view.
Tis paper presents the comparison of various evolutionary computation optimization techniques that are applied
to IIR model identifcation. In the comparison, special
attention is paid to recently developed algorithms such as
the cuckoo search (CS) and the fower pollination algorithm
(FPA), including also popular approaches as the particle
swarm optimization (PSO), the artifcial bee colony (ABC)
optimization, and the electromagnetism-like optimization
(EM) algorithm. Results over several models with diferent
2. Evolutionary Computation
Techniques (ECT)
In the real world, many optimization problems can be
considered as black box challenges. Ofen, less information
is available about an optimization problem itself unless the
information emerges from function evaluations. In the worst
case, nothing is known about the characteristics of the ftness
function, for example, whether it is unimodal or multimodal.
On the other hand, ECT are used to estimate the solution
to complex optimization problems since they adapt easily to
black-box formulations and extremely ill-behaved functions.
ECT are based on a collective learning process within a
population of candidate solutions. Te population in ECT
is usually arbitrarily initialized while each iteration (also
called a generation) evolves towards better solution regions
by means of randomized processes with several operators
being applied to each candidate solution. ECT have been
applied to many engineering optimization problems ensuring
an efective solution for some specifc problems, including
multimodal optimization, dynamic optimization, noisy optimization, multiobjective optimization, and others [1517].
Terefore, ECT are becoming increasingly popular tools
to solve various hard optimization problems. Tis section
presents a brief description of fve evolutionary computation
techniques: swarm optimization (PSO), artifcial bee colony
(ABC) optimization and electromagnetism-like optimization
(EM), cuckoo search (CS), and fower pollination algorithm
(FPA), which have been all employed in our comparative
study.
2.1. Particle Swarm Optimization (PSO). PSO, proposed by
Kennedy and Eberhart in 1995 [10], is a population-based
stochastic optimization technique that is inspired on the
social behavior of bird focking or fsh schooling. Te algorithm searches for the optimum using a group or swarm
formed by possible solutions of the problem, which are
called particles. From the implementation point of view, in
the PSO operation, a population P ({p1 , p2 , . . . , p}) of
particles (individuals) evolves from the initial point ( =
0) to a total gen number of iterations ( = gen). Each
particle p ( [1, . . . , ]) represents a -dimensional vector
{,1
, ,2
, . . . , ,
} where each dimension corresponds to a
decision variable of the optimization problem at hand. Te
quality of each particle p (candidate solution) is evaluated
by using an objective function (p ) whose fnal result
represents the ftness value of p . During the evolution
process, the best global position g (1 , 2 , . . . ) seen so far
14
g=
arg min
{1,2,...,},{1,2,...,}
( (p )) ,
= arg min ( (p )) .
(1)
{1,2,...,}
= V,
+ 1 1 (,
,
) + 2 2 ( ,
);
V,
+1
+1
= ,
+ V,
,
,
(2)
, ,2 , . . . , ,
} where
represents a -dimensional vector {,1
each dimension corresponds to a decision variable of the
optimization problem to be solved. Afer initialization, an
objective function evaluates each food location to assess
whether it represent an acceptable solution (nectar-amount)
or not. Guided by the values of such an objective function,
the candidate solution l is evolved through diferent ABC
operations (honeybee types). In the main operator, each
food location l generates a new food source t in the
neighborhood of its present position as follows:
t = l + (l l ) ,
, (1, 2, . . . , ) ,
(3)
1
,
{
1
+
(l )
={
{1 + abs ( (l )) ,
if (l ) 0,
if (l ) < 0,
(4)
, . . . , ,
})
the parameter values to be optimized (x = {,1
whereas denotes the iteration (or generation) number. Te
, ,2
, . . . , ,
} where each dimension corresponds to a
{,1
decision variable of the optimization problem to be solved.
Te quality of each egg e (candidate solution) is evaluated
by using an objective function (e ) whose fnal result
represents the ftness value of e . Tree diferent operators
defne the evolution process of CS: (A) Levy fight, (B) the
replacing of nests operator for constructing new solutions,
and (C) the elitist selection strategy.
(A) Te Levy Flight. One of the most powerful features
of cuckoo search is the use of Levy fights to generate
15
|k|1/
(5)
= (
V (0, V2 ) ,
(1 + ) sin( /2)
)
((1 + )/2) 2(1)/2
1/
V = 1,
(6)
(7)
calculated by using
= e + c .
e+1
(8)
e + rand (e e ) ,
e ,
with probability ,
with probability (1 ) ,
(9)
either
(C) Te Elitist Selection Strategy. Afer producing e+1
e , then e+1
is
accepted
as
the
fnal
solution;
otherwise,
e
e+1
,
e ,
if (e+1
) < (e ) ,
otherwise.
(10)
}) of fower
FPA operation, a population F ({f1 , f2 , . . . , f
positions (individuals) is evolved from the initial point ( =
0) to a total gen number of iterations ( = gen). Each
fower f ( [1, . . . , ]) represents a -dimensional vector
{,1
, ,2
, . . . , ,
} where each dimension corresponds to a
decision variable of the optimization problem to be solved.
In FPA, a new population F+1 is produced by considering
two operators: local and global pollination. A probabilistic
global pollination factor is associated with such operators.
In order to decide which operator should be applied to each
current fower position f , a uniform random number is
generated within the range [0, 1]. If is less than , the
global pollination operator is applied to f . Otherwise, the
local pollination operator is considered.
Global Pollination Operator. Under this operator, the original
position f is displaced to a new position f+1 according to
the following model:
f+1 = f + (f g) ,
(11)
, , (1, 2, . . . , ) ,
(12)
16
d(t) +
Plant
4. Experimental Results
e(t)
u(t)
y(t)
IIR model
Evolutionary computation
technique
(13)
where and are the number of numerator and denominator coefcients of the transfer function, respectively, and,
and are the pole and zero parameters of the IIR model
( [1, . . . , ], [1, . . . , ]). Equation (13) can be written as
diference equation of the form:
=1
=0
() = ( ) + ( ) ,
(14)
1
2
( () ()) ,
=1
(15)
(16)
0.05 0.41
,
1 1.13141 + 0.252
.
( ) =
1 1
1
(17)
17
5
4 4
3
0.5
8
6
f()
0
4
0.5
2
1
0
b
1
0.5
1
1
1
0.5
0.5
0.5
(a)
(b)
Figure 2: Multimodal error surface () for the frst experiment: (a) tridimensional fgure and (b) contour.
1 + 1
+ 2
.
2
(18)
0.9125
0.1420
0.9034
0.9173
0.9364
ABP
0.3012
0.3525
0.3030
0.2382
0.2001
AV
SD
0.0284
0.0197
0.0165
0.0101
0.0105
0.0105
0.0015
0.0012
3.118 004
5.103 004
+ 1 + 2 2 + 3 3 + 4 4
( ) = 0 1 1
.
1 + 1 + 2 2 + 3 3 + 4 4
1
(19)
18
Table 2: Performance results of the second experiment.
ABP
2
0.4925
0.6850
0.4802
0.4900
0.4900
1
1.4024
1.2138
1.0301
1.400
1.400
Algorithms
PSO
ABC
EM
CS
FPA
0.9706
0.2736
1.0091
1.000
1.000
AV
SD
4.0035 005
0.3584
3.9648 005
0.000
4.6246 32
1.3970 005
0.1987
8.7077 005
0.000
2.7360 31
1
0.3683
1.1634
0.4950
0.9599
0.0328
2
0.7043
0.6354
0.7049
0.0248
0.1059
3
0.2807
1.5182
0.5656
0.0368
0.0243
4
0.3818
0.6923
0.2691
0.0002
0.7619
AV
5.8843
7.3067
0.0140
6.7515 004
0.0018
SD
3.4812
4.3194
0.0064
4.1451 004
0.0020
PSO
ABC
EM
FPA
0.1011
ABP
0
0.9939
0.5214
1.0335
0.2377
1.0171
1
0.6601
1.2703
0.6670
0.0031
0.0038
2
0.8520
0.3520
0.4682
0.3579
0.2374
3
0.2275
1.1816
0.6961
0.0011
0.0259
4
1.4990
1.9411
0.0673
0.5330
0.3365
5. Conclusions
Te results show that CS presents better precision (AV value)
and robustness (SD value). Tese results also indicate that CS,
FPA, and EM are able to identify the sixth-order plant under
diferent accuracy levels. On the other hand, PSO and ABC
obtain suboptimal solutions whose parameters weakly model
the unknown system.
4.2. Statistical Analysis. In order to statistically validate the
results, a nonparametric statistical signifcance-proof which
is known as Wilcoxons rank sum test for independent samples [31, 32] has been conducted over the the average ()
value (AV) data of Tables 1, 2, and 4 with a 5% signifcance
level. Te test has been conducted considering 30 diferent
executions for each algorithm. Table 5 reports the values
Tis paper presents a comparison study between fve evolutionary algorithms for the IIR-based model identifcation.
Under this research, the identifcation task is considered as an
optimization problem. In the comparison, special attention
is paid to recently developed algorithms such as the cuckoo
search (CS) and the fower pollination algorithm (FPA), also
including popular approaches such as the particle swarm
optimization (PSO), the artifcial Bee colony optimization
(ABC), and the electromagnetism-like (EM) optimization
algorithm.
Te comparison has been experimentally evaluated over a
test suite of three benchmark experiments that produce multimodal functions. Te experiment results have demonstrated
that CS outperforms PSO, ABC, and EM in terms of both the
19
accuracy (AV values) and robustness (SD values), within a
statistically signifcant framework (Wilcoxon test). However,
there is not statistical evidence that CS surpasses the FPA
performance.
Te remarkable performance of CS and FPA is explained
by two diferent features: (i) operators (such as Levy fight)
that allow a better exploration of the search space, increasing
the capacity to fnd multiple optima, and (ii) their exploitation operators that allow a better precision of previously
found solutions.
Conflict of Interests
Te authors declare that there is no confict of interests
regarding the publication of this paper.
References
[1] X. Zhou, C. Yang, and W. Gui, Nonlinear system identifcation
and control using state transition algorithm, Applied Mathematics and Computation, vol. 226, pp. 169179, 2014.
[2] M. Albaghdadi, B. Briley, and M. Evens, Event storm detection
and identifcation in communication systems, Reliability Engineering and System Safety, vol. 91, no. 5, pp. 602613, 2006.
[3] P. Frank Pai, B.-A. Nguyen, and M. J. Sundaresan, Nonlinearity
identifcation by time-domain-only signal processing, International Journal of Non-Linear Mechanics, vol. 54, pp. 8598, 2013.
[4] H.-C. Chung, J. Liang, S. Kushiyama, and M. Shinozuka,
Digital image processing for non-linear system identifcation,
International Journal of Non-Linear Mechanics, vol. 39, no. 5, pp.
691707, 2004.
[5] J. Na, X. Ren, and Y. Xia, Adaptive parameter identifcation
of linear SISO systems with unknown time-delay, Systems &
Control Letters, vol. 66, no. 1, pp. 4350, 2014.
[6] O. Kukrer, Analysis of the dynamics of a memoryless nonlinear
gradient IIR adaptive notch flter, Signal Processing, vol. 91, no.
10, pp. 23792394, 2011.
[7] T. Mostajabi, J. Poshtan, and Z. Mostajabi, IIR model identifcation via evolutionary algorithmsa comparative study,
Artifcial Intelligence Review, 2013.
[8] C. Dai, W. Chen, and Y. Zhu, Seeker optimization algorithm
for digital IIR flter design, IEEE Transactions on Industrial
Electronics, vol. 57, no. 5, pp. 17101718, 2010.
[9] W. Fang, J. Sun, and W. Xu, A new mutated quantum-behaved
particle swarm optimizer for digital IIR flter design, Eurasip
Journal on Advances in Signal Processing, vol. 2009, Article ID
367465, 7 pages, 2009.
[10] J. Kennedy and R. C. Eberhart, Particle swarm optimization,
in Proceedings of the IEEE International Conference on Neural
Networks, vol. 4, pp. 19421948, December 1995.
[11] D. Karaboga, An idea based on honey bee swarm for numerical optimization, Tech. Rep. TR06, Computer Engineering
Department, Engineering Faculty, Erciyes University, 2005.
[12] B. Ilker, S. Birbil, and F. Shu-Cherng, An electromagnetismlike mechanism for global optimization, Journal of Global
Optimization, vol. 25, no. 3, pp. 263282, 2003.
[13] X.-S. Yang and S. Deb, Cuckoo search via Levy fights, in
Proceedings of the World Congress on Nature and Biologically
Inspired Computing (NABIC 09), pp. 210214, December 2009.
20
Research Article
Analysis of Repairable Geo/G/1 Queues with Negative Customers
Doo Ho Lee1 and Kilhwan Kim2
1
2
1. Introduction
Afer the introduction in the work of Gelenbe [1], there has
been a rapid increase in the literature on queueing systems
with negative customers due to their applications to neural
networks, communication systems, and manufacturing systems. If a negative customer arrives at a queueing system, it
removes one ordinary customer (called a positive customer)
according to a predetermined removal discipline. Tere are
two typical removal disciplines: (i) a negative customer
removes the customer being served (RCH); (ii) a negative
customer removes the customer who arrived most recently
(RCE). Te negative customers cannot accumulate in the
queue and do not receive services. Te negative customers
have a rich and wide range of real applications such as job canceling signals accompanied by machine breakdowns [2], network routing control [3], and load balancing [4]. For example,
if a certain virus enters a computer system, the virus interrupts system operations, deletes fles, and infects other fles.
In telecommunication systems, the negative customers can be
used to model cell losses caused by the arrival of a corrupted
cell, when the preceding cells of a packet would be discarded.
Te concept of the negative customer has been applied to
the M/M/1 queue [5], the M/G/1 queue [6], and the GI/M/1
queue [7]. Excellent surveys on negative customers have been
provided by Artalejo [8], Do [9, 10], and Gelenbe [1113].
22
to restore the server. Similarly, when a canceling signal for the
job in progress arrives at a database server, a system cleaning
time such as a rollback time for incomplete data operation
may be needed. For this reason, we consider the Geo/G/1
queues with negative customers under the assumption that
the server is repairable. Since the performance measure of
such a system may be infuenced by breakdown and repair
time, the repairable system is well worth analyzing in view of
queueing theory along with reliability theory.
Tere are few researches on repairable queueing systems
with negative customers. Wang and Zhang [16] considered
the Geo/Geo/1 retrial queue with negative customers and an
unreliable server, where server breakdown occurs whenever
negative customers arrive at a system. Lee et al. [21] presented
the results of repairable Geo/G/1 queues with disasters, where
if a disaster arrives at a system, all present customers (i.e., a
customer in service plus customers in queue) are forced to
leave the system. In [21], the authors considered a disaster as
server failure which leads to the destruction of all work in
process in a system. However, neither of them considered the
case of repairable queueing systems with negative customers
which cause server breakdown as well as canceling a single
ordinary customer.
Systems under our study have following features. Positive
customers arrive at a single server queue, according to a
Bernoulli process. A server provides a service to each customer on an FCFS basis. Te service times are independent
and identically distributed (i.i.d.) random variables that are
generally distributed. Negative customers arrive when the
server is busy according to a Bernoulli process. Negative
customers do not have an efect on the system if the server
is idle or under repair. Each time a negative customer arrives
at the system while the server is busy, the server fails and the
customer in service leaves the system. At a failure epoch, the
server is turned of and a repair period immediately begins.
Te server repair times are i.i.d. random variables that are
generally distributed. During the repair period, the stream of
new arrivals continues. Depending on whether new arrivals
are allowed to enter the system during the repair time, there
can be one of the two following systems. In System 1, the
customers newly arriving during the repair period cannot
enter the system and are blocked. Tis is the case when a
virus-infected computer server is isolated from the network
during the repair time without any backup server. In System 2,
they join the queue and wait for the server to be repaired. Tis
is the case when a backup server just stores requests during
the repair time and passes them to the original server when
it is recovered. As soon as the repair period ends, the server
promptly becomes available. Both Systems 1 and 2 have RCH
discipline. For each system, using the probability generating
function (PGF) technique, we present PGFs of the system size
distribution and the sojourn time distribution.
In this paper, we frst describe the mathematical model in
Section 2. In Sections 3 and 4, general results on the system
size and the sojourn time of System 1 and System 2 are
presented. Section 5 deals with numerical experiments which
we conducted to investigate the infuence of the arrival of
negative customers on the mean system size of each system.
2. Model Description
Troughout this paper, we adopt the late arrival system (LAS)
[22]. Let the time axis be marked by = 0, 1, 2, . . .. According
to the LAS model, the potential arrival of a positive customer
takes place during the interval ( , ) and the potential service
completion occurs during the interval (, + ), where + and
represent lim 0 (+||) and lim 0 (||), respectively.
Since the arrivals of a positive customer and a negative customer can occur simultaneously at a slot boundary, the order
of these events must be stated. We assume that the potential
arrival of a negative customer occurs during the interval
( , ) and immediately before the potential arrival of a positive customer. We also assume that the potential completion
of a repair time occurs during the interval ( , ) and immediately before the potential arrival of a positive customer.
We further assume that a negative customer arrival and a
repair completion do not occur at the same slot boundary
simultaneously.
We defne commonly used notations to analyze both Systems 1 and 2. Interarrival times of positive customers { }
=1
are i.i.d. discrete random variables and follow a geometric
distribution with parameter :
Pr{ = } =
= 1 ,
[ ] = 1 .
1;
0 < < 1;
(1)
[ ] = 1 .
