Book of Abstracts
EASIAM 2011 The 7th East Asia SIAM Conference & RIMS Workshop on Methods in Industrial and Applied Mathematics
Waseda University, Kitakyushu Campus, June 27-29, 2011 http://www.oishi.info.waseda.ac.jp/~easiam2011/ The 7th East Asia SIAM Conference & RIMS Workshop on Methods in Industrial and Applied Mathematics EASIAM 2011
Kitakyushu Campus, Waseda University, Japan June 27-29, 2011
Book of Abstracts
Organaizer East Asia SIAM
Co-Sponsors Research Institute for Science and Engineering, Waseda University Research Institute for Mathematical Sciences (RIMS), Kyoto University Numerical Analysis Group, Department of Computational Science and Engineering, Nagoya University Kitakyushu City West Japan Industry and Trade Convention Association Scientific Committee President Hisashi Okamoto (Kyoto University) Vice-President Ming-Chih Lai (National Chiao Tung University) Secretary Shao Liang Zhang (Nagoya University) Ex-President Dongwoo Sheen (Seoul National University)
Member Victor Didenko (Universiti Brunei Darussalam) Zhong-Ci Shi (Chinese Academy of Sciences) Tao Tang (Hong Kong Baptist University) Wai-Ki Ching (University of Hong Kong) L. H. Wiryanto (Bandung Institute of Technology) Hendra Gunawan (Bandung Institute of Technology) Hyung-Chun Lee (Ajou University) Byeong-Chun Shin (Chonnam National University) Xiao-Qing Jin (University of Macau) Haiwei Sun (University of Macau) Mohd Omar (University of Malaya) Abdul Rahni (Universiti Sains Malaysia) Rafael P. Saldana (Ateneo de Manila University) Weizhu Bao (National University of Singapore) Zuowei Shen (National University of Singapore) I-Liang Chern (National Taiwan University) Yongwimon Lenbury (Mahidol Univeristy) Pham Ky Anh (Vietnam National University) Nguyen Van Huu (Hanoi University of Science)
1 Greeting
The 7th East Asia SIAM Conference & RIMS Workshop on Methods in Industrial and Applied Mathematics will be held at Kitakyushu Campus, Waseda University, Japan. The EASIAM conference gathers researchers in applied mathematics with respect to science, engineering and technology. The scope of the EASIAM conference covers all aspects of applied mathe- matics to science, industry, engineering and technology in the East and South East Asia. Previous EASIAM meetings were held in Hong Kong (1st), Sap- poro (2nd), Xiamen (3rd), Daejeon (4th), Brunei (5th) and Kuala Lumpur (6th). EASIAM is a section of Society for Industrial and Applied Mathemat- ics (SIAM). It promotes basic research and education in mathematics, which supports science and technology. Topics of interest include applications of partial differential equations; applied and numerical linear algebra; fluid mechanics; mathematical analy- sis of engineering problems; mathematical aspects of geophysical problems; mathematical biology and/or medicine; mathematics of material sciences; numerical analysis; operations research; optimization. We are grateful to the distinguished SIAM keynote speakers & invited speakers, who have accepted to give plenary talks:
SIAM Keynote Speakers Mitchell Luskin (University of Minnesota, USA) • Andrea Bertozzi (University of California, Los Angeles, USA) • Invited Speakers Ji Hui (National University of Singapore, Shingapore) • Yasumasa Nishiura (Hokkaido University, Japan) • Tony Wen-Hann Sheu (National Taiwan University, Republic of China) • Roslinda Nazar (National University of Malaysia, Malaysia) • Eun-Jae Park (Yonsei University, Korea) • We wish you a productive, rich, instructive edition of the EASIAM con- ference and we thank you for your participation.
Hisashi Okamoto President of the East Asia section of SIAM Shin’ichi Oishi Chair of local organizing committee
2 Acknowledfements
It is my great pleasure to thank all the participants of the 7th East Asia SIAM conference, whose latest achievements are reported here. I am particularly grateful to them since they do that when our country faces an unprecedented difficulty caused by the earthquake in March 11th, 2011, and the nuclear plant disaster right after that. On behalf of the committee of East Asia SIAM, I thank very much the keynote lecturers and the invited speakers. Their determination to give a talk in Japan was vitally important in holding the conference in this hard time. Last but not least, I would like to express my deep gratitude to organizers of Waseda University, particularly Professor Shin ’ichi Oishi, without whose efforts this conference would not be held.
Hisashi Okamoto President of the East Asia section of SIAM Kyoto University
On behalf of the local organizing committee, we would like to express out sincere thanks to Kitakyushu city for supporting the conference, and we are also grateful to Research Institute of Science and Engineering, Waseda University for supporting the conference.
Shin’ichi Oishi Chair of local organizing committee Waseda University
3 Local Organizing Committee
Chair Shin’ichi Oishi (Waseda University) Member Takeshi Ogita (Tokyo Woman’s Christian University) Katsuhisa Ozaki (Shibaura Institute of Technology) Xuefeng Liu (Waseda University) Naoya Yamanaka (Waseda University) Akitoshi Takayasu (Waseda University)
4 Free Shuttle Bus Information
Free shuttle buses are arranged for all conference participants.
Direction Hotel (Comfort Hotel Kurosaki) Conference (Collaboration Center) ↔ Departure Time Date Time Direction 27th, June 7:45–8:00 am Hotel Conference 19:45–20:00 pm Conference → Hotel 28th, June 8:30–8:45 am Hotel → Conference 17:40–18:00 pm Conference → Banquet Hall After Banquet Banquet Hall → Hotel 29th, June 8:30–8:45 am Hotel → Conference 19:15–19:30 pm Conference → Hotel →
5 Public Transportation
No. Direction Way Time Fare 1 Kitakyushu Airport Conference Place Kitakyushu Airport Access Bus 80 min 700 JPY 2 Kitakyushu Airport ↔ Kokura Sta. City Bus 40 min 600 JPY 3 Kokura Sta. ↔ Kurosaki Sta. Kagoshima Main Line 20 min 270 JPY *4 Kurosaki Sta. ↔ Conference Place Public Bus 30 min 270 JPY 5 Fukuoka Airport ↔ Hakata Sta. Fukuoka City Subway 5 min 250 JPY 6 Hakata Sta. ↔ Orio Sta. Kagoshima Main Line 47 min 910 JPY ↔ Limited Express Kirameki 30 min 1910 JPY 7 Orio Sta. Conference Place Public Bus 20 min 200 JPY 8 Orio Sta. ↔ Kurosaki Sta. Kagoshima Main Line 4 min 200 JPY * For more information,↔ see next page.
Taxi Direction Fare (about) Fukuoka Airport Conference Place 13,000 JPY Kitasyushu Airport ↔ Conference Place 14,000 JPY Hakata Sta. ↔ Conference Place 19,000 JPY Kokura Sta. ↔ Conference Place 8,000 JPY Kurosaki Sta. ↔ Conference Place 4,000 JPY Orio Sta. ↔ Conference Place 2,000 JPY ↔
6 *4. Kurosaki Station Conference Place ↔
Direction Kurosaki Station Gakkentoshi Hibikino (Conference Place) → Hour Departure Time 8 (Free Shuttle Bus Available) 9 47 10 11 12 12 52 13 14 22 15 52 16 17 07
Direction Gakkentoshi Hibikino (Conference Place) Kurosaki Station → Hour Departure Time 9 48 10 11 12 12 47 13 14 17 15 16 02 17 18 18 18 19 (Free Shuttle Bus Available)
7 Banquet
Date 28th June, 19:00–21:30 Place Hotel Harmonie Cinq ( TEL : 093-592-5401 ) Address 12-3 Otemachi Kokurakita Kitakyushu Fukuoka, Japan, 803-0814. From Conference Place • – Free Shuttle Bus (Direct only) Departure Time : 18:00 on 28th June @ Collabolation center From JR Kokura Station (2km away) • 1. Bus No.27, No.110. Bus Stop: ”Soleil-hall-move mae”, 10min. 2. Taxi 10min, about 1,000 JPY.
8 1 2 3 4 27th, June 9:00-- 9:30-- 10:30 10:40-- 12:00-- 13:20-- 15:20 15:40-- 18:00-- Keynote Lecture Invited Lecture Student Prize Media Center Room A Opening 27-1-A 27-2-A (1)(2) 27-3-A General Session Room B 27-4-B Collaboration General Session Room C Center 27-4-C General Session Room D 27-4-D
Conference Center Lunch Reception
28th, June 9:00-- 9:30-- 10:30 10:40-- 12:00-- 13:20-- 15:20 15:40-- 18:00-- Invited Lecture Invited Lecture Media Center Room A 28-1-A 28-2-A (1)(2) General Session General Session Room B 28-3-B 28-4-B Collaboration General Session General Session Banquet Room C Center 28-3-C 28-4-C Time Registration Desk General Session General Session Room D 28-3-D 28-4-D 27th, June 8:30--15:40 @RoomA 15:40--18:00 @Conference Center Conference Center Lunch
29th, June 9:00-- 9:30-- 10:30 10:40-- 12:00-- 13:20-- 15:20 15:40-- 18:00-- 28th, June 9:00--12:00 @RoomA Keynote Lecture 12:00--18:00 @Conference Center Media Center Room A 29-1-A General Session General Session General Session Room B 29-2-B 29-3-B 29-4-B 29th, June 9:00--10:40 @RoomA Collaboration General Session General Session General Session 10:40--18:00 @Conference Center Room C Center 29-2-C 29-3-C 29-4-C General Session General Session General Session Room D 29-2-D 29-3-D 29-4-D
Conference Center Lunch Dinner
1 2 3 4 Program
10 Monday, June 27, 2011
Opening Ceremony 09:00∼09:20 Room A
Chair: Shao-liang ZHANG
◦ Opening talk by the president
◦ Some remarks
27-1-A SIAM Key Note Speaker 09:30∼10:30 Room A
Chair: Hisashi OKAMOTO ◦ Mitchell Luskin : Atomistic-to-Continuum Coupling Methods ...... p.24
27-2-A(1) Invited Speaker 10:40∼11:20 Room A
Chair: I-Liang Chern ◦ Wen-Hann Sheu : On a symplectic scheme that optimizes the dispersion-relation equation of the Maxwell’s equations ...... p.26
27-2-A(2) Invited Speaker 11:20∼12:00 Room A
Chair: Xiao-qing Jin
◦ HUI JI : Sparsity-based regularizations for image restoration ...... p.27
11 27-3-A Student prize winner presentations 13:20∼15:20 Room A
Chair: Dongwoo Sheen ◦ Yongyong Cai, Weizhu Bao and Hanquan Wang : Efficient numerical methods for com- puting ground states and dynamics of dipolar Bose-Einstein condensates ...... p.28
◦ Lei Du, Tomohiro Sogabe and Shao-Liang Zhang : A variant of the IDR(s) method with quasi-minimal residual strategy ...... p.30
◦ Chiun-Chang Lee : Limit Problems of Solutions for the Coupled Nonlinear Schr¨odinger Equations and Steady-state Solutions of the Poisson-Nernst-Planck Systems ...... p.32
◦ Akitoshi Takayasu and Shin’ichi Oishi : A method of computer assisted proof for nonlinear two-point boundary value problems using higher order finite elements ...... p.33
27-4-B Numerical Linear Algebra 15:40∼17:40 Room B
Chair: T. Miyata and K. Ozaki ◦ Takafumi Miyata, Lei Du, Tomohiro SOGABE, Yusaku YAMAMOTO and Shao-Liang ZHANG : A Projection Approach to Complex Eigenvalues of a Specified Absolute Value p.34
◦ Takeshi Fukaya, Yusaku Yamamoto and Shao-Liang Zhang : Automatic Performance Tuning for the Blocked Householder QR Algorithm ...... p.36
◦ Xiao-Qing Jin, Wei-ping Shen and Chong Li : A Ulm Method for Inverse Eigenvalue Problems ...... p.38
◦ Yung-Ta Li : Folding a linear system to a second-order form and its application to structure preserving model order reduction ...... p.39
◦ Roberd Saragih : Model reduction order of bilinear system using balanced singular per- turbation ...... p.41
◦ Katsuhisa Ozaki, Takeshi Ogita and Shin’ichi Oishi : Simplified semi-static floating-point filter for 2D orientation Problem ...... p.42
12 27-4-C Mathematical Analysis 15:40∼17:40 Room C
Chair: C-A Wang and C-H. Cho ◦ Ching-An Wang and Un-Un Kuo : On the existence of steady flows in a rectangular channel with slip effect on one porous wall ...... p.44
◦ Yukihito Suzuki, Yukio Takemoto and Jiro Mizushima : Exchange of two modes of instability leading to sustained oscillations in the flow past a cylinderical obstacle . . . . p.47
◦ Chien-Hong Cho and Marcus Wunsch : Global and singular solutions to the generalized Proudman-Johnson equation ...... p.49
◦ Hendra Gunawan and Yoshihiro Sawano : Application of RKHS Theory to a Minimization Problem with Prescribed Nodes ...... p.51
◦ Kwok Kin Wong, Robert Conte and Tuen-Wai Ng : Exact meromorphic solutions of the real cubic Swift-Hohenberg equation ...... p.53
27-4-D Mathematics for Flow 15:40∼17:40 Room D
Chair: L.H.Wiryanto and D. Victor ◦ Leo Hari Wiryanto : Free-surface flow under a sluice gate of an inclined wall from deep water ...... p.55
◦ Syakila Binti Ahmad, Ioan Pop and Kuppalapalle Vajravelu : Free Convection Boundary Layer Flow from a Vertical Cone in a Porous Medium Filled with a Nanofluid ...... p.56
◦ Victor Didenko and Johan Helsing : Features of the Nystroem Method for the Sherman- Lauricella Equation on Piecewise Smooth Contours ...... p.58
◦ Changtong Luo, Shao-Liang Zhang and Zonglin Jiang : A New Evolutionary Algorithm for Ground to Flight Data Correlation ...... p.60
◦ Swaroop Nandan Bora and Santu Das : Ocean wave scattering within or over a porous medium ...... p.62
13 Tuesday, June 28, 2011
28-1-A Invited Speaker 09:30∼10:20 Room A
Chair: Hiroshi Kokubu ◦ Yasumasa Nishiura, Takeshi Watanabe and Makoto Iima : Spatially localized traveling structures and the asymptotic behavior in binary fluid convection ...... p.64
28-2-A(1) Invited Speaker 10:40∼11:20 Room A
Chair: Hyung-Chun Lee
◦ Eun-Jae Park : Adaptive mixed FEM and new Discontinuous Galerkin methods . . . p.66
28-2-A(2) Invited Speaker 11:20∼12:00 Room A
Chair: Mohd Omar ◦ Roslinda Mohd Nazar : Numerical investigation of mixed convection flow over a horizontal circular cylinder and a solid sphere in nanofluids ...... p.68
14 28-3-B Numerical Algorithm I 13:20∼15:20 Room B
Chair: H.Yasuda and C.Luo ◦ Hidenori Yasuda : Lamellar pattern and two-phase shallow water equations ...... p.70
◦ Tanay Deshpande : Implementing the Finite Volume Method on a Rectangular Grid p.72
◦ Naoya Yamanaka, Masahide Kashiwagi and Shin’ichi Oishi : Accurate and Rigorous Exponential Algorithm in Round to Nearest ...... p.73
28-3-C Phenomena, Model, Simulation I 13:20∼15:20 Room C
Chair: Tomoeda and C. Mendoza ◦ Mohd Omar : An integrated manufacturing system for time-varying demand process p.75
◦ Takanori Katsura : Numerical analysis of a stationary transport equation by upwind scheme ...... p.76
◦ Akiyasu Tomoeda, Ryosuke Nishi, Kazumichi Ohtsuka and Katsuhiro Nishinari : Stable Traffic Jam Occurs from Nonlinear Saturation ...... p.78
◦ Carolina Mendoza and A.M. Mancho : Lagrangian Skeleton on the Gulf of Mexico Oil Spill ...... p.80
◦ Keita Iida, Hiroyuki Kitahata, Satoshi Nakata, Masaharu Nagayama, : Mathematical studies on the self-motion of surfactant scrapings at the air-water interface ...... p.82
◦ Hisashi Okamoto : A curious bifurcation diagram arising in a scalor reaction-diffusion equation with an integral constraint ...... p.84
15 28-3-D Applied Mathematics I 13:20∼15:20 Room D
Chair: K. Matsue and T. Mizuguchi ◦ Hiroshi Kokubu : Application of topological computation method for global dynamics of an associative memory model of the Hopfield type ...... p.85
◦ Kaname Matsue : Rigorous verification of equilibria for evolutionary equations - Existence, Uniqueness and Hyperbolicity - ...... p.86
◦ Tsuyoshi Mizuguchi, and Makoto Yomosa : Role of unstable symmetric solutions in symmetry restoring process ...... p.88
◦ Edi Cahyono and Buyung Sarita : A Diffusion Equation Representing the Dynamics of the Jakarta Composite Index (JSX) ...... p.90
◦ Buyung Sarita, Femy Puspita Arisanty and Edi Cahyono : The Trend and Probability Density Function of the Jakarta Composite Index (JSX) ...... p.91
◦ Masataka Kuwamura : Dormancy of predators in prey-predator systems ...... p.92
28-4-B Numerical Algorithm II 15:40∼17:40 Room B
Chair: M. Kimura and K. H. Kwon ◦ Masato Kimura and Kiyohiro Ishijima : Truncation error analysis for particle methods p.94
