Plenary Lecture in Symposium: New Development of Simulational Science in Statistical Physics
Recent Development of Markov-Chain Monte Carlo in Computational Physics
Synge Todo, Professor
Department of Physics, University of Tokyo
Institute for Solid State Physics, University of Tokyo
Mathematics and Informatics Center, University of Tokyo
CMI2, MaDIS, National Institute for Materials Science
Recent development of the Markov-chain Monte Carlo (MCMC) method is discussed. Although the MCMC is a versatile and powerful tool in many applications, it often suffers from slow convergence and reduction in the number of effective samples. This problem can be solved by introducing several novel techniques such as the non-local global update, worm algorithm, irreversible kernel, event-driven technique, geometric allocation scheme, and so on.
Synge Todo received his Ph.D. from University of Tokyo in 1996. After PostDoc and Research Associate at Institute of Solid State Physics, University of Tokyo, PostDoc at Theoretical Physics, ETH Zurich, Lecturer at Department of Applied Physics, University of Tokyo, and Project Professor at Institute of Solid State Physics, University of Tokyo, he became Associate Professor at Department Physics, University of Tokyo in 2014.
Plenary Lecture in Symposium: Numerical Simulation and Visual Analytics of Nonlinear Problems
Visual Causality Exploration and its applications
Koji Koyamada, Professor
Academic Center for Computing and Media Studies, Kyoto University, Japan
In the big data era, it is expected that every citizen can access the open data and participate in a scientific research. For realizing such a data science, visualization techniques will play an important role since they will enable the big data to be transmitted to the brain efficiently.
It is highly expected for visualization to facilitate an exploration of a causality in a data science field. Although it is possible to calculate a correlation between data items (variables) by using a statistic method, the causality is a feature which domain experts can clarify by making the best of their professional knowledge. There are several examples of illogically inferring causation from correlation. That is why a visualization plays an important role in the scientific discovery.
In this talk, we would like to introduce our activities on visual causality exploration. They include an interactive specification of a latent variable which explains several observable variables by using a causality graph in a phenotypic expression network, and an interactive exploration of a causality between two time-varying variables defined on computational grids.
Prof. Koji Koyamada is currently a professor at the Academic Center for Computing and Media Studies, Kyoto University, Japan. His research interest includes modeling & simulation and visualization. He is an associate member of the Science Council of Japan, a former president of the Visualization Society Japan, and a former president of Japan Society of Simulation Technology. He received the IEMT/IMC outstanding paper award in 1998, the VSJ contribution award in 2009 and the VSJ outstanding paper award in 2010. He received his B.S., M.S. and Ph.D. degrees in electronic engineering from Kyoto University, Japan in 1983, 1985 and 1994, respectively, and worked for IBM Japan from 1985 to 1998. From 1998 to 2001 he was an associate professor at the Iwate Prefectural University, Japan. From 2001 to 2003, he was an associate professor at Kyoto University, Japan.
Plenary Lecture in Symposium on Simulation Technology for Safe, Secure and Resilient Society
Cyber-physical System and Industrial Applications of Large-Scale Graph Analysis and Optimization Problem
Katsuki Fujisawa, Professor
Institute of Mathematics for Industry, Kyushu University
Artificial Intelligence Research Center, Advanced Industrial Science and Technology
Global Scientific Information and Computing Center, Tokyo Institute of Technology
Research Center for Statistical Machine Learning, The Institute of Statistical Mathematics
In this talk, we present our ongoing research project. We have started the research project for developing the Urban OS (Operating System) on a large-scale city from 2013. The Urban OS, which is regarded as one of emerging applications of cyber-physical system (CPS), gathers big data sets of people and transportation movements by utilizing different sensor technologies and storing them to the cloud storage system. We have another research project whose objective is to develop advanced computing and optimization infrastructures for extremely large-scale graphs on post peta-scale supercomputers. For example, our project team was a winner of the 8th and 10th to 14th Graph500 benchmark. The Urban OS employs the graph analysis system developed by this research project and provides a feedback to a predicting and controlling center to optimize many social systems and services.