1;
0 < < 1;
(2)
Service times { }
=1 are i.i.d. general discrete random variables. A distribution and its PGF are denoted by Pr{ = } =
( 1) and () =
=1 , respectively. Repair times
() =
=1 . We assume that { }=1 , { }=1 , { }=1 ,
0,
1,
(3)
23
0.
() = lim Pr{( ) = , ( ) = 0, ( ) = },
() = lim Pr{(+ ) = , (+ ) = 1, (+ ) = },
=1
(5)
0,
(11)
(, ) = (, + 1) + 1 (, 1).
(12)
(, )(1 1 ) = 1 () (, 1) (, 1).
(, ) =
(, ) = () ,
|| 1,
(, ) = () ,
|| 1,
=0
=1
(, ) = (, ) ,
|| 1,
(, ) = (, ) ,
|| 1.
=1
=1
(, 1) = []1 (, 1).
(15)
(16)
(, )
= 1 (, + 1)
(17)
+ [0 (, 1) + 1 1 (, 1) 0 (1 )],
3. System 1
= 1 1 (, 1)(() )
(() 1)
(, 1).
( 1)
1.
(14)
(13)
2,
1, 1,
+ ( + 1) + +1 (1) ),
(10)
0, 1,
+ (1 ( + 1) + (1)
(4)
Since we are interested in the steady-state behavior of the systems, we employ the following limiting probabilities throughout this paper to manipulate the Kolmogorov equations of the
Markov chain:
+
() = 1 (1) + (1)
0,
0 = (0 (1) + 0 + 1 (1)),
1 () = 0 (1) + 1 (1) + 0
+ (1 (1) + 1 ( + 1) + 2 (1) ),
(7)
(8)
(9)
(18)
+ ()[0 (, 1) 0 (1 )].
(1 )[0 (1 ) 0 (, 1)]
.
1 ((1 ) )
(19)
24
0 (1 )(1 (1 ))
.
0 (1 (1 )) + (1 1 )((1 ) )
(, 1) =
(21)
0 (1 )( + [])(1 (1 ))
,
0 (1 (1 )) + (1 1 )((1 ) )
(22)
=
1 + [](1 ())
[
(23)
(1 + [])(1 () ())
( + ())
],
() = Pr{ = }
=
1 ()
0 + (1 0 )(1 )
.
1 1
2,
(26)
=1
() + (1 ())()
.
1
(24)
() = Pr{ = } =
=1
1,
+ (1 )()
.
1
,1 ()
(27a)
= 0 .
Case 2 yields
(25)
,2 ()
=0
25
=0
(27b)
=1 =0
( ) = 1, (+ ) = 1}[ ()]
=1
= ( + 1)[ ()]
=1 =0
= ()[ ()]
=1 =1
+ ()[ ()]
=1 =1
()
1
[1 ( (), )
()
[ +
[] =
=1
()]
()
(1 ()
1
()].
1
1 ()
(1 + [])(1 () ())
( + ())
(30)
.
+ ( ( (), 1) 1 ( (), ))
(29)
+ Pr{ = | ( ) = } ()]
1
(28)
( (), 1)
( +
())],
()
1
[Pr{ ( ) < | ( ) = }
=1
( (), )
(1
())
()
1
+ Pr{ = | ( ) = } ()]
=1 =0
1
1
[Pr{ ( ) < | ( ) = }
[0 +
= lim Pr{( ) = , ( ) = ,
,1 () + ,2 () + ,3 ()
() =
=
Let denote the unconditional waiting time of a TC. Combining (27a), (27b), and (27c) with (14), we have () given
by
4. System 2
(27c)
26
0 = (0 (1) + 0 + 1 (1)),
(32)
(33)
1 () = 0 (1) + 1 (1) + 0
+ (1 (1) + 1 ( + 1) + 2 (1) ),
(34)
+ ( + 1) + +1 (1) ),
(35)
2,
0 + (1, 1) + (1, 1) = 1.
(36)
(, ) = 0 ((, + 1) + 1 (, 1)),
(37)
= 0 (1 () (, 1) (, 1)).
(38)
(39)
0 (() (0 ))
(, 1).
( 0 )
+ ()[0 (, 1) 0 (1 )].
0 (1 (0 ))
(, 1).
(1 0 )
(40)
(41)
(, )
(42)
+ [0 (, 1) + 1 (, 1) 0 (1 )],
1
(44)
(, 1)
=
0 (1 )(1 (1 ))
.
0 (0 )(1 (1 )) + (1 1 )((1 ) )
(46)
0 (1 (1 ))[(1 ) + 0 (1 (0 ))]
,
0 (0 )(1 (1 )) + (1 1 )((1 ) )
(47)
(1 ())
(48)
(1 )[0 (1 ) 0 (, 1)]
.
1 ((1 ) )
(43)
= 1 1 (, 1)(() )
(, )
() = 1 (1) + (1)
+ (1 ( + 1) + (1)
(, )(1 1 1 )
(31)
() = 1 ( + 1) + ( + 1) +
+ +1 ,
0 (0 ) + (1 1 0 (0 ))(1 )
.
1 1
(49)
27
(50)
(51)
(1 )(1 )
=
,
(1 )(1 ) (1 ())(1 ())
(52)
1 ()
For the negative binomial case, the PMF of the service time is
defned by
= (
1 1 4 1 4
)( ) ( ) ,
3
2
2
= 4, 5, . . . .
(55)
For the geometric mixture case, the PMF of the service time
is defned by
4 1 2 1 1 1
27 1
( )( ) + ( )( ) ,
5 3 3
5 28 28
= 1, 2, . . . .
(56)
= 1, 2, . . . .
(57)
For the negative binomial case, the PMF of the repair time is
defned by
(53)
5. Numerical Examples
Te purpose of this section is to study the infuence of negative customer arrival on the mean system size and the service
canceling probability of both System 1 and System 2. We
also investigate the infuence of the type of the repair time
distribution on the mean system size. In every example,
positive customers arrive at a system, according to a Bernoulli
process at a rate of 0.1. We consider three types of service time
distributions: a geometric distribution, a negative binomial
distribution, and a mixture of two diferent geometric distributions. Specifcally, for the geometric case, the probability
mass function (PMF) of the service time is defned by
= 1, 2, . . . .
1 2 2 3 2
)( ) ( ) ,
1
5
5
= 2, 3, . . . .
(58)
For the geometric mixture case, the PMF of the repair time is
defned by
1 7 1
= ( )( ) ,
8 8
= (
(54)
7 3
17 1
( )( )
10 20 20
3 9
1 1
+ ( )( ) ,
10 10 10
= 1, 2, . . . .
(59)
28
1
Service canceling probability
1.5
1.25
1
0.75
0.5
0.25
0.9
0.8
0.7
0.6
0.5
0.4
0
0
0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
Geo service
System 1
System 2
Geo service
System 1
System 2
Figure 1: Mean system size and service canceling probability over with geometric service time distribution.
1
Service canceling probability
2.5
2
1.5
1
0.5
0.9
0.8
0.7
0.6
0.5
0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
Figure 2: Mean system size and service canceling probability over with negative binomial service time distribution.
0.8
0.6
0.4
0.2
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.4
0.6
0.8
0.2
0.4
0.6
0.8
Figure 3: Mean system size and service canceling probability over with geometric mixture service time distribution.
29
References
2.5
2
1.5
1
0.2
0.4
0.6
0.8
Geometric mixture
6. Conclusion
We considered two repairable Geo/G/1 queueing systems
with negative customers, in which a negative customer arrival
causes the customer in service to leave the system and the
server to fail. At a failure epoch, the repair process immediately begins. Tose arriving during the repair period are
blocked in System 1, while they remain in System 2. For
these systems, we derived some important results of system
characteristics such as the PGFs of the system sizes and
FCFS sojourn times. Finally, some numerical examples were
performed to illustrate the infuence of the negative customer
arrival rate on the systems. Our research presents an extension of the discrete-time repairable queueing model and the
analysis of the model may provide a decision making tool for
repair policies arising in many practical systems.
For further studies, we can extend this model to more
complex situations such as the discrete-time repairable
queueing systems with the phase type arrival process of either
positive customers or negative customers.
Conflict of Interests
Te authors declare that there is no confict of interests
regarding the publication of this paper.
[1] E. Gelenbe, Random neural networks with negative and positive signals and product form solution, Neural Computation,
vol. 1, no. 4, pp. 502510, 1989.
[2] R. J. Boucherie and O. J. Boxma, Te workload in the M/G/1
queue with work removal, Probability in the Engineering and
Informational Sciences, vol. 10, no. 2, pp. 261277, 1996.
[3] C. Morfopoulou, Network routing control with G-networks,
Performance Evaluation, vol. 68, no. 4, pp. 320329, 2011.
[4] J. Wu and X. Yin, An M/G/1 retrial G-Queue with non-exhaustive random vacations and an unreliable server, Computers &
Mathematics with Applications, vol. 62, no. 5, pp. 23142329,
2011.
[5] P. G. Harrison and E. Pitel, Sojourn times in single-server
queues with negative customers, Journal of Applied Probability,
vol. 30, no. 4, pp. 943963, 1993.
[6] P. G. Harrison and E. Pitel, Te M/G/1 queue with negative
customers, Advances in Applied Probability, vol. 28, no. 2, pp.
540566, 1996.
[7] W. S. Yang and K. C. Chae, A note on the GI/M/1 queue with
Poisson negative arrivals, Journal of Applied Probability, vol. 38,
no. 4, pp. 10811085, 2001.
[8] J. R. Artalejo, G-networks: a versatile approach for work
removal in queueing networks, European Journal of Operational Research, vol. 126, no. 2, pp. 233249, 2000.
[9] T. V. Do, An initiative for a classifed bibliography on G-networks, Performance Evaluation, vol. 68, no. 4, pp. 385394, 2011.
[10] T. V. Do, Bibliography on G-networks, negative customers and
applications, Mathematical and Computer Modelling, vol. 53,
no. 1-2, pp. 205212, 2011.
[11] E. Gelenbe, Product-form queueing networks with negative
and positive customers, Journal of Applied Probability, vol. 28,
no. 3, pp. 656663, 1991.
[12] E. Gelenbe, G-networks: a unifying model for neural and
queueing networks, Annals of Operations Research, vol. 48, no.
5, pp. 433461, 1994.
[13] E. Gelenbe, Te frst decade of G-networks, European Journal
of Operational Research, vol. 126, no. 2, pp. 231232, 2000.
[14] I. Atencia and P. Moreno, A single-server G-queue in discretetime with geometrical arrival and service process, Performance
Evaluation, vol. 59, no. 1, pp. 8597, 2005.
[15] W. H. Zhou, Performance analysis of discrete-time queue
GI/G/1 with negative arrivals, Applied Mathematics and Computation, vol. 170, no. 2, pp. 13491355, 2005.
[16] J. Wang and P. Zhang, A discrete-time retrial queue with negative customers and unreliable server, Computers and Industrial
Engineering, vol. 56, no. 4, pp. 12161222, 2009.
[17] H. M. Park, W. S. Yang, and K. C. Chae, Te Geo//1 queue
with negative customers and disasters, Stochastic Models, vol.
25, no. 4, pp. 673688, 2009.
[18] K. C. Chae, H. M. Park, and W. S. Yang, A GI/Geo/1 queue
with negative and positive customers, Applied Mathematical
Modelling, vol. 34, no. 6, pp. 16621671, 2010.
[19] J. Wu, Z. Liu, and Y. Peng, On the BMAP/G/1 G-queues
with second optional service and multiple vacations, Applied
Mathematical Modelling, vol. 33, no. 12, pp. 43144325, 2009.
[20] J. Wu, Z. Liu, and G. Yang, Analysis of the fnite source $MAP/
PH/N$ retrial $G$-queue operating in a random environment,
Applied Mathematical Modelling: Simulation and Computation
for Engineering and Environmental Systems, vol. 35, no. 3, pp.
11841193, 2011.
30
[21] D. H. Lee, W. S. Yang, and H. M. Park, Geo/G/1 queues
with disasters an d general repair times, Applied Mathematical
Modelling, vol. 35, no. 4, pp. 15611570, 2011.
[22] H. Takagi, Queueing Analysis: A Foundation of Performance
Evaluation. Vol. 3, North-Holland, Amsterdam, Te Netherlands, 1993.
Research Article
Radial Basis Point Interpolation Method with Reordering
Gauss Domains for 2D Plane Problems
Shi-Chao Yi,1 Fu-jun Chen,1 and Lin-Quan Yao2
1
2
1. Introduction
In the past two decades, the development and application of
meshfree methods have attracted much attention. One of the
reasons is the versatility of meshfree methods for complex
geometry of solids and fexibility for diferent engineering
problems [1]. Element-free Galerkin (EFG) method, which is
originated by Belytschko et al. [25], is one of the most widely
used meshfree methods. Te key advantage of EFG method is
that only nodal data is required and no element connectivity
is needed, when moving least-squares (MLS) interpolation is
used to construct trial and test functions. It is currently widely
used in computational mechanics and other areas, such as by
Sarboland and Aminataei [6] for nonlinear nonhomogeneous
Burgers Equation and Pirali et al. [7] for crack discontinuities
problem. In the meantime, some new techniques are used
to improve the performance of the MLS, complex variable
moving least-squares [810] and improved complex variable
moving least-squares [11], the moving least-squares with singular weight function [12], and so forth. But shape functions
constructed by MLS interpolation do not possess Kronecker
delta function property; the treatment of essential boundary
conditions is one of the typical drawbacks. Tus, many special
techniques have been proposed to impose essential boundary
conditions [1315], such as point collocation [13], Lagrange
multipliers [2], singular weighting functions [14], and penalty
method [15]. None of these methods is fully satisfactory, as
32
In fact, common meshless method is based on the numerical integration for Gauss domains with RPIM. It is a tedious
process when the integration points are extremely more than
the nodes. In practical work, some Gauss point has the same
nodes as some neighboring computational points. Tus, they
only need one shape function with them. A process will be
needed to fnd these nodes. Tis scheme is called researching
Gauss points method (RGP). But the storage of searching
program will increase quickly with the increasing number of
nodes and Gauss points. Tus, more computational time is
spent. To avoid this, reconstructing Gauss domains (RGD) is
presented. All computational points share one same infuence
domain (the same nodes) in the domain. Tis scheme ofsets
the RGPs disadvantage without the searching program. Te
reason why we call it researching Gauss points method is
that the RGP method needs a searching nodes program same
as in the RPIM. Te two techniques are collectively named
reordering Gauss domain (RG) method.
Te remainder of the paper is arranged as follows.
In Section 2, the radial basis point interpolation method
(RPIM) is presented to get the shape functions. In Section 3,
a Galerkin weak form and its numerical algorithm are
studied for 2D solid mechanics problems. Te reordering
Gauss domains methods are then constructed in Section 4.
In Section 5, several examples are presented to show the
efectiveness of the proposed method and some parameters
performances of the proposed method are also investigated.
Finally, we end this paper with some conclusions in Section 6.
(4)
R0 P0 a
u
] [ ] = [ ] ,
0
PT0 0 b
(5)
aT0 = [aT bT ] = {1 , 2 , . . . , , 1 , 2 , 3 } ,
T
=1
=1
() = () + ()
(1)
( ) = 0,
=1
= 1, 2, . . . , ,
(2)
(2
+ ) ,
(3)
uT = {1 , 2 , . . . , } ,
1 (1 ) 2 (1 )
[1 (2 ) 2 (2 )
[
R0 = [ ..
..
[ .
.
[1 ( ) 2 ( )
(1 )
(2 )]
]
.. ] ,
d
. ]
( )]
(6)
1 1 1
PT0 = [1 2 ] .
[1 2 ]3
Te matrix R0 is symmetric, and thus the matrix G is
symmetric. Ten, the interpolation equation (1) is fnally
expressed as
() = [RT () pT ()] G1 u = () u ,
(7)
1,
0,
= ,
= .
(9)
33
in ,
(10)
where is the stress tensor corresponding to the displacement feld u and b is a body force vector, and boundary
conditions are as follows:
n=t
on ,
on ,
u=u
1
T D uT b uT t
2
(12)
0 { }
(13)
u = { } = [ ] { } = u .
=1
=1
{ }
Substituting (13) into (12) leads to the following total potential
energy in the matrix form:
1 T
(14)
u K u uT f,
2
and invoking = 0 results in the following linear system of
u:
=
Ku = f,
(15)
, = 1, 2, . . . , ,
f = b + t ,
where
= 1, 2, . . . , ,
(16)
, 0
B = [ 0 , ] ,
[, , ]
1 ] 0
[] 1 0 ]
D=
[
]
1 ]2 0 0 1 ]
[
2 ]
34
Nodes
Gauss points
Te center of the Gauss domain
Nodes
Gauss points
(a)
(b)
Figure 1: Te distribution of the nodes and the Gauss points: fnd the 1-1 mapping (a); assign the 1-1 mapping (b).
Tus, is written as
[1
[
S=[
[
[1
[
0]
]
]
.
]
0]
]
(19)
5. Numerical Experiments
In this section, several numerical examples are selected to
demonstrate the applicability of the RG meshless methods.
Te numerical results for these examples are compared
with the analytical solutions and the RPIM solution. Square
support domains are used for calculations in the present
paper. stands for the average node distance. Te MQ radial
basis function and the linear polynomial are used to construct
shape functions. All runs are performed in MATLAB 7.0 on
an Intel Pentium 4 (2 GB RAM) Windows XP system.
5.1. Patch Test. Te frst numerical example is the standard
patch test shown in Figure 2. Te patch test consists of thirteen nodes including fve interior nodes. A 2 2 rectangular
background mesh is used for numerical integration and 4 4
Gauss points are used in each mesh. In this patch test, the
displacements are prescribed on all outside boundaries by a
linear function of and on the patches of the dimension.