◦ Elliott Ginder, Seiro Omata and Karel Svadlenka : A numerical method for multiphase volume-preserving mean curvature flow ...... p.96
◦ Min-Hung Chen and Rong-Jhao Wu : A High-Order Discontinuous Galerkin Method for Elliptic Interface Problems ...... p.98
◦ Teng-Yao Kuo, Hsin-Chu Chen and Tzyy-Leng Horng : A Coarse-Grain Parallel Scheme for Solving Poisson Equation by Chebyshev Pseudospectral Method ...... p.99
◦ Haiwei Sun and Qin Sheng : On the stability of an oscillation-free ADI method in laser beam propagation computations ...... p.101
16 28-4-C Phenomena, Model, Simulation II 15:40∼17:40 Room C
Chair: H. Suito and S.R. Pudjaprasetya
◦ Hiroshi Suito, Takuya Ueda and Daniel Sze : Flows in thoracic aorta with torsion p.103
◦ Alex Chang : A Pore-scale Network Flow Model for Two Phase Flow: The Genesis and Migration of Gas Phase ...... p.105
◦ Santiago Madruga and Santiago Madruga : Convective transport and stability in films of binary mixtures ...... p.106
◦ Sri Redjeki Pudjaprasetya, and Elis Khatizah : Longshore Wave Breaker with Reflected Beach ...... p.108
◦ Ikha Magdalena and S.R. Pudjaprasetya : Wave Energy Dissipation in Porous Media p.110
◦ G Sharma and Madhu Jain : Multi-Compartment Modeling of Tumor Cells Interacting With Dynamic Chemotherapeutic Drug ...... p.112
28-4-D Applied Mathematics II & Networks 15:40∼17:40 Room D
Chair: K. Miura and W-K Ching ◦ Matthew Min-Hsiung Lin : On LaSalle’s Invariance Principle and Its Application to Synchronize Hyperchaotic Systems ...... p.113
◦ Keiji Miura : An unbiased estimator of noise correlations under signal drift ...... p.115
◦ Wai-Ki CHING : Construction of Probabilistic Boolean Networks: A Maximum Entropy Rate Approach ...... p.117
◦ Inseok Yang, Donggil Kim and Dongik Lee : Compensation Technique for Transmission Delay and Packet Loss in Networked Control System based on Lagrange Interpolation p.119
17 Wednesday, June 29, 2011
29-1-A SIAM Key Note Speaker 09:30∼10:30 Room A
Chair: Ming-Chih Lai
◦ Andrea Bertozzi : Mathematics of Crime ...... p.121
29-2-B Numerical Algorithms III 10:40∼12:00 Room B
Chair: T. Ichinomiya and C-T. Wu
◦ David Ni : Numerical Studies of Lorentz Transformation ...... p.122
◦ Chin-Tien Wu, Ming-Chiang Jiang and Yu-Lin Tsai : Numerical Studies on Monge Ampere Equation Arising from Free-Form Design of Geometric Optics ...... p.124
◦ Takashi Ichinomiya : Renormalization approach to solve Langevin’s equation . . . . . p.126
◦ Kil H Kwon and Lee Jaekyu : Consistent approximation-sampling with multi pre and post filtering ...... p.128
29-2-C Error Analysis 10:40∼12:00 Room C
Chair: K. Kobayashi and T. Kinoshita ◦ Yintzer Shih and R. Bruce Kellogg : A Tailored Finite Point Method for Convection Diffusion Reaction Problems with Variable Coefficients ...... p.130
◦ Kenta Kobayashi : On the interpolation constants over triangular elements ...... p.131
◦ Takehiko Kinoshita, T. Kimura and M.T. Nakao : A posteriori estimates of inverse linear ordinary differential operators ...... p.132
◦ Xuefeng LIU , Akitoshi Takayasu and Shin’ichi Oishi : Numerical verification for solution existence of elliptic PDE on arbitrary polygonal domain ...... p.133
18 29-2-D Applied Mathematics III 10:40∼12:00 Room D
Chair: C. Y. Han and K.K. Viswanathan ◦ Naoki Wada : Reconstruction of the density of tree-shaped networks from boundary measurements of waves ...... p.135
◦ Chang Yong Han and Song-Hwa Kwon : Cubic helical splines with Frenet-frame continuity p.137
◦ Viswanathan Kodakkal Kannan, Zainal Abdul Aziz and Saira Javed : Free vibration of symmetric angle-ply laminated cylindrical shells of variable thickness including shear deformation theory: spline method ...... p.139
29-3-B Bifurcation 13:20∼15:20 Room B
Chair: T. Sakajo and M. Shoji
◦ Takashi Sakajo : Point vortex equilibria enhancing forces over two parallel plates . p.141
◦ Tomoyuki Miyaji, Isamu Ohnishi and Yoshio Tsutsumi : Bifurcation analysis for the Lugiato-Lefever equation in a square ...... p.143
◦ Kuo-Chih Hung and Shin-Hwa Wang : A theorem on S-shaped bifurcation curve for a positone problem with convex-concave nonlinearity and its applications to the perturbed Gelfand problem ...... p.145
◦ Shin-Hwa Wang, Kuo-Chih Hung and Chien-Shang Yu : On the existence of a double S-shaped bifurcation curve with six solutions for a combustion problem ...... p.146
◦ Mayumi Shoji and Chika Shimizu : Appearance and disappearance of non-symmetric progressive capillary-gravity waves of deep water ...... p.148
◦ Takeshi WATANABE, Makoto IIMA and Yasumasa NISHIURA : Bifurcation structure and spontaneous pattern formation in binary fluid convection ...... p.150
19 29-3-C Optimization I 13:20∼15:20 Room C
Chair: W.J.Leong and W-H Kuo ◦ Wen-Hung Kuo : Single-machine group scheduling problems with time-dependent learning effect and position-based setup time learning effect ...... p.152
◦ Seyoung Park, Bong-Gyu Jang and Yuna Rhee : Optimal Retirement with Unemployment Risks and Market Completion ...... p.154
◦ Wah June Leong, Mahboubeh Farid and Malik Abu Hassan : Rescaled Gradient-based Methods with Fixed Stepsize for Large-scale Optimization ...... p.156
29-3-D Numerical Analysis I 13:20∼15:20 Room D
Chair: T. Ishiwata and H. Notsu ◦ Tetsuya Ishiwata : Behavior of polygonal curves by crystalline curvature flow with a driving force ...... p.157
◦ Guanyu Zhou and Norikazu Saito : Some remarks on the fictitious domain method with penalty for elliptic problems ...... p.159
◦ Norikazu Saito : Analysis of the finite volume approximation for a degenerate parabolic equation ...... p.161
◦ Takahito Kashiwabara : FEM analysis of the Stokes equations under boundary conditions of friction type ...... p.163
◦ Hirofumi Notsu, Masahiro Yamaguchi and Daishin Ueyama : A mesh generator using a self-replicating system ...... p.165
20 29-4-B Accurate and High Performance Computing 15:40∼17:40 Room B
Chair: H. Fujiwara and D. Sheen ◦ Hiroshi Fujiwara : A Remark on Numerical Instability of Complex Inverse Laplace Trans- forms using Multiple-Precision Arithmetic ...... p.167
◦ Dongwoo Sheen and Jiwoon Kim : Numerical Laplace inversion using mult-precision p.169
◦ Chenhan Yu, Weichung Wang and Dan’l Pierce : CPU-GPU Hybrid Approaches in Multifrontal Methods for Large and Sparse Linear System ...... p.170
◦ Shugo Manabe : Design and implementation of a multiple-precision system on GPU p.172
29-4-C Optimization II 15:40∼17:40 Room C
Chair: G. M. Lee and M. Jain ◦ Gue Myung Lee : On Optimality Theorems for Robust Multiobjective Optimization problems ...... p.174
◦ Sangho Kum and Yongdo Lim : The Resolvent average on symmetric cones ...... p.176
◦ Dai-Ni Hsieh, Ray-Bing Chen, Ying Hung and Weichung Wang : Optimizing Latin Hy- percube Designs by Particle Swarm with GPU Acceleration ...... p.178
◦ Madhu Jain : Maximum Entropy Approach For Optimal Repairable Mx/G/1 Queue With Bernoulli Feedback And Setup ...... p.180
◦ Afshin Ghanbarzadeh and Abbas : Application of the Bees Algorithm to Multi-Objective Optimization Engineering Problems ...... p.181
21 29-4-D Numerical Analysis II 15:40∼17:40 Room D
Chair: C. Park and S-M. Chang ◦ Chunjae Park and Dongwoo Sheen : A quadrilateral Morley element for biharmonic equations ...... p.183
◦ Shu-Ming Chang : Applying Snapback Repellers in Ecology ...... p.185
◦ Yusuke Morikura, Katsuhisa Ozaki and Shin’ichi Oishi : Verified solutions of linear systems on GPU ...... p.186
22
Abstracts
23
Atomistic-to-Continuum Coupling Methods Mitchell Luskin
School of Mathematics, University of Minnesota Minneapolis, MN 55455, USA
e-mail: [email protected] http://www.umn.edu/~luskin/
Abstract Materials often have highly singular strain fields at crack tips, disloca- tions, and grain boundaries that require the accuracy of atomistic modeling in small regions surrounding these defects. However, these localized defects interact through long-ranged elastic fields with a much larger region where the strain gradients are sufficiently small to allow accurate approximation by coarse-grained continuum finite element models. The size of the region requiring atomistic modeling is typically several orders of magnitude smaller than the size of the region that can be adequately approximated by coarse- grained continuum finite element models [6], which has motivated the devel- opment of numerical methods that couple atomistic regions with continuum regions to compute up to length scales that are sufficiently large for accurate and reliable scientific and engineering application.
Figure 1: The atomistic (red) region surrounding an edge dislocation is cou- pled to a continuum (grey) region. The O(1/r2) magnitude of the strain gradient (where r is the distance to the dislocation) implies that a small region (the core) around the dislocation requires atomistic modeling, but a coarse-grained continuum model can be accurately used outside the dis- location core. In the continuum region, the deformation of the atoms are approximated by piecewise linear interpolation from the deformation of the blue atoms, thus reducing the computational degrees of freedom and allowing significantly increased length scales to be computed.
24
During the past several years, we have developed an analysis of the sta- bility, modeling error, and iterative solution of prototypical atomistic-to- continuum coupling methods [1–3,5,6]. The formation and motion of defects such as cracks, dislocations, or grain boundaries, occurs when the configura- tion loses stability, that is, when an eigenvalue of the Hessian of the energy functional becomes negative. It is thus essential for the evaluation of the predictive capability of atomistic-to-continuum coupling methods that our theoretical analyses have investigated accuracy and stability for strains up to the onset of instability of the atomistic energy [2,6]. We have initiated a computational study of the accuracy of atomistic- to-continuum coupling methods for several benchmark problems [4]. Our theoretical analysis can successfully explain most of the computational results for these multi-dimensional benchmark problems. However, we also observe the need for additional theoretical analysis and benchmark problems to more thoroughly understand the accuracy of atomistic-to-continuum methods. Joint work with M. Dobson, B. Van Koten, Xingjie Li, and C. Ortner.
References [1] M. Dobson, M. Luskin, and C. Ortner. Stability, instability, and error of the force-based quasicontinuum approximation. Archive for Rational Mechanics and Analysis, 197:179–202, 2010. [2] M. Dobson, M. Luskin, and C. Ortner. Accuracy of quasicontinuum approximations near instabilities. Journal of the Mechanics and Physics of Solids, 58:1741–1757, 2010. arXiv:0905.2914v2. [3] M. Dobson, M. Luskin, and C. Ortner. Iterative methods for the force- based quasicontinuum approximation. Computer Methods in Applied Mechanics and Engineering, to appear. arXiv:0910.2013v3.
[4] B. Van Koten, Xingjie Li, M. Luskin, and C. Ortner. A computational and theoretical investigation of the accuracy of QC methods. In Num. Anal. of Multi. Prob.. Springer, to appear. arXiv:1012.6031. [5] M. Luskin and C. Ortner. Linear stationary iterative methods for the force-based quasicontinuum approximation. In Numerical Analysis and Multiscale Computations, volume 82 of Lect. Notes Comput. Sci. Eng. Springer Verlag, to appear. arXiv:1104.1774. [6] B. Van Koten and M. Luskin. Development and Analysis of Blended Quasicontinuum Approximations. arXiv:1008.2138.
25
On a symplectic scheme that optimizes the dispersion- relation equation of the Maxwell’s equations Tony W. H. Sheu123*, L. Y. Liang1, C. M. Mei1, J. H. Li1 1. Department of Engineering Science and Ocean Engineering, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei, Taiwan, Republic of China 2. Taida Institute of Mathematical Science (TIMS), National Taiwan University 3. Center for Quantum Science and Engineering (CQSE), National Taiwan University
Abstract
In this talk I will present an explicit finite-difference time-domain scheme in Cartesian and curvilinear coordinates for solving the Maxwell’s equations in non-staggered grids. The proposed scheme for solving the Faraday’s and Ampere’s equations theoretically preserves the discrete zero-divergence for the electrical and magnetic fields. The inherent local conservation laws in Maxwell’s equations are also preserved discretely all the time using the explicit second-order accurate symplectic partitioned Runge-Kutta temporal scheme. The remaining spatial derivative terms in the semi-discretized Faraday’s and Ampere’s equations are then discretized to provide an accurate functional relation that governs the numerical angular frequency and the wave numbers in two space dimensions. To achieve the goal of getting a better dispersive characteristics, a fourth-order accurate space centered scheme which minimizes the difference between the exact and numerical dispersion relation equations is proposed. Through the computational exercises, the proposed dual-preserving solver is computationally demonstrated to be efficient for use to predict the long-term accurate Maxwell’s solutions.
26
Sparsity-based regularizations for image restoration Hui Ji Department of Mathematics, National University of Singapore, 10 Lower Kent Ridge Road, Singapore 119076 E-mail: [email protected] Abstract Image restoration is about how to recover images distorted and degraded in the formation or transmission stage. Most image restoration problems are challenging ill-posed inverse problems that require certain prior information of images to obtain satisfactory solutions. In recent years, the sparsity as- sumption of image/video data under suitable dictionaries, i.e. most of impor- tant information of images can be represented by only few coefficients in some system, has emerged as one powerful prior for most nature images. Com- bined with rapid progresses of `1 norm related minimization approaches for finding the sparse approximation of a given signal in redundant systems (e.g. split Bregman iteration), sparsity-based regularization has been one highly efficient technique for solving many challenging image restoration problems. In this talk, we will begin with a brief review on mathematical theory and numerical algorithms of sparse approximation under redundant tight frame. Then, several applications of sparsity-based regularization in image restora- tion will be presented, including: blind motion deblurring ([4]), blind image inpainting ([1]), surface fitting in range imaging ([3]) and action analysis in video ([2]).
References
[1] B. Dong, H. Ji, J. Li and Z. Shen and Y.-H. Xu: Wavelet Frame Based Blind Image Inpainting, Technical report, NUS, 2011.
[2] Y. Li, C. Fermuller, Y. Aloimonos and H. Ji: Learning shift-invariant sparse representation of actions, IEEE Conf. on Computer Vision and Pattern Recog- nition (CVPR), San Francisco, 2010. [3] H. Ji, Z. Shen and Y.-H. Xu: Wavelet frame based scene reconstruction from range data, Journal of Computational Physics, 229 (6), March 2010, 2093– 2018. [4] J. Cai, H. Ji, C. Liu and Z. Shen: Blind motion deblurring from a single image using sparse approximation, IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Miami, 2009.