Katsuki Fujisawa has been a Full Professor at the Institute of Mathematics for Industry (IMI) of Kyushu University. He had also been a research director of the JST (Japan Science and Technology Agency) CREST (Core Research for Evolutional Science and Technology) post-Peta High Performance Computing from 2011 to 2017. He received his Ph. D. from the Tokyo Institute of Technology in 1998. The objective of the JST CREST project is to develop an advanced computing and optimization infrastructure for extremely large-scale graphs on post peta-scale supercomputers. His project team has challenged the Graph500 benchmark, which is designed to measure the performance of a computer system for applications that require irregular memory and network access patterns. In 2014 to 2016, his project team was a winner of the 8th and 10th to 14th Graph500 benchmark. In 2017, He received the Prize for Science and Technology (Research Category), Commendation for Science and Technology by the Minister of Education、Culture、Sports、Science and Technology, Japan
Invited Talk #1 in Symposium: New Development of Simulational Science in Statistical Physics
Nonequilibrium relaxation method in random systems
Shibaura Institute of Technology
A system with random interactions shows many exotic and nontrivial phenomena. However, its simulation studies are plagued by severely slow dynamics. Various algorithms have been proposed to accelerate the dynamics. However, we still cannot reach a conclusion of the spin-glass transition, which is a problem of random magnets first investigated more than 30 years ago.
In this talk, we introduce an alternative approach to investigate such slow-dynamic systems, namely a non-equilibrium relaxation method. This method uses data in a non-equilibrium relaxation process that have been discarded in the conventional equilibrium method. The slow dynamics caused a waste of computational time but it now becomes an advantage because we can use more data. It can be regarded as an exchange of size and time limit to the thermodynamic limit as shown in the figure. Its application to the spin-glass problem coupled with the Bayesian inference successfully clarified the phase transition relevant to the real spin-glass materials.
Prof. Tota Nakamura is currently a professor at the school of arts and sciences, Shibaura Institute of Technology, Japan. His research interest includes a phase transition in random magnets, developments of new simulation algorithms and Bayesian inferences. He received his B.S., M.S. and Ph.D. degrees in physics from Tokyo Institute of Technology, Japan in 1988, 1990, and 1993, respectively. From 1993 to 1995, he was a post-doc researcher at the University of Tsukuba, and research associate at the Tohoku University from 1995 to 2004. From 2004 to 2010, he was an associate professor at Shibaura Institute of Technology, Japan.
Invited Talk #2 in Symposium: New Development of Simulational Science in Statistical Physics
Emergence of Antiferroelectric Phase in Spheroidal Dipolar Particles
Institute of Industrial Science, University of Tokyo
We demonstrate ferroelectricity and antiferroelectricity in crystals composed of spheroidal dipolar particles by varying the aspect ratio of the particles. We perform molecular dynamics simulations to study phase behavior, where the electrostatic dipolar inter-action is calculated by smooth particle mesh Ewald method. We obtain phase diagram of the system by calculating the chemical potential of liquids, paraelectric crystals, ferroelectric crystals, and antiferroelectric crystals. It is found that the ferroelectric phase is realized for the particles with aspect ratio close to unity, whereas antiferroelectric phase is realized for large aspect ratio.
Kyohei Takae is a research associate at Institute of Industrial Science, University of Tokyo. He received his Ph. D. from Kyoto University in 2013. His research interest includes phase transition dynamics and mesoscopic heterogeneity in condensed matter, particularly in soft matter, glasses and dielectrics.
Invited Talk #3 in Symposium: New Development of Simulational Science in Statistical Physics
Simulations of word popularity dynamics observed from large scale social data
National Institute of Informatics, JST Presto
We show the dynamics of frequency of trending words observed from a large- scale blog database, which are characterized by an exponential function and a power function. We reproduce these dynamics by an agent-based model based on the SIR (Susceptible- Infected-Recovered) model which is well known in mathematical epidemiology to clarify the origin of these dynamics from the view point of bloggers interactions. In our social model, we introduce a “ground”, “excited” and “final” state respectively instead of susceptible, infected and recovered state. Agents move from the ground state to the excited state and from the excited state to the final state according to transition probabilities. Our model reasonably reproduces the dynamics observed from our data. In addition we applied our model to analysis of false rumor spreading.