Te parameters are taken as = 1.0 and ] = 0.3. Te linear
35
Coordinates
(1.0, 1.0)
10
(0.65, 1.0)
11
(0.70, 1.5)
12
(1.3, 1.2)
13
(1.2, 0.6)
( , )
( , , )
( , )
( , , )
( , )
( , , )
( , )
( , , )
( , )
( , , )
11
12
2
10
[3]2 ( ) + (4 + 5])
+ (3 ) 2 ] ,
6
4
(20)
RGD
(0.601, 0.599)
(0.853, 0.853, 0)
(0.391, 0.599)
(0.861, 0.853, 0.002)
(0.421, 0.897)
(0.856, 0.855, 0.001)
(0.779, 0.718)
(0.846, 0.852, 0.002)
(0.720, 0.359)
(0.854, 0.856, 0)
2
[(6 3) + (2 + ]) (2
)] ,
6
4
13
RGP/RPIM
(0.601, 0.599)
(0.853, 0.853, 0)
(0.391, 0.599)
(0.861, 0.853, 0.002)
(0.422, 0.897)
(0.856, 0.855, 0.001)
(0.779, 0.718)
(0.846, 0.852, 0.002)
(0.720, 0.359)
(0.854, 0.856, 0)
Exact
(0.600, 0.600)
(0.857, 0.857, 0)
(0.390, 0.600)
(0.857, 0.857, 0)
(0.420, 0.900)
(0.857, 0.857, 0)
(0.780, 0.720)
(0.857, 0.857, 0)
(0.720, 0.360)
(0.857, 0.857, 0)
x
P
( )
,
2
(
2 ) ,
2 4
= 0.
(21)
2
u u
,
u2
,
2
(22)
36
(a)
(b)
Figure 4: Meshless models with irregular data points: 325 nodes (a), 481 nodes (b).
103
1
20
1
20
2
40
3
4
60
80
6
100
7
120
8
9
0
10
20
30
40
50
140
6
RPIM/RGP
RGD
Analytical
RPIM/RGP
RGD
Analytical
(a)
(b)
Figure 5: Results for 325 irregular nodes: (a) displacement and (b) shear stress.
RGP
(Part I + the rest)
RGD
325
CPU
3.034
0.0046
0.0013
0.0793
0.0582
1.162
481
CPU
RPIM
4.521
0.0080
0.0019
0.0824
0.0472
1.668
2.5
3.0
3.5
4.0
4.5
5.0
37
0.0513
0.0129
0.0109
0.0006
0.0004
0.0002
RGD
0.2033
0.1000
0.0945
0.0285
0.0274
0.0202
CPU
0.413
0.646
0.733
1.182
1.336
2.112
0.0074
0.0031
0.0031
0.0002
0.0021
0.0004
0.0522
0.0620
0.0305
0.0533
0.0191
0.0294
CPU
RGP/RPIM
RPIM CPU
RGP CPU (Part I + the rest)
13 4
25 7
37 10
49 13
61 16
0.0384
0.1891
0.440
0.0063
0.0399
0.839
0.0035
0.0676
1.972
0.0025
0.0668
2.730
0.0015
0.0182
5.420
0.0358
0.2057
1.864
0.219 + 0.350
0.0028
0.0522
2.450
1.404 + 0.168
0.0024
0.0559
5.637
5.943 + 2.147
0.0196
0.1434
5.981
9.975 + 3.049
0.0240
0.1738
12.302
59.036 + 6.468
CPU
RGP/RPIM
RPIM CPU
RGP CPU (Part I + the rest)
52
175
370
637
976
0.0413
0.1836
0.419
0.0062
0.0399
0.831
0.0021
0.0578
2.055
0.0055
0.0832
2.627
0.0175
0.1138
4.387
0.0386
0.1758
1.637
0.203 + 0.332
0.0003
0.0553
2.486
0.832 + 0.734
0.0020
0.0422
5.600
5.989 + 1.667
0.0038
0.0610
6.178
10.018 + 2.480
0.0003
0.0108
12.814
58.798 + 3.672
38
1.6
1.5
1.8
2
Log10 (ru )
Log10 (ru )
2
2.2
2.5
2.4
3
2.6
3.5
2.8
3
0.1
0.1
0.2
0.3
0.4
Log10 (h)
0.5
0.6
0.7
RPIM/RGP
RGD
4
0.1
0.1
0.2
0.3
0.4
Log10 (h)
0.5
0.6
0.7
RPIM/RGP
RGD
(a)
(b)
Figure 6: Convergence study with error of displacement: regular nodes (a), irregular nodes (b).
(Tables 4 and 5 and Figure 6). Tus, the RGP method is not
considered in the next section.
(a)
(b)
+ preprocess + postprocess)
(23)
(NG ( (( + )3 ) + 16 (( + )2 )
+16 ( + ))
+ preprocess + postprocess) .
39
2000
RPIM/RGP solution x
Analytical solution x
15
15
1500
1500
10
1000
500
10
1000
500
500
500
1000
10
1000
10
1500
15
1500
15
2000
2000
0
10
20
30
40
10
20
(a)
30
40
(b)
2000
RGD solution xy
15
1500
10
1000
500
500
10
1000
15
1500
2000
0
10
20
30
40
(c)
Figure 8: Stress : analytical results (a), RGP numerical results (b), and RGD numerical results (c).
+ 16 (( + )2 )
+16 ( + ))
( (( + )3 ) + 16 (( + )2 )
+16 ( + ))
16.
(24)
2 3
34
[
cos 4,
cos
2
+
cos
4]
+
2 2
24
2 1
34
[
cos 4,
cos
2
cos
4]
2 2
24
2 1
34
[
sin 4,
sin
2
+
sin
4]
+
2 2
24
(25)
where (, ) are polar coordinates, is the measured counterclockwise from the positive -axis, and is the radius of
40
Analytical solution xy
RPIM/RGP solution xy
15
15
20
20
10
10
40
40
5
5
0
60
60
80
80
10
100
10
100
15
15
120
120
0
10
20
30
40
10
20
30
(a)
40
(b)
RGD solutionx
15
20
10
40
5
0
60
80
10
100
15
120
0
10
20
30
40
(c)
Figure 9: Shear stress : analytical results (a), RGP numerical results (b), and RGD numerical results (c).
1
1
2
{ [
+ cos 2] +
[1 + (1 + ) cos 2]
4
2
4
cos 2} ,
3
2
1
4
[(1 )
3 ] sin 2,
4
(26)
Gauss points
Nodes
41
Table 6: Te complexity of the integration for a Gauss domain.
Part I
(( + ))
16 times
16 times
Complexity
RPIM
RG
Part II
(( + )3 )
16 times
1 time
Part III
(( + )2 )
16 times
16 times
Sum
16 (( + )3 ) + 16 (( + )2 ) + 16 ( + )
(( + )3 ) + 16 (( + )2 ) + 16 ( + )
Table 7: Efect of the size of support domain for the RG method (Example 4.2).
Schemes
RG
RPIM
Efciency ratio
Grid
25 13
4.7796 004
0.0204
37 17
CPU
1.267
14.991
11.832
7.3559 005
CPU
2.373
31.708
13.362
0.0054
5m
2a
x
o
5m
(a)
(b)
Figure 11: Plate with a central circular hole and its model problem: (a) a plate with a hole, (b) a quarter of the plate and boundary conditions.
1.1
1.2
1.5
1.8
2.0
3.0
0.0149
0.0097
0.0094
0.0056
0.0054
0.0044
RGD
0.1101
0.0881
0.0623
0.0886
0.1154
0.0590
CPU
0.880
0.963
1.175
1.580
2.043
4.600
0.0219
0.0205
0.0271
0.0133
0.0115
0.0057
RPIM
0.1005
0.1004
0.1526
0.2052
0.2226
01500
CPU
0.968
1.052
1.547
2.383
3.052
9.286
CPU
RPIM
CPU
99 nodes
289 nodes
625 nodes
1089 nodes
0.0149
0.1101
0.880
0.0053
0.0534
2.030
0.0025
0.0384
3.480
0.0016
0.0343
13.516
0.0219
0.1005
0.968
0.0057
0.0541
4.523
0.0031
0.0343
17.419
0.0022
0.0283
55.530
42
4.5
4.5
3.5
3.5
2.5
2.5
1.5
1.5
0.5
0.5
0
0
0
1
2
3
4
Nodes and background cells distribution
Analytical
Numerical (scaled 80)
(a)
(b)
Figure 12: Plate with a central circular hole: (a) 99 nodes distribution, (b) analytical and RGD solution of displacement.
Analytical solution xx
Numerical solution xx
5
3
4.5
4.5
2.5
4
3.5
2.5
3.5
2
3
3
1.5
Y 2.5
Y 2.5
1.5
2
1
1
1.5
1.5
1
0.5
0.5
0.5
0.5
0
0
0
0
0
0
(a)
(b)
Figure 13: Stress : analytical results (a) and RGD numerical results (b).
6. Conclusions
In this paper, we have studied researching Gauss points and
reordering Gauss domain technique (GD methods), where
the computational processes are more reasonable. Secondly,
this technique is compatible with most of the common
RPIM methods, and this aspect is worth studying. Finally,
43
Analytical solution xy
Numerical solution xy
5
0.1
4.5
4.5
0
0.1
0.1
3.5
3.5
0.2
0.2
Y 2.5
0.3
3
0.3
Y 2.5
0.4
0.4
0.5
1.5
1.5
0.6
0.5
1
1
0.7
0.6
0.5
0.5
0.8
0
0
X
(a)
(b)
Figure 14: Shear stress : analytical results (a) and RGD numerical results (b).
0.2
3.5
0
Stress y along y = 0 section
2.5
1.5
0.2
0.4
0.6
0.8
1
1
1.2
1
1.5
2.5
3
y
3.5
4.5
1.5
2.5
3
x
3.5
4.5
RGD
Analytical
RPIM
RGD
Analytical
RPIM
(a)
(b)
Figure 15: Comparisons of RGD and analytical solution: stress at the stations along the lef edge of the plate (a) and stress at the stations
along the bottom edge of the plate (b).
Conflict of Interests
References
Acknowledgments
Tis work is supported by the National Natural Science
Foundation of China under Grant nos. 11172192 and 11301290.
Finally, the authors wish to thank the editors and the
44
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
Research Article
Initial Time Difference Stability of Causal Differential Systems
in terms of Lyapunov Functions and Lyapunov Functionals
CoGkun Yakar1 and Mustafa Bayram Gcen2
1
2
1. Introduction
Lyapunovs second method is a standard technique used in
the study of the qualitative behavior of causal diferential
systems along with a comparison result [1] that allows the
prediction of behavior of a causal diferential system when
the behavior of the null solution of a comparison system is
known. However, there has been difculty with this approach
when trying to apply it to unperturbed causal diferential
systems and associated perturbed causal diferential systems
with an initial time diference. Te difculty arises because
there is a signifcant diference between initial time diference
(ITD) stability and the classical notion of stability. Te classical notions of stability are with respect to the null solution, but
ITD stability is with respect to unperturbed causal diferential
system where the perturbed causal diferential system and the
unperturbed causal diferential system difer in both initial
position and initial time [24].
In this paper, we have resolved this difculty and have a
new comparison result which again gives the null solution
a central role in the comparison diferential system. Tis
result creates many paths for continuing research by direct
application and generalization.
In Section 2, we present defnitions and necessary background material. In Section 3, we discuss and compare the
diferences between the classical notion of stability and
the recent notion of initial time diference (ITD) stability.
In Section 4, we have comparison theorems via Lyapunov
functions with initial time diference. In Section 5, we have
our main results of initial time diference stability criteria and
asymptotic stability via Lyapunov functions and we fnally
have our main result of initial time diference uniformly
asymptotic stability via Lyapunov functionals.
2. Preliminaries
In order to investigate the theory of stability for causal
diferential equations, we need comparison results in terms
of Lyapunov-like functions.
Consider the causal diferential systems
() = () () ,
() = () () ,
(0 ) = 0
(0 ) = 0
(0 ) = 0
for 0 0
(1)
for 0 0
(2)
for 0 ,
(3)
46
where , : = ([0 , ], R ) is a causal operator and = [ R : < < ]. A special case of (3) is where ()() = ()() + ()() and
()() is the perturbation term. We assume the existence
and uniqueness of solutions () and () of (1) and (3)
through (0 , 0 ) and (0 , 0 ), respectively. We need to have
some classes of functions to utilize Lyapunov-like functions
for the generalized derivative of Lyapunov functions which
has to satisfy suitable conditions as follows:
= { : (, ()) () (, ()) ()
for 0 }
+ (, )
= lim+ sup
1
[ ( + , () + () ()) (, ())]
= lim inf
1
[ ( + , () + () ()) (, ())]
(, )
0
(4)
+ (, )
= lim+ sup
0
Before we can establish our comparison theorem and Lyapunov stability criteria for initial time diference we need to
introduce the following defnitions.
Defnition 1. Te solution (, 0 , 0 ) of system (3) through
(0 , 0 ) is said to be initial time diference stable with respect
0 , 0 ) = ( , 0 , 0 ), where (, 0 , 0 )
to the solution (,
is any solution of system (1) for 0 0, 0 R+ , and
= 0 0 if and only if given any > 0 there exist 1 =
1 (, 0 ) > 0 and 2 = 2 (, 0 ) > 0 such that
(, 0 , 0 ) ( , 0 , 0 ) <
for 0 .
(5)
lim (, 0 , 0 ) ( , 0 , 0 ) = 0
(6)
(7)
for (, ) R+ R .
(b) One defnes the generalized derivatives (Dini-like
derivatives) as follows:
1
[ ( + , ()
()
+ (() () (
) ()))
(, )
= lim inf
0
(, () ())]
(8)
1
[ ( + , ()
()
+ (() () (
) ()))
(, () ())]
[ ( + , ( + , , ) ( + , , ))
(, () ())]
(, () ())
= lim inf
0
(9)
[ ( + , ( + , , ) ( + , , ))
(, () ())] ,
47
(0 ) = 0
for 0 0,
(10)
(17)
implies (, 0 , 0 ) <
for 0 .
(11)
for 0 .
(12)
for 0 . Ten
(, 0 , 0 0 ) = (, 0 , 0 ) (, 0 , 0 )
(0 0 ) ()
= ( (0 0 ) + (0 )) () ( (0 )) ()
(0 0 )) () .
(0 0 ) () = (
(13)
(14)
(, 0 , 0 0 ) = (, 0 , 0 ) ( , 0 , 0 )
(0 0 )) (; ) .
(0 0 ) () = (
(18)
()
= ( , 0 , 0 ) where (, 0 , 0 ) is any solution of
system (1).
Teorem 6. Assume that
(i) (, ) [R+ R , R+ ] and (, ) is locally
Lipschitzian in ;
(, (, () ())) ,
+ (, )
(19)
where
(15)
(16)
+ (, () ())
= lim+ sup
0
1
[ ( + , () ()
)
()))
+ (() () (
(, () ())]
(20)
(, ) R+ R and [R+ R+ , R+ ];
with (, ),
48
existing on [0 , ).
(21)
Ten, if ()
= ( , 0 , 0 ), = 0 0 , where
(, 0 , 0 ) is any solution of the causal diferential system (1)
and () = (, 0 , 0 ) is any solution of causal diferential
system (3) existing on [0 , ) such that (0 , 0 0 ) 0
implies
(, () ()) ()
0 .
(22)
= ( , 0 , 0 ), = 0 0 , and let (, 0 , 0 )
Proof. Let ()
be any solution of the causal diferential system (1) and let
() = (, 0 , 0 ) be any solution of causal diferential system
(3) for 0 . Defne
() = (, () ())
for 0
(23)
0 )) 0 . For some
so that (0 ) = (0 , (0 ) (
sufciently small > 0, consider the diferential equation
= (, ) + ,
(0 ) = 0 +
for 0
(24)
for 0 .
(25)
( + ) ()
= ( + , ( + ) ( + )) (, () ())
( + , ( + ) ( + ))
)
()))
( + , () () + (() () (
)
()))
+ ( + , () () + (() () (
(, () ())
[ ( + ) ()] [ ( + ) ()]
)
())
(() () (
)
()))
+ ( + , () () + (() () (
(, () ()) .
[1 () 2 ()]
)
()))
+ ( + , () () + (() () (
for 1 > 0 ,
(1 ) = (1 , ) .
1
= lim+ sup [1 () 2 ()]
0
+ lim+ sup
0
(26)
Consequently, ( )()
1 .
for 1 0 .
)
()))
+ (() () (
(31)
(, (, () ()))
= (, ())
for 0 1 <
1
[ ( + , () ()
(, () ())]
(30)
(, () ()) ,
(29)
(28)
+ (1 ) + (1 , (1 ) (1 ))
(1 , (, (1 ) (1 ))) = (1 , (1 ))
< (1 , (1 , )) +
(32)
for 0 . (33)
Journal of Applied
AppliedMathematics
Mathematics
14
5
0 ,
(34)
0 1 2 < .
(35)
(, (, () ()) ()) ,
= lim+ sup
0
)
()))
+ (() () (
(37)
(, ) R+ R and [R+ R+ , R+ ];
with (, ),
(iii) () = (, 0 , 0 ) is the maximal solution of the scalar
diferential equation
existing on [0 , ).
(38)
= ( , 0 , 0 ), = 0 0 , where () is any
Ten, if ()
solution of the system (1) and () = (, 0 , 0 ) is any solution
of the system (3) existing on [0 , ) such that (0 ) (0 , 0
0 ) 0 implies
(, () ()) () ()
0 .