27
Efficient numerical methods for computing ground states and dynamics of dipolar Bose-Einstein condensates
1),2) 2), 4) Weizhu Bao , Yongyong Cai ∗, Hanquan Wang
1) Department of Mathematics, National University of Singapore, 117543, Singapore 2) Center for Computational Science and Engineering, National University of Singapore, 117543, Singapore 3) School of Statistics and Mathematics, Yunnan University of Finance and Economics, PR China
e-mail: [email protected] ∗ Abstract New efficient and accurate numerical methods are proposed to compute ground states and dynamics of dipolar Bose-Einstein condensates (BECs) described by a three-dimensional (3D) Gross-Pitaevskii equation (GPE) with a dipolar interaction potential. Due to the high singularity in the dipolar interaction potential, it brings significant difficulties in mathematical analysis and numerical simulations of dipolar BECs. In this paper, by decoupling the two-body dipolar interaction potential into short-range (or local) and long-range interactions (or repulsive and attractive interactions), the GPE for dipolar BECs is reformulated as a Gross-Pitaevskii-Poisson type system. Based on this new mathematical formulation, we prove rigorously existence and uniqueness as well as nonexistence of the ground states, and discuss the existence of global weak solution and finite time blowup of the dynamics in different parameter regimes of dipolar BECs. In addition, a backward Euler sine pseudospectral method is presented for computing the ground states and a time-splitting sine pseudospectral method is proposed for computing the dynamics of dipolar BECs. Due to the adaption of new mathematical formulation, our new numerical methods avoid evaluating integrals with high singularity and thus they are more efficient and accurate than those numerical methods currently used in the literatures for solving the problem. Extensive numerical examples in 3D are reported to demonstrate the efficiency and accuracy of our new numerical methods for computing the ground states and dynamics of dipolar BECs.
28
References
[1] W. Bao, Y. Cai and H. Wang, Efficient numerical methods for computing ground states and dynamics of dipolar Bose-Einstein condensates, J. Comput. Phys., 229 (2010), 7874-7892.
29
A variant of the IDR(s) method with quasi-minimal residual strategy Lei Du1),∗, Tomohiro Sogabe2), Shao-Liang Zhang1)
1) Department of Computational Science and Engineering Nagoya University 2) Graduate School of Information Science and Technology Aichi Prefectural University
∗ e-mail: [email protected]
Abstract We consider the solution of large, sparse and nonsymmetric linear systems of the form Ax = b (1) where A is a nonsingular n × n real matrix, and b is a real vector of order n. This problem arises in numerous applications such as discretizations of partial differential equations, Newton-type methods for nonlinear linear systems and optimization. The IDR(s) method proposed by Sonneveld and van Gijzen [4] can solve problem (1) effectively. Recently, it has received extensive attention and sev- eral variants have been developed. In this talk, we will describe a new variant of the IDR(s) method. The IDR(s) algorithm can converge faster than most product-type methods, but also usually with irregular convergence behavior. We reformulate the relations of residuals and their auxiliary vectors gener- ated by the IDR(s) method in matrix form. Then, by this new formulation and motivated by methods in [1, 2, 3], we present the variant of IDR(s), named QMRIDR(s), to improve the demerit of its irregular convergence be- havior. Both the fast and smooth convergence behaviors of the QMRIDR(s) method can be exhibited. Two numerical results are reported to show the comparison between IDR(s) and QMRIDR(s) in Figs. 1–2.
References
[1] T.F. Chan, E. Gallopoulos, V. Simoncini, T. Szeto and C. Tong, A quasi- minimal residual variant of the Bi-CGSTAB algorithm for nonsymmetric systems, SIAM J. Sci. Comput., 15(1994), 338–347.
30
[2] R.W. Freund, QMR: a quasi-minimal residual method for non-Hermitian linear systems, Numer. Math., 60(1991), 315–339. [3] R.W. Freund, A transpose-free quasi-minimal residual algorithm for non-Hermitian linear systems, SIAM J. Sci. Comput., 14(1993), 470– 482.
[4] P. Sonneveld and M.B. van Gijzen, IDR(s): a family of simple and fast algorithms for solving large nonsymmetric systems of linear equations, SIAM J. Sci. Comput., 31(2008), 1035–1062.
Figure 1: PDE2961(n=2961, nnz=14585)
Figure 2: SHERMAN4(n=1104, nnz=3786)
31
LIMIT PROBLEMS OF SOLUTIONS FOR THE COUPLED NONLINEAR SCHRODINGER¨ EQUATIONS AND STEADY-STATE SOLUTIONS OF THE POISSON-NERNST-PLANCK SYSTEMS CHIUN-CHANG LEE DEPARTMENT OF MATHEMATICS, NATIONAL TAIWAN UNIVERSITY
Abstract. In this talk we present results from my Ph.D. thesis, which is joint works with my advisor Professor Tai-Chia Lin and my Host Professor Chun Liu, when I visited the Institute of Mathematics and Its Applications (IMA) in Minnesota. The Poisson-Boltzmann (PB) equation is conventionally used to model the equilibrium of bulk ionic species in different media and solvents. We study a New Poisson-Boltzmann (PB n) equation in 1-dimensional interval (−1, 1):
N N 1 akφ 2 −blφ 2 00 X akαke X blβle φ = 1 − 1 R akφ(y) R −blφ(y) k=1 −1 e dy l=1 −1 e dy with a small dielectric parameter 2 and nonlocal nonlinearity which takes into consideration of the preservation of the total amounts of each individual ion. This equation can be derived from the original Poisson-Nernst-Planck (PNP) system. Under Robin type boundary conditions with various coefficient scales, we demonstrate the asymptotic behaviors of one dimensional solutions of PB n equations as the parameter approaches zero. In particular, we show that in case of electro- PN1 PN2 neutrality, i.e., k=1 akαk = l=1 blβl, we prove that φ’s solutions of 1-D PB n equations may tend to a nonzero constant c at every interior point as goes to zero. The value c can be uniquely determined by ak, bl’s valences of ions, αk, βl’s total concentrations of ions and the limit of φ’s at the boundary x = ±1. Such a result can not be found in conventional 1-D Poisson-Boltzmann (PB) PN1 PN2 equations. On the other hand, as k=1 akαk 6= l=1 blβl (non-electroneutrality), solutions of 1-D PB n equations may have blow-up behavior which also may not be obtained in 1-D PB equations. We also study the compressible and incompressible limits of the solution of the non-linear Schr¨odingersystem. Recently, a rich variety of dynamical phenomena and a turbulent relaxation have been observed in rotating Bose-Einstein condensates depicted by Gross-Pitaevskii equations coupled with rotating fields and trap potentials. The dynamical phenomena range from shock-wave formation to anisotropic sound propagation. The turbulent relaxation leads to the crystallization of vortex lattices. To see the dynamical phenomena and the turbulent relaxation of two-component rotating Bose-Einstein condensates, we study the incompressible and the compressible limits of two- component systems of Gross-Pitaevskii equations. We define ”H-function” a modulated energy functional which may control the propagation of densities and linear momentums under the effect of rotating fields and trap potentials.
1
32
A method of computer assisted proof for nonlinear two-point boundary value problems using higher order finite elements
1), 2,3) Akitoshi Takayasu ∗, Shin’ichi Oishi 1) Graduate School of Fundamental Science and Engineering,
Waseda University 2) Department of Applied Mathematics, Faculty of Science and Engineering, Waseda University 3) CREST, JST e-mail: [email protected] ∗ Abstract Present authors have presented with Takayuki Kubo at University of Tsukuba a method of a computer assisted proof for the existence and unique- ness of solutions to two-point boundary value problems of nonlinear ordi- nary differential equations in the paper submitted for NOLTA, IEICE. This method uses piecewise linear finite element base functions and sometimes requires fine mesh. To overcome this difficulty, in this paper, an improved method is presented for the norm estimation of the residual to the operator equation. In this refined formulation, piecewise quadratic finite element base functions are used. A kind of the residual technique works sophisticatedly well. It is stated that the estimation of the residual can be expected smaller than that of the previous method. Finally, four examples are presented. Each result demonstrates that a remarkable improvement is achieved in accuracy of the guaranteed error estimation.
References
[1] S. Oishi: Numerical verification of existence and inclusion of solutions for nonlinear operator equations, Journal of Computational and Applied Mathematics, 60 (1995), 171–185. [2] M. Plum: Computer-assisted existence proofs for two-point boundary value problems, Computing, 46 (1991), 19–34. [3] N. Yamamoto and M.T. Nakao: Numerical verifications for solutions to elliptic equations using residual equations with a higher order finite element, Journal of Computational and Applied Mathematics, 60 (1995), 271–279.
33
A Projection Approach to Complex Eigenvalues of a Specified Absolute Value
1), 1) 2) Takafumi MIYATA ∗, Lei DU , Tomohiro SOGABE , Yusaku YAMAMOTO3), Shao-Liang ZHANG1)
1) Graduate School of Engineering, Nagoya University 2) Graduate School of Information Science & Technology, Aichi Prefectural University 3) Graduate School of System Informatics, Kobe University
e-mail: [email protected] ∗ Abstract Given n n matrices A and B, we consider computing a few eigenpairs (λ , x ×n) satisfying ∈ C ∈ C A x = λ B x ( x = 0 ). ̸
This problem arises in many applications of scientific computing, such as structural analysis and stability analysis of fluid dynamics [1]. It is often needed to find a few eigenvalues near a specified point in the complex plane and for the problem several iterative methods are available [2, 3]. In recent years, so called the photonic crystals [4] have attracted much at- tention due to their potential applications in future integrated circuits based on light, instead of electrons. In these practical problems, it is needed to find all the complex eigenvalues such that λ = 1, see Fig. 1. The number of desired eigenvalues are small. However,| they| are usually scattered, not located near one point in the complex plane. In this talk, we present an approach to complex eigenvalues of a specified absolute value. This approach extends the projection method of Sakurai and Sugiura [5] to a new method for the practical problems of photonic crystals. In contrast to existing methods, our method can avoid computing almost all the undesired eigenvalues and thus can be efficient.
34
1
0 Im.
-1
-1 0 1 Re. Figure 1: Distribution of eigenvalues in the complex plane. The desired eigenvalues on the unit circle are shown by , the other ones are shown by . • ◦
References [1] Z. Bai, D. Day, J. W. Demmel, J. J. Dongarra, A test matrix collection for non-Hermitian eigenvalue problems, 1996.
[2] Z. Bai, J. Demmel, J. Dongarra, A. Ruhe, H. van der Vorst, eds., Tem- plates for the solution of Algebraic Eigenvalue Problems: A Practical Guide, SIAM, Philadelphia, 2000.
[3] D. R. Fokkema, G. L. G. Sleijpen, H. A. van der Vorst, Jacobi-Davidson style QR and QZ algorithms for the reduction of matrix pencils, SIAM J. Sci. Comput., 20 (1998), 94–125.
[4] Y. Huang, Y. Y. Lu, S. Li, Analyzing photonic crystal waveguides by Dirichlet-to-Neumann maps, J. Opt. Soc. Am. B, 24 (2007), 2860–2867.
[5] T. Sakurai and H. Sugiura, A projection method for generalized eigen- value problems using numerical integration, J. Comput. Appl. Math., 159 (2003), 119–128.
35
Automatic Performance Tuning for the Blocked Householder QR Algorithm
1), 2) 1) Takeshi FUKAYA ∗, Yusaku YAMAMOTO , Shao-Liang ZHANG 1) Graduate School of Engineering, Nagoya University 2) Graduate School of System Informatics, Kobe University
e-mail: [email protected] ∗ Abstract In developing software for scientific computations, it is important to optimize the software in order to fully exploit the potential performance of hardware. Such optimization has conventionally been done by humans based on their empirical knowledge. However, as the computer architecture becomes complex, cost for op- timizing becomes too high to be done only by humans. In addition, such optimiza- tion is very difficult for researchers in application fields who lack the knowledge about hardware. These issues indicate the necessity of systematic and nonempiri- cal frameworks for the performance optimization. Under this situation, a framework named Software Automatic Tuning has re- cently been extensively researched [1]. In this framework, the target software is required to be tunable, that is, it has controllable variability or adaptability in it. Additionally a tuning mechanism tunes the target software automatically so that it runs in good performance under given conditions. Considering these two points, we have studied on the automatic tuning especially for the dense matrix computa- tions such as LU or QR decomposition [2]. In this talk, we introduce our approach to automatic tuning for the blocked Householder QR algorithm. The QR decomposition of A using the Householder transformations is written in T T (I tny y ) (I t1y y )A = R − n n ··· − 1 1 where ti is a scalar, yi is a vector, and R is an upper triangular matrix. In the blocked algorithm, we first partition the target matrix into blocks and compute partial decomposition. Next, we aggregate the Householder transformations used in the partial decomposition into a matrix form as T T T (I tpy y ) (I t1y y ) = I Y1T1Y1 . − p p ··· − 1 1 − After that, we update the remaining blocks using matrix-matrix multiplication. An overview of these computations is shown in Figure 1. Though we pay ad- ditional computation cost for aggregating the transformations, by using matrix- matrix multiplications, which have high data reusability, we can obtain advan- tage on recent architecture where floating point operation is much faster than data transfer.
36
©¨ ¢¡
¢ ¢£
¢¤£ ¢¦¥ §¨£©¢¦¥ ¢
¥
¡ ¡
§
¤ ¤ ¤ ¥¦¥¦¥ ¥¦¥¦¥
Figure 1: Overview of the blocked algorithm Figure 2: Binary tree
The performance of the blocked algorithm depends on how to partition the tar- get matrix into blocks. This is because total computation cost and the performance of each matrix-matrix multiplication which affected by the matrix size are fixed by the partitioning way. Therefore, it is important to optimize how to partition the matrix. For this subject, optimization by man power has been done [3], however, there are few studies on systematic optimization. Our goal is automatic tuning of the way of partitioning. In our approach, a binary tree shown in Figure 2 represents the way of partitioning. By using a binary tree as a parameter, various kinds of partitioning way can be controlled systematically. Moreover taking advantage of the recursive structure in binary trees, we construct the mechanism based on the dynamic programming to find the optimal binary tree. Through the performance evaluation on some computational environments, it is shown that the performance obtained by our approach is as good as that obtained by human tuning.
References
[1] K. Naono, K. Teranishi, J. Cavazos, and R. Suda: Software Automatic Tun- ing: From Concepts to State-of-the-Art Results, Springer, 2010.
[2] T. Fukaya, Y. Yamamoto, and S. L. Zhang: A Dynamic Programming Ap- proach to Optimizing the Blocking Strategy for the Householder QR De- composition, Proceedings of IEEE Cluster 2008, 2008, 402–410.
[3] E. Elmroth, and F. Gustavson: Applying Recursion to Serial and Parallel QR Factorization Leads to Better Performance, IBM Journal of Research and Development, 44 (2000), 605–624.
37
A Ulm-like Method for Inverse Eigenvalue Problems
1), 2) 3) Xiao-qing Jin ∗, Wei-ping Shen , Chong Li ,
1) Department of Mathematics, University of Macau 2) Department of Mathematics, Zhejiang Normal University 3) Department of Mathematics, Zhejiang University
e-mail: [email protected] ∗ Abstract We propose a Ulm-like method for solving inverse eigenvalue problems, which avoids solving approximate Jacobian equations comparing with other known methods. A convergence analysis of this method is provided and the root quadratic convergence property is proved under the assumption of the distinction of given eigenvalues. Numerical experiments are given in comparison with the inexact Newton-like method. This paper is concerned with [1].
References
[1] W. P. Shen, C. Li, X. Q. Jin: A Ulm-like method for inverse eigenvalue problems, Applied Numerical Mathematics, 61(3) (2011), 356-367.
38
Folding a linear system to a second-order form and its application to structure preserving model order reduction Yung-Ta Li
Department of Applied Mathematics, National Chiao Tung University, Hsinchu, Taiwan 30010
e-mail: [email protected]
Abstract A strategy that folds a linear system to a second-order representation is presented in [1]. The strategy begins with the observation that the inverse matrix of the transformation matrix permits a specific form. However, not all matrices that promise the specific form are of full rank. Consequently one might need to search for all possibilities. In other words, the construction of the transformation matrix is combinatorially hard. We introduce a new method to extract second-order system matrices by applying a similarity transformation to a linear state space system. Instead of taking advantage of the specific form of the inverse of the transformation matrix, we obtain the transformation matrix in a factored form. The new method solves the hard combinatorial problem for the construction of the transformation matrix proposed in [1]. An application of the folding technique is for structure preserving reduc- tion of large-scale second-order systems. The main idea is to reduce a linear state space system and then fold it into a second-order form. A second-order system with the output being a linear combination of displacement and veloc- ity is considered in [1]. The output associated with displacement or velocity only is considered in [2] or [3], respectively. Focusing on the folding process, we find that the construction of the transformation matrices in [2] and [3] suffers the same challenge as in [1]. We will show that a combination usage of the new method and the procedure in [2] or [3] solves the hard combinatorial problem.
References
[1] D. G. Meyer and S. Srinivasan: Balancing and Model Reduction for Second-Order Form Linear Systems, IEEE Trans. Aut. Control, 41 (1996), 1632–1644.
39
[2] B. Salimbahrami and B. Lohmann: Order reduction of large scale second-order systems using Krylov subspace methods, Linear Algebra Appl., 415 (2006), 385–405.