Kenta Yamada is a research assistant professor of the National Institute of Informatics(NII) in Japan. He received his Ph. D in Science from Tokyo Institute of Technology in 2009. His research interests lie in the area of socio-physics and econo-physics, especially analyzing and modeling big social and financial data by using statistical physics methods.
Invited Talk #4 in Symposium: New Development of Simulational Science in Statistical Physics
Machine learning the quantum phase transitions in random systems
Physics Division, Sophia University
Quantum phase transition is a zero temperature phase transition caused by the change of quantum fluctuations with varying the parameter(s) of Hamiltonian. In the presence of randomness such as random potentials and lattice defects, the system shows various quantum phases. Here we apply the convolutional neural network to draw the phase diagram.
Tomi Ohtsuki, Doctor of Science (University of Tokyo, 1989), is Professor of physics at Sophia University, Tokyo, where he conducts theoretical and computational researches in condensed matter physics. His recent research focuses on quantum transport phenomena such as the Anderson transition, conductance fluctuations, Hall and spin Hall effects in nanoscale systems. His research has been published by Physical Review Letters, Physical Review B, Physics Reports, Journal of the Physical Society of Japan, and others.
Invited Talk in Symposium: Numerical Simulation and Visual Analytics of Nonlinear Problems
Energy-Based Multiscale Modeling of Magnetic Material
Tetsuji Matsuo, Professor
The development of magnetic-material simulator is a challenging task because of its multiscale nature where domain-wall behavior in nm-scale affects macroscopic property in mm/cm-scale. Based on mesoscopic magnetic-domain modeling in crystal-grain scale, we developed a physical multiscale model of magnetic material. It is an energy-based model that can take into account influence of physical factors in their energy forms. For example, the magneto-mechanical interaction can be analyzed by including the magneto-elastic energy as an energy component. The multi-scale model successfully reconstructed stress-dependent properties of silicon steel and magnetization properties of a thin-film magneto-impedance element. Magnetization analysis [Fig. 1(a)] of silicon steel under mechanical stress predicts the stress dependence of hysteresis loss [Fig. 1(b)], which quantitatively agrees with the measured loss.
Tetsuji Matsuo received the B.E., M.E., and Dr. Eng. degrees from Kyoto University, Japan, in 1986, 1988 and 1991, respectively. He became a Research Associate, a Lecturer, and an Associate Professor at Kyoto University in 1991, 2001, and 2003, respectively. He is currently a Professor in the Department of Electrical Engineering, the Graduate School of Engineering, Kyoto University. His current research interests include computational electromagnetics and magnetic material modeling.
Invited Talk in Symposium on Simulation Technology for Safe, Secure and Resilient Society
Factors of Security Breach
Cybersecurity Advisory, KPMG Consulting
Today we hear Cyber attacks news all around the world. And in Japan, we recently had major PII breach in June, 2015. In this talk, I will examine the factors of Security Breach focusing on 3 major factors why breach happens when Cyber attacks hits corporations. Companies can make effort to protect against from Cyber attacks but with attackers growing sophisticated more and more there is no single bullet for Cyber attacks. By focusing on the factors of security breach we can help the corporations to limit the damage of security breach at minimum if not none. Three 3 major factors: People, Process, and Technology.
Ms. Yamashita is a Senior Manager in KPMG Consulting Japan. She is a member of the Security Advisory Group where she focuses on security advisory and security program assessments. Prior to joining KPMG, Ms. Yamashita spent 10 years in Microsoft engineering Windows. Then Ms. Yamashita has gained 7 years of experience in security field where she was Threat research engineer, Sales Engineer, then consultant providing security transformation services to global corporations.