(39)
() + (, () ()) + (, () ()) ()
(, (, () ()) ())
= (, (, () ())) .
(40)
(, () ()) () = (, () ()) ()
(42)
for 0 .
1
[ ( + , () ()
)
()))
+ (() () (
(43)
(, () ())]
(, () ())]
(0 ) = 0 0
0,
+ (, )
+ (, () ())
1
[ ( + , () ()
= (, ) ,
+ (, () ())
0
(36)
where
= lim+ sup
() + (, () ()) + (, () ()) ()
(, ) R+ ();
with (, ),
(, ) R+ () ,
, K.
(44)
(1 ) (1 ) = ,
() ()
(2 ) (2 ) = ,
for 1 2 .
(45)
(2 , (2 ) (2 )) (1 , (1 ) (1 ))
for 1 2 < .
(46)
50
() = ( (2 ) (2 )) (2 , (2 ) (2 ))
(1 , (1 ) (1 )) ( (1 ) (1 ))
(47)
= () < ()
+ (, () ()) () + (, () ()) () 0,
(48)
where () is continuously diferentiable for 0 with
(0 ) = 1, () 1, and () as ,
+ (, () ())
= lim+ sup
0
1
[ ( + , () ()
)
()))
+ (() () (
(49)
(, ) R+ ();
with (, ),
, K.
(50)
( () ())
(, () ()) (0 , 0 0 ) 1 ()
(0 0 ) 1 () < (0 ) 1 () <
(52)
(, () ())]
() (, ) () ,
for 0 . (51)
By Teorem 9, we have the solution (, 0 , 0 ) of unperturbed causal system of (1) which is stable for 0 . We
only have to prove quasiasymptotic stability. In order to do
that, let = so that 0 = (, 0 ) where (, 0 ) =
max [1 (, 0 ), 2 (, 0 )] > 0. Choose 0 0 < 0 and
|0 0 | < 2 . Ten, in view of (ii), (iii), and (51), it follows
(() ()),
K;
(, ) R+ ,
, K.
(53)
(54)
51
(1 , 0 , 0 ) (1 , 0 , 0 ) = ,
(, 0 , 0 ) (, 0 , 0 )
(55)
for 0 1 .
for 0 1 < .
(1 , (1 , 0 , 0 ) (1 , 0 , 0 ))
(0 , 0 0 ) (0 0 ) = () < ()
(57)
(58)
(0 , 0 0 ) ( (, 0 , 0 ) (, 0 , 0 ))
(59)
0 (0 + , (0 + , 0 , 0 ) (0 + , 0 , 0 ))
(0 , 0 0 )
0 +
(0 ) () < 0
for 0 + .
for 0 .
(61)
Conflict of Interests
0 0 < 0 ,
0 0 < 0
implies (, 0 , 0 ) ( , 0 , 0 ) <
0 0 < ,
0 0 < ,
(, 0 , 0 ) ( , 0 , 0 ) <
(56)
() = ( (1 , 0 , 0 ) (1 , 0 , 0 ))
( (, 0 , 0 ) (, 0 , 0 ))
(60)
Acknowledgments
Tis work has been supported by Te Yldz Technical University and Yeditepe University Department of Mathematics
and Te Scientifc and Technological Research Council of
Turkey. Te authors also would like to thank V. Lakshmikantham and the referees for their insightful comments and
detailed suggestions which improved the quality of the paper.
References
[1] V. Lakshmikantham, S. Leela, Z. Drici, and F. A. McRae, Teory
of Causal Diferential Equations, vol. 5 of Atlantis Studies in
Mathematics for Engineering and Science, 2010.
[2] V. Lakshmikantham and S. Leela, Diferential and Integral
Inequalities, vol. 1, Academic Press, New York, NY, USA, 1969.
[3] V. Lakshmikantham and A. S. Vatsala, Diferential inequalities
with initial time diference and applications, Journal of Inequalities and Applications, vol. 3, no. 3, pp. 233244, 1999.
[4] M. D. Shaw and C. Yakar, Generalized variation of parameters
with initial time diference and a comparison result in term
Lyapunov-like functions, International Journal of Nonlinear
Diferential Equations, Teory Methods and Applications, vol. 5,
pp. 86108, 1999.
[5] F. Brauer and J. Nohel, Te Qualitative Teory of Ordinary
Diferential Equations, W.A. Benjamin, New York, NY, USA,
1969.
Research Article
Bifurcation of Safe Basins and Chaos in Nonlinear Vibroimpact
Oscillator under Harmonic and Bounded Noise Excitations
Rong Haiwu,1 Wang Xiangdong,1 Luo Qizhi,1 Xu Wei,2 and Fang Tong2
1
2
1. Introduction
Nonsmooth factors arise naturally in engineering applications, such as impacts, collisions, and dry frictions [1].
A considerable amount of research activities have focused
on nonsmooth dynamical systems, including vibroimpact
systems, collisions dynamics, chattering dynamics, and stickslip motions, in recent years. In practice, engineering structures are ofen subjected to time dependent loadings of
stochastic nature, such as the natural phenomena due to wind
gusts, earthquakes, ocean waves, and random disturbance
or noise which always exists in a physical system. Te
infuence of random excitation on the dynamical behavior of
a vibroimpact dynamical system has caught the attention of
many researchers. Many efective methods have been developed, for example, linearization method used by Metrikyn
[2], quasistatic approach method used by Stratonovich [3],
exponential polynomial ftting method proposed by Zhu
[4], Markov process method used by Jing and Young [5],
stochastic averaging method used by Xu et al. [6, 7], variable
transformation method used by Zhuravlev [8], energy balance method used by Iourtchenko and Dimentberg [9], mean
impact Poincare map method used by Feng and He [10], path
integration method used by Iourtchenko and Song [11], and
53
|| = 1,
|| < 1,
(1)
(2)
1
1
2
+
].
[
2 4 ( 2 )2 + 4 4 ( + 2 )2 + 4
(3)
Obviously |()| 2 is a bounded random process. Periodicin-time excitation with a random disorder, or random phase
modulation appears in structural dynamics with traveling
loadings and/or structures, having certain imperfect spatial
periodicity in certain problems of aeroelasticity [19]. Tis
process will be assumed to be narrow-band, which is clearly
seen to be in the case provided that 0. We assume
that 2 in this paper. Te bounded noise model
(2) is a suitable model and many researches have been
done on the responses of nonlinear system under bounded
noise excitation [20, 21]. Obviously, bounded noise () has
2
sech ( 0 ) tanh ( 0 )) ,
2
( , ) = ( sech ( + 0 ) ,
0,
(4)
2
sech ( + 0 ) tanh ( + 0 )) ,
0,
where 0 = arcsech(/2).
Equation (4) is exactly the solution of the unperturbed
deterministic system (1) as = 0 without vibroimpact,
and similar system has been discussed in [17] by using the
Melnikov method. Using the results in [17, 22], one obtains
the stochastic Melnikov function for homoclinic orbits of
system (1):
() = 1 () + 2 ()
+ 1 3 () + 2 () ,
(5)
1
2
( 3 ) = 21 ,
2
1 (/2) 0
2 () =
0
+
1
[ 2 () + 2 () ]
1 (/2)
0
2
1 (/2)
sech2 ( 0 ) tanh2 ( 0 )
sech2 ( + 0 ) tanh2 ( + 0 ) ]
4
(1 tanh3 0 )
3 1 (/2)
4.5
4
3/2
=
[1 (1 ) ] ,
2
31 (/2)
3.5
0
1
[ cos (1 + 1 ) ()
1 (/2)
cos (1 + 1 ) () ]
1.5
1
0.5
1
sin (1 )
(2/) 1
() ( + ) ] .
(6)
2
,
[ () [ ()]]2 = [2 ()] = 22
[0 [ ()]]2 = [ ()]2 ,
0.4
0.3
(8)
0.5
f2
0.6
0.7
0.8
0.9
|()|2 ()
2 +
1
| ()|2 [
2
2
4 ( 2 ) + 4
() =
= 2
4 ( + 2 ) + 4
() exp () +
] ,
() exp ()
2
sech ( 0 ) tanh ( 0 ) exp ()
2
sech ( + 0 ) tanh ( + 0 ) exp ()
2 +
sech () tanh () sin ( 0 ) .
0
(9)
1 () + 2 () + 2
3 ()
(7)
= (1 () + 2 () + 1 3 ()) ,
0.2
() ( + )
= [ ()]
0.1
where
1
() =
1 (/2)
2
22
Nonchaotic area
2.5
2
(2/) 1
Chaotic area
f1
3 () =
54
(10)
= 2.4,
1 = 1.5,
= 0.1,
2 = 1.0,
= 1.8,
= 0.4.
(11)
55
1
0.8
0.4
0.2
0.4
Velocity y(t)
Displacement x(t)
0.6
0.2
0
0.2
0.4
0
0.2
0.4
0.6
0.6
0.8
1
800
820
840
860
880
900
920
940
960
980 1000
Times
0.8
1
0.2
0.8 0.6 0.4 0.2 0
Displacement x(t)
0.4
0.6
0.8
= 2.5).
Figure 2: Numerical results of (1) (1 = 2.5, 2 = 0.0, (0) = 0.8, (0)
56
0.6
0.8
0.4
0.2
0.4
Velocity y(t)
Displacement x(t)
0.6
0.2
0
0.2
0.4
0
0.2
0.4
0.6
0.6
0.8
1
800
820
840
860
880
900
920
940
960
0.8
1
980 1000
Times
0.6
0.8
0.6
0.8
= 2.5).
Figure 3: Numerical results of (1) (1 = 2.5, 2 = 0.0, (0) = 0.4, (0)
0.6
1
0.8
0.4
0.2
0.4
0.2
Velocity y(t)
Displacement x(t)
0.6
0
0.2
0.4
0
0.2
0.4
0.6
0.6
0.8
1
800
820
840
860
880
900 920
Times
940
960
980 1000
0.8
1
0.2
0.4
Displacement x(t)
= 2.5).
Figure 4: Numerical results of (1) (1 = 2.5, 2 = 0.05, (0) = 0.8, (0)
57
0.6
0.8
0.4
0.2
0.4
Velocity y(t)
Displacement x(t)
0.6
0.2
0
0.2
0.4
0
0.2
0.4
0.6
0.6
0.8
1
800
820
840
860
880
900 920
Times
940
960
0.8
1
980 1000
0.6
0.8
0.6
0.8
0.8
0.8
0.6
0.6
0.4
0.4
Velocity y(t)
Displacement x(t)
= 2.5).
Figure 5: Numerical results of (1) (1 = 2.5, 2 = 0.05, (0) = 0.4, (0)
0.2
0
0.2
0
0.2
0.2
0.4
0.4
0.6
0.6
0.8
0.8
1
800
820
840
860
880
900
920
940
960
980 1000
Times
1
1
= 2.5).
Figure 6: Numerical results of (1) (1 = 5.5, 2 = 0.05, (0) = 0.8, (0)
within the safe basin will be chaotic. According to [25, 26], the
safe basins of the system may be defned using a bounded area
in the space of phase trajectories. Te trajectory starting
from the safe basins will be stay in the area when the time
tends to infnity. Otherwise, the trajectory starting beyond the
safe basins will escape the area ; such trajectory is unstable
and may destroy or collapse the system. Te structure of the
safe basins is similar to some attractor basins. Te acreage and
shape of the safe basins will change when the parameter of the
system changes.
3.1. Bifurcation without Bounded Noises. In this paper,
the evolution of the safe basins of system (1) is studied
numerically when the parameter 1 changes its value in the
1 = 1.5,
= 0.9,
2 = 1.0,
= 0.1,
= 0.4,
= 1.8,
2 = 0.
(12)
(13)
then is divided into 100 100 lattices and the lattice points
are taken as the initial values for the solutions of system (1).
If the solution of system (1) stays in area for a long enough
time up to = 2000, such a solution can be approximately
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0.8
0.8
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.6
0.4
0.4
0.4
0.2
0.2
0.4
0.6
0.8
0.6
0.4
0.2
0.6
0.4
0.2
0.2
0.4
0.6
0.8
0.6
0.4
0.2
0.2
0.4
0.8
0.6
0.8
0.6
0.8
0.8
0.6
1
0.4
0.6
0.8
0.4
0.6
0.4
0.4
0.2
0.8
0
0.2
0.2
0.2
0.4
0.2
y
0
0.2
0.8
0.6
0.2
0.8
0.8
0.4
0.6
0.6
0.8
0.8
0.8
0.6
0.4
0.2
0.8
0.2
0.4
0.6
0.2
0.8
0.8
0.8
0.6
0.8
0.4
0.8
0.2
0.6
0.6
0.2
0.4
0.6
0.4
0.4
0.8
0.4
0.6
0.2
0.2
0.2
0.2
y
0.8
0.6
0.2
0.6
0.4
0.8
0.6
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
0.2
0.8
0.8
0.6
0.8
0.4
0.8
0.6
0.2
0.4
0.6
0.4
0.4
0.6
0.8
0.2
0.6
0.2
0.4
0
0.2
0.8
58
59
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
y
0.4
0.2
0.2
0.6
0.6
0.4
0.2
0.2
0.4
0.6
0.8
0.6
0.4
0.2
0.2
0.4
0.8
0.6
0.8
0.6
0.8
0.8
0.6
0.8
0.6
0.6
0.4
0.4
0.2
0.4
0.8
0.4
0.2
0.4
0.2
y
0
0.2
0.2
0.8
0.2
0.4
0.4
0.8
0.8
0.6
0.4
0.6
0.2
0.8
0.8
0.6
0.6
0.2
0.8
0.4
0.8
0.6
0.8
0.6
0.4
0.2
0.2
0.4
0.8
0.6
0.8
0.6
0.4
0.8
0.2
0.8
0.2
0.6
0.8
0.4
0.4
0.6
0.6
0.4
0.6
1
0.4
0.8
0.2
0.2
0.2
0.2
0.8
0.6
0.4
0.2
y
0
0.2
0.4
0.8
0.6
0.4
0.2
0.2
0.4
0.8
0.6
0.8
0.6
1 = 1.5,
= 0.9,
2 = 1.0,
= 0.1,
= 0.4,
= 1.8,
2 = 2.0,
(14)
60
an alternative defnition for stochastic bifurcation based on
the analysis of the safe basins of a sofening Dufng oscillator
subject to deterministic harmonic and bounded random excitations, which focuses on a sudden change in the character
of the safe basins of the dynamical system as the bifurcation
parameter passes through a critical value. Tis defnition
applies equally well either to the stochastic bifurcation or to
the deterministic bifurcation. However, the application of the
defnition for real systems needs more efect. Te analysis
shows that the random noise causes the two bifurcation
points and decrease; the parametric threshold for
chaotic motions varies in a larger region, therefore making
the system more unsafely and making the chaotic motions
occur more easily. From physical point of view, the results
of this paper can help one to better design the system, such
that the system operates in a nonchaotic state and then can
be controlled more easily, meanwhile reducing the erosion of
the safe basins which makes the system more secure.
In the paper, Melnikovs methods and bifurcation of safe
basins are the main research methods. In the fact, there
are other efective method to verify the chaos, for instance,
topological horseshoes method which has been successfully
applied in many works [2830].
Conflict of Interests
Te authors declare that there is no confict of interests
regarding the publication of this paper.
Acknowledgments
Te work reported in this paper was supported by the
National Natural Science Foundation of China under Grant
no. 11401096 and 11326123, the Natural Science Foundation of
Guangdong Province under Grant no. S2013010014485 and
S201310012463, and Special fund of the Guangdong College
discipline construction under Grant nos. 2013KJCX0189 and
2013B020314020.
References
[1] B. Brogliato, Non-smooth Mechanics: Models, Dynamics and
Control, Springer, London, UK, 1999.
[2] V. S. Metrikyn, On the theory of vibro-impact devices with
randomly varying parameters, Izvestiya Vysshikh Uchebnykh
Zavedenii, Radiofzika, vol. 13, pp. 412, 1970 (Russian).
[3] R. L. Stratonovich, Topics in the Teory of Random Noise, vol.
1-2, Gordon and Breach, New York, NY, USA, 1963.
[4] H. T. Zhu, Stochastic response of vibro-impact Dufng oscillators under external and parametric Gaussian white noises,
Journal of Sound and Vibration, vol. 333, no. 3, pp. 954961, 2014.
[5] H.-S. Jing and M. Young, Random response of a single-degreeof-freedom vibro-impact system with clearance, Earthquake
Engineering & Structural Dynamics, vol. 19, no. 6, pp. 789798,
1990.
[6] M. Xu, Y. Wang, X. L. Jin, Z. L. Huang, and T. X. Yu, Random response of vibro-impact systems with inelastic contact,
International Journal of Non-Linear Mechanics, vol. 52, pp. 26
31, 2013.
61
[7] C. Li, W. Xu, J. Feng, and L. Wang, Response probability density
functions of Dufng-Van der Pol vibro-impact system under
correlated Gaussian white noise excitations, Physica A, vol. 392,
no. 6, pp. 12691279, 2013.
[8] V. F. Zhuravlev, A method for analyzing vibration-impact
systems by means of special functions, Mechanics of Solids, vol.
11, pp. 2327, 1976.
[9] D. V. Iourtchenko and M. F. Dimentberg, Energy balance for
random vibrations of piecewise-conservative systems, Journal
of Sound and Vibration, vol. 248, no. 5, pp. 913923, 2001.
[10] Q. Feng and H. He, Modeling of the mean Poincare map
on a class of random impact oscillators, European Journal of
Mechanics, A: Solids, vol. 22, no. 2, pp. 267281, 2003.