[3] B. Salimbahrami, B. Lohmann and A. Bunse-Gerstner: Passive reduced order modelling of second-order systems, Mathematical and Computer Modelling of Dynamical Systems, 14 (2008), 407–420
40
Model Order Reduction for Bilinear System using Balanced Singular Perturbation
Roberd Saragih
Industrial and Financial Mathematics Research Group, Faculty of Mathematics and Natural Sciences, Institut Teknologi Bandung, INDONESIA e-mail: [email protected]
Abstract Many problems in science and engineering have model in the bilinear system such as electrical networks, mechanical links, nuclear fission, cardiovascular regulator, urban process, hormone regulation, predator-prey models, etc. In some applications, it has the high–order degree of freedom (state). In such case, it is difficult to implement the designed controller since it can cause the numerical difficulties and high computational cost. Hence, it is desired the low-order controller. Meanwhile, the low-order controller can be obtained from the low-order model. In this paper, we concerned with reduction order of the bilinear system model by using the balanced singular perturbation approach. Some approximation methods for the bilinear system, such as the balanced truncation and the projection method, have been developed by authors [2,3]. The singular perturbation approach has been applied to the linear system by Liu [1]. As given in [2], we first transform the bilinear system into the balanced system. The concept of the singular perturbation is based on the rejection of very fast modes from a model. Based on this concept, we then divide the state of the balanced bilinear system into the fast modes and the slow modes. The low-order model can be obtained by replacing the small parameter by zero to the fast mode. The advantage of the singular perturbation is better than the balanced truncation in the low frequencies.
References
1] Liu. Y. and Anderson, B.D.O. (1989), Singular perturbation approximation of balanced system, International Journal of Control, Vol.50, pp. 1379-1405. 2] Liqian Zhang and James Lam, (2002), On H2 model reduction of bilinear systems, Automatica, 38, pp. 205-216. [3] Zhaojun Bai and Daniel Skoogh (2006), A projection method for model reduction of bilinear dynamical systems, Linear Algebra and Its Applications, 415, pp. 406-425
41
Simplified semi-static floating-point filter for 2D orientation problem
1),4) 2).4) 3),4) Katsuhisa Ozaki ∗, Takeshi Ogita , Shin’ichi Oishi ,
1) Department of Mathematical Sciences, Shibaura Institute of Technology 2) Department of Mathematical Sciences, Tokyo Woman’s Christian University 3) Faculty of Science and Engineering, Waseda University 4) Japan Science and Technology Agency, CREST
e-mail: [email protected] ∗ Abstract This talk is concerned with robustness problems in computational ge- ometry, especially 2D orientation problem is focused on. The 2D orientation problem is one of basic problems in the field of computational geometry. The following is brief explanation of this problem: An oriented line and a point are given on the two-dimensional surface. The aim is to clarify that the point is left or right, or on the line. It can be boiled down to a sign of the 3-by-3 matrix determinant. If the sign of the determinant is positive / negative, the point lies on the left / right to the line. If the determinant is zero, the point is on the line. Based on this problem, segment intersection and a point-in polygon problem can be solved. Although the computation is small-scale, the 2D orientation problem plays important roles in computational geometry. Floating-point arithmetic is fast performed by modern architectures. How- ever, the number of significand bits of a floating-point number is finite so that it may cause rounding errors in each operation. In the worst case, unreliable results may be obtained due to accumulation of rounding errors. As for the 2D orientation problem, the computed determinant may be opposite in sign to the exact determinant. The accuracy is especially worried when the point is very close to the line. Classical algorithms in computational geometry correctly work if the all computations are exact. Namely, it is assumed that all expressions are evalu- ated by rational arithmetic (symbolic computations). Once an inexact result is obtained, then an algorithm may output a meaningless result. For ex- amples, there are several algorithms outputting a convex hull which is the minimum convex enclosing all points. However, meaningless results are some- times output, for example, a resultant convex does not contain all points or the result is not convex. Such problems are called robustness problems and many examples can be shown in [1].
42
If the accuracy of the computed result is a concern, using multi-precision arithmetic or rational arithmetic is an option. However, these approaches take much computing time. In addition, many problems are correctly solved by floating-point arithmetic. Therefore, straightforward use of robust com- putations is not effective. There are so-called floating-point filters. They quickly guarantee that the computed sign of the determinant is correct if the problem is well-conditioned. It is a kind of self-validating methods. If the filter cannot guarantee the correctness of the result, then robust computations should be applied. As for earlier researches for the floating-point filters, for example, semi-static filters [2, 3] and dynamic filters [4, 5] have been developed. We suggest a simplified semi-static floating-point filter for the 2D orienta- tion problem. When the filter handles floating-point exceptions like overflow and underflow, then the checks of the order of the magnitude for input coor- dinates seem to be necessary. Several branches are required so that it makes the performance slow down. Our filter handles any floating-point exceptions with only a branch so that it is possible to work fast. Finally, numerical examples are shown to illustrate the efficiency of the proposed filter.
References
[1] L. Kettner, K. Mehlhorn, S. Pion, S. Schirra, C. Yap: Classroom Ex- amples of Robustness Problems in Geometric Computations, Computa- tional Geometry, 40, 61–78, 2008. [2] J. R. Shewchuk: Adaptive Precision Floating-point Arithmetic and Fast Robust Geometric Predicates, Discrete & Computational Geometry, 18, 305–363, 1997. [3] G. Melquiond, S. Pion: Formally certified floating-point filters for ho- mogenous geometric predicates, Theoretical Informatics and Applica- tions, Special issue on Real Numbers, 41:57–69, 2007. [4] V. Y. Pan, Y. Yu: Certified Computation of the Sign of a Matrix Deter- minant, Proc. 10th Annual ACM-SIAM Symposium on Discrete Algo- rithms, ACM Press, New York, and SIAM Publications, Philadelphia, 715-724, 1999. [5] H. Br¨onnimann,C. Burnikel, S. Pion: Interval Arithmetic Yields Ef- ficient Dynamic Filters for Computational Geometry, Discrete Applied Mathematics, 109:25–47, 2001.
43
On the existence of steady flows in a rectangular channel with slip effect on one porous wall Un-Un Kuo and Ching-An Wang∗
Department of Mathematics, National Chung Cheng University, TAIWAN
∗ e-mail: [email protected]
Abstract We study a boundary layer problem arising from the study of steady laminar flows in channels with one porous wall with the slip effect. By applying a similarity transformation [1], the governing equation of the Navier- Stokes system for steady, incompressible laminar flows in this channel could be reduced to f 000(η) + R((f 0(η))2 − f(η)f 00(η)) = K. (1) Here K is an integration constant and R corresponds to the cross flow
Reynolds number based on wall velocity (filtration Reynolds number), while positive (negative) R represents the suction (injection) through the walls. The function f is related to the stream function, η is the normalized trans- verse coordinate, namely, η = 1 at the the porous wall, while the position at the insulated wall is denoted by η = −1. By imposing the slip effect on the porous wall, the steady flows could be studied from the boundary value problem (BVP) of (1), subjects to the boundary conditions:
f(−1) = f 00(−1) = 0, f(1) = 1, f 0(1) + ϕf 00(1) = 0, (2) where positive ϕ is the slip coefficient. Following the frame work in [9], it is our purpose to explore the solutions of (BVP). It is clear that the given problem possesses the unique solution when R = 0. For nonzero R, as in [2, 4], suppose we set f(η) = bG(ξ)/2R, ξ = b(η + 1)/2 for some positive b which is to be determined. Then, G(ξ) satisfies the following associated boundary value problem(BVP1):
G000(ξ) + (G0(ξ))2 − G(ξ)G00(ξ) = 16RK/b4 ≡ β, (3)
G(0) = G0(0) = 0, (4) G(b) = 2R/b, G0(b) + ϕbG00(b)/2 = 0. (5)
44
By assigning values to
G00(0) = α, G000(0) = β, (6) we let G(ξ; α, β) be the solution of the initial value problem (IVP) of (??), (??) and (??). Suppose, for a prescribed positive ϕ, G0(b; α, β)+ϕbG00(b; α, β)/2 = 0 holds at ξ = b∗, then (BVP) will possess a solution with R = b∗G(b∗; α, β)/2 and K = (b∗)4β/16R. From the homogeneity of G(ξ; α, β),
R(α, β) = R(α/λ3, β/λ4), (7)
K(α, β) = K(α/λ3, β/λ4), (8) for all λ > 0, we are able to verify that given any position ϕ > 0, (BVP) can only possess either nonnegative and monotone, or non-monotone solutions. In particular, there exists a positive number R(ϕ) such that (BVP) possesses at least one nonnegative and monotone solutions for R < R(ϕ), and at least non-monotone solution for R ≥ R(ϕ). Moreover, the given problem possesses no solution for R ≤ 0, K ≥ 0.
References
[1] A.S. Berman, Laminar flow in channels with porous walls, J. Appl. Phy., 1953, 24, 1232-1235.
[2] F. M. Skalak and C. Y. Wang, On the nonunique solutions of laminar flow through a porous tube or channel SIAM J. Appl. Math., 1978, 34, 535-544.
[3] S. M. Cox, Analysis of steady flow in channel with porous walls, or with accelerating walls, SIAM J. Appl. Math., 1991, 51, 429-438.
[4] T.-W. Hwang and C.-A. Wang, On multiple solutions for Berman’s prob- lem, Proc. Roy. Soc. Edinburgh, 1992, 121A: 219-230.
[5] S. Chellaml, M. Wiesnerl and C. Dawson2,Slip at a unifromly porous boundary: effect on fluid flow and mass transfer, Journal of Engineering Mathematics, 1992, 26, 481-492.
[6] C.-A. Wang and Y.-Y. Chen, On existence of similarity solutions for Laminar flow in a channel with one porous wall, Bulletin Inst. Math. Academia Sinica, 1992, 20, 299-320.
45
[7] C.-A. Wang and T.-C. Wu, Similarity solution of steady flows in a chan- nel with accelerating walls, Computers Math. Applic., 1995, 30, 1-16.
[8] S. Chellam and M. Liu, Effect of slip on existence, uniqueness, and behavior of similarity solutions for steady incompressible lami- nar flow in porous tubes and channels, Phys. Fluid, 18(8), 2006, DOI:10.1063/1.2236302.
[9] U.-U. Kuo and C.-A. Wang, Multiple solutions of steady flows in a rect- angular channel with slip effect on two equally porous walls, to appear in Taiwanese J. Math.
46
Exchange of two modes of instability leading to sustained oscillations in the flow past a cylinderical obstacle
1), 2) 2) Yukihito Suzuki ∗, Yukio Takemoto , Jiro Mizushima
1) School of Fundamental Science and Engineering, Waseda University 2) Faculty of Science and Engineering, Doshisha University
e-mail: [email protected] ∗ Abstract Flows past cylindrical obstacles often exhibit staggered arrangement of localized vortices, called B´enard-K´arm´an’svortex street. The most typical and simplest example is the flow around a circular cylinder placed in a uni- form flow. The vortex street was conjectured to arise due to instability of the otherwise steady flow and the classical stability theory, based on a parallel flow approximation, was applied to the flow[1], which predicted a so small value such as Rec = 3.2 for the critical Reynolds number of instability. Here the Reynolds number is defined with the cylinder diameter and the uniform velocity. On the other hand, experiments and numerical simulations show that the flow becomes oscillatory for Re & 40 50. In fact, the global sta- bility analysis of the flow without resorting the− parallel flow approximation revealed that the critical Reynolds number Reg is 46.184[2]. The difference between Rec = 3.2 and Reg = 46.184 is beyond allowance of error arising from the parallel flow approximation in the classical stability theory. In order to resolve the disagreement between the classical theory and the global stability analysis, a packet of disturbance is considered instead of a monochromatic wave[3, 4, 5]. For disturbances having a packet shape, two distinct instability modes are identified, one of which is called convective instability, and the other absolute instability. The flow is said convectively unstable when a packet of growing disturbances has a positive group velocity, and the absolute instability is defined if there exists a packet of growing dis- turbance with zero group velocity. Many numerical works have been reported on the convective and absolute instabilities of the cylinder wake based on the parallel flow approximation and it was found that the value Rec = 3.2 from the classical stability theory indicates the critical condition for the convective instability, while the criticality for the absolute instability is determined as Rea 25[4]. It is a common understanding at present that the global insta- bility∼ occurs after the region of the absolute instability has a greater extent in the streamwise direction than 3.5 times the cylinder diameter[4] so that there exists a ‘pocket’ between Rea 25 and Reg = 46.184[5]. ∼ 47
We aim to resolve the gap between the absolute and global instabilities and clarify the mechanism of sustained oscillations in the flow, in which im- pulsive force is added at a certain point in the flow and subsequent spatiotem- poral development of induced disturbance is observed. In order to clarify the underlying physics of the transition from globally stable to unstable flow, we introduce the notions of passive and active modes of instability, extending the convective and absolute instabilities to non-parallel flows, respectively. Our numerical analysis brought us a conclusion that the oscillation is sus- tained by superiority of the growth of disturbance due to instability over the decrease due to advection of the packet. According to our definitions of the passive and active instabilities, the flow becomes unstable to active mode of disturbance simultaneously in the whole flow field at the critical Reynolds number Reg 47 for the global instability, whereas some extent of the region ∼ is passively unstable above a critical Reynolds number Rep. When the flow is passively unstable, the flow can be oscillatory if some external disturbance is continuously added anywhere in the flow field. Otherwise, any localized disturbance is swept out to far downstream at Re < 47. At supercritical Reynolds numbers, any packet of disturbance added in any location will ex- pand over the entire region behind the cylinder, proceeding upstream to the region just behind the cylinder as well as being advected far downstream. Once the packet of disturbance arrives at the vicinity of the cylinder, it acti- vates oscillation leading to the global instability and becomes an oscillation source. We will propose an one-dimensional model equation, like a Ginzburg- Landau equation, which illustrates the mechanism of sustained oscillation and the transition from the passively unstable to actively (globally) unstable flow. The equation and its solutions will be presented at the conference.
References
[1] S. Taneda, J. Phys. Soc. Japan., 18, (1963), 288. [2] C. P. Jackson, J. Fluid Mech., 182, (1987), 23. [3] H. Oertel, Jr, Annu. Rev. Fluid Mech., 22, (1990), 539. [4] P. Huerre and P. A. Monkewitz, Annu. Rev. Fluid Mech., 22, (1990), 473. [5] J. M. Chomaz, Annu. Rev. Fluid Mech., 37, (2005), 357. [6] Y. Takemoto and J. Mizushima, Phys. Rev. E, 82, (2010), 056316.
48
Global and singular solutions to the generalized Proudman-Johnson equation Chien-Hong Cho1),∗, Marcus Wunsch2)
1) Department of Mathematics, National Chung Cheng University 2) Research Institute for Mathematical Sciences, Kyoto University
∗ e-mail: [email protected]
Abstract We consider the Generalized Proudman-Johnson equation
f + ff = af f , x ∈ , t > 0 txx xxx x xx R , (1) f(0, x) = f0(x) on the whole real line as well as the periodic Generalized Proudman-Johnson equation
f + ff = af f , x ∈ = / , t > 0 txx xxx x xx S R Z . (2) f(0, x) = f0(x), x ∈ S The generalized Proudman-Johnson equation (1), (2) can be motivated in several ways. The first motivation comes from the study of the two- dimensional incompressible Euler equations. For parameters a ∈ R ∪ {∞}, Eq. (1) was first derived by Okamoto and Zhu [7]. Remarkably, the param- eter a interconnects several well-studied equations within the framework of the generalized Proudman-Johnson equation (1), (2). For a = 1, it reduces to the original Proudman-Johnson equation [8], which is obtained via the separation of spatial variables ψ(t, x, y) = yf(t, x) for the stream function ψ of the two-dimensional incompressible Euler equations, where x is contained in a finite interval, and y ∈ . Moreover, the generalized Proudman-Johnson R equation encompasses the well-known Burgers equation for a = −3, which is a successful model in gas dynamics; if a = −2, the generalized Proudman- Johnson equation becomes the Hunter-Saxton equation [5] arising in the study of nematic liquid crystals.
The existence of global and singular solutions were widely investigated recently, see for instance [1], [2], [6], [9]. In this talk, we construct a weak solution for (1) for −2 ≤ a < −1 by the modified characteristic method. Then by using an abstract lemma in [3], we explore the blow-up solutions for (2) for a < −1 and their exact blow-up rate.
49
References
[1] A. Bressan, A. Constantin: Global solutions of the HunterVSaxton equa- tion, SIAM J. Math. Anal., 37 (2005) 996-1026.
[2] X.-F. Chen and H. Okamoto: Global existence of solutions to the gener- alized Proudman-Johnson equation, Proc. Japan Acad. Ser. A, 78 (2002) 136-139.
[3] A. Constantin and J. Escher: Wave breaking for nonlinear nonlocal shallow water equations, Acta Math., 181 (1998) 229-243.
[4] C.-H. Cho and M. Wunsch: Global and singular solutions to the general- ized Proudman-Johnson equations, J. Differential Equations, 249 (2010) 392-413.
[5] J.K. Hunter, R. Saxton: Dynamics of director fields, SIAM J. Appl. Math., 51 (1991) 1498-1521.
[6] H. Okamoto: Well-posedness of the generalized ProudmanVJohnson equation without viscosity, J. Math. Fluid Mech., 11 (2009) 46-59.