Tutorial in Symposium: New Development of Simulational Science in Statistical Physics
Worldline Monte-Carlo methods of huge systems
The quantum Monte Carlo method based on the Feynman’s path integral representation, called the worldline Monte Carlo method, is a numerical method for exactly solving large quantum many-body systems in the field of statistical physics. The worm algorithm is one of most effective algorithm for globally updating the worldline configuration and this algorithm has wide range of applicability to quantum magnets (i.e. quantum spin systems) and bosonic particles in lattice systems. In this talk we will introduce the basics of the worldline Monte Carlo method and recent developments of the worm algorithm for huge systems. We will show the benchmark of a worm algorithm with the nontrivial parallelization for the bosonic particles in huge two-dimensional square and three-dimensional cubic lattices.
Dr. Akiko Masaki-Kato is a special postdoctoral researcher at Riken. She received her B.S., M.S. and Ph. D degrees in Physics from Tokyo Metropolitan University, Japan in 2007, 2009, 2013 respectively. Her current research interests include development of quantum Monte Carlo algorithm, quantum manybody phenomena (phase transition, superfluidity, excitation dynamics) of bosons and quantum spin system.
Tutorial in Symposium: Numerical Simulation and Visual Analytics of Nonlinear Problems
Introduction to Visualization System, KVS
College of Information Science & Engineering, Ritsumeikan University
Kyoto Visualization System (KVS) is a multi-platform, open source C++ Toolkit for developing scientific visualization applications. KVS provides useful classes to quickly implement surface rendering, volume rendering, particle-based rendering, and others. KVS users are required to have knowledge on C++ languages, but not on OpenGL, GPU, or details of volume-rendering algorithms. In this tutorial, we show how to implement visualization applications using KVS by examples such as volume rendering, surface rendering and fused visualization (These figures show examples of particle-based rendering created by using KVS).
Kyoko Hasegawa received her PhD in engineering from the University of Tsukuba, Japan, in 2004. She has been a lecture at College of Information Science and Engineering, Ritsumeikan University, Japan, since 2017.
Kyoko Hasegawa has received many academic awards. For example, Best Paper Award in Asia Simulation Conference 2012 (on visualization of surgery simulation), Best Art Work Award from the Visualization Society of Japan in 2014 (on visualization of cultural assets), Best Paper Award from Japan Society for Simulation Technology in 2015 (on visualization of large-scale particle fluid simulation).
Tutorial in Symposium on Simulation Technology for Safe, Secure and Resilient Society
A study on traffic flow control using traffic simulator UC-win/Road
This paper considers traffic control problem using a traffic simulator. Parameters of traffic signals to increase traffic flows are designed. The traffic simulator, UC-win/Road is used to confirm an effect of the design of the parameters. The traffic simulator is constructed using some actual data. A comparison with a result obtained by actual parameters was carried out. From the simulation results the proposed parameters provide increasing the traffic flows in comparison with the result obtained by actual parameters.
Professor Mukai received the B.E., M.E. and Dr. of Engineering degrees in Electrical Engineering from Kanazawa University, Japan, in 2000, 2002, and 2005, respectively. He was with the Graduate School of Information Science and Electrical Engineering, Kyushu University from 2005 to 2014. He is currently with Department of Electrical and Electronic Engineering, Kogakuin University. His research interests include receding horizon control and its applications. He is a member of the SCIE, ISCIE, IEEJ, and IEEE.
Tutorial in JSST Student Research Symposium
Construction of muscle activity model based on Bayesian network and kinematic evaluation
Tokyo Denki University
We build foot muscular activity model based on Bayesian network and evaluate kinematic aspect. We aimed to enable quantitative selection of lower foot orthoses based on a patient’s muscular activity in the lower foot. Physical models require three dimensional motion analy-sis and force plate. These measurement systems cannot be used clinically, so they are not suitable for making these clinical use. Therefore, we chose Bayesian network to construct a model for estimating the muscular activity from parameters that can be measured easily, such as joint angle and sole pressure. Here, we verify the kinematic validity of the model parent node for foot muscles.
Jun Inoue received him PhD in engineering from the Waseda University, Japan, in 2013. He is currently an assistant professor at Tokyo Denki University. His current research interests include development of welfare equipment, human motion analysis, construction of muscle activity model using information engineering approach.