[11] D. V. Iourtchenko and L. L. Song, Numerical investigation of a
response probability density function of stochastic vibroimpact
systems with inelastic impacts, International Journal of NonLinear Mechanics, vol. 41, no. 3, pp. 447455, 2006.
[12] M. F. Dimentberg, O. Gaidai, and A. Naess, Random vibrations with strongly inelastic impacts: response PDF by the
path integration method, International Journal of Non-Linear
Mechanics, vol. 44, no. 7, pp. 791796, 2009.
[13] M. F. Dimentberg and D. V. Iourtchenko, Random vibrations
with impacts: a review, Nonlinear Dynamics, vol. 36, no. 24,
pp. 229254, 2004.
[14] J. Guckenherner and P. Holmes, Nonlinear oscillations, in
Dynamical Systems and Bifurcations of Vector Fields, Springer,
New York, NY, USA, 1983.
[15] S.-N. Chow and S. W. Shaw, Bifurcations of subharmonics,
Journal of Diferential Equations, vol. 65, no. 3, pp. 304320,
1986.
[16] S. W. Shaw and R. H. Rand, Te transition to chaos in a
simple mechanical system, International Journal of Non-Linear
Mechanics, vol. 24, no. 1, pp. 4156, 1989.
[17] Z. Du and W. Zhang, Melnikov method for homoclinic
bifurcation in nonlinear impact oscillators, Computers and
Mathematics with Applications, vol. 50, no. 3-4, pp. 445458,
2005.
[18] R. Stratonovich, Topics in the Teory of Random Noise, vol. 1,
Gordon and Breach Science Publishers, New York, NY, USA,
1963.
[19] Y. K. Lin and G. Q. Cai, Probabilistic Structural Dynamics:
Advanced Teory and Applications, Mc-Graw-Hill, New York,
NY, USA, 1995.
[20] M. F. Dimentberg, D. V. Iourtchenko, and O. van Ewijk,
Subharmonic response of a quasi-isochronous vibroimpact
system to a randomly disordered periodic excitation, Nonlinear
Dynamics, vol. 17, no. 2, pp. 173186, 1998.
[21] H. Rong, X. Wang, W. Xu, and T. Fang, Subharmonic response
of a single-degree-of-freedom nonlinear vibroimpact system to
a randomly disordered periodic excitation, Journal of Sound
and Vibration, vol. 327, no. 1-2, pp. 173182, 2009.
[22] E. Simiu, Chaotic Transitions in Deterministic and Stochastic
Systems, Princeton Series in Applied Mathematics, Princeton
University Press, Princeton, NJ, USA, 2002.
[23] A. Wolf, J. B. Swif, H. L. Swinney, and J. A. Vastano, Determining Lyapunov exponents from a time series, Physica D, vol. 16,
no. 3, pp. 285317, 1985.
[24] H. Rong, X. Wang, W. Xu, and T. Fang, Resonant response of
a non-linear vibro-impact system to combined deterministic
harmonic and random excitations, International Journal of
Non-Linear Mechanics, vol. 45, no. 5, pp. 474481, 2010.
Research Article
A General Setting and Solution of
Bellman Equation in Monetary Theory
Xiaoli Gan and Wanbo Lu
School of Statistics, Southwestern University of Finance and Economics, Chengdu 610074, China
Correspondence should be addressed to Wanbo Lu; luwanbo.swufe@gmail.com
Received 28 September 2014; Revised 22 November 2014; Accepted 22 November 2014; Published 21 December 2014
Academic Editor: Zhihua Zhang
Copyright 2014 X. Gan and W. Lu. Tis is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
As an important tool in theoretical economics, Bellman equation is very powerful in solving optimization problems of discrete time
and is frequently used in monetary theory. Because there is not a general method to solve this problem in monetary theory, it is
hard to grasp the setting and solution of Bellman equation and easy to reach wrong conclusions. In this paper, we discuss the rules
and problems that should be paid attention to when incorporating money into general equilibrium models. A general setting and
solution of Bellman equation in monetary theory are provided. Te proposed method is clear, is easy to grasp, is generalized, and
always leads to the correct results.
1. Introduction
In recent years, many economists applied business cycle
approaches to macroeconomic modeling so that monetary
factors could be modeled into dynamic general equilibrium
models. As an important method of monetary economics
modeling, infnite horizon representative-agent models provide a close link between theory and practice; its research
framework can guide practice behavior and be tested by
actual data. Tis method can link monetary economics and
other popular models for studying business cycle phenomena
closely. Tere are three basic monetary economics approaches
introducing money to economic general equilibrium models
in the infnite horizon representative-agent framework. First,
it is assumed that utility could be yielded by money directly
so that the money variable has been incorporated into utility
function of the representative-agent models [1]. Second,
assuming asset exchanges are costly, transaction costs of some
form give rise to a money demand [2, 3]. Clower [4] considers
that money is used for some types of transactions. Brock [5],
McCallum and Goodfriend [6], and Croushore [7] assume
that time and money can be combined to yield transaction
services. Tird, money is treated as an asset that can be used
to transfer resources between generations [8].
63
= max { (1 , 2 , . . . , )
()
()
()
, 2,+1
, . . . , ,+1
)}
+ (1,+1
+ 1 1 (1 , . . . , , +1, , . . . , )
+ + (1 , . . . , , +1, , . . . , ) ,
(1 , 2 , . . . , )
()
+ [
[
= max (1 , 2 , . . . , )
=0
1 (1 , 2 , . . . , , +1, , . . . , ) = 0,
{
{
{
..
s.t. {
.
{
{
{ (1 , 2 , . . . , , +1, , . . . , ) = 0,
(1)
,
,
(2)
()
()
()
(1,+1
, . . . , ,+1
) 1,+1
()
()
1,+1
+ +
(3)
+ 1
()
()
()
(1,+1
, . . . , ,+1
) ,+1
]
()
()
,+1
1 (1 , . . . , )
()
(1 , . . . , )
()
= 0,
= 1, 2, . . . , .
(4)
64
()
= 1
(1 , . . . , )
()
+ +
(1 , . . . , )
()
= 1, 2, . . . , .
(5)
()
()
()
Still, it does not need to expand 1,+1
, 2,+1
, . . . , ,+1
afer the multipliers during the derivations. Requirements are
the same as Step 3.
=
=0 ( , ). Usually the per capita version of
intertemporal budget constraint is
(
1
1
) + + (
)
1+
1 + 1
(1 + 1 ) 1 + 1
+
= + + + ,
(1 + ) (1 + )
(6)
(7)
= max { ( , )
+ ( (
)
1+
+ +1 + (
+
(8)
1
)
1+
(1 + ) +
)} .
(1 + +1 ) (1 + )
65
( )
= ( , )
( ) = max { ( , )
+ ( (
) + +1
1+
+(
+
1
) ( )
1+
+ (+1 )
(9)
(1 + ) +
)} .
(1 + +1 ) (1 + )
( ) + 1
(+1 ) [
] = 0,
1+
(10)
( )
= ( , ) = ,
(11)
( ) = max { ( , )
1
) + +1 + (
)
1+
1+
(1 + ) +
)} .
(1 + +1 ) (1 + )
(14)
(+1 )
= 0.
(1 + +1 ) (1 + )
+ ( ) .
+ ( (
+ ( , )
( ) = max { ( , ) + (+1 )}
( ) + 1
] = 0,
1+
( ) = ( , ) .
( )
= ( , )
( ) = ( , ) .
(+1 )
( )
= ( , ) +
(1 + +1 ) (1 + )
(+1 ) [
(13)
[ ( ) + 1 ] = 0,
1+
( )
1 +
= (+1 )
(1 + +1 ) (1 + )
[ ( ) + 1 ] = 0,
1+
( )
= ( , ) (+1 )
1 +
= 0,
(1 + +1 ) (1 + )
( )
= ( , ) + (+1 )
(12)
( )
1 +
= ,
= (+1 )
(1 + +1 ) (1 + )
( )
1
= (+1 )
[ ( ) + 1 ] = , (16)
1+
( )
= ( , ) + (+1 )
1
= .
(1 + +1 ) (1 + )
(17)
66
1
.
(1 + +1 ) (1 + )
(18)
=0
s.t.
(19)
(1 , ) + + (1 ) 1
+
(1 + 1 ) 1 + 1
= + + + .
1 +
Because controllable factors, labor supply, and consumption afect labor supply, output is afected not only by state
(20)
= max {V [ , 1 ( , )]
+ [+1 +
(1 + ) +
, (1 , )
1 + +1
+ (1 ) 1 + ]} .
(21)
Computing the partial derivatives with respect to control
variables, we get
( , 1 )
= V [ , 1 ( , )]
V [ , 1 ( , )] ( , )
(+1 , ) = 0,
( , 1 )
= V [ , 1 ( , )]
+ (+1 , ) (1 , ) = 0,
(22)
( , 1 )
= V [ , 1 ( , )] ( , )
+ (+1 , )
1
1 + +1
(+1 , ) = 0.
= max V (+ , + )
( , 1 ) = max {V ( , ) + (+1 , )} .
( , 1 ) = (+1 , )
V (+ , 1 + (+ , + ))
=0
(23)
+ [ + (1 , ) + (1 ) 1
] ,
(24)
67
( , 1 )
1 +
= (+1 , )
= ,
1 + +1
( , 1 )
= (+1 , ) = ,
( , )
( , 1 )
= V ( , ) ( , ) + +1 = .
1 + +1
(25)
Using the envelope theorem and computing the derivatives with respect to the state variables , 1 , we get
( , 1 ) = ,
( , 1 ) = [ (1 , ) + 1 ] .
(26)
(27)
(( (+1 , )) / (1 + +1 ))
=1
=1
(( (+1 , )) / (1 + +1 ))
(+1 , ) ((1 + ) / (1 + +1 ))
1
= .
1 + 1 +
(1 ) + + (1 ) 1
+
(1 + 1 ) 1 + 1
+ + + .
1 +
In monetary theory, constraints are expressed as inequality frequently. Tis constraint describes that the representative agents time real resources should be more than
or equal to the use of it, that is, purchasing consumption,
capital, bonds, and money holdings that are then carried into
period + 1. Because of the assumption of rational agents,
the certainty problems we are discussing, and the positive
opportunity cost of money holdings, the constraints become
equations in equilibriums.
Output is only afected by state variable 1 without
considering the efect of labor supply. Following the solution
methods of Walsh [9], let the lef side of the budget constraint
be a state variable = (1 )+ +(1)1 +(((1+1 )1 +
1 )/(1 + )). It will be very tedious adopting expansion
method with two constraints. Setting Bellman equation with
multipliers will be better. Let state variable 1 in value
function get the economic signifcance of money in frstorder conditions. However, there is an overlapped setting
with ,
( , 1 ) = max { ( ) + (+1 , )}
+ ( )
+ ( 1 + ) .
1 +
(28)
(29)
(30)
1
)
1 +
(31)
1
+ ) .
1 +
68
Using the envelope theorem and computing the derivatives with respect to state variables, we get
( , 1 , 1 )
= ( , 1 , 1 ) = + ,
( , )
=0
(1 ) + + (1 ) 1
+
(1 + 1 ) 1 + 1
+ + + ,
1 +
(32)
1
+ + .
1 +
+ [ + (1 ) 1 ] .
= [ (1 , ) + (1 )]
+ (1 ) ,
( , 1 , 1 )
1
= ( , 1 , 1 ) = ( + )
.
1
1 +
(35)
When the question is to derive the efect of infation rate
on the steady-state capital-labor ratio, that is, the steadystate relationship of / and , assume that the aggregate
(1 , )
+
= ([ (+1 , , ) + (+1 , , )
(1 , ) ) ( + )
= ([ ( +1 + +1 )
= max { ( , ) + (+1 , , )}
+ [ + (1 , ) + (1 ) 1
( , 1 , 1 )
= ( , 1 , 1 )
1
(33)
( , 1 , 1 )
1 +
= (+1 , , )
= ,
1 + +1
( , 1 , 1 )
= (+1 , , ) = + ,
( , 1 , 1 )
= (+1 , , )
1
+ (+1 , , )
= ,
1 + +1
( , 1 , 1 )
= ( , ) + (1 , ) = 0.
(34)
1
1
+ ( +1 + +1 )
]
1 + +1
1 + +1
(1 , ) ) ( + )
= 2
Ten
1
]
1 + +1
+ +1
1
(1 , ) +1
.
1 + +1
+
+1 + +1
1
= 2
.
+
1 + +1 +
From (35),
(1 , ) =
( , 1 , 1 ) (1 ) ( + )
+
1
= [ (1 )]
.
(36)
(37)
(38)
1
= ()1 ( 1 ) [ (1 )] .
+
(39)
69
1 1 1
) [ (1 )]
(40)
+1 + +1
1
.
= 2
1 + +1 +
2 1
1
=[
( 1 + )
1 + +1
+1 + +1
]
+
1/(1)
. (41)
In steady-state, = +1 = , = +1 = , +1 =
, 1 = , = , from (41), we have
1/(1)
1 + 1
=[
.
( 1 + )]
(42)
= max { ( , ) + (+1 , , )}
+ [ +
1
+ (1 , )
1 +
+ (1 ) 1 ]
+ [
(1 + 1 ) 1
+ 1
1 +
1 +
+ (1 ) 1 ] .
( , 1 , 1 )
1 +
= (+1 , , )
= ,
1 + +1
( , 1 , 1 )
= (+1 , , ) = + ,
( , 1 , 1 )
= ( , ) = + ,
(43)
( , 1 , 1 )
= (+1 , , ) = ,
( , 1 , 1 )
= ( , ) + (1 , ) = 0.
(44)
Tird, compute the partial derivatives with respect to ,
1 , and 1 and the envelope theorem yields
( , 1 , 1 )
= ( , 1 , 1 ) = + ,
( , 1 , 1 )
1
= ( , 1 , 1 )
= [ (1 , ) + (1 )] + (1 ) ,
( , 1 , 1 )
1
= ( , 1 , 1 ) = ( + )
.
1
1 +
(45)
Finally, derive the steady-state capital-labor ratio by the
results above:
1/(1)
1 + 1
=[
.
( 1 + )]
(46)
7. Conclusions
As an important tool in theory economics, Bellman equation
is very powerful in solving optimization problems of discrete
time and is frequently used in monetary theory. It is hard to
grasp the setting and solution of Bellman equation and easy to
reach wrong conclusions since there is no general method to
set Bellman equation or the settings of Bellman equation are
excessively fexible. In this paper, we provide a set of general
setting and solution methods for Bellman equation with
multipliers. In the processes of solving monetary problems,
comparing with other current methods in classic reference,
our proposed method demonstrates its features of clarity,
validity, correct results, easy operation, and generalization.
Bellman equation is used not only in monetary problems
but also in almost every dynamic programming problem
associated with discrete time optimization. Our future work
is to study the applicability of the proposed method in this
paper in other areas.
Conflict of Interests
Te authors declare that there is no confict of interests
regarding the publication of this paper.
Acknowledgments
Wanbo Lus research is sponsored by the National Science
Foundation of China (71101118) and the Program for New
Century Excellent Talents in University (NCET-13-0961) in
China.
References
[1] M. Sidrauski, Rational choice and patterns of growth in a monetary economy, American Economic Review, vol. 57, no. 2, pp.
534544, 1967.
[2] W. Baumol, Te transactions demand for cash, Quarterly Journal of Economics, vol. 67, no. 4, pp. 545556, 1952.
[3] J. Tobin, Te interest elasticity of the transactions demand for
cash, Review of Economics and Statistics, vol. 38, no. 3, pp. 241
247, 1956.
[4] R. W. Clower, A reconsideration of the microfoundations of
monetary theory, Western Economic Journal, vol. 6, no. 1, pp.
19, 1967.
[5] W. A. Brock, Money and growth: the case of long run perfect
foresight, International Economic Review, vol. 15, pp. 750777,
1974.
[6] B. T. McCallum and M. S. Goodfriend, Demand for money:
theoretical studies, in Te New Palgrave Dictionary of Economics, pp. 775781, Palgrave MacMillan, Houndmills, UK,
1987.
[7] D. Croushore, Money in the utility function: functional equivalence to a shopping-time model, Journal of Macroeconomics,
vol. 15, no. 1, pp. 175182, 1993.
70
[8] P. A. Samuelson, An exact consumption-loan model of interest
with or without the social contrivance of money, Journal of
Political Economy, vol. 66, no. 6, pp. 467482, 1958.
[9] C. E. Walsh, Monetary Teory and Policy, Te MIT Press,
Cambridge, Mass, USA, 3rd edition, 2010.
[10] L. Gong, Optimization Methods in Economics, Peking University
Press, Beijing, China, 2000.
[11] D. Romer, Advanced Macroeconomics, McGraw-Hill, New York,
NY, USA, 4th edition, 2011.
[12] L. E. O. Svensson, Money and asset prices in a cash-in-advance
economy, Journal of Political Economy, vol. 93, no. 5, pp. 919
944, 1985.
Research Article
1 : 3 Resonance and Chaos in a Discrete Hindmarsh-Rose Model
Bo Li and Zhimin He
School of Mathematics and Statistics, Central South University, Changsha, Hunan 410083, China
Correspondence should be addressed to Zhimin He; hezhimin@mail.csu.edu.cn
Received 10 July 2014; Accepted 30 November 2014; Published 17 December 2014
Academic Editor: Zhidong Teng
Copyright 2014 B. Li and Z. He. Tis is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1 : 3 resonance of a two-dimensional discrete Hindmarsh-Rose model is discussed by normal form method and bifurcation theory.