[7] H. Okamoto, J. Zhu: Some similarity solutions of the NavierVStokes equations and related topics, in: Proceedings of 1999 International Con- ference on Nonlinear Analysis (Taipei), Taiwanese J. Math., 4 (2000) 65-103.
[8] I. Proudman, K. Johnson: Boundary-layer growth near a rear stagnation point, J. Fluid Mech., 12 (1962) 161-168.
[9] R. Saxton, F. Tiglay: Global existence of some infinite energy solutions for a perfect incompressible fluid, SIAM J. Math. Anal., 4 (2008) 1499- 1515.
50
Application of RKHS Theory to a Minimization Problem with Prescribed Nodes Hendra GUNAWAN1),∗, Yoshihiro SAWANO2) 1) Department of Mathematics, Institut Teknologi Bandung, Indonesia 2) Department of Mathematics, Kyoto University, Japan ∗ e-mail: [email protected]
Abstract d Let 2 < α < ∞. We define a Hilbert space Hα to be the set of functions f on [0, 1]d of the form X
f(x1, . . . , xn) := am1···md sin(m1πx1) ··· sin(mdπxd) m1,...,md∈N for which π2α X kfk := (m2 + ··· + m2)α|a |2 < ∞. Hα 2d 1 d m1···md m1,...,md∈N The above norm is induced from the inner product π2α X hf, gi = (m2 + ··· + m2)αa b , Hα 2d 1 d m1···md m1···md m1,...,md∈N where am1···md and bm1···md are the coefficients of f and g, respectively. We are then interested in studying the following problem:
Minimize kfkHα subject to the prescribed nodes:
f(pk) = ck, k = 1,...,N,
d where pk := (pk1, . . . , pkd) ∈ (0, 1) and ck ∈ R are given. [Note here that d the points pk’s are ‘inside’ the unit cube [0, 1] .] The 1- and 2-dimensional case have been studied by Gunawan et al. [1, 2]. In this talk, we shall present the use the theory of reproducing kernel Hilbert spaces to study the problem (in a more general setting). One of our results is the following: Proposition. The solution to the minimization problem
Minimize kfkHα
51 subject to f(p1, . . . , pd) = 1, is given by F (x1, . . . , xd) :=
X sin(m πp ) ··· sin(m πp ) A 1 1 d d sin(m πx ) ··· sin(m πx ). (m2 + ··· + m2)α 1 1 d d m1,...,md∈N 1 d
P 2 −1 (sin(m1πp1)··· sin(mdπpd)) where A := 2 2 α . (m1+···+md) m1,...,md∈N A more general result will be presented. In addition, we shall also consider the H¨oldercontinuity of the solution, by using properties of Besov spaces.
References
[1] H. Gunawan, F. Pranolo and E. Rusyaman, “An interpolation method that minimizes an energy integral of fractional order”, in D. Kapur (ed.), ASCM 2007, LNAI 5081 (2008), 151–162.
[2] H. Gunawan, E. Rusyaman and L. Ambarwati, “Surfaces with pre- scribed nodes and minimum energy integral of fractional order”, sub- mitted.
[3] G.G. Lorentz, Approximation of Functions, AMS Chelsea Publishing, Providence, 1966.
[4] M. Sugimoto and N. Tomita, “The dilation property of modulation spaces and their inclusion relation with Besov spaces”, J. Funct. Anal. 248 (2007), 79–106.
[5] M. H. Taibleson, “On the theory of Lipschitz spaces of distributions on Euclidean n-space. I. Principal properties”, J. Math. Mech. 13 (1964), 407–479.
52
Exact meromorphic solutions of the
real cubic Swift-Hohenberg equation
Robert Conte1,2∗, Tuen-Wai Ng1†and Kwok-Kin Wong1‡
1. Department of Mathematics, The University of Hong Kong. 2. LRC MESO, Centre de math´ematiqueset de leurs applications (UMR 8536) et CEA-DAM, Ecole´ normale sup´erieurede Cachan.
E-mail: [email protected], [email protected], [email protected]
Abstract
We considered the real cubic Swift-Hohenberg (RCSH) equation
∂u ∂2 ∂2 2 = εu − 1 + + u − u3, ε ∈ . (1) ∂t ∂x2 ∂y2 R The equation and its generalizations have been used in various areas, such as laser [4] and nonlinear optics [5] and the theory of pattern formation [2], [6]. The equation (1) admits the following traveling wave reduction [7],
0000 00 3 0 d U + aU + U − U = 0, := . (2) dZ In our paper, we considered the equation (2) to be defined on complex domain. By using a method of Eremenko [3], we show that all meromorphic solutions of the traveling wave reduction (2) of (1) are elliptic or degenerate
∗Partially supported by PROCORE - France/Hong Kong joint research grant F- HK29/05T and RGC grant HKU 703807P. †Partially supported by PROCORE - France/Hong Kong joint research grant F- HK29/05T and RGC grant HKU 703807P. ‡Partially supported by RGC grant HKU 703807P and a post-graduate studentship at HKU.
531 elliptic. The method is based on the analysis using Nevanlinna theory from complex analysis. We then obtain all the meromorphic solutions explicitly by the subequation method of Conte and Musette [1]. This is an algorithmic method, so is systematic. One of the solutions appears to be a new elliptic solution.
References
[1] R. Conte and M. Musette, Analytic solitary waves of nonintegrable equa- tions, Phys. D 181 (2003) pp. 70–79.
[2] M.C. Cross and P.C. Hohenberg, Pattern formation outside of equilib- rium, Rev. Modern Phys. 65 (1993) pp. 851–1112.
[3] A.E. Eremenko, Meromorphic traveling wave solutions of the Kuramoto- Sivashinsky equation, J. Math. Phys., Anal. Geom. 2 (2006) pp. 278– 286.
[4] J. Lega, J.V. Moloney and A.C. Newell, Swift-Hohenberg equation for lasers, Phys. Rev. Lett. 73 (1994) pp. 2978–2981.
[5] S. Longhi and A. Geraci, Swift-Hohenberg equation for optical paramet- ric oscillators, Phys. Rev. A 54 (1996) pp. 4581–4584.
[6] R. Hoyle, Pattern formation: an introduction to methods (Cambridge University Press, Cambridge, 2006).
[7] L.A. Peletier and W.C. Troy, Spatial patterns: higher order models in physics and mechanics (Springer, 2001).
54
Free-surface flow under a sluice gate of an inclined wall from deep water
L.H. Wiryanto
Faculty of Mathematics and Natural Sciences, Bandung Institute of Technology e-mail: [email protected]
Abstract Nonlinear solutions of free surface flow under a sluice gate are studied in this paper. Upstream, the fluid is assumed to be infinitely in depth, and the gate makes an angle β to the horizontal axis. Therefore, the flow emerges from the gate and produces uniform stream far downstream. The problem is solved numerically by a boundary element method derived from an integral equation along the free surface. An analytical function is constructed, relating to the upstream flow, so that the integral equation is solvable. As the result, a free surface flow with smooth detachment from the edge of the gate is obtained for relatively large upstream Froude numbers, otherwise a free surface with back flow near the edge of the gate is indicated, and it tends to a stagnation point for a certain Froude number.
55
Free Convection Boundary Layer Flow from a Vertical Cone in a Porous Medium Filled with a Nanofluid S. Ahmad1),∗, I. Pop2), K. Vajravelu3)
1) School of Mathematical Sciences, Universiti Sains Malaysia, 11800 USM, Penang, Malaysia 2) Faculty of Mathematics, University of Cluj, R-400082 Cluj, CP 253, Romania 3) Department of Mathematics, University of Central Florida, Orlando, FL 32816, USA
∗ e-mail: [email protected]
Abstract
A study of convective flow in porous media filled with conventional heat transfer fluids has been widely conducted in the recent years due to its wide applications in engineering as post-accidental heat removal in nuclear reac- tors, solar collectors, drying processes, heat exchangers, geothermal and oil recovery, building construction, etc. (see [1], [2], [3], [4]). It is well known that conventional heat transfer fluids including water, oil and ethylene glycol mixture are poor heat transfer fluids, since the thermal conductivity of these fluids plays an important role on the heat transfer coefficient between the heat transfer medium and the heat transfer surface. In order to improve the efficiency of the fluids in the term of heat transfer, a new kind of heat trans- fer fluids known as nanofluid which contains a small quantity of nanosized particles (usually less than 100nm) that are uniformly and stably suspended in a liquid has been introduced by Choi [5]. In this paper, a problem of steady free convection boundary layer flow near a vertical cone (with half angle σ) embedded in a saturated porous medium filled with a nanofluid is theoretically studied using three types of metallic and non-metallic nanoparticles such as copper (Cu), alumina (Al2O3) and titania (T iO2) in the base fluid of water. It appears that this problem have been investigated by Cheng et al. [6], Yih [7] and Sohouli et al. [8] but only for the conventional heat transfer fluids. Therefore, our aim is to extend the problem by considering a nanofluid with constant surface heat flux. Under the assumption of using the model proposed by Tiwari and Das [9] along with the assumptions of the Darcy-Boussinesq and bound- ary layer approximations, the basic partial equations along with the bound- ary conditions are obtained. Then, these equations are reduced to ordinary 56 differential equations (ODEs) by introducing the similarity variables. The ODEs are solved numerically using shooting method for some values of the nanoparticle volume fraction parameter ϕ. The local Nusselt number and the temperature profiles are presented and discussed in detail. It is found that the local Nusselt number and the temperature profiles increase as the volume fraction parameter increases.
References
[1] D.A. Nield, A. Bejan: Convection in Porous Media, 3rd ed., Springer, New York, 2006. [2] D.B. Ingham, I. Pop (Eds.): Transport Phenomena in Porous Media, Vol. III, Elsevier, Oxford, 2005. [3] K. Vafai (Ed.): Handbook of Porous Media, 2nd ed., Taylor & Frabcis, New York, 2005.
[4] P. Vadasz: Emerging Topics in Heat and Mass Transfer in Porous Media, Springer, New York, 2008. [5] S.U.S. Choi: Enhancing Thermal Conductivity of Fluids with Nanopar- ticles, in: D.A. Siginer, H.P. Wang (Eds.), Development and Applica- tions of non-Newtonian Flows, ASME MD-vol. 231 and FED-vol. 66 (1995), 99–105. [6] P. Cheng, T.T. Le, I. Pop: Natural Convection of a Darcian Fluid about a Cone, International Communications in Heat and Mass Transfer 12 (1985), 705–717. [7] K.A. Yih: Uniform Transpiration Effect on Combined Heat and Mass Transfer by Natural Convection over a Cone in Saturated Porous Media: Uniform Wall Temperature/Concentration or Heat/Mass Flux, Interna- tional Journal of Heat and Mass Transfer 42 (1999), 3533–3537. [8] A.R. Sohouli, M. Famouri, A. Kimiaeifar, G. Domairry: Application of Homotopy Analysis Method for Natural Convection of Darcian fluid about a vertical full cone embedded in Porous Media Prescribed Surface Heat Flux, 15 (2010), 1691–1699. [9] R.K. Tiwari, M.K. Das: Heat Transfer Augmentation in a Two-sided Lid-driven differentially Heated Square Cavity Utilizing Nanofluids, In- ternational Journal of Heat and Mass Transfer 50 (2007), 2002–2018. 57
Features of the Nystr¨omMethod for the Sherman-Lauricella Equation on Piecewise Smooth Contours
1), 2),⋆ Didenko Victor ∗, Helsing Johan
1) Faculty of Science, University of Brunei Darussalam 2) Centre for Mathematical Sciences, Lund University
e-mail: [email protected] ⋆ ∗e-mail: [email protected]
Abstract Let Γ be a simple closed contour in the complex plane C. The Sherman- Lauricella equation 1 τ t 1 τ t ω(t) + ω(τ) d ln − ω(τ) d − = f(t), t Γ. 2πi Γ τ t − 2πi Γ τ t ∈ ∫ ( − ) ∫ ( − ) (1) plays an important role in various applications. Thus it is used in radar imaging, in theory of viscous incompressible flows, as well as in plane elas- ticity and in other problems of mathematical physics [1, 2]. However, the solution of equation (1), if it exists, is not unique and there is no analytic formula to obtain it. In this work we study the stability of an approximation method in the space L2(Γ) for equation (1) in the case where contour Γ has corner points. More precisely, we consider the Nystr¨ommethod based on composite Gauss- Legendre quadrature formula
1 n 1 d 1 − − u(s) ds wpu(slp)/n, (2) ≈ ∫0 l=0 p=0 ∑ ∑ where l + εp slp = , l = 0, 1, . . . , n 1; p = 0, 1, . . . , d 1, (3) n − − and wp and 0 < ε0 < ε1, . . . < εd < 1 are the weights and the Gauss-Legendre points on the interval [0, 1]. It is worth noting that for smooth contours Γ the integral operators in the left-hand side of (1) are compact, and the method proposed is always stable. On the other hand, if Γ possesses angular points τk, k = 1, 2, . . . , m, the sta- bility of the method under consideration is connected to the invertibility of certain operators Aτk from an algebra of Toeplitz operators. The operators
58
Aτk depend on the above parameters εp and on the magnitude of the corre- sponding angles at the points τk. These operators Aτk arise if one applies quadrature formulas similar to formulas (2)-(3) to Mellin operators with the kernels defined by the kernels of the integral operators in equation (1). They have a complicated structure and, at the moment, there is no analytic tools to study their invertibility. However, each such operator can be associated with the corresponding Nystr¨ommethod for the Sherman-Laurichella equa- tion on a special model curve Γ0 with only one corner point. On the other hand, the stability of the method in those model situations is connected to the behaviour of condition numbers of systems of algebraic equations and it can be tested numerically. Thus, considering the sequences of the con- dition numbers we obtain certain information about the invertibility of the operators Aτk .
References
[1] S. G. Mikhlin: Integral Equations and Their Applications to Certain Problems in Mechanics, Mathematical Physics and Technology, 2nd ed.,
Pergamon Press, 1964.
[2] N. I. Muskhelishvili: Fundamental Problems in the Theory of Elasticity, Nauka, 1966.
59
A New Evolutionary Algorithm for Ground to Flight Data Correlation Changtong Luo1),∗, Shao-Liang Zhang2), Zonglin Jiang1) 1) Institute of Mechanics, Chinese Academy of Sciences 2) Department of Computational Science and Engineering, Nagoya University ∗ e-mail: [email protected]
Abstract In hypersonic vehicle design, it is very difficult/expensive to simulate the environment in flight. Therefore, it is of key important to predict the aerodynamic coefficients of the flight vehicle using wind-tunnel and flight- test data. The task is called ground to flight data correlation, also shorten as ground/flight correlation. A number of methods have been applied to ground/flight correlation, including extrapolation [1], least squares regres- sion [2], artificial neural network [3, 4] and maximum likelihood method [5]. However, all these methods are based on a certain model structure and op- timize the parameters in the model. This is, in many cases, unreliable. To find an optimal model structure will definitely improve the predicted results. In this study, we suggest searching the best model structure from a func- N tional space F with collected data {{xi1, xi2, ··· , xin; yi}}i=1, which can be N ∗ X described as f = arg min kf(xi1, xi2, ··· , xin) − yik . f∈F i Genetic programming [6] is a candidate for this optimization problem. However, its tree-based representation makes it difficult (although not im- possible) to be implemented in general-purpose programming languages such as C/C++ and Fortran. To overcome this difficulty, M. O’Neill and C. Ryan proposed a grammatical evolution (GE) [7] in 2001. However, GE is still not so easy to use. First, the implementation of GE is complicated because it needs an additional function parser for the encoding and decoding process. Next, the incomplete mapping and extra codons problems [7, 8] are common but difficult to handle. In this paper, a new evolutionary algorithm, parse-matrix evolution (PME), for flight data correlation is proposed. A chromosome in PME is a parse- matrix with integer entries. The mapping process from a chromosome to its model function is based on a mapping table. PME can easily be implemented in any programming language and free to control. Furthermore, it does not need any additional function parsing process. Numerical results show that PME can solve the symbolic regression problems effectively (see Table 1).