Numerical simulations are presented to illustrate the theoretical analysis, which predict the occurrence of a closed invariant circle,
period-three saddle cycle, and homoclinic structure. Furthermore, it also displays the complex dynamical behaviors, especially the
transitions between three main dynamical behaviors, namely, quiescence, spiking, and bursting.
1. Introduction
When the bifurcation problem of a system
(, ) ,
R , R ,
(1)
proximity of the nullclines can be exploited to give a qualitative explanation for burst generation. Te Hindmarsh-Rose
model is known to reproduce all dynamical behaviors, such
as quiescence, spiking, bursting, irregular spiking, and irregular
bursting [4, 6]. Bifurcation analysis is examined once more in
the past, with respect to one or two bifurcation parameters
[711]. Local bifurcations and global bifurcations are also
analysed and these bifurcation phenomena can be used to
explain the transitions between the dynamical behaviors. For
example, the transition between spiking and bursting in the
model can be understood by homoclinic bifurcations [12, 13].
More information on bifurcation can be found in [1, 4, 8
11, 1422].
Recently, X. Liu and S. Liu [8] discussed the codimension2 bifurcations of the following two-dimensional HindmarshRose model:
= 3 + 2 ,
= 2 ,
(2)
72
),
( )
(
( + 2 + )
3
(4)
= .
(5)
1 + (3 + 2 )
),
2
1
= 3 + 2 1,
= 3 2 + 2 .
(8)
( 2 4) .
2
(9)
3
(2 + 3)2 + 3
= , =
, , , , , > 0} .
24
(10)
3. 1 : 3 Resonance
( , ) = (
(7)
where
+ 2 + = 0.
() := 2 (2 + ) + (1 + + 2 ) = 0,
(3)
where > 0 is the step size. In [21, 27], we proved that map
(3) possesses fip bifurcation, Neimark-Sacker bifurcation,
and 1 : 1, 1 : 2, and 1 : 4 resonance. Te aim of this paper is to
prove that this discrete model possesses the 1 : 3 resonance.
Te method we used is based on the normal form method
and bifurcation theory of discrete dynamical system (see
Kuznetsov, Sections 4 and 9 in [2]).
Tis paper is organized as follows. In Section 2, we present
the existence and local stability of fxed points for map (3). In
Section 3, we show that there exist some values of parameters
such that map (3) undergoes 1 : 3 resonance. In Section 4, we
present numerical simulations, which not only illustrate our
results with the theoretical analysis but also exhibit complex
dynamical behaviors. Finally, a brief discussion is given in
Section 5.
3 + 2 = 0,
(6)
( )
).
(
0 ( + 0 2 + )
(11)
+ ( + )
),
( ) (
( + 2 + )
(12)
(1 + 11 ) + V + 13 2 3
( ) (
),
V
21 + (1 ) V 2
(13)
73
where = (3 1)/2, we get
where
2
+ 211 ,
11 = 311
13 = 311 + ,
21 = 211 .
(1
(14)
) (0 , 0 ) , (0 , 0 ) = 0,
(23)
( ) (, ) ( ) + (, V, , ) ,
V
V
(0 , 0 ) , (0 , 0 ) = 0.
(15)
(24)
where
(, ) = (
1 + 11
),
21 1
2 3
(, V, , ) = ( 13
).
2
= (0 , 0 ) , .
(16)
3 1
(0 , 0 ) .
2
3 + 1
(0 , 0 ) (0 , 0 ) =
(0 , 0 ) ,
2
(18)
(0 , 0 ) , (0 , 0 ) = 1,
(19)
= (0 , 0 ) , (0 , 0 ) + ( , 0 , 0 )
= (0 , 0 ) , (0 , 0 ) ( (0 , 0 ) + (0 , 0 ))
+ ( (0 , 0 ) + (0 , 0 ), 0 , 0 )
(17)
= (0 , 0 ) + (0 , 0 ),
= (0 , 0 ) , (0 , 0 ) (0 , 0 ) + (0 , 0 )
+ ( (0 , 0 ) + (0 , 0 ), 0 , 0 )
3 1
+ (0 , 0 ) ,
2
( (0 , 0 )+ (0 , 0 ), 0 , 0 ) .
(26)
(20)
(0 , 0 ) ,
= (0 , 0 ) , (0 , 0 ) + (0 , 0 )
= (0 , 0 ) , (0 , 0 ) + (0 , 0 ) , (0 , 0 ) .
(21)
1
(0 , 0 ) , (0 , 0 ) = (0 , 0 ) , (0 , 0 )
(0 , 0 ) , (0 , 0 )
(0 , 0 ) , (0 , 0 ) ,
(22)
3 3
0 11 ) ,
2
3(3 + 20 11 ) + 3 3
(
,
)
60
3
Since
(25)
(27)
as (0 , 0 ) and (0 , 0 ), respectively.
By (26), map (15) can be transformed into the complex
form
3 1
1
+
(0 , 0 ) ,
2
!!
2+3
(28)
where
20 = 11 = 02
= 02 13 +
3 2
(20 313 20 11 13 ) ,
3 0
30 = 21 = 12 = 03
= 303 + 303 (3 + 20 11 ) .
(29)
74
(30)
1
1
2
+ ( 11 20 + 11
+ 02 11 + 20 02 ) 2
2
2
1
( + 11 02 ) 3 + (||4 ) .
2 02 20
03 =
3
(1 + 3) 02 20 + 311 02 + 320 02
2
+
3
3
(1 + 3) 11 02 + 03 + (3 3) 02 20 .
2
2
By setting
20 =
3
,
3 20
30 =
3 3
2
+ 30 ,
11 02 + 320
2
12 =
3 1
1
+
+ (||4 ) ,
2
!!
2+3
(32)
where
20 = 20 + 320 ,
11 = 211 + (3 3) 11 ,
02 = 02 ,
3
30 = (3 3) 20 20 + 311 02
2
21 =
3
3
2
+ 30 ,
( 3 1) 11 02 + (3 + 3) 20
2
2
1
(5 + 3) 11 20 + (2 3) 20 11
2
+ 211 11 + (1 3) 11 11 + 02 02
+
+
1
(1 + 3) 02 02 + 21
2
1
(5 3 3) 11 20 + (3 3) 11 11 ,
2
12 = 20 02 +
1
(1 3) 11 20
2
+ (20 + (3 + 3) 11 ) 11 + (3 + 1) 02 11 + 12
1
+ ( (1 3) 20 + 211 ) 02
2
1
2
(3 + 3) 11 20 (3 + 3) 11
2302 11 ,
2
3 + 3
11 ,
6
02 = 0,
3 + 23
3 3 2
20 11 +
11 + 21 ,
3
3
3 + 3
3 3
11 02
20 02 +
6
3
+
11 =
(34)
21 =
(31)
(33)
(35)
3 + 3 2 3
+ 12 ,
11
3
3 20 11
03 = 311 02 320 02 + 03 .
(36)
3 1
2
1
4
+ 02 +
+ ( ) , (37)
2
2
!!
+=3
where
30 = 30 +
3 3
30 ,
2
12 = 12 + 312 ,
By setting
30 =
3 + 3
30 ,
6
03
21 = 21 ,
3 3
03 = 03 +
03 .
2
21 = 0,
12 =
3 + 3
=
03 ,
6
3
,
3 12
(38)
(39)
3 1
2
(0 , 0 ) 2 + (4 ) ,
+ (0 , 0 ) +
2
(40)
75
1.4
1.4
1.3
1.3
1.2
1.2
1.1
1.1
x
1
0.9
0.9
0.8
0.8
0.7
1.4
1.5
1.6
1.7
1.8
1.9
0.7
1.4
1.5
1.6
(a)
1.7
1.8
1.9
(b)
1.3
1.2
1.1
x
1
0.9
0.8
0.7
0.16
2
0.15
1.9
1.8
0.14
1.7
0.13
1.6
1.5
(c)
Figure 1: 1 : 3 resonance bifurcation diagram at with = 2, = 3, = 0.8, = 0.12428002, and = 1.6771238. (a) In (, ) plane with
= 0.12428002. (b) In (, ) plane with = 0.14. (c) In (, , ) plane.
where
(0 , 0 ) =
(0 , 0 ) =
20 11 (3 + 23)
6
02
,
2
2
(3 3) 11
6
21
.
2
(41)
If 1 (0 , 0 ) = (3/2)(3 + 1)(0 , 0 ), 1 (0 , 0 ) =
3|(0 , 0 )|2 (3(1 + 3)/2)(0 , 0 ), a similar argument
as in Lemma 9.13 in [2] can be obtained.
Teorem 2. Let (0 , 0 ) . If 1 (0 , 0 ) = 0 and
Re(1 (0 , 0 )) = 0, then map (3) has the the following complex
dynamical behaviors:
(a) there is a Neimark-Sacker bifurcation curve at the
trivial fxed point 0 of map (40);
4. Numerical Simulations
In this section, the 2-dimensional and 3-dimensional bifurcation diagrams show that the 1 : 3 resonance is the degenerate
case of Neimark-Sacker bifurcation. So there exists a closed
invariant circle near the 1 : 3 resonance. Here, we provide the
following case to illustrate the dynamic behaviors of map (3).
Take parameters = 2, = 3, = 0.8, = 0 =
0.12428002, and = 0 = 1.6771238 in map (3). We know
that map (3) has a fxed point (1.1176267, 0.95523687).
Te eigenvalues of the corresponding Jacobian matrix ()
76
0.2
0.4
0.2
0.4
0.2
0
0.2
0.4
0.6
0.6
1.4
1.5
1.6
1.7
1.8
1.9
0.8
1.4
1.5
1.6
1.7
1.8
1.9
(a)
(b)
0.14
0.2
0.138
0.2
0.136
0.4
0
0.134
0.2
0.2
0.2
d 0.132
0.4
0.13
0.4
0.2
0.6
0.128
0.6
0.4
0.14
0.135
0.126
1.8
0.13
d
1.7
0.125
1.6
0.12 1.5
0.8
0.8
(c)
0.124
1.5
1.55
1.6
1.65
1.7
1.75
1.8
(d)
Figure 2: Maximum Lyapunov exponents of map (3) near 1 : 3 resonance as and vary. (a) and (b) are maximum Lyapunov exponents
corresponding to (a) and (b) in Figure 1. (c) and (d) are 3-dimensional maximum Lyapunov exponents in [1.5, 1.82] [0.124, 0.14] and
2-dimensional projection onto (, ).
77
0.9548
0.9585
0.959
0.955
0.9595
0.9552
0.9554
0.96
0.9556
0.9605
y
0.9558
0.961
0.956
0.9615
0.9562
0.962
0.9564
0.9625
0.9566
0.963
1.106
1.108
1.11
1.112
0.9568
1.1155 1.116 1.1165 1.117 1.1175 1.118 1.1185
1.114
(a)
(b)
0.955
0.86
0.9551
0.88
0.9552
0.9
0.9553
0.92
0.9554
y
y
0.9555
0.94
0.9556
0.96
0.9557
0.98
0.9558
0.9559
1.1166
1.117
1.1174
1.1178
1
0.95
1.1182
1.05
(c)
1.1
1.15
1.2
1.25
1.1
1.15
1.2
1.25
(d)
0.84
0.82
0.86
0.84
0.88
0.86
0.88
0.9
0.9
y
0.92
y
0.92
0.94
0.94
0.96
0.96
0.98
0.98
1
0.95
1.05
1.1
1.15
1.2
1
0.95
1.25
(e)
1.05
(f)
0.8
0.86
0.82
0.88
0.84
0.9
0.86
0.92
0.88
y
y
0.9
0.94
0.92
0.96
0.94
0.98
0.96
0.98
0.85 0.9 0.95
1
1
1.05
1.1
1.15
x
(g)
(h)
Figure 3: Continued.
1.2
1.25
78
0.82
0.84
0.85
0.86
0.88
0.9
0.9
y
0.92
0.95
0.94
0.96
0.98
1
1.02
0.85
0.9 0.95
1.05
1.1
1.15
1.2
1.05
0.7
1.25
0.8
0.9
(i)
1.1
1.2
1.3
1.4
(j)
0.8
0.88
0.9
0.85
0.92
0.94
y
0.9
0.96
y
0.95
0.98
1
1
1.02
1.05
0.85 0.9 0.95
1.04
1
1.05
1.1
1.15
1.2
(k)
(l)
0.9762
0.96
0.9763
0.965
0.9764
0.9765
0.97
0.9766
y
0.9767
0.975
0.9768
0.98
0.9769
0.977
0.985
0.9771
0.9772
1.0848
1.0852
1.0856
1.086
0.99
1.06
1.0864
(m)
1.07
1.08
1.09
1.1
1.11
(n)
0.92
0.93
0.94
0.95
0.96
y 0.97
0.98
0.99
1
1.01
1.02
1
1.1
x
(o)
Figure 3: Phase portraits corresponding to Figure 1 for diferent and . (a) = 1.6771238, = 0.13; (b) = 1.6771238, = 0.125; (c)
= 1.6771238, = 0.12457; (d) = 1.6771238, = 0.12; (e) = 1.6771238, = 0.115; (f) = 1.6771238, = 0.1134; (g) = 1.6771238,
= 0.09; (h) = 1.685, = 0.12428002; (i) = 1.78, = 0.12428002; (j) = 1.83, = 0.12428002; (k) = 1.75, = 0.14; (l) = 1.9,
= 0.14; (m) = 1.76, = 0.15; (n) = 1.77, = 0.15; (o) = 1.83, = 0.15.
5. Conclusion
In this paper, we investigated the 1 : 3 resonance of a discrete
Hindmarsh-Rose model. Here, we examined the fxed points
of the model in detail and showed that the model exhibits the
1 : 3 resonance. Furthermore, near 1 : 3 resonance point, the
Neimark-Sacker bifurcation and the homoclinic bifurcation
can occur. Te onset of 1 : 3 resonance means that, in some
cases, there is a region such that the model will have an
invariant circle from three-saddle cycle.
Here, we want to note that 1 : 3 resonance involves the
bifurcations of Z3 symmetric system, which are not discussed
in this paper. From the presented phase portraits, some
symmetric phenomena can be observed. Te homoclinic loop
can explain the transition between spiking and bursting.
Conflict of Interests
Te authors declare that there is no confict of interests
regarding the publication of this paper.
Acknowledgments
Te authors thank the editor and the referees for their valuable suggestions and comments which led to the improvement of the paper.
References
[1] J. Guckenheimer and P. Holmes, Nonlinear Oscillations, Dynamical Systems, and Bifurcations of Vector Fields, Springer, Berlin,
Germany, 1983.
[2] Y. A. Kuznetsov, Elements of Applied Bifurcation Teory,
Springer, New York, NY, USA, 3rd edition, 2004.
[3] S. Wiggins, Introduction to Applied Nonlinear Dynamical Systems and Chaos, vol. 2, Springer, New York, NY, USA, 2nd
edition, 2003.
[4] J. L. Hindmarsh and R. M. Rose, A model of the nerve impulse
using two frst-order diferential equations, Nature, vol. 296,
no. 5853, pp. 162164, 1982.
[5] R. FitzHugh, Impulses and physiological state in theoretical
models of nerve membrane, Biophysical Journal, vol. 1, pp. 445
467, 1961.
[6] J. Nagumo, S. Arimoto, and S. Yoshizawa, An active pulse
transmission line simulating nerve axon, Proceedings of the IRE,
vol. 50, pp. 20612070, 1962.
[7] J. M. Gonzalez-Miranda, Complex bifurcation structures in
the Hindmarsh-Rose neuron model, International Journal of
Bifurcation and Chaos, vol. 17, no. 9, pp. 30713083, 2007.
[8] X. Liu and S. Liu, Codimension-two bifurcation analysis in
two-dimensional Hindmarsh-Rose model, Nonlinear Dynamics, vol. 67, no. 1, pp. 847857, 2012.
[9] M. Storace, D. Linaro, and E. de Lange, Te HindmarshRose neuron model: bifurcation analysis and piecewise-linear
approximations, Chaos, vol. 18, no. 3, Article ID 033128, 2008.
79
[10] S. Nikolov, An alternative bifurcation analysis of the RoseHindmarsh model, Chaos, Solitons and Fractals, vol. 23, no. 5,
pp. 16431649, 2005.
[11] S. Tsuji, T. Ueta, H. Kawakami, H. Fujii, and K. Aihara, Bifurcations in two-dimensional Hindmarsh-Rose type model, International Journal of Bifurcation and Chaos, vol. 17, no. 3, pp. 985
998, 2007.
[12] D. Linaro, A. Champneys, M. Desroches, and M. Storace,
Codimension-two homoclinic bifurcations underlying spike
adding in the Hindmarsh-Rose burster, SIAM Journal on
Applied Dynamical Systems, vol. 11, no. 3, pp. 939962, 2012.
[13] A. Shilnikov and M. Kolomiets, Methods of the qualitative
theory for the Hindmarsh-Rose model: a case study. A tutorial,
International Journal of Bifurcation and Chaos, vol. 18, no. 8, pp.
21412168, 2008.
[14] Q. Chen, Z. Teng, L. Wang, and H. Jiang, Te existence of
codimension-two bifurcation in a discrete SIS epidemic model
with standard incidence, Nonlinear Dynamics. An International
Journal of Nonlinear Dynamics and Chaos in Engineering Systems, vol. 71, no. 1-2, pp. 5573, 2013.
[15] Z. He and X. Lai, Bifurcation and chaotic behavior of a discretetime predator-prey system, Nonlinear Analysis. Real World
Applications, vol. 12, no. 1, pp. 403417, 2011.
[16] Z. He and J. Qiu, Neimark-Sacker bifurcation of a third-order
rational diference equation, Journal of Diference Equations
and Applications, vol. 19, no. 9, pp. 15131522, 2013.
[17] J. L. Hindmarsh and R. M. Rose, A model of neuronal
bursting using three coupled frst order diferential equations,
Proceedings of the Royal Society of London Series B: Biological
sciences, vol. 221, no. 1222, pp. 87102, 1984.
[18] A. L. Hodgkin and A. F. Huxley, A quantitative description
of membrane current and its application to conduction and
excitation in nerve, Te Journal of Physiology, vol. 117, no. 4, pp.
500544, 1952.
[19] G. Innocenti, A. Morelli, R. Genesio, and A. Torcini, Dynamical phases of the Hindmarsh-Rose neuronal model: studies of
the transition from bursting to spiking chaos, Chaos, vol. 17, no.
4, Article ID 043128, 2007.
[20] Z. Jing, Y. Chang, and B. Guo, Bifurcation and chaos in discrete
FitzHugh-Nagumo system, Chaos, Solitons & Fractals, vol. 21,
no. 3, pp. 701720, 2004.
[21] B. Li and Z. He, Bifurcations and chaos in a two-dimensional
discrete Hindmarsh-Rose model, Nonlinear Dynamics, vol. 76,
no. 1, pp. 697715, 2014.
[22] X. Liu and D. Xiao, Complex dynamic behaviors of a discretetime predator-prey system, Chaos, Solitons & Fractals, vol. 32,
no. 1, pp. 8094, 2007.
[23] S.-S. Chen, C.-Y. Cheng, and Y.-R. Lin, Application of a
two-dimensional Hindmarsh-Rose type model for bifurcation
analysis, International Journal of Bifurcation and Chaos in
Applied Sciences and Engineering, vol. 23, no. 3, Article ID
1350055, 2013.
[24] S. Ma and Z. Feng, Fold-Hopf bifurcations of the RoseHindmarsh model with time delay, International Journal of
Bifurcation and Chaos, vol. 21, no. 2, pp. 437452, 2011.
[25] D. Terman, Te transition from bursting to continuous spiking
in excitable membrane models, Journal of Nonlinear Science,
vol. 2, no. 2, pp. 135182, 1992.
[26] X.-J. Wang, Genesis of bursting oscillations in the HindmarshRose model and homoclinicity to a chaotic saddle, Physica D:
Nonlinear Phenomena, vol. 62, no. 14, pp. 263274, 1993.
80
[27] B. Li and Z. M. He, 1 : 2 and 1 : 4 resonances in a twodimensional discrete Hindmarsh-Rose model, Nonlinear
Dynamics, 2014.
[28] A. D. Polyanin and A. I. Chernoutsan, A Concise Handbook of
Mathematics, Physics, and Engineering Science, CRC Press, New
York, NY, USA, 2011.
Research Article
Exponential Convergence for Numerical Solution of
Integral Equations Using Radial Basis Functions
Zakieh Avazzadeh,1 Mohammad Heydari,2 Wen Chen,1 and G. B. Loghmani3
1
State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering, College of Mechanics and Materials,
Hohai University, Nanjing 210098, China
2
Young Researchers and Elite Club, Islamic Azad University, Ashkezar Branch, Ashkezar 8941613695, Iran
3
Department of Mathematics, Yazd University, P.O. Box 89195-741, Yazd, Iran
Correspondence should be addressed to Zakieh Avazzadeh; z.avazzadeh@yahoo.com
Received 4 July 2014; Revised 31 October 2014; Accepted 1 November 2014; Published 10 December 2014
Academic Editor: Xiao-wei Gao
Copyright 2014 Zakieh Avazzadeh et al. Tis is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.
We solve some diferent type of Urysohn integral equations by using the radial basis functions. Tese types include the linear
and nonlinear Fredholm, Volterra, and mixed Volterra-Fredholm integral equations. Our main aim is to investigate the rate of
convergence to solve these equations using the radial basis functions which have normic structure that utilize approximation in
higher dimensions. Of course, the use of this method ofen leads to ill-posed systems. Tus we propose an algorithm to improve
the results. Numerical results show that this method leads to the exponential convergence for solving integral equations as it was
already confrmed for partial and ordinary diferential equations.
1. Introduction
Integral equations have been solved by many diferent methods [1, 2]. In [3] integral equations and methods of their
solving are classifed. Tis reference includes some traditional
methods for solving integral equations. But some recent
methods are Adomian decomposition method (ADM) [4],
homotopy perturbation method (HPM) [5], Hes variational
iteration methods [6], optimal control [7], wavelets [811],
neural networks [12], simulation method [13], block-pulse
method [14], and so forth. Also, there are many other articles
which contain new approaches in solving integral equations
[1518]. It is necessary to recall that most of the mentioned
methods are not easy for solving integral equations in higher
dimensions [1924] and also for solving the mixed VolterraFredholm cases [6, 2527]. However, in the present paper,
we restrict ourselves to the method of radial basis functions
(RBFs).
Te RBF methodology was introduced by Hardy [28].
At frst, it was popular in multivariate interpolation [2934].
In 1990, Kansa introduced a way to use these functions for
82
() = ( ) ,
(1)
c = f,
(2)
=0
c = [0 , . . . , ] ,
= ( ) ,
f = [0 , . . . , ] ,
, = 0, 1, . . . , .
(3)
Gaussian (GA): () ,
Multiquadric (MQ): (1 + ()2 )/2 , ( = 0, = 2N),
2
Madych and Nelson have proved exponential convergence property of multiquadratic approximation [31, 50]. He
has shown that under certain conditions the interpolation
error is = (/ ) (note that MQ RBF has been redefned
from Hardys original defnition by the transformation =
1/), is the mesh size, and 0 < < 1 is a constant. As is
said in [51], this implies that there are two ways to improve
the approximation: either by reducing the size of or by
increasing the magnitude of . It means that if then
0. While reducing leads to the heavy computations,
increasing is without the extra computational cost. However, according to uncertainty principle of Schaback [52],
as the error becomes smaller, the matrix becomes more illconditioned; hence the solution will break down as becomes
too large. Nevertheless, there exists a wide range of that high
accurate results can be produced. So, if we could solve the illconditioned system, we could increase and obtain the best
approximation [50]. Tere are some experimental trials about
the shape parameter, ill-conditioning, and convergence [53
55].
Tere are some methods for trade-of between and
error [56, 57]. Te golden section algorithm [56] as a new
method for fnding a good shape parameter can be efective
but ofen it is expensive. Baxter [58] investigated the preconditioned conjugate gradient technique. Casciola et al. [59]
regularized the solutions with changing the Euclidean norm
to the anisotropic norm. Karageorghis et al. [60] applied the
matrix decomposition algorithm for improving 3D elliptic
problems. Also, there are the regularization techniques for
solving ill-conditioned systems such as truncated singular
value decomposition (TSVD) and Tikhonov regularization
method. Reader can see details in [5860] and the references
there in.
3. Integral Equation
3.1. Fredholm Integral Equation. Consider the following
Fredholm integral equation of the Urysohn form:
() (, , ()) = () ,
(4)
() ( ) = () ,
=0
(5)
where = [0 , 1 , . . . , ] and () = [( 0 ), (
1 ), . . . , ( )] . Now, by replacing (5) in (4) we obtain
() (, , ()) () .
(6)
83
( ) ( , , ()) ( ) .
(7)
( ) ( , , ( )) ( ) .
=0
(8)
(, ) (, , , , (, ))
= (, ) ,
(, ) [, ] [, ] ,
(9)
() ( ) = () ,
=0
(10)
(, ) (, , , , (, ))
(, ) ,
(, ) [, ] [, ] .
(11)
( , ) ( , , , , (, ))
( , ) .
(12)
( , ) ( , , , , ( , ))
=0 =0
( , ) .
(13)
() (, , ()) = () ,
(14)
( ) ( , , ()) ( ) .
(15)
+
+
,
2
2
(16)
1
( , () , ( ())) ( ) .
2
1
(17)
( (, )) (, , , , (, ))
= (, ) ,
(, ) [, ] [, ] ,
(18)
where (, , , , (, )) and (, ) are given analytic functions. Similarly, we substitute (10) in (18) and collocate the
points ( , ), , = 0, 1, . . . , . So we have
( ( , )) ( , , , , (, ))
( , ) .
(19)
84
= () =
(20)
( ) ( )
(21)
] () , ( () , ] ()))
( , ) .
( (, )) (, , , , (, ))
= (, ) ,
(, ) [, ] [, ] ,
(22)
+
= () =
+
,
2
2
+
+
,
= ] () =
2
2
(23)
( , , ( ) ,
=0 =0
( , ) .
9
16
25
36
49
( , , () ,
1 1
] ( ) , ( ( ) , ] ( )))
(24)
GA
7.6 10
MQ
9.0 10
2.1 10
1.6 105
3.0 107
2.0 10
1.1 105
7.0 107
6.6 108
2.5 1010
4.9 108
3.7 1010
IQ
7.5 103
1.6 104
1.4 105
2.6 107
6.2 108
2.8 1010
4. Numerical Example
In this section, some examples are given to show validity and
efciency of the mentioned method. In this paper, the computations have been done using the Maple 13 with 100
digits precision. Note that digits are important factor because
the obtained systems are ill-conditioned. For improving the
results we have two ways: increasing digits and applying a
regularization method. Since decreasing of digits leads to
intensive decreasing of accuracy [54], we preferred to use
high digits instead of applying complicated regularization
methods. It can be a trade-of between increasing of complexity of mathematical operations in applying a regularization
algorithm and increasing digits.
In our practical experiments with 100 digits, even applying SVD, QR, and iterative refnement methods did not
afect the results for solving the linear systems. It must be
mentioned that although it occurred in our experiments,
it can be related to the rate of ill-conditioning. Hence,
we reported the result of solving the obtaining nonlinear
systems by Newtons iteration method without applying any
regularization method. In this study, the criterion of accuracy
is the value of infnity norm of the error function.
Example 1. Consider the following Volterra integral equation
[63]:
(, ) (2 + cos ) (, )
0
1
1
1
= sin 5 + 5 cos 2 sin (2 ) ,
4
4
4
(25)
(26)
85
Example 1
10
103
104
105
106
10
7
108
8
109
9
1010
10
20
30
40
10
20
30
40
n
Example 1
Example 2
Example 3
(b)
(a)
Example 3
Example 2
2
3
4
3
5
6
7
4
8
9
10
11
10
20
30
40
10
20
30
40
n
(c)
(d)
Figure 1: Te horizontal axis is related to and the vertical axis shows the corresponding error values. It is perceived that the relation between
and log (error) is linear. It confrms can afect error exponentially. We emphasize the spectral accuracy is the most important feature of
the method. Te relevant data is presented in Tables 1, 2, and 3.
(, ) [(, )]2
0
1
2
1
= + 5 + 3 3 5 ,
5
9
5
2
(, ) [0, 1]2 ,
(27)
86
10
Error
104
9
16
25
106
10
36
49
1010
GA
2.2 10
2.5 10
3.2 107
7.1 108
6.0 1011
1.5 1011
MQ
2.6 10
1.2 10
3.0 107
7.4 108
4
5.1 1011
2.1 1011
IQ
2.2 102
5.2 104
4.4 107
8.0 108
5.9 1011
2.0 1011
0.5
5
c
10
50 100
(, )
0
= ln (1 +
(1 2 )
(1 + ) (1 + 2 )
(1 exp(,) )
)+
,
2
1+
8 (1 + ) (1 + 2 )
where
(, ) = (, )
(, , , , (, )) (, ) ,
4
9
16
25
36
49
(28)
(29)
(30)
GA
MQ
IQ
1.2 102
2.1 103
8.0 104
1.7 102
2.4 104
7.7 104
1.2 102
2.4 103
6.5 104
5.2 106
4.9 106
7.2 106
1.0 104
1.6 105
3.1 104
2.8 105
3.4 104
1.4 105
4
9
16
25
36
49
Example 1
Example 2
Example 3
1.4 107
3.1 108
8.3 104
7.5 106
1.0 10
1.8 105
9.3 10
8.9 105
8.5 108
6.8 109
2.0 1011
5.2 1012
1.8 105
1.4 106
1.2 103
2.1 104
5.9 106
8.2 107
5. Conclusion
First, we exploit some technical methods that can be used to
improve results.
Conflict of Interests
Te authors declare that there is no confict of interests
regarding the publication of this paper.
Acknowledgments
Te work described in this paper was supported by the
National Basic Research Program of China (973 Project
no. 2010CB832702), the National Science Funds for Distinguished Young Scholars of China (1077403), NSFC Funds
(no. 11372097), the 111 project under Grant B12032. Also,
the authors greatly appreciate the precious time of Professor
Mahdi Dehghan (from Amirkabir University, Iran) who help
them generously.
References
[1] K. E. Atkinson, Te Numerical Solution of Integral Equations of
the Second Kind, Cambridge University Press, Cambridge, UK,
1997.
[2] A. J. Jerri, Introduction to Integral Equations with Applications,
John Wiley & Sons, New York, NY, USA, 1999.
[3] A. D. Polyanin and A. V. Manzhirov, Handbook of Integral Equations, Chapman & Hall/CRC, 2008.
87
[4] A.-M. Wazwaz, A First Course in Integral Equations, World Scientifc, Singapore, 1997.
[5] S. Abbasbandy, Application of Hes homotopy perturbation
method to functional integral equations, Chaos, Solitons &
Fractals, vol. 31, no. 5, pp. 12431247, 2007.
[6] S. A. Yousef, A. Lotf, and M. Dehghan, Hes variational iteration method for solving nonlinear mixed Volterra-Fredholm
integral equations, Computers & Mathematics with Applications, vol. 58, no. 11-12, pp. 21722176, 2009.
[7] S. A. Belbas, A new method for optimal control of Volterra
integral equations, Applied Mathematics and Computation, vol.
189, no. 2, pp. 19021915, 2007.
[8] Y. Mahmoudi, Wavelet Galerkin method for numerical solution of nonlinear integral equation, Applied Mathematics and
Computation, vol. 167, no. 2, pp. 11191129, 2005.
[9] K. Maleknejad, R. Mollapourasl, and M. Alizadeh, Numerical
solution of Volterra type integral equation of the frst kind with
wavelet basis, Applied Mathematics and Computation, vol. 194,
no. 2, pp. 400405, 2007.
[10] G. Vainikko, A. Kivinukk, and J. Lippus, Fast solvers of integral
equations of the second kind: wavelet methods, Journal of
Complexity, vol. 21, no. 2, pp. 243273, 2005.
[11] S. Yousef and M. Razzaghi, Legendre wavelets method for the
nonlinear Volterra-Fredholm integral equations, Mathematics
and Computers in Simulation, vol. 70, no. 1, pp. 18, 2005.
[12] A. Golbabai, M. Mammadov, and S. Seifollahi, Solving a system
of nonlinear integral equations by an RBF network, Computers
& Mathematics with Applications, vol. 57, no. 10, pp. 16511658,
2009.
[13] M. H. Kalos and P. A. Whitlock, Monte Carlo Methods, WileyBlackwell, Weinheim, Germany, 2nd edition, 2004.
[14] K. Maleknejad, M. Shahrezaee, and H. Khatami, Numerical
solution of integral equations system of the second kind by
block-pulse functions, Applied Mathematics and Computation,
vol. 166, no. 1, pp. 1524, 2005.
[15] Z. Avazzadeh, M. Heydari, and G. B. Loghmani, Numerical
solution of Fredholm integral equations of the second kind
by using integral mean value theorem, Applied Mathematical
Modelling, vol. 35, no. 5, pp. 23742383, 2011.
[16] K. Maleknejad and H. Derili, Numerical solution of integral
equations by using combination of spline-collocation method
and Lagrange interpolation, Applied Mathematics and Computation, vol. 175, no. 2, pp. 12351244, 2006.
[17] E. Mammen and K. Yu, Nonparametric estimation of noisy
integral equations of the second kind, Journal of the Korean
Statistical Society, vol. 38, no. 2, pp. 99110, 2009.
[18] H. K. Pathak, M. S. Khan, and R. Tiwari, A common fxed point
theorem and its application to nonlinear integral equations,
Computers & Mathematics with Applications, vol. 53, no. 6, pp.
961971, 2007.
[19] S. A. Belbas, Optimal control of Volterra integral equations in
two independent variables, Applied Mathematics and Computation, vol. 202, no. 2, pp. 647665, 2008.
[20] H. Guoqiang, K. Hayami, K. Sugihara, and W. Jiong, Extrapolation method of iterated collocation solution for twodimensional nonlinear Volterra integral equations, Applied
Mathematics and Computation, vol. 112, no. 1, pp. 4961, 2000.
[21] G. Han and J. Wang, Extrapolation of Nystrom solution for two
dimensional nonlinear Fredholm integral equations, Journal of
Computational and Applied Mathematics, vol. 134, no. 1-2, pp.
259268, 2001.
88
[22] M. Heydari, Z. Avazzadeh, H. Navabpour, and G. B. Loghmani,
Numerical solution of Fredholm integral equations of the
second kind by using integral mean value theorem II. High
dimensional problems, Applied Mathematical Modelling, vol.
37, no. 1-2, pp. 432442, 2013.
[23] K. Maleknejad, S. Sohrabi, and B. Baranji, Application of
2D-BPFs to nonlinear integral equations, Communications in
Nonlinear Science and Numerical Simulation, vol. 15, no. 3, pp.
527535, 2010.
[24] A. M. Wazwaz, A reliable treatment for mixed VolterraFredholm integral equations, Applied Mathematics and Computation, vol. 127, no. 2-3, pp. 405414, 2002.