60
Table 1: Test models and performance of PME no. Dim Target model Domain no.samples ave no.eval ave no.start 1 1 x2 − sin x [1, 3] 8 3230 4.8 2 1 sin x + 2x [1, 3] 8 2071 1.1 3 1 cos x2 − x [1, 3] 8 2603 2.3 4 1 sin x + cos x − x [1, 3] 8 3646 7.4 2 5 2 x1 + 2x2 [1, 3] 16 1047 1.0 2 6 2 (x1 + x2)/x2 [1, 3] 16 1701 1.3 2 7 2 sin(x1 + x2) [1, 3] 16 1537 1.0 2 8 2 cos(x1 + x2) − x2 [1, 3] 16 2798 1.5 2 9 2 ln(x1 + x2) [1, 3] 16 1584 1.2 x1+x2 2 10 2 x1 − e [1, 3] 16 3593 1.8 2 11 2 ln(x1 + x2) + sin(x1 + x2) [1, 3] 16 4911 5.7 2 2 12 2 sin(x1 − x2) + ln(x1 + x2) [1, 3] 16 7637 13.2 3 13 3 x1 + x2 − x3 [1, 3] 64 4505 1.9 3 14 3 x1x2 − x3 [1, 3] 64 3164 2.7 3 15 3 sin(x1x2) + x3 [1, 3] 64 3302 11.6
References
[1] J.B. Peterson, Jr., M.J. Mann, R.B. Sorrells III, W.C. Sawyer, and D.E. Fuller: Extrapolation of Wind-Tunnel Data to Full-Scale Conditions, NASA TP-1515, 1980. [2] E.A. Morelli, R. DeLoach: Wind Tunnel Database Development Using Mod- ern Experiment Design and Multivariate Orthogonal Functions, AIAA paper, AIAA 2003-0653, 2003. [3] M. Norgaard, C. Jorgensen, J. Ross: Neural Network Prediction of New Aircraft Design Coefficients, NASA Technical Memorandum 112197, 1997. [4] T. Rajkumar, Jorge Bardina: Prediction of Aerodynamic Coefficients using Neural Networks for Sparse Data, in: Proc. 15th Int’l Florida Artificial Intel- ligence Research Society Conf., 242-246, 2002.
[5] J.-H. Lee, E.T. Kim, B.-H. Chang, I.-H. Hwang, D.-S. Lee, The Accuracy of the Flight Derivative Estimates Derived from Flight Data, World Academy of Science, Eng. Technol., 58 (2009), 843-849. [6] J. R. Koza, Genetic programming: on the programming of computers by means of natural selection. Cambridge, MA: MIT Press, 1992. [7] M. O’Neill and C. Ryan, Grammatical evolution, IEEE Trans. Evol. Comput. 5 (2001), 349-358. [8] M. O’Neill, A. Brabazon, Grammatical swarm: the generation of programs by social programming. Nat. Comput. 5 (2006), 443-462.
61
Ocean wave scattering within or over a porous medium Swaroop Nandan Bora1),∗, Santu Das2)
1) Department of Mathematics, Indian Institute of Tech. Guwahati, India 2) Department of Mathematics, Indian Institute of Tech. Guwahati, India
∗ e-mail: [email protected]
Abstract Water wave scattering within a porous medium on a sea-bed is considered based on linear wave theory. Using the model of wave-induced flow within a porous medium and Galerkin eigenfunction expansions, refraction-diffraction equations for surface waves are derived whenever needed. The reflection and transmission of waves are considered mainly for two cases: (i) the porous medium is obstructed on the right by a solid wall which does not allow any transmission for the case of horizontal bottom, (ii) and when the bottom is approximated by a line of constant slope or by a raised horizontal bot- tom. Further, by introducing a non-porous medium again after the porous medium, subsequent scattering is investigated. The fluid region is divided into two or three sub-regions according to the points of discontinuity. For the first case, the wave motion inside the porous structure tends to decay as it propagates through the pores. The wave, on encountering the solid wall, gets reflected into the porous structure. For the second case, refraction-diffraction equations for surface waves are derived for the porous region. The scattering is now due to both the porous structure and the structure at the bottom. A boundary value is set up for a flat sea-bed and the analytical solu- tion obtained and Galerkin eigenfunction method is introduced in order to expand the spatially dependent potential in terms of depth-dependent func- tions. The oblique incident wave propagates at an angle θ to the x-direction. By appropriate matching of the properties such as velocity potential, velocity components of two nearby regions, the reflection phenomenon is studied. For a water depth of h, which may be uniform (even bed) or non-uniform (uneven bed), the potential φ in each region satisfies Laplace’s equation
∇2φ = 0, −h ≤ z ≤ 0. (1)
The free surface boundary condition will be ∂φ − µφ = 0, at z = 0, in the non-porous medium, and, (2) ∂z
62
∂φ − iµR = 0, at z = 0, in the porous medium. (3) ∂z where µ = σ2/g with σ as the wave frequency, g the gravity constant, R = f − iS the impedance of the porous medium with f as friction factor and S as inertial coefficient. Considering the depth-dependent part of the potential cosh kn(hj + z) as Zj,n(z) = , n = 0, 1, 2,... with j = 1, 2,... denoting the cosh knhj region number and the wave number kn satisfying the complex dispersion relation iRµ = kn tanh knhj. In the first region prior to the location of the porous structure, the velocity potential is X∞ iK1,0 x −iK1,0 x −iK1,n x φ(x, z) = (e + R0e )Z1,0(h1, z) + Rne Z1,n(h1, z), (4) n=1 q 2 2 where K1,n = k1,n − λ , with λ = k0 sin θ and R0 the reflection coefficient. The potentials in the other regions can be found by looking at the relevant physics. Ultimately the reflection coefficient can be obtained with the help of coefficients through the orthogonal functions Zn(z). Some values for the relevant parameters are assumed and the reflection is studied for various depths, porosity factor etc. When the wave is allowed to propagate over an uneven porous sea-bed, perturbation technique is used to reduce the original problem to a simpler one for the first order correction of the potentials. The solution of this problem is obtained by an approximate use of Green’s integral theorem. Sinusoidal ripples is considered as an example of the sea-bed. The reflection and trans- mission coefficients are evaluated against various values of parameters and it is observed that very few ripples are needed to generate significant reflection.
References
[1] Martha SC, Bora SN, Chakrabarti A. Oblique water wave scattering by small undulations on a porous sea-bed. Applied Ocean Research, 2007; 29: 86-90. [2] Dalrymple RA, Losada MA, Martin PA. Reflection and transmission from porous structures under oblique wave attack. J. Fluid Mech., 1991; 224: 625-644. [3] Zhu S. Water waves within a porous medium on an undulating bed. Coastal Engineering, 2001; 42: 87-101.
63
Spatially localized traveling structures and the asymptotic behavior in binary fluid convection 1),* 2) 2) Yasumasa Nishiura , Takeshi Watanabe , Makoto Iima
1) Research Institute for Electronic Science, Hokkaido University 2) Department of Mathematical and Life Sciences, Hiroshima University * e-mail: [email protected]
Abstract We study spontaneous pattern formation and its asymptotic behavior in binary fluid flow driven by a temperature gradient [1]. When the conductive state is unstable and the size of the domain is large enough, finitely many number of spatially localized time-periodic traveling structures (PTP) [3] are generated spontaneously in the conductive state. They collide and merge via strong collisions, and finally, they are arranged at non-uniform intervals and move in the same direction as in Fig.2. Although there are many other stable patterns, it is remarkable that PTPs end up the winner for survival via interactions as in Fig.1. We found that the role of PTP solutions and their strong interactions (collision) [2] are important to characterize the asymptotic state and hence, pulse-pulse interactions are investigated in detail.
Fig.1: Collision between stable stationary convective pattern (middle) and PTP (from right). They merge into one body, but PTP persists after strong interaction.
64
Fig.2: Time evolution for two generic initial data. A point disturbance is added at t = 0, x = 500 (right), and small random disturbances are added in the whole domain (left). Time interval of the visualization is t = 250.
References
[1] P. Kolodner, Collisions between pulses of traveling-wave convection, Phys. Rev. A 44, (1991) 6466–6479. [2] Y. Nishiura, T. Teramoto, and K. Ueda, Scattering of traveling spots in dissipative system, Chaos 15, (2005) 047509. [3] T. Watanabe, K. Toyabe, M. Iima and Y. NishiuraTime-periodic traveling solutions of localized convection cells in binary fluid mixture, Theoretical and Applied Mechanics Japan, 59, (2010), 211–219.
65
1
Adaptive Mixed Finite Element Methods and New Discontinuous Galerkin methods
1), Eun-Jae Park ∗,
1) Department of Computational Science and Engineering, Yonsei University, Seoul 120-749, Korea
e-mail: [email protected] ∗ Abstract This extended abstract deals with locally conservative methods and con- sists of two parts: first part on adaptive mixed finite element methods and second on new discontinuous Galerkin methods. First part is based on joint work with Dongho Kim [2]. We study adap- tive mixed finite element discretizations for nonlinear elliptic problems. The mixed finite element method has two important features; it conserves the mass locally and produces accurate flux even for highly nonhomogeneous media with large jumps in the physical properties. In this paper, we take an approach based on the Brezzi-Rappaz-Raviart framework rather than using the Brouwer fixed point theory. Compared with existing literature several improvements are made. First, the Raviart-Thomas finite element space of all orders including the lowest order case is treated. Existence and unique- ness of the approximate solution to the model problem is proved. Next, we point out that compared with the work [5], the required H5/2(Ω) regularity of the solution weakens to H2+ϵ(Ω). Next, we drive an optimal order a priori error estimate measured in the Lm(Ω)-norm in the framework of Brezzi-Rappaz-Raviart. The Brezzi- Rappaz-Raviart theory requires of the C1,1-regularity of the nonlinear func- tion; we extend the theory to allow more general C1,α class of the function for 0 < α 1. Finally,≤ we derive reliable and efficient a posteriori error estimators in Lm(Ω) for the error control of our approximation to the nonlinear problem under consideration. To the best of the authors’ knowledge, this is the first result obtaining a posteriori error estimator measured in the Lm(Ω)-norm which is reliable and efficient for a mixed approximation of nonlinear elliptic problems.
Second part is based on joint work with Youngmok Jeon [4]. A new class of hybrid discontinuous Galerkin methods is introduced and analyzed for second-order elliptic equations. Recently, the hybridizable discontinuous
66
2
Galerkin (HDG) method is introduced and developed by Cockburn and his collaborators [1]. The main feature of the HDG method is that their approxi- mate solutions can be expressed in an element-by-element fashion in terms of an approximate trace uh satisfying a global weak formulation, which reduces globally coupled degrees of freedom dramatically. In this sense, our method can be understood asb a HDG method.
On the other hand, our method can be viewed as a generalization of so called cell boundary element (CBE) by Jeon and his coauthors [3], therefore, the method is named as GCBE. The CBE method is a flux conserving method without a dual partition, but it has the same disadvantages as the finite volume method, namely that high-order accurate methods are not easy to construct [3]. The motivation of the GCBE method is to construct a flux conserving method with high order approximation property. More details can be found in the recent paper [4].
References
[1] , Unified hy- B. Cockburn, J. Gopalakrishnan and R. Lazarov bridization of discontinuous Galerkin, mixed and continuous Galerkin methods for second order elliptic problems, SIAM J. Numer. Anal. 47 (2009), pp. 1319–1365.
[2] D. Kim and E.-J. Park, A Priori and A Posteriori Analysis of Mixed Fi- nite Element Methods for Nonlinear Elliptic Equations. SIAM J. Numer. Anal. Volume 48 (2010), no. 3, pp. 1186-1207.
[3] Y. Jeon, E.-J. Park, Nonconforming cell boundary element methods for elliptic problems on triangular mesh, Appl. Numer. Math. 58 (6) (2008) 800–814.
[4] Y. Jeon and E.-J. Park, A hybrid discontinuous Galerkin method for elliptic problems, SIAM J. Numer. Anal. Volume 48 (2010), no. 5, pp. 1968-1983
[5] F. A. Milner, E.-J. Park, Mixed finite element methods for Hamilton- Jacobi-Bellman-type equations, IMA J. Numer. Anal. Vol.16 (1996), pp. 399–412.
67
Numerical Investigation of Mixed Convection Flow over a Horizontal Circular Cylinder and a Solid Sphere in Nanofluids Roslinda Nazar
School of Mathematical Sciences, Faculty of Science & Technology Universiti Kebangsaan Malaysia
e-mail: [email protected]
Abstract The recent discovery of nanofluids by Choi [1], which is a new kind of fluid suspension consisting of uniformly dispersed and suspended nanometer-sized (10-50 nm) particles and fibers in the base fluid, marks the next approach as a cooling technology. This is because nanofluids exerting nano effects have special features. One such characteristic of nanofluids is the anomalous high thermal conductivity at very low concentration of nanoparticles and the considerable enhancement of forced convective heat transfer, as it was pointed out by many researchers. Nanofluids usually contain the nanoparticles such as metals, oxides, or carbon nanotubes, whereby these nanoparticles have unique chemical and physical properties. In this study, the steady mixed convection boundary layer flow of nanoflu- ids, formed by the dilution of nanoparticles, past a horizontal circular cylinder and about a solid sphere with constant surface temperature is investigated numerically for both cases of assisting and opposing flows. The nanofluid equations model proposed by Tiwari and Das [2] is used for the analysis. The resulting system of nonlinear partial differential equations is solved us- ing an implicit finite-difference scheme known as the Keller-box method. The solutions for the flow and heat transfer characteristics are evaluated numer- ically for various values of the parameters, namely the nanoparticle volume fraction ϕ and the mixed convection parameter λ at Prandtl number Pr =
0.7, 1, 6.2 and 7. Three different types of nanoparticles, namely Al2O3, Cu and T iO2 (by using water-based fluid with Pr = 6.2) have been considered in this study and representative results for the local skin friction coefficient and the local heat transfer coefficient or local Nusselt number have been ob- tained for the following range of nanoparticle volume fraction: 0 ≤ ϕ ≤ 0.2 (see Abu-Nada and Oztop [3]). We have used data related to thermophysical properties of fluid and nanoparticles as presented in [3] to compute each case of nanofluid. In order to verify the accuracy of the present method, the values of the local skin friction coefficient and the local Nusselt number are compared
68 with those reported by Merkin [4] for ϕ = 0 (regular Newtonian fluid) at Pr = 1 and various values of x and λ, for the horizontal circular cylinder case; while the values of the local skin friction coefficient and the local heat transfer coefficient are compared with those reported by Nazar et al. [5] for ϕ = 0 at Pr = 0.7 and Pr = 7 and various values of x and λ, for the solid sphere case. The comparison of the previous results obtained in [4, 5] and the present ones are found to be in excellent agreement. It is found that for each particular nanoparticle, as the nanoparticle volume fraction ϕ increases, the skin friction coefficient and the heat transfer rate at the surface also increase, and it also leads to the increase of the value of the mixed convection parameter λ which first gives no separation and the value of λ below which a boundary layer solution does not exist. It is also found that an increase in the Prandtl number Pr leads to an increase of the local skin friction coefficient and the local Nusselt number. Finally, it is found that the velocity and temperature profiles satisfy the far field boundary conditions asymptotically, and as such, this support the validity of the numerical results obtained. The results obtained can be used to explain the characteristics and applications of nanofluids, which are widely used as coolants, lubricants, heat exchangers and micro-channel heat sinks.
References
[1] S. Choi: Enhancing thermal conductivity of fluids with nanoparticles, ASME International Mechanical Engineering Congress and Exposition, FED 231/MD, 66 (1995) 99–105.
[2] R.K. Tiwari, M.K Das: Heat transfer augmentation in a two-sided lid- driven differentially heated square cavity utilizing nanofluids, Interna- tional Journal of Heat and Mass Transfer, 50 (2007), 2002–2018.
[3] E. Abu-Nada, H.F. Oztop: Effects of inclination angle on natural convec-
tion in enclosures filled with Cu-water nanofluids, International Journal of Heat and Mass Transfer, 30 (2009), 669–678.
[4] J.H. Merkin: Mixed convection from a horizontal circular cylinder, In- ternational Journal of Heat and Mass Transfer, 20 (1977), 73–77.
[5] R. Nazar, N. Amin, I. Pop: Mixed convection boundary layer flow about an isothermal sphere in a micropolar fluid, International Journal of Ther- mal Sciences, 42 (2002) 283–293.
69
Lamellar pattern and two-phase shallow water equations
1) Hidenori Yasuda ∗
1) Department of Mathematics, Josai University
e-mail: [email protected] ∗ Abstract We propose a numerical method to simulate the phase separation of block copolymer in one-dimensional thin liquid film. Phase separation of polymer blend: a mixture of homopolymers, finally becomes a single circle [1]. However, phase separation of block copolymer: chemically connected two homopolymers, is qualitatively different. In one- dimensional system, periodic microscopic scale pattern called lamellar pat- tern appears. Phenomena of phase separation of copolymer is modeled by two-phase shallow water equations and Ohta-Kawasaki potential. Ohta-Kawasaki po- tential governs the phase separation of block copolymer [2]. However, some difficulty appears in the computation of Ohta-Kawasaki potential. A mod- ification of the potential is proposed to circumvent the difficulty. Two- phase shallow water equations are discretized by an invariant finite difference method. And the potential is integrated by the trapezoidal rule. Lamellar pattern and wavy surface developed using the proposed method [3].
Models and numerical method 1. Two-phase shallow water equations Two-phase flow in the liquid film is described by two-phase shallow water equations, ∂ (α h) ∂ d + (α hu ) = 0, (1) ∂t ∂x d d ∂ (α h) ∂ c + (α hu ) = 0, (2) ∂t ∂x c c 2 ∂ud ∂ 1 ∂ ∂ + ud ud = (ρmgh) + ν 2 ud + fd, (3) ∂t ∂x −ρd ∂x ∂x 2 ∂uc ∂ 1 ∂ ∂ + uc uc = (ρmgh) + ν 2 uc + fc, (4) ∂t ∂x −ρc ∂x ∂x
αd + αc = 1. (5)
70
Here, α is the volume fraction, u the velocity, h surface height, ρ density, d is the suffix of minor phase, c is the suffix of major phase, and ρm is the density of mixture:
Two-phase shallow water equations is discretized by an invariant finite differ- ence scheme which highly resolve the interface region of phase separations [1].
2. Ohta-Kawasaki potential Phase separation of block copolymer is governed by Ohta-Kawasaki poten- tial [2], F ψ = Fs ψ + Fl ψ (6) { } { } { } ψ = αd αc (7) − a 2 b 4 γ 2 Fs ψ = dx ψ + ψ + ( ψ) (8) { } ∫ [−2 4 2 ∇ ] ε ¯ ¯ Fl ψ = dx dx0G (x, x0) ψ (x) ψ ψ (x0) ψ (9) { } 2 − − ∫ ∫ ( )( ) ∆G (x, x0) = δ (x x0) . (10) − − Here, F means free energy of block copolymer; ψ is the order parameter: ψ¯ is the average of ψ. G is the fundamental solution of Laplace equation.
Ohta and Kawasaki potential has some difficulty to compute numerically. For lamellar patterns, we propose a modification to circumvent the difficulty. Numerical integration of the integration of the potential is performed by the trapezoidal rule.
In simulations using the proposed method, lamellar pattern and wavy surface consistent with the specified period developed [3].
References
[1] H. Yasuda, Two-phase shallow water equations and phase separation in thin immiscible liquid films, J. Sci. Comput. 43 (2010) 471-487.
[2] T.Ohta, K. Kawasaki, Equilibrium morphology of block copolymer melts, Macromolecules, 19 (1986) 2621-2632.
[3] H. Yasuda, Simulation of micro-phase separation of block copolymer in one-dimensional thin liquid films, submitted.
71
IMPLEMENTING THE FINITE VOLUME METHOD ON A RECTANGULAR GRID
Tanay Milind Deshpande* Undergraduate, B.E.(Hons.) Mechanical Engineering Birla Institute of Technology and Science (BITS) – Pilani K. K. Birla Goa Campus Goa, India *[email protected]
Abstract
The Finite Volume Method (FVM) is commercially the most widely used CFD technique to convert the non-linear partial differential equations that govern fluid flow into a linear algebraic system of equations. This paper formulates an algorithm to solve steady-state, laminar and incompressible fluid flow equations with a structured collocated grid arrangement in a rectangular domain and describes its implementation using C++. The chief aim of this algorithm is to considerably simplify and optimize existing FVM techniques such as the Semi Implicit Method for Pressure Linked Equations (SIMPLE) algorithm (Patankar, 1972) for low Peclet number problems in regular geometries. After setting boundary conditions for driven-cavity flow as an instance, the Central Differencing scheme and the divergence theorem are invoked. The Navier-Stokes equations are discretized at each cell in the grid into a system of coefficients of neighbouring cell velocities. The algebraic equations representing each line in the grid are then fed into a solver subroutine that implements the Tridiagonal Matrix Algorithm. After two Alternating Direction Implicit traversals for velocities, the continuity equation is discretized in a similar manner for evaluating pressure correction terms which update the pressure and velocities at every node for the next iteration. At this crucial juncture where the SIMPLE takes contributions of pressure errors at neighbouring cells to be zero while calculating that at a single cell, this algorithm departs from the procedure and instead accommodates for neighbouring pressure errors.
References-
1. Versteeg H., Malalasekera W., ‘Introduction to Computational Fluid Dynamics: The Finite Volume Method’, 1995, Longman Scientific & Technical 2. Anderson J., ‘Computational Fluid Dynamics: The Basics with Applications’, 1995, McGraw-Hill 3. Patankar S., ‘Numerical Fluid Flow and Heat Transfer’, Hemisphere Publishing Corporation, 1980 4. Ferziger J., Peric M., ‘Computational Methods for Fluid Dynamics, Springer-Verlag, 2001 5. The SIMPLE Algorithm, Patakar S., Spalding B., 1972 & The Tridiagonal Matrix Algorithm, Thomas, 1949
72
Accurate and Rigorous Exponential Algorithm in Round to Nearest
1), 2) 2) Naoya Yamanaka ∗, Masahide Kashiwagi , Shin’ichi Oishi
1) Research Institute for Science and Engineering, Waseda University 2) Faculty of Science and Engineering, Waseda University
e-mail: naoya [email protected] ∗ Abstract This talk is concerned with numerical algorithms retaining high reliability, high accuracy and high portability. In this talk, an algorithm with high reliability means a numerical algorithm which outputs a mathematically- rigorous result. An algorithm with high accuracy represents an algorithm which returns a result with high accuracy. Furthermore, an algorithm with high portability indicates an algorithm which calculates a result without relying on any numerical environment. In this talk, numerical exponential algorithm retaining high reliability, high accuracy and high portability is discussed. Todays libraries for the approximation of elementary functions are very fast and the results are mostly of very high accuracy [1, 2]. The achieved accuracy does not exceed one or two ulp for almost all input arguments; how- ever, there is no proof for that. In this talk, we discuss a reliable, accurate, portable implementation of the exponential function. We intend to utilize the marvelous accuracy by a table approach. Proposed algorithm delivers rigorous bounds for the result for all floating point input arguments. The order of evaluation of the formula is carefully chosen to diminish accumula- tion of rounding errors. As a result we obtain the relative accuracy of the exponential function value is better than 1ulp. Throughout this talk, we assume floating-point arithmetic adhering to IEEE standard 754-1985. IEEE standard 754-1985 is one of a technical standard established by the IEEE and the most widely-used standard for floating point computation, followed by many hardware and software imple- mentations. Proposed algorithm is based on computer interval arithmetic adhering to the standard. Interval arithmetic is a numerical method devel- oped by mathematicians since the 1950s and 1960s as an approach to putting bounds on rounding errors and measurement errors in mathematical compu- tation and thus developing numerical methods that yield reliable results. It represents each value as a range of possibilities. The interval arithmetic on floating-point arithmetic with changing rounding mode is widely used.
73
Rounding mode is one of defined parameter in IEEE standard 754-1985. The standard defines four rounding algorithms: rounding to nearest, round to- ward zero, round toward plus-infinity, round toward minus-infinity. However, changing rounding mode takes some computational costs, and the commands to change the mode vary widely depending on the numerical environment. Besides, numerical environments which do not have the commands to change the mode exist. For these problems, proposed algorithm only in rounding to nearest is proposed.
References
[1] J. M. Muller: Elementary Functions, Algorithm and Implementation, Second Edition, Birkh¨auser,2006.
[2] S. M. Rump: Rigorous and portable standard functions. BIT Numerical Mathematics, 41(3):540–562, 2001.
74
An integrated manufacturing system for time-varying demand process M.Omar
Institute of Mathematical Sciences, University of Malaya, Kuala Lumpur
e-mail: [email protected]
Abstract This paper considers a manufacturing system in which a single-manufacturer procures raw material from a single-supplier at multiple installments and process them to make finished products and deliver to a single-buyer at mul- tiple shipments to satisfy a time-varying demand rate. In this system, the manufacturer must deliver the products in small quantities to minimize the buyer’s holding cost, and accept the supply of small quantities of raw ma- terial to minimize its own holding cost. We develop a mathematical model for this problem and illustrate the effectiveness of the model with numerical examples.
75
Numerical analysis of a stationary transport equation by upwind scheme Takanori Katsura Graduate School of Informatics, Kyoto University e-mail: [email protected] Abstract The aim of this study is to simulate the stationary transport equation numerically as a fundamental study of the diffuse optical tomography(DOT). DOT is a new biomedical imaging technique, which is hoped to come after X-ray CT and MRI, and it has been extensively studied for practical use in medical sciences. It is considered that light propagation in biomedical tissue is described as photon density by the transport equation, and DOT is mathematically formulated as an inverse problem of the equation [1, 2]. This mathematical model of DOT is commonly accepted but should be more verified in detail to put DOT to practical use. To this end, we propose a new numerical scheme based on the finite difference method which is helpful for a fundamental study of the equation. The comparison of the numerical simulations and the experimental measurements will give us some suggestions for the mathematical modeling. In this presentation, we consider the following boundary value problem of stationary transport equation:
ξ xu(x, ξ) + µt(x)u(x, ξ) = µs(x) p(ξ ξ0)u(x, ξ0) dσξ0 + f(x)(x, ξ) Q ·∇ 1 · ∈ ∫S u(x, ξ) = α u(x, ξ0) + q(x, ξ) (x, ξ) Γ−. ∈ Here Ω is a square domain (0, 1) (0, 1) and n(x) is the outward unit normal + × vector on ∂Ω, Γ is the outward boundary, Γ− is the inward boundary; 1 Γ± := (x, ξ) ∂Ω S ξ n(x) < 0 , α is reflectance, ξ0 = ξ 2 (ξ n(x)) n(x). To{ discretize∈ the× transport| · equation,} we apply the trapezoidal− · formula to the integral term, and the upwind difference to x. Let K be the divi- ∇ sion number of velocity and h be the the spatial mesh size. We set ξk := 2π 2π 1 (cos K , sin K ) S , xi,j := (ih, jh) Ω, s(k, i) := sgn(ξk,i). Then we pro- pose the following∈ upwind scheme (1)∈ (4): − ui,j,k ui s(k,1),j,k ui,j,k ui,j s(k,2),k ξ − − + ξ − − + µ (x )u k,1 s(k, 1)h k,2 s(k, 2)h t i,j i,j,k K 1 (1) 2π − = µs(xi,j) p(ξk ξl)ui,j,l + f(xi,j)(xi,j, ξk) Q K · ∈ l=0 ∑ 76
ui,j,k ui 1,j,k ui 1,j,k ui 1,j s(k,2),k ξk,1 − ± + ξk,2 ± − ± − + µt(xi 1,j)ui 1,j,k h s(k, 2)h ± ± ∓ K 1 2π − (2) = µs(xi 1,j) p(ξk ξl)ui 1,j,l + f(xi 1,j) ± K · ± ± l=0 ∑ + (xi,j, ξk) Γ (x1 = 0, x1 = 1) ∈ ui,j 1,k ui s(k,1),j 1,k ui,j,k ui,j 1,k ξk,1 ± − − ± + ξk,2 − ± + µt(xi,j 1)ui,j 1,k s(k, 1)h h ± ± K ∓1 2π − (3) = µs(xi,j 1) p(ξk ξl)ui,j 1,l + f(xi,j 1) ± K · ± ± l=0 ∑ + (xi,j, ξk) Γ (x2 = 0, x2 = 1) ∈ u = α u + q(x , ξ ) i,j,k i,j,l i,j k (4) ξl = ξk 2 (ξk n(xi,j)) n(xi,j), (xi,j, ξk) Γ−. − · ∈ We prove the convergency of this upwind scheme. Moreover we propose a iterative scheme based on the upwind scheme and prove the convergency. The following figures are numerical examples of the light intensity U(x) by our proposed scheme. Light scattering is described by the scattering kernel p(ξ, ξ0) and the anisotropy factor g controls how strongly the scattering is forward-peaked; g = 0, isotropic scattering; g = 0.9, strongly forward-peaked scattering.
g = 0.0 g = 0.5 g = 0.9
Fig. 1 Numerical examples of the light intensity U(x)
References
[1] S. R. Arridge, M. Schwiger: Image reconstruction in optical tomography Philosophical Transactions of the Royal Society B, 352 (1997), 717-726
[2] A. D. Klose, E. W. Larsen: Light transport in biological tissue based on the simplified spherical harmonic equations, Journal of Computational Physics, 220 (2006), 441-470
77
Stable Traffic Jam Occurs from Nonlinear Saturation Akiyasu Tomoeda1,2),∗, Ryosuke Nishi3,4), Kazumichi Ohtsuka5,6), Katsuhiro Nishinari3,5)
1) Meiji Institute for Advanced Study of Mathematical Sciences, Meiji University 2) JST, CREST 3) School of Engineering, The University of Tokyo 4) Japan Society for the Promotion of Science 5) Research Center for Advanced Science and Technology, The University of Tokyo 6) Economics and Social Research Institute, Cabinet office, Government of Japan
∗ e-mail: [email protected]
Abstract
Among various kinds of jamming phenomena, a traffic jam of vehicles is a very familiar phenomenon and causes several losses in our daily life such as decreasing efficiency of transportation, waste of energy, serious environmental degradation, etc. In particular, the dynamics of traffic flow on highway has attracted many researchers and has been investigated as a non-equilibrium system of interacting particles for the last few decades [1, 2] . The dynamics of traffic flow is often treated as an effectively one-dimensional compressible fluid by focusing on the collective behavior of vehicles, so-called macroscopic models [3, 4] , which consist of two equations: conservation law and mo- tion equation of traffic vehicles. In previous fluid models, one does not have any choice to introduce the diffusion term into the models such as Kerner and Konh¨auser model [4] , in order to represent the stabilized density wave, which indicates the formation of traffic jam. However, it emerges as a se- rious problem that some vehicles move backward even under heavy traffic, as mentioned in [5] . Thus, unfortunately we would have to conclude that traffic models which include the diffusion term are not reasonable for the re- alistic expression of traffic flow. Given these factors, we suppose that traffic jam forms as a result of the plateaued growth of small perturbation by the nonlinear saturation effect. In this contribution, we have proposed a new compressible fluid model for the one-dimensional traffic flow taking into account a variation of the reaction time of drivers, which is based on the actual measurements. Our new model
78 is a generalization of the Payne model [3] by introducing a density-dependent function of reaction time as a reasonable assumption, and given by ∂ρ ∂ + (ρv) = 0, (1) ∂t ∂x ∂v ∂v 1 1 dV (ρ) ∂ρ + v = V (ρ) − v + opt , (2) ∂t ∂x τ(ρ) opt 2ρτ(ρ) dρ ∂x where ρ(x, t) and v(x, t) correspond to the spatial vehicle density and the average velocity at position x and time t, respectively. τ(ρ) and Vopt(ρ) are the reaction time of drivers and the optimal velocity function under the current density ρ, respectively. The linear stability analysis of this new model shows the instability of a free flow state caused by the enhancement of fluctuations around a cer- tain critical density of vehicles, which is observed in real traffic flow [6] . Moreover, the condition of the nonlinear saturation of density against small perturbation is theoretically derived from the reductive perturbation method.
References
[1] D. Chowdhury, L. Santen and A. Schadschneider: Statistical physics of vehicular traffic and some related systems, Phys. Rep. 329 (2000), 199–329.
[2] D. Helbing: Traffic and related self-driven many-particle systems, Rev. Mod. Phys. 73 (2001), 1067–1141.
[3] H. J. Payne: Mathematical Models of Public Systems 1, ed G. A. Bekey (La Jolla,CA: Simulation Council) (1971), 51–61.
[4] B. S. Kerner and P. Konh¨auser:Cluster effect in initially homogeneous traffic flow, Phys. Rev. E 48 (1993), R2355–R2338. [5] C. F. Daganzo: Requiem for second-order fluid approximation of traffic flow, Trans. Res. B 29B (1995), 277-286.
[6] Yuki Sugiyama, et al.: Traffic jams without bottlenecks – experimental evidence for the physical mechanism of the formation of a jam, New J. Phys. 10 (2008), 033001.
79
Lagrangian Skeleton on the Gulf of Mexico Oil Spill C. Mendoza1),∗, A.M. Mancho2)
1) Universidad Polit´ecnica de Madrid. ETSI Navales. Av. Arco de la Victoria s/n. 28040 Madrid, Spain 2) Instituto de Ciencias Matem´aticas. CSIC-UAM-UC3M-UCM. Serrano 121. 28006 Madrid, Spain
∗ e-mail: [email protected]
Abstract The understanding of the circulation of ocean currents, the exchange of CO2 between atmosphere and oceans, and the influence of the oceans on the distribution of heat on a global scale is key to our ability to predict and assess the future evolution of climate [1, 2]. Global climate change is affecting sea breathing through mechanisms not yet understood [3]. The ocean is impor- tant in the regulation of heat and moisture fluxes, and oceanic physical and bio-geochemical processes are major regulators of natural greenhouse gases. Understanding how oceans mix their waters is key to provide sound fore- casts on the climate [1]. Global change also affects marine biodiversity and threatens the survival of ecosystems and exploitable resources. To predict not only the effects of global change on the oceans, but also the response time of climate feedback requires to improve detection systems and to open new lines of research. We use a novel Lagrangian descriptor (function M, introduced in [4, 5]). It is based on the measure of the arclength of particle trajectories on the ocean surface at a given time. In [6, 7, 8] this technique has been proven to be successful for characterizing the Kuroshio current. We employ this tool on velocity data sets on the Gulf of Mexico obtained from HYCOM project. The tool identifies the underlying Lagrangian struc- tures in the oceanic currents. In particular invariant manifolds, hyperbolic and non-hyperbolic flow regions are detected. We study the influence of these structures on particle motions on the oil spill. We acknowledge to CESGA for support with the supercomputer FINIS TERRAE. Thanks for support by grants: CSIC-OCEANTECH, I-Math C3- 0104, MICINN-MTM2008-03754 and MTM2008-03840-E, and C. Madrid- SIMUMAT.