[25] M. A. Abdou, K. I. Mohamed, and A. S. Ismail, On the numerical solutions of Fredholm-Volterra integral equation, Applied
Mathematics and Computation, vol. 146, no. 2-3, pp. 713728,
2003.
[26] A. A. Badr, Numerical solution of Fredholm-Volterra integral
equation in one dimension with time dependent, Applied
Mathematics and Computation, vol. 167, no. 2, pp. 11561161,
2005.
[27] A. Yildirim, Homotopy perturbation method for the mixed
Volterra-Fredholm integral equations, Chaos, Solitons and
Fractals, vol. 42, no. 5, pp. 27602764, 2009.
[28] R. L. Hardy, Multiquadratic equation of topology and other
irregular surface, Journal of Geophysical Research, vol. 76, pp.
19051915, 1971.
[29] M. D. Buhmann, Multivariate interpolation in odd-dimensional Euclidean spaces using multiquadrics, Constructive
Approximation, vol. 6, no. 1, pp. 2134, 1990.
[30] M. D. Buhmann and C. A. Micchelli, Multiquadric interpolation improved, Computers & Mathematics with Applications,
vol. 24, no. 12, pp. 2125, 1992.
[31] W. R. Madych and S. A. Nelson, Multivariate interpolation
and conditionally positive defnite functions, II, Mathematics
of Computation, vol. 54, no. 189, pp. 211230, 1990.
[32] W. R. Madych and S. A. Nelson, Multivariate interpolation
and conditionally positive defnite functions, Approximation
Teory and its Applications, vol. 4, no. 4, pp. 7789, 1988.
[33] C. A. Micchelli, Interpolation of scattered data: distance matrices and conditionally positive defnite functions, Constructive
Approximation, vol. 2, no. 1, pp. 1122, 1986.
[34] G. B. Wright, Radial basis function interpolation: numerical and
analytical developments [Ph.D. Dissertation], University of Colorado, 2003.
[35] E. J. Kansa, Multiquadricsa scattered data approximation
scheme with applications to computational fuid-dynamicsII
solutions to parabolic, hyperbolic and elliptic partial diferential
equations, Computers & Mathematics with Applications, vol. 19,
no. 8-9, pp. 147161, 1990.
[36] A. Makroglou, Radial basis functions in the numerical solution
of Fredholm integral and integro-diferential equations, in
Advances in Computer Methods for Partial Diferential Equations, VII, R. Vichnevetsky, D. Knight, and G. Richter, Eds., pp.
478484, Karen Hahn Imacs, New Brunswick, NJ, USA, 1992.
[37] E. A. Galperin and E. J. Kansa, Application of global optimization and radial basis functions to numerical solutions of weakly
singular Volterra integral equations, Computers & Mathematics
with Applications, vol. 43, no. 35, pp. 491499, 2002.
[38] A. Alipanah and M. Dehghan, Numerical solution of the nonlinear Fredholm integral equations by positive defnite functions, Applied Mathematics and Computation, vol. 190, no. 2,
pp. 17541761, 2007.
89
Research Article
Working with Missing Data: Imputation of
Nonresponse Items in Categorical Survey Data with
a Non-Monotone Missing Pattern
Machelle D. Wilson1 and Kerstin Lueck2,3
1
Department of Public Health Sciences, Division of Biostatistics, University of California, Davis, Davis, CA 95616, USA
Social Psychology, Te University of Adelaide, Adelaide, SA 5005, Australia
3
Department of Integration and Confict, Max Planck Institute, 06017 Halle, Germany
2
1. Introduction
Te problem of bias due to missing data has received a good
deal of attention over the last 20 years and the correction of
bias due to item and unit nonresponse remains an important
problem for investigators using survey data [19]. For data
missing because of item nonresponse, imputation of the
missing data is ofen the best solution. However, methods
for imputing categorical data are still experimental in some
sofware releases. Many sofware packages will automatically
remove cases with missing values from the analysis, greatly
reducing the sample size, ofen causing a drastic loss of information. Additionally, if the data are not missing completely
at random, removing cases with missing items will result in
biased parameter estimates in subsequent analyses.
Durrant [10] conducted an extensive review of various
imputation methods. She showed that parameter estimates
91
the missing data process. However, in many situations, the
use of model-based approaches such as Markov chain Monte
Carlo methods (MCMC) is superior. Tese methods defne
a model or distribution for the missing data (the missing
data model) and sample from this distribution to impute the
missing values [9], hence simulating or mimicking random
sampling from the population of interest. Tese methods
have been shown via theory and simulation to converge to
the true target distribution. Sampling from the distribution
of the missing variable reduces the underestimation of the
population variance, as well as the standard errors of the
parameter estimates for the substantive model, which would
result from most other single-imputation methods such
as mean imputation or regression imputation alone. Ofen
the model-based approaches are combined with regression
modelling to perform the imputation so as to also more
fully exploit the information in the data. For categorical
data, logistic regression is a natural choice and has the
advantage of accurately modelling the distribution of the
missing data given the observed data. Te parameters are
easily estimated via the incomplete observed data [11, 12].
However, the model-based approaches in conjunction with
logistic regression become problematic for data sets with a
large number of variables. Tis is because the relationships
between the variables are modelled via cross-tabulation and
the size of the contingency table grows exponentially with
additional variables, ofen creating a situation which exceeds
the limitations of the sofware package [13]. Te limitation
can be circumvented if the missing data pattern is monotone.
However, assessing the existence of a monotone missing
pattern is equally problematic for a large number of variables.
Furthermore, for the typical researcher in education and
the social sciences, ease of implementation of the imputation
so as to devote time and energy to the substantive model
is of more interest than programming advanced statistical
algorithms. Hence, our goal here is to provide a simple
method for exploiting the modelling capabilities of SAS or
other sofware packages and circumventing the difculties for
the case of a large number of categorical variables with a nonmonotone missing data pattern.
In this paper we will review the degrees of randomness
and the implications for imputation. We will discuss MCMC
implementation of multivariate normal data and MCMC
combined with logistic regression for imputation of categorical data, the monotone missing requirement, creation of
a monotone missing pattern, and how to perform multiple
imputations using this method. Finally, we will apply the
method to data drawn from the National Latino and Asian
American Study (NLAAS) and show the use of multiple
imputations for an example substantive model using these
data.
1.1. Degrees of Randomness in Missing Data. A missing data
process is said to be missing completely at random (MCAR)
if the probability that a subject is missing a variable is
completely independent of the value of the variable and of
the values of any other variables. For example, missing data
from lost survey pages are MCAR because, presumably, the
probability that a page was lost is not in any way related
92
{ | } ()
,
{ | } ()
(1)
93
(2)
= Pr ( = 1 | 1 , 2 , . . . , )
(3)
).
(4)
1
2.1
3.0
1.9
3.2
2 3 4
0 10.2 0.5
1 11.0
1
(5)
94
1
,
=1
(6)
1
.
=1
(7)
2
1
( ) .
1 =1
= + (1 +
1
) .
(9)
1/2 ( ) (V ) ,
(10)
] .
(1 + 1/)
(11)
2. Methods
(8)
95
immigrants who resided in the contiguous United States.
Regarding the Latino sample, there were 577 Cubans, 868
Mexicans, and 614 other Latinos. Te subcategory other
Latinos included immigrants from Colombia, the Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras,
Nicaragua, and Peru. Te Asian sample consisted of 600
Chinese, 508 Filipino, and 520 Vietnamese participants and
467 other Asians. Te subcategory other Asians consisted
of Koreans, Japanese, Asian Indians, and individuals of other
Asian backgrounds.
Te NLAAS data set had one of the most comprehensive
and advanced designs ever developed. A detailed description
of the NLAAS methods of data collection has been previously
documented [2932].
Te sampling techniques consisted of three major
approaches. First, core and secondary sampling units were
selected according to probability proportionate to size, from
which household members in the continental United States
were sampled. While primary sampling units were defned as
metropolitan statistical areas, secondary sampling units were
formed from contiguous groupings of census blocks. Second,
high-density supplemental sampling was applied, using a
greater than 5% density criterion, in which Asian and Latino
groups were oversampled. Asian and Latino individuals who
did not belong to the target groups under which these
geographical areas were classifed were still eligible to take
part in NLAAS. For example, Vietnamese individuals living
in a Chinese high-density census block were eligible. Tird,
secondary respondents were recruited from households in
which one eligible participant had already been recruited and
interviewed. Secondary respondents sampling was used to
further increase the number of study participants. In all three
sampling procedures explained above weighting corrections
were applied to take into account joint probabilities of
selection.
Te NLAAS instruments were available in Cantonese,
Mandarin, Tagalog, Vietnamese, Spanish, and English. Tey
were translated using standard translation as well as backtranslation techniques. All participants received an introductory letter and the study brochure in their preferred
language. Tose who gave their consent to take part in the
study were screened and interviewed by professionals who
had linguistic and cultural backgrounds similar to those
of the sample population. Interviews were conducted with
computer-assisted interviewing sofware in the preferred
language of the participants. Face-to-face interviews with
the participants were administered in the core and highdensity samples. Exceptions were made when respondents
specifcally requested a telephone interview or when faceto-face interviewing was prohibitive. Te average length of
each interview was 2.4 hours. As a measure of quality control,
a randomly selected sample of participants with completed
interviews was contacted to validate the data.
Written consent was obtained for all study participants,
protocols, and procedures. Human subject approval was
given by the Cambridge Health Alliance, Harvard University,
the University of Michigan, and the University of Washington.
96
250
60
Frequency
Frequency
200
40
150
100
20
50
0
0
100
200
300
Missing
400
500
600
0.00
10.00
15.00
20.00
25.00
5.00
Number of years mother attended school
30.00
Frequency
Missing
400
300
Frequency
200
100
Variable
Frequency
?1
?4
?7
?10
?13
?16
?19
?22
?25
?28
?31
?34
?37
?40
?43
?46
?49
?52
?55
?58
?61
?64
?67
?72
?75
Poor
Good
Fair
How well do you write in English?
500
200
150
100
50
0
200
150
100
50
0
200
150
100
50
0
200
150
100
50
0
Excellent
Frequency
600
0.00
1.00
2.00
3.00
4.00
5.00
6.00
400
Frequency
300
100
0
0
10
20
30
97
Table 1: Results of logistic regression substantive analyses on 10 imputed data sets. Te far right column reports the percentage of the imputed
data sets which resulted in rejecting the null hypothesis that the parameter is equal to zero versus the two-sided alternative. Te other columns
represent the parameter estimates and values for the -test described in Section 1.6. = 4649, 9 (0.05) = 2.26.
Index
Child labor
English language profciency
Parents education
Race
Sex
4.40 01
1.98 01
6.83 02
3.67 01
2.25 01
Good
Fair
How well do you write English?
Poor
120
100
80
60
40
20
0
120
100
80
60
40
20
0
120
100
80
60
40
20
0
120
100
80
60
40
20
0
Excellent
Frequency
Frequency
Frequency
Frequency
1.17 02
1.65 04
8.86 05
1.26 03
1.11 03
0.000
6.000
3.2. Multiple Imputation. Te results for the multiple imputations are shown in Table 1. Findings indicate that, for all
variables in the model, all of the single tests were consistent
with the combined test.
Our approach represents a simple and very efective
method for imputation of survey data, which are ofen
ordinal or nominal. Our method combines the capability of
modelling the missing data distribution of the automated
model-based procedures, such as Bayesian MCMC methods,
commonly available in many sofware packages, while circumventing the current limitations in many of these packages
for the imputation of a large number of categorical variables.
Conflict of Interests
Te authors declare there is no confict of interests regarding
the publication of this paper.
Acknowledgment
Te project described was supported in part by the National
Center for Advancing Translational Sciences (NCATS),
3.15 03
4.00 04
2.59 04
3.72 04
3.13 03
1.51 02
6.05 04
3.73 04
1.67 03
4.55 03
3.57
8.07
3.54
8.98
3.33
% Reject
100
100
100
100
100
References
[1] P. D. Allison, Multiple imputation for missing data: a cautionary tale, Sociological Methods & Research, vol. 28, no. 3, pp. 301
309, 2000.
[2] P. D. Allison, Missing Data, Sage, Tousand Oaks, Calif, USA,
2001.
[3] P. A. Gimotty and M. B. Brown, Imputation procedures
for categorical data: their efects on the goodness-of-ft chisquare statistic, Communications in Statistics: Simulation and
Computation, vol. 19, no. 2, pp. 681703, 1990.
[4] D. B. Rubin, Multiple Imputation for Nonresponse in Surveys,
John Wiley & Sons, New York, NY, USA, 1987.
[5] D. B. Rubin, Multiple imputation afer 18+ years, Journal of the
American Statistical Association, vol. 91, no. 434, pp. 473489,
1996.
[6] U. Kohler, Surveys from inside: an assessment of unit nonresponse bias with internal criteria, Survey Research Methods, vol.
1, no. 2, pp. 5567, 2007.
[7] T. W. Smith, Survey non-response procedures in crossnational perspective: the 2005 ISSP non-response survey,
Survey Research Methods, vol. 1, no. 1, pp. 4554, 2007.
[8] A. W. Hoogendoorn and J. Daalmans, Nonresponse in the
recruitment of an internet panel based on probability sampling,
Survey Research Methods, vol. 3, no. 2, pp. 5972, 2009.
98
Author Guidelines
Submission
Manuscripts should be submitted by one of the authors of the manuscript through the online
Manuscript Tracking System. Regardless of the source of the word-processing tool, only electronic
PDF (.pdf) or Word (.doc, .docx, .rtf) files can be submitted through the MTS. There is no page
limit. Only online submissions are accepted to facilitate rapid publication and minimize
administrative costs. Submissions by anyone other than one of the authors will not be accepted. The
submitting author takes responsibility for the paper during submission and peer review. If for some
technical reason submission through the MTS is not possible, the author can contact
jam@hindawi.com for support.
Terms of Submission
Papers must be submitted on the understanding that they have not been published elsewhere and are
not currently under consideration by another journal published by Hindawi or any other publisher.
The submitting author is responsible for ensuring that the articles publication has been approved by
all the other coauthors. It is also the authors responsibility to ensure that the articles emanating
from a particular institution are submitted with the approval of the necessary institution. Only an
acknowledgment from the editorial office officially establishes the date of receipt. Further
correspondence and proofs will be sent to the author(s) before publication unless otherwise
indicated. It is a condition of submission of a paper that the authors permit editing of the paper for
readability. All inquiries concerning the publication of accepted papers should be addressed to
jam@hindawi.com
Peer Review
All manuscripts are subject to peer review and are expected to meet standards of academic
excellence. If approved by the editor, submissions will be considered by peer-reviewers, whose
identities will remain anonymous to the authors.
Concurrent Submissions
In order to ensure sufficient diversity within the authorship of the journal, authors will be limited to
having two manuscripts under review at any point in time. If an author already has two manuscripts
under review in the journal, he or she will need to wait until the review process of at least one of
these manuscripts is complete before submitting another manuscript for consideration. This policy
does not apply to Editorials or other non-peer reviewed manuscript types.
Article Processing Charges
Journal of Electrical and Computer Engineering is an open access journal. Open access charges
allow publishers to make the published material available for free to all interested online visitors.
For more details about the article processing charges of Journal of Electrical and Computer
Engineering, please visit the Article Processing Charges information page.
Units of Measurement
Units of measurement should be presented simply and concisely using System International (SI)
units.
images should be of 300 dpi resolution at least unless the resolution is intentionally set to a lower
level for scientific reasons. If a bitmap image has labels, the image and labels should be embedded
in separate layers.
Preparation of Tables
Tables should be cited consecutively in the text. Every table must have a descriptive title and if
numerical measurements are given, the units should be included in the column heading. Vertical
rules should not be used.
Proofs
Corrected proofs must be returned to the publisher within 2-3 days of receipt. The publisher will do
everything possible to ensure prompt publication. It will therefore be appreciated if the manuscripts
and figures conform from the outset to the style of the journal.
Copyright
Open Access authors retain the copyrights of their papers, and all open access articles are
distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution and reproduction in any medium, provided that the original work is
properly cited.
The use of general descriptive names, trade names, trademarks, and so forth in this publication,
even if not specifically identified, does not imply that these names are not protected by the relevant
laws and regulations.
While the advice and information in this journal are believed to be true and accurate on the date of
its going to press, neither the authors, the editors, nor the publisher can accept any legal
responsibility for any errors or omissions that may be made. The publisher makes no warranty,
express or implied, with respect to the material contained herein.
Disclosure Policy
A competing interest exists when professional judgment concerning the validity of research is
influenced by a secondary interest, such as financial gain. We require that our authors reveal any
possible conflict of interest in their submitted manuscripts.
If there is no conflict of interest, authors should state that The author(s) declare(s) that there is no
conflict of interest regarding the publication of this paper.
Contact Information
Email Addresses
Editorial Office:jam@hindawi.com
Postal Addresses
Hindawi Publishing Corporation
410 Park Avenue
15th Floor, #287 pmb
New York, NY 10022
USA
Hindawi Publishing Corporation
Cairo Free Zone
Road 5, Block E
Nasr City, Cairo 11816
Egypt
Fax Numbers
New York, USA
Philadelphia, USA
Amsterdam, The Netherlands
Brussels, Belgium
Paris, France
Milan, Italy
Zurich, Switzerland
London, England
Germany
Sydney, Australia
Tokyo, Japan
+1-212-409-8818
+1-215-893-4392
+31-20-5248282
+32-2-7062424
+33-1-53013737
+39-02-95441264
+41-44-2742823
+44-2076917066
+49-3221-10899-1460
+61-29-4751375
+81-3-44965310