80
References
[1] S. Bowen, M.S. Lozier, S. F. Gary, and C.W. Bning. Nature 459, 243- 247, 2009.
[2] K. Katija, and J.O. Dabiri. Nature 460, 624-626, 2009.
[3] Q. Schiermeier. Nature 447, 522-524, 2007.
[4] J. A. Jim´enezMadrid, A.M. Mancho. Distinguished trajectories in time dependent vector fields. Chaos 19 (2009), 013111-1-013111-18.
[5] C. Mendoza, A.M. Mancho. The hidden geometry of ocean flows. Phys- ical Review Letters 105 (2010), 3, 038501-1-038501-4.
[6] A.M. Mancho, S. Wiggins, A. Turiel, E. Hern´andez-Garca,C. L´opez, E. Garc´ıa-Ladona.Nonlinear Processes in Oceanic and Atmospheric Flows. Nonlinear Proc. Geoph 17 (2010), 3, 283-285.
[7] C. Mendoza, A.M. Mancho, Marie-H´eleneRio. The turnstile mechanism across the Kuroshio current: analysis of dynamics in altimeter velocity fields. Nonlinear Proc. Geoph 17 (2010), 2, 103-111.
[8] C. Mendoza, A.M. Mancho. The Lagrangian description of aperiodic flows: a case study of the Kuroshio Current. preprint (2011).
81
Mathematical studies on the self-motion of surfactant scrapings at the air-water interface
1), 2,5) Keita Iida ∗, Hiroyuki Kitahata , Satoshi Nakata3) and Masaharu Nagayama4,6)
1) Graduate School of Natural Science and Technology, Kanazawa University, 2) Graduate School of Science, Chiba University, 3) Graduate School of Science, Hiroshima University, 4) Institute of Science and Engineering, Kanazawa University, 5) PRESTO, Japan Science and Technology Agency, 6) CREST, Japan Science and Technology Agency
e-mail: [email protected] ∗ Abstract Several types of autonomous motors are known to exhibit various man- ners of “self-motion” at air/water interfaces. Spontaneous motion of sur- factant scrapings at the water surface (henceforth surfactant-water system) is a representative example of self-motion, which involves a variety of rich topics and which was introduced in 1997 [1,2]. As presented in [2], camphor scrapings (diameter, ca. 1 mm) exhibit differing manners of sustained motion on water under almost isothermal and non-equilibrium conditions, of which the working mechanisms are similar to living organisms. Hence, from the physicochemical point of view, they have been considered as a novel efficient chemo-mechanical transducer which mimics living things. From the analytical point of view, we would like to discover the math- ematics and physics that underlie the surfactant-water system and to un- derstand the essential mechanisms of the self-motion.*1) One proposition on a camphor-water system which has been verified to be true, by means of experiments and analyses, is that a pitch-fork bifurcation of travelling pulse solution occurs depending on the viscosity of the aqueous phase in a one- dimensional problem [3]. For the one-dimensional problem, we can now see such mathematical structure at the point of demarcation between self-motion and the rest state. How about the two-dimensional problem? The next step we discuss be- low is to consider both translational and rotational motion induced by the * 1)Our real goal is to understand the principles and structures that living and nonliving systems have in common and to realize the autonomic motor system in living things at their most fundamental level.
82 anisotropic shape of the camphor scraping. Although a mathematical model consisting of translational and rotational equations, coupled with a reaction- diffusion equation, has been proposed in 2005 [4], the characteristics of the two-dimensional camphor motion depending on its shape has not yet been clarified. For example, let us consider a camphor scraping with elliptical shape. Which direction does the camphor move, in the long axis or in the short axis? Or, does the elliptical camphor rotate? In fact, we found it difficult to investigate the above question by brief experiments and simula- tions due to difficulties originating from the elliptical shape (e.g., we have to compute the elliptic integral in the model). To tackle the above questions, we apply bifurcation theory and conduct proper experiments to determine the mode of motion of elliptical camphor scrapings. In the presentation, we will introduce our mathematical findings, numerical simulations and experiments of the camphor-water system.
References
[1] A. Mikhailov and D. Meink¨ohn,Self-Motion in Physico-Chemical Sys-
tems Far from Thermal Equilibrium, Springer, Berlin Heidelberg, 484 (1997).
[2] S. Nakata and Y. Iguchi, S. Ose, M. Kuboyama, T. Ishii and K. Yoshikawa, Self-Rotation of a Camphor Scraping on Water: New In- sight into the Old Problem, Langmuir, 13 (1997), 4454–4458.
[3] M. Nagayama, S. Nakata, Y. Doi and Y. Hayashima, A theoretical and experimental study on the unidirectional motion of a camphor disk, Physica D (Amsterdam), 194 (2004), 151–165.
[4] H. Kitahata and K. Yoshikawa, Chemo-mechanical energy transduc- tion through interfacial instability, Physica D (Amsterdam), 205 (2005), 283–291.
83
A curious bifurcation diagram arising in a scalar reaction-diffusion equation with an integral constraint
Hisashi OKAMOTO
Research Institute for Mathematical Sciences, Kyoto University, Kyoto 606-8502 * e-mail: [email protected]
Abstract
2 π We consider uxx + Ru = constant + f (x), together with ∫ u(x) dx = 0. A -π strange bifurcation diagram was found. The external force f is chosen so that u = cos kx becomes a solution (k = 2, 3, 4). Solving the equation by the spectral method, we obtained non-trivial solutions. In the figure, where k = 2, the solution branch comes back to the trivial solution at the points indicated by the arrows. As R increases, the profile of the solution becomes complicated, but they also lose their complexity when they come back to the trivial solution. In my talk, I will explain these phenomena, emphasizing the contrast with the bifurcating solutions appearing in the Kolmogorov flows of 2D Navier-Stokes equations [1].
References
[1] Sun-Chul Kim & Hisashi Okamoto, Vortices of large scale appearing in the 2D stationary Navier-Stokes equations at large Reynolds numbers, Japan J. Indust. Appl. Math., vol. 27 (2010), 47--71. 84
Application of topological computation method for global dynamics of an associative memory model of the Hopfield type
1), Hiroshi Kokubu ∗,
1) Department of Mathematics, Kyoto University / JST CREST
e-mail: [email protected] ∗ Abstract A new rigorous topological computation method for global analysis of dynamical systems was introduced in [1]. I would like to revisit the clas- sical associative memory model of the Hopfield type, and study its global phase space structure from the point of view of this topological computation approach.
References [1] Z. Arai, W. Kalies, H. Kokubu, K. Mischaikow, H. Oka, and P. Pilar- czyk: A database schema for the analysis of global dynamics of mul- tiparameter systems, SIAM Journal on Applied Dynamical Systems, 8 (2009), 757–789.
85
Rigorous verification of equilibria for evolutionary equations - Existence, Uniqueness and Hyperbolicity -
1), Kaname Matsue ∗,
1) Department of Mathematics, Kyoto University
e-mail: [email protected] ∗ Abstract The aim of this talk is to provide a method for investigating a precise structure of equilibria of (infinite dimensional) dynamical systems, such as the local uniqueness and the hyperbolicity. Recently several rigorous computation methods for verifying stationary solutions of partial differential equations are studied (e.g. [8] or [10]). Ex- istence and local uniqueness are mainly investigated therein. Moreover, in [2], one of applications of [10], the global structure of all bounded global solutions for a partial differential equations is studied by using computer as- sisted proof and an algebraic-topological invariant of invariant sets, called the Conely index. This invariant gives us the information of an invariant set, like the existence and its stability. Hyperbolicity, which is related to the structural stability is known as one of the most important properties of dynamical systems, and is stronger than the robustness provided by the Conley index. An equilibrium u of a PDE is hyperbolic if the linearized operator at u has no spectrum on the imaginary axis. This property is important from the viewpoint of dynamical systems, such as the verification of local or global bifurcations (e.g. [3]).
In order to verify the hyperbolicity, one may need analytic methods for investigating the spectrum of a linear operator in general. It is sometimes the case, however, that a linearized operator is self-adjoint, in which case the invertibility of the linearized operator may be sufficient to yield the hyperbol- icity of an equilibrium. Even if the linearized operator is not self-adjoint, we can prove the hyperbolicity of equilibria for a broad class of infinite dimen- sional dynamical systems, called dynamical systems generated by parabolic evolutionary equations ([5]). The author shows that the specific Conely index and the suitable hypoth- esis of the linearized operator yield the local uniqueness and the hyperbolicity of an equilibrium. Using the preceding work by Nakao and his collaborators ([7]) and by Zgliczy´nskiand Mischaikow ([10]) or the author ([6]), we can
86 verify the hyperbolicity of equilibria of parabolic evolutionary equations by computer assisted analysis. At the end of this talk, the author will show several sample rigorous numerical results about the existence, local uniqueness and hyperbolicity of equilibria.
References
[1] C.Conley, Isolated Invariant Sets and the Morse Index, CBMS Reg. Con. Ser. Math., Vol.38, Amer. Math. Soc. Providence, RI, 1978.
[2] S. Day, Y. Hiraoka, K. Mischaikow and T. Ogawa, Rigorous Numerics for Dynamics: A Study of the Swift-Hohenberg Equation, SIAM J. Applied Dynamical Systems, 4(2005), 1–31.
[3] J. Guckenheimer and P. Holmes, Nonlinear Oscillations, Dynamical Sys- tems, and Bifurcations of Vector Fields, Springer, 1983.
[4] J.K. Hale, Asymptotic behavior of dissipative systems, Mathematical Surveys and Monographs, 25. AMS, Providence, 1988.
[5] D. Henry, Geometric Theory of Semilinear Parabolic Equations, Springer-Verlag, 1981.
[6] K. Matsue, Rigorous verification of equilibria for evolutionary equa- tions - Existence, Uniqueness and Hyperbolicity - Part 1. Gradient case, preprint.
[7] M.T. Nakao, K. Hashimoto and Y. Watanabe, A Numerical Method to Verify the Invertibility of Linear Elliptic Operators with Applications to Nonlinear Problems, Computing, 75(2005), 1–14.
[8] M.T. Nakao and N. Yamamoto, Numerical verifications for solutions to elliptic equations using residual iterations with higher order finite element, J. Comput. Appl. Math., 60(1995), 271–279.
[9] K.P. Rybakowski, The Homotopy Index and Partial Differential Equa- tions, Springer-Verlag, 1987.
[10] P. Zgliczy´nskiand K. Mischaikow, Rigorous numerics for partial differ- ential equations: The Kuramoto-Sivashinsky Equation, Found. Comput. Math. 1(2001), 255–288.
87
Role of unstable symmetric solutions in symmetry restoring process
1),2) 1) Tsuyoshi Mizuguchi ∗, Makoto Yomosa
1) Department of Mathematical Sciences, Osaka Prefecture University. 2) PRESTO, Japan Science and Technology Agency (JST).
e-mail: [email protected] ∗ Abstract In the study of dynamical systems, symmetries are one of the most im- portant concepts for characterising and/or classifying systems and solutions. For example, some bifurcations involve spontaneous breakdown or restora- tion of symmetry. Several chaotic systems having a discrete symmetry often exhibit a bifurcation that breaks or restores the symmetry while maintaining its aperiodic behaviour, i.e., an attractor merging crisis (AMC). On the asymmetric side of the AMC bifurcation point, a pair of strange attractors exists, one of which is chosen by an initial condition. At the bifurcation point, they merge with each other. On the symmetric side, a single strange attractor that satisfies the symmetry is observed. Let us focus on singularities near the AMC bifurcation point. Intermittent behaviours are reported in several systems. Namely, the inter-burst interval time tends to diverge when the system approaches the bifurcation point. If a state is translated to a motion in space, the singularity corresponds to divergence of the diffusion coefficient of the motion. Therefore, one can forecast the bifurcation by observing these quantities. The singularity described above is observed when the system approaches from the symmetric side. On the other hand, from the asymmetric side, a specific unstable symmetric solution so-called mediating unstable periodic orbit, or simply a mediating solution M plays an important role to charac- terize the symmetry restoration process. The mediating solution is defined as that with which a pair of strange attractors simultaneously collides at the AMC point when the system approaches from the asymmetric side. Near this AMC point, the outermost part of the strange attractor in each parameter passes close to solution M. Therefore, the distance dM between the strange attractor and M is an appropriate quantity to characterise the closeness to the bifurcation point. This approach is also characterised by a duration in which a state is close to solution M. We define a state whose distance to M is less than a threshold value as a temporarily sticking state, and we focus on the maximum
88 duration τM of this temporarily sticking state during the measurement time. A logarithmic singular behaviour of τM with periodic modulation is predicted near the bifurcation point. In this paper, we exemplify several dynamical systems exhibiting AMC transition, such as Ginzburg Landau type equation with external force term or coupled Stuart Landau type equation. The distance d and the maxi- M mum duration τM are directly measured near the bifurcation point. A linear decrease in dM and the logarithmic divergence of τM are confirmed numeri- cally. We suggest a time series analysis method to measure these quantities without knowing the precise bifurcation point or concrete function form of the mediating solution M.
References
[1] M. U. Kobayashi, T. Mizuguchi, “Chaotically oscillating interfaces in a parametrically forced system”, Phys. Rev. E, 73 (2006) 016212.
[2] M. U. Kobayashi, T. Mizuguchi, “Chaotic interfaces in a parametrically forces system”, Prog. Theor. Phys. Supplement, 161 (2006) 228–231.
[3] Y. Morita, N. Fujiwara, M. U. Kobayashi and T. Mizuguchi, “Scytale decodes chaos: A method for estimating unstable symmetric solutions”, Chaos, 20 (2010) 013126.
89
A Diffusion Equation Representing the Dynamics of the Jakarta Composite Index (JSX)
1),2)* 1), 3) Edi Cahyono , Buyung Sarita 1) Graduate Program, Universitas Haluoleo 2) Department of Mathematics FMIPA, Universitas Haluoleo 3) Faculty of Economics, Universitas Haluoleo * e-mail: [email protected]
Abstract We cosider the dynamics of the Jakarta Composite Index (JSX). The dynamics is presented as time series. The dynamics may be charaterized by the trend and the probability density function (pdf) based on the assumption that at a fixed time it is normally distributed with respect to ‘spatial’ variable. Details of the trend and the pdf for the case of the dynamics of JSX will be presented in the succesive article [2]. Following the work of [1], we seek a type of diffusion equation which has fundamental solution representing the trend and the probability density function of such dynamics.
References
[1] E. Cahyono, J.R. Juliana, R. Raya: Asset value dynamics and the fundamental solution of a modified heat equation, 4th Intl Conf. Math. Stat., Bandar Lampung, Indonesia (2009), 312--318. [2] B. Sarita, F. Arisanti, E. Cahyono: The trend and probability density function of Indonesia Stock Index (IHSG), submitted.
90
The Trend and Probability Density Function of the Jakarta Composite Index (JSX)
1),2)* 3) 1), 3) Buyung Sarita , Femy Puspita Arisanty , Edi Cahyono 1) Graduate Program, Universitas Haluoleo 2) Faculty of Economics, Universitas Haluoleo 3) Department of Mathematics FMIPA, Universitas Haluoleo * e-mail: [email protected]
Abstract In this paper we discuss the dynamics of the Jakart Composite Index (JSX), which is related to the work of [3]. The dynamics is presented as time series. To predict the dynamics is still difficult. In general it is almost impossible to predict such dynamics for the case of high frequency data [2]. Hence, we do not predict the dynamics. Rather, we seek the trend and the probability density function based on the assumption that the dynamics is normally distributed with respect to time variable. According to [1], the trend and the probability density function of such dynamics is the fundamental solution of a type of diffusion equation.
References
[1] E. Cahyono, J.R. Juliana, R. Raya: Asset value dynamics and the fundamental solution of a modified heat equation, Proc. 4th Intl Conf. Math. & Stat., Bandar Lampung, Indonesia (2009), 312--318. [2] T. Ito, K. Sato: Exchange rate change and inflation in post-crisis Asian economies: Vector autoregression analysis of exchange rate pass-through, J. Money, Credit and Banking, (2008) 40(7), 1407--1437. [3] B. Sarita, Agusrawati, L.D. Suryadi, S. Wahab, E. Cahyono: Dynamics of gold price relative to Indonesian rupiah, Proc. 3rd Intl. Conf. Quantitative Methods, Bandar Lampung, Indonesia (2010) 42--46.